Escolar Documentos
Profissional Documentos
Cultura Documentos
Edited by
A.R. Kitai
Department of Materials Science and Engineering and Engineering Physics
McMaster University, Ontario, Canada
Apart from any fair dealing for the purposes of research or private study, or
criticism or review, as permitted under the UK Copyright Designs and
Patents Act, 1988, this publication may not be reproduced, stored, or
transmitted, in any form or by any means, without the prior permission in
writing of the publishers, or in the case of reprographic reproduction only in
accordance with the terms of the Iicences issued by the Copyright Licensing
Agency in the UK, or in accordance with the terms of Iicences issued by
the appropriate Reproduction Rights Organization outside the UK.
Enquiries conceming reproduction outside the terms stated here should be
sent to the publishers at the London address printed on this page.
The publisher makes no representation, express or implied, with regard to
the accuracy of the information contained in this book and cannot accept
any legal responsibility or Iiability for any errors or omissions that may be
made.
A catalogue record for this book is available from the British Library
Library of Congress Cataloging-in-Publication data available.
Principles of luminescence 1
A.H. Kitai
2 Luminescent centres in insulators 21
G. Blasse
3 Luminescence spectroscopy 53
U. W. Pohl and H.-E. Gumlich
4 One-photon rare earth optical transitions: recent theoretical
developments 97
G. W. Burdick and M.e. Downer
5 Thin film electroluminescence 133
G.O. Muller
6 Powder electro luminescence 159
S.S. Chadha
7 Thin film electroluminescence devices 229
R Mach
8 Light emitting diodes: materials growth and properties 263
S.P. DenBaars
9 Atomic layer epitaxy of phosphor thin films 293
B.W Sanders
10 Lamp phosphors 313
T.E. Peters, RG. Pappalardo and RB. Hunt, Jr
11 Phosphors for other applications 349
G. Blasse
Index 373
v
Contributors
G. Blasse
Debye Research Institute
University of Utrecht
Utrecht
The Netherlands
G.W. Burdick
The University of Texas at Austin
Texas
USA
S.S. Chadha
School of Biological and Chemical Sciences
University of Greenwich
London
UK
S.P. DenBaars
Materials Department
University of California, Santa Barbara
California
USA
M.C. Downer
The University of Texas at Austin
Texas
USA
H.-E. Gumlich
Technical University of Berlin
Berlin
Germany
Vll
Vlll Contributors
R.B. Hunt, Jr
GTE Laboratories Inc., Danvers
Massach usetts
USA
A.K. Kitai
Departments of Materials Science and Engineering Physics
McMaster University, Hamilton
Ontario
Canada
R. Mach
Central Institute for Electron Physics
Berlin
Germany
G.O. Muller
Central Institute for Electron Physics
Berlin
Germany
R.G. Pappalardo
GTE Laboratories Inc., Danvers
Massachusetts
USA
T.E. Peters
GTE Laboratories Inc., Chelmsford
Massachusetts
USA
U.W. Pohl
Technical University of Berlin
Berlin
Germany
B.W. Sanders
Institute for Environmental Chemistry
National Research Council Canada, Ottawa
Ontario
Canada
Preface
Historically, black body radiation in the tungsten filament lamp was our
primary industrial means for producing 'artificial' light, as it replaced gas
lamps. Solid state luminescent devices for applications ranging from lamps
to displays have proliferated since then, particularly owing to the develop-
ment of semiconductors and phosphors. Our lighting products are now
mostly phosphor based and this 'cold light' is replacing an increasing
fraction of tungsten filament lamps. Even light emitting diodes now chal-
lenge such lamps for automotive brake lights.
In the area of information displays, cathode ray tube phosphors have
proved themselves to be outstandingly efficient light emitters with excellent
colour capability. The current push for flat panel displays is quite intense,
and much confusion exists as to where development and commercialization
will occur most rapidly, but with the need for colour, it is now apparent that
solid state luminescence will play a primary role, as gas phase plasma
displays do not conveniently permit colour at the high resolution needed
today. The long term challenge to develop electroluminescent displays
continues, and high performance fluorescent lamps currently illuminate
liquid crystal monochrome and colour displays. The development of tri-
component rare earth phosphors is of particular importance.
This book begins with a chapter on the physics of luminescence, covering
the classical and quantum mechanical theory of radiation in atoms and
solids. Chapter 2 focuses on phosphors, describing the fundamental pro-
cesses and models that are useful to sort out the rather complex electronic
and vibrational interactions. Chapter 3 deals with state of the art experimen-
tal methods and gives examples of fundamental luminescence processes in
solids. Chapter 4 presents a current understanding of rare earth ions in
crystalline fields.
The next section of the book focuses on different device and material
types: thin film electroluminescence is described from a physics perspective
in Chapter 5 and Chapter 6 traces the rather painful but important
development of powder phosphors for electroluminescence.
IX
x Preface
A.H. Kitai
Ontario
Canada
1
Principles of luminescence
A.H. Kitai
1.1 INTRODUCTION
Technologically important forms of luminescence may be broken into
several categories, as shown in Table 1.1. Although the means by which the
luminescence is excited varies, all luminescence is generated by means of
accelerating charges. The portion of the electromagnetic spectrum visible to
the human eye has wavelengths from 400 to 700 nm. The evolution of the
relatively narrow sensitivity range of the human eye is a complex subject,
but is intimately related to the solar spectrum, the absorbing behaviour of
the terrestrial atmosphere, and the reflecting properties of organic materials,
green being the dominant colour in nature and, not surprisingly, the
wavelength at which the human eye is most sensitive. In this chapter, we
cover the physical basis for radiation and radiation sources in solids that
produce visible light.
Table 1.1 Luminescence types, applications and typical efficiencies (visible output
power/electrical input power)
Luminescence type Typical application Luminous efficiency
1
2 Principles of luminescence
Fig. 1.2 The lines of magnetic field B due to a point charge q moving into the page
with uniform velocity.
Both electric and magnetic fields store energy; the total energy density is
given by
It is important to note that the energy density moves with the charge so long
as the charge is either stationary or undergoing uniform motion; this is
evident since a new reference frame may be constructed in which the
observer is stationary with respect to the charge.
For an accelerated charge, however, energy continuously leaves the
charge to compensate exactly the work done in causing the charge to
accelerate. Consider the charge q in Fig. 1.3. Initially at rest in position A,
it accelerates to position B and then stops there. The electric field lines now
emanate from position B, but would, further out, have emanated from
position A, since the field lines cannot convey information about the
location of the charge at speeds greater than the velocity of light c. This
results in kinks in the lines of electric field which propagate away from q
with velocity c. Each time q accelerates, a new series of propagating kinks
is generated. Each kink is made up of a component of E that is transverse
to the direction of expansion, which we call E 1-' If the velocity of the charge
during its acceleration does not exceed a small fraction of c, then for r large,
qa . ()
E1- = 2 sm
4n8oc r
Radiation theory 3
-
a
Fig. 1.3 Lines of electric field emanating from an accelerating charge. (After
Eisberg and Resnick [1].)
Here, a is acceleration, and r is the distance between the charge and the
position where the electric field is evaluated. The strongest transverse field
occurs in directions normal to the direction of acceleration, as suggested by
Fig. 1.3.
Likewise, a transverse magnetic field B 1- is generated during the acceler-
ation of the charge as shown in Fig. 1.4, given by
Iloqa .
B1-=--sm8
4ncr
The two transverse fields propagate outward with velocity c each time q
undergoes an acceleration, giving rise to the electromagnetic radiation
whose frequency matches the frequency with which q acclerates. Note that E.l
and B.l are perpendicular to each other. The energy density of the radiation is
.1. 2 1 2
Iff = 21::0E.l + -2 B.l
f.lo
The Poynting vector or energy flow per unit area (radiation intensity) is
1
S=-E.l xB.l
f.lo
2 2
-q-=---a-=--=-sin 2 0 f
16nl::oc 3 r 2
f
P= S(O) dA= f: S(0)2nr2 sin 0 dO
and
2q2 A 2W4 sin 2wt
P=----,----:;,---
4nl::03c 3
Now, average power P is the root-mean-square power which gives
_ q2 A2w4
P=-=-----;;- (1.1)
4nl::03c 3
Quantum description 5
If we now consider that an equal and opposite stationary charge - q is
located at x=o then we have a dipole radiator with electric dipole moment
of amplitude p = qA. Now we may write
_ p2W 4
P=---:o
12n80c3
Non-oscillatory radiation does exist also; the synchroton radiation source is
an example of a radiator that relies on the constant centripetal acceleration of
an orbiting charge. Quadrupole and higher-order poles may exist even in the
absence of a dipole moment; however, they have lower rates of energy release.
We let
rnn,=(nlrln')=
J
r
all space
I/I:rl/ln,dV
(1.3)
11 = f 0
21[
cos ei(m-m')</> d
12 = f 0
21[
sin ei(m -m')</> d
13= f 0
21[
ei(m-m')</> d
The integral
many atoms
many atoms
Fig. 1.5 The decay of an electron from state k to state I results in the release of a
photon.
'( ) {O
H r,t = t<O
2H'(r) cOSOJt t~O
where
and
t/I(r, t) = cPk(r )e iwkt
The probability of a transition from the initial eigenstate t/ll(r, t) to a new
eigenstate t/lk(r, t) is given simply by ICk(tW. We write
Pl_k(t)=ICk(t)12=(~~ly 1 t eiwkI''j(t')dt' 12
Harmonic perturbation 11
Because of the weak electromagnetic field, f(t)=2 cos wt, and therefore
Resonance occurs when Wkl = w. The two signs signify either a sti-
mulated absorption process (Wkl = w) or a stimulated emission process
(Wkl= -w) since energy is then released (Ekl negative). If Wkl~W,
(1.5)
Thus it is evident that when P 1k OCt 2, then the transition rate increases
linearly with time. This situation obtains for small t, since from equation
(1.5) we see that
P Ik OC 11m sin 2[!-(Wkl-W)tJ IHkti 2 1. 2
= 2 4t
n-+oo (Wkl-W)
Of course, for long times, W1k becomes constant as equilibrium is reached.
Note that Blk = Bkl since P1k=Pkl .
Therefore
NI Ak,+Bk1u(v)
Nk Bk1u(v)
and
it follows that
AlB (1.6)
u(v)= ehrojkT -1
Note that the frequencies are quantized and may be counted using integers
nx Similar expressions may be written for Ey and Ez Consider an artificial
z
>-~~--------~~y
x
Fig. 1.7 Cavity of cubic shape with edge length a. (After Solymar and Walsh, [3].)
/C-----h4-----1~ ny
nx
Fig. 1.8 Spherical shell enclosing points in (nx, ny, nz ) space lattice that represent
standing waves that range in frequency from v to v+dv. (After Solymar and Walsh
[3].)
Therefore
and the final result, valid over all v and T, is Planck's famous blackbody
radiation energy density function
8nv 3
U(V)=-3-
c ehv/kT - 1
12,0000 K
--.....
\r\
I \ / "
,
/ ........ 3,OoooK
I 'I
I
I
" ........
........
I "
I
I
~ ~ ~ ~ 1~ 1~ 1~ 1~ 1~ ~
Wavelength Injlm
Fig. 1.9 Blackbody radiation spectrum showing spectral power density for sources
at temperatures of 3000 K, 6000 K and 12000 K. Note that the 6000 K curve
matches the visible range best, and is similar to the solar spectrum. The three curves
are artificially normalized to appear identical in height.
16 Principles of luminescence
For visible light sources, tungsten filament lamps which are blackbody
radiators are limited in filament temperature to somewhat below the melting
point of tungsten, or '" 3000 K. As is clear from Fig. 1.9, only a small
fraction of the area under the curve corresponding to this temperature is in
the visible range: the physical basis for the low efficiency of such lamps. A
considerable attenuation of short-wavelength (blue-violet) compared with
long-wavelength (red) visible is also evident. A lamp operating at 6000 K
would approximately match the sun's surface temperature (5700 K) and be
far more efficient. The tungsten halogen lamp allows for a modest gain in
performance over a regular tungsten lamp by chemically stabilizing the
tungsten filament, allowing for higher filament temperature.
RSA ------I
Fig. 1.10 Excited atom S a distance RSA from atom A.
electric field E due to its oscillating dipole. This field falls off as l/r 3 [4]
and, provided that its energy does not have time to escape as a photon, it
will directly stimulate a transition in A by means of the same process
described by equation (1.4). Since E falls off as l/r 3 , then the energy density
in the electric field
U(V)=16 0 E 2
falls off as l/r6. From equation (1.4), therefore, the rate (or the probability)
of energy transfer from S to A depends on R;1. Energy transfer is discussed
further in Chapter 2.
and
En=E~O) + <~O)IH'I~O
Note that the new eigenstates n and eigenenergies En are based on the
eigenstates ~O) and eigenenergies E~O) of the unperturbed system. To
account for the crystal field, we simply substitute H'(8, ) = Vs(8, ) and
determine the new states due to the crystal field.
If the symmetry of the crystal field is different from eigenstates involved,
which is usually the case, then degenerate states will probably split because
of the crystal field. This is known as crystal field splitting. In this case,
E ~O) - E lO) = 0 for some n, i, n 01= i. In order to determine the n and En, we
must first make the matrix element <lO) IH' I ~O zero whenever ElO) = E~O)
such that these singular terms vanish. This may be accomplished by
diagonalizing the submatrix of H;n which contains the degenerate states. The
energy splittings are then obtained directly.
For transition metal ions placed in a crystal field, the three-dimensional
states may be more influenced by the crystal field than by He. This is known
as the strong field scheme [5]. For such an ion in an octahedral crystal field,
for example, the five-fold degenerate three-dimensional orbitals split into a
two-fold degenerate eg state and a three-fold degenerate t 2g state. The energy
separation between the two states is called lODq where Dq is a parameter
that is determined by the crystal field strength. Now, He may be accounted
for in a manner analogous to free-ion LS coupling.
In Cr3+, for example, there is a 3d 3 configuration. If placed in an octa-
hedral crystal field, the splitting depends on the term Dq/B which is a
measure of the crystal field influence. A theoretical treatment [6] gives the
splittings. Results for Cr3 + are shown in Fig. 2.16.
Transitions that were forbidden in the free ion may become dipole
transitions with a crystal field. The lower symmetry allows a dipole moment
References 19
to exist, and these new transitions can take place, although with small rates.
Radiative lifetimes of 10- 3 s are not unusual for transition metal ions such
as Mn2+ in a tetrahedral crystal field.
REFERENCES
1. Eisberg, R and Resnick, R (1985) Quantum Physics of Atoms, Molecules, Solids,
Nuclei and Particles, 2nd edn, Wiley, New York.
2. Liboff, RL. (1980) Introductory Quantum Mechanics, Holden-Day.
3. Solymar, L. and Walsh, D. (1985) Lectures on the Electrical Properties of
Materials, 3rd edn, Oxford University Press.
4. Kip, A.F. (1969) Fundamentals of Electricity and Magnetism, 2nd edn, McGraw
Hill, New York.
5. Yen, W.M. and Selzer, P.M. (1986) Laser Spectroscopy of Solids, Vol. 49, 2nd edn,
Springer, Berlin.
6. Tanabe, Y. and Sugano, S. (1954) J. Phys. Soc. Japan, 9, 766.
2
Luminescent centres in insulators
G. Blasse
2.1 INTRODUCTION
The existence of optically active centres in solids, liquids and molecules is
now well known. Their properties have been studied intensively, not only
for fundamental reasons, but also in view of their potential applications.
Energy-saving fluorescent lamps, X-ray photography, and television display
tubes are well-known examples [1].
This can be hardly better illustrated than by the case of ruby
(AI 2 0 3 :Cr3+). The optical properties of ruby have been studied for over a
century starting with the work of Becquerel in 1867 [2] who excited ruby
with sunlight. He claimed that the properties of ruby were intrinsic, but it
soon became clear that they are due to the Cr 3 + ion that is an optically
active centre in the nonabsorbing Al 2 0 3 host. Later crystal field theory was
able to explain the spectroscopy of Cr 3 + in Al 2 0 3 in detail. The application
followed rather soon: the first solid state laser was based on a ruby crystal.
In the development of tunable infrared lasers the Cr 3 + ion plays an
important role [3,4].
The purpose of this chapter is to present the theories which are at present
in use to describe the luminescent properties of optically active centres in
solids. This will be done in such a way that even the unexperienced reader
can apply the theoretical results to practical cases. In order to do so, section
2.2 treats the interaction between an optical centre and its immediate
surroundings by presenting the configurational coordinate diagram. Radi-
ative as well as nonradiative transitions will be dealt with. In Section 2.3 we
will consider the mutual interactions between optical centres, and especially
those leading to energy transfer and energy migration. In section 2.4 we will
illustrate how theory applies to specific cases by discussing some carefully
selected case studies relating to important optically active centres.
For those who are not familiar with these types of phenomena, this
introduction gives now a schematic, first picture of the physical properties
of optically active centres. Figure 2.1 shows an optical centre (an ion or a
21
22 Luminescent centres in insulators
x M
q H
x M
Fig. 2.2 Energy transfer between two centres in a solid. The excitation X excites
centre S which transfers its excitation energy to A (T). Finally A shows emission (M).
can also be used. A more chemical approach to these topics was presented
by ourselves [7,8].
Q'
o
parabola v. The excited state parabola is drawn in such a way that the force
constant is weaker than in the ground state. Since the excited state is usually
more weakly bound than the ground state, this is a representative situation.
Optical absorption corresponds to a transition from the u to the v state
under absorption of electromagnetic radiation. Emission is the reverse
transition. Let us now consider how these transitions have to be described
in the configurational coordinate model. It is essential to remember that the
wavefunction of the lowest vibrational state (i.e. n=O or m=O) is Gaussian;
that is, the most likely value of Q is Qo (or Qo in the excited state). For the
higher vibrational states, however, the most likely value is at the edges of
the parabola (i.e. at the turning points, as in the classical pendulum).
The most probable transition in absorption at low temperatures is from
the n = 0 level in u, starting at the value Qo. Optical absorption corresponds
to a vertical transition, because the transition u--+v on the dopant ion occurs
so rapidly that the surrounding lattice does not change during the transition
(Born-Oppenheimer approximation). Our transition will end on the edge of
The configurational coordinate diagram 25
parabola v, since it is there that the vibrational states have their highest
amplitude. This transition, drawn as a solid line in Fig. 2.3, corresponds to
the maximum in the absorption band. However, we may also start at Q
values different from Qo, although the probability is lower. This leads to the
width of the absorption band, indicated in Fig. 2.3 by broken lines. It can
be shown that the probability of the optical transition between the n=O
vibrational level of the ground state and the mth vibrational level of the
excited state is proportional to
(2.3)
where r represents the electric dipole operator and Vm and Uo the vibrational
wavefunctions. The first term, the electronic matrix element, is independent
of the vibrational levels; the second term gives the vibrational overlap. The
transition from n = 0 to m = 0 does not involve the vibrations. It is called the
zero-vibrational transition (or no-phonon transition). Equation (2.3) shows
that the effect of the vibrations is mainly to change the shape of the
absorption line (or band), but not the strength of the transition (which is
given by the electronic matrix element).
What happens after the absorption transition? First we return to the
lowest vibrational level of the excited state; that is, the excited state v relaxes
to its equilibrium position, giving up the excess energy as heat to the lattice.
The system of dopant ion and surroundings is then in the relaxed excited
state. The emission transition can be described in exactly the same way as
the absorption transition. This is indicated in Fig. 2.3 in the same way as
for the absorption transition. Finally the system relaxes within the u
parabola to the lowest vibrational level.
If the temperature is not low, higher vibrational levels may be occupied
thermally, so that we start the process not only from n = 0, but also from
n = 1, and possibly from even higher levels. This leads to a further broaden-
ing of the absorption and emission bands, but does not change our
arguments essentially.
The emission transition will usually be situated at lower energy than the
absorption transition. This phenomenon is known as the Stokes shift. Only
the zero-vibrational transition is expected to occur at the same energy in the
absorption and emission spectra. The Stokes shift is a direct consequence of
the relaxation processes that occur after the optical transitions. It is obvious
that the larger Qo - Qo is, the larger the Stokes shift will be. If the two
parabolae have the same shape and vibrational frequency, it is possible to
define a parameter S (the so-called Huang-Rhys parameter) as follows:
tk(Qo - QO)2 = Shw (2.4)
where hw is the energy difference between the vibrational levels. The Stokes
shift is then given by
!lE. = k(Qo - Q)2 - hw = 2Shw (2.5)
340 310nm
(a)
500 600nm
(b)
ss
x M
300 500nm
(e)
The configurational coordinate diagram 27
The parameter S measures the interaction between the dopant ion and the
vibrating lattice. Equation (2.5) shows that, if S is large, the Stokes shift is
also large. Equation (2.4) shows that S is immediately related to the offset
of the parabolae in the configurational coordinate diagram (Fig. 2.3). This
offset, LlQ = Qo - Qo, may vary considerably as a function of the dopant ion
and as a function of the vibrating lattice, as we will see below.
It can be shown that the relative intensity of the zero-vibrational
transition (n=O~m=O) is exp( -S) [5]. We can now divide our lumines-
cent centres into three classes:
1. those with weak coupling (i.e. S < 1), so that the zero-vibrational transi-
tion dominates the spectrum;
2. those with intermediate coupling (i.e. 1 < S < 5), so that the zero-vibra-
tional transition is observable, but not the strongest line in the absorp-
tion or emission band;
3. those with strong coupling (i.e. S> 5), so that the zero-vibrational
transition is so weak that it is not observable in the spectra. (This case
is also characterized by large Stokes shifts.)
Figure 2.4 shows three emission spectra that are representative of the
three cases. Characteristic examples of case 1 are the trivalent rare earth
ions. The value of S is so small for these ions that the spectra consist in good
approximation of the zero-vibrational transitions only. Figure 2.4(a) gives as
an example the emission spectrum of the Gd3+ ion in LaB 3 0 6 It consists
of one strong electronic line at about 310 nm, whereas the weak repetition
at about 325 nm is a vibronic transition. Actually the energy difference
between these two lines corresponds to the vibrational stretching frequency
of the borate group in LaB 3 0 6 .
A characteristic example of case 2 is the uranyl ion (UO~ +). The m = 0--+
n = 2 line dominates in the spectrum (Fig. 2.4(b)). The tungstate ion (WOi -)
is a good example of case 3. The very broad emission spectrum (see Fig.
2.4(c)) does not show any vibrational structure at all, the Stokes shift is very
large (,...., 16 000 cm -1) and the zero-vibrational transition is not observable,
not even at the lowest possible temperatures, or for the highest possible
resolving powers.
Fig. 2.4 (a) An example of an emission spectrum in the weak-coupling case. The
spectrum consists of a line which corresponds to the zero-phonon transition. The
weak line at about 325 nm is a vibronic transition. This spectrum is the Gd 3 +
emission spectrum of LaB 3 0 6 :Gd 3 +. (b) An example of an emission spectrum in the
medium-coupling case. There is a progression of vibronic lines. The line on the
left-hand side is the zero-phonon line. This spectrum is the uranyl (UO~ +) emission
spectrum. (c) An example of an emission spectrum (M) in the strong-coupling case.
The spectrum consists of a broad, structureless band which shows a large Stokes shift
(SS) relative to the absorption (excitation) band (X). The spectra relate to the
tungstate luminescence of CaW0 4 .
28 Luminescent centres in insulators
Fig. 2.5 A configurational coordinate diagram in which the excited state v has its
equilibrium position outside the u ground state parabola. Return from the v
minimum to the ground state u is difficult and therefore slow. See also text.
Q
(a) (b)
E
Q
(c)
Fig. 2.6 Nonradiative transitions in the configurational coordinate diagram: (a)
strong coupling; (b) weak coupling; (c) combination of both.
crossing, it is possible to return to the ground state in a nonradiative
manner. The excitation energy is then completely given up as heat to the
lattice. This model accounts for the thermal quenching of luminescence.
In Fig. 2.6(b) the parabolae are parallel (S=O) and will never cross. It is
impossible to reach the ground state in the way described for Fig. 2.6(a).
However, nonradiative return to the ground state is possible if certain
conditions are fulfilled; that is, the energy difference ~E should be equal to
or less than 4-5 times the highest vibrational frequency of the surroundings.
In that case this amount of energy can excite simultaneously a few high
energy vibrations and is then lost for the radiative process. Usually this
nonradiative process is called multi phonon emission.
The configurational coordinate diagram 31
12
Q
oL---~~~~------~---
between the electronic states U and v cannot be neglected. For Cr 3 + this has
been shown theoretically as well as experimentally [16, 17].
However, let us take the approximation that the electronic parts of
equations (2.7) and (2.8) can be considered to be constants. The total
radiative rate is obtained by summing over all initial and final states Un and
Vm' For nonradiative transitions the rate is obtained by summing over all
nearly resonant Vm, Un states.
Figure 2.8 shows the results of a model calculation for a red-emitting
phosphor [18]. Ruv and N uv are taken to be 104 s- 1 and 10 14 s- 1 respect-
ively. Further, hw=500cm- 1 in the U and v states, and auv =7.746 (large
offset). The distance between the parabolae minima is 25000 cm -1. Such
a phosphor has a thermal quenching temperature T1/2 of 450 K (at T1/2
the emission intensity at 0 K has dropped to 50%). The results show how
a variation of one of the parameters influences the thermal quenching,
i.e. the nonradiative rate. Shifting the excited parabola to lower energy
The configurational coordinate diagram 33
Fig. 2.8 Calculations with the Struck-Fonger approach on a model phosphor. The
quantum efficiency q is calculated as function of temperature. In each of the three
pictures only one parameter is varied: in the picture on top the energy of the zero-
phonon transition (E zp ), in the centre the vibrational frequency (hw.) and in the lower
one the parabolae offset a. v The latter two parameters in particular show a drastic
influence on the temperature dependence of q. (After Bleijenberg and Blasse [18].)
with a and /3 constants and W max the highest available vibrational frequency
of the surroundings of the rare earth ion. This is the energy gap law in the
revised form of van Dijk and Schuurmans [20] that makes it possible to
calculate N with an accuracy of one order of magnitude.
In solids recently a new mechanism for loss of excitation energy gained
interest, although in principle this was already known for molecular species
in solution. Let us start with the latter where this mechanism is known as
quenching by electron transfer [21]. Figure 2.9 shows schematically the
essentials of this process. Parabolae a, band c represent energy levels of a
species A, parabola d that of a state A + B -. This state is obtained by
electron transfer from species A to a nearby species B. The offset of parabola
d relative to the other parabolae is large, and its position is at not too high
an energy. It is clear that from parabola c we can reach the ground state
/
/ d
/
/
/
/
----5d
CB
----5d
----4f
VB
Fig. 2.10 Schematic representation of the Ce3+ energy levels in Y3Als012' VB and
CB indicate the top of the valence band and the bottom of the conduction band of
Y3Als012 respectively. 4f gives the ground state of the Ce 3+ ion. The lowest excited
5d s-tate of Ce3+ is situated in the forbidden zone, but the lowest-but-one state is in
the conduction band making photoionization of Ce 3+ possible when the ion is
promoted into that state.
36 Luminescent centres in insulators
observed. Because these higher levels are situated in the conduction band,
photoionization of the Ce 3 + ion can take place. In analogy to the electron
transfer in ion pairs described above, the electron-transfer state involved can
be indicated as Ce 4 + --(HL)-, where HL denotes the host lattice.
Pedrini et al. have shown that this is an important mechanism for other
ions too, for example Eu 2 +, Sm 2 + and Yb 2 + [24,25]. Not always does this
mechanism prevent the occurrence of emission. In several cases they observe
what is called impurity-bound exciton emission. A nice example is Yb 2 + in
the fluorides CaF 2, SrF 2 and BaF 2. The former two yield impurity-bound
exciton emission, but in BaF 2 there is no emission at all [25]. The emission
is ascribed to exciton recombination: the hole resides on the (photoionized)
Yb 2 + ion, the electron on the surroundings. In Fig. 2.9 this emission
transition would be described by a transition from parabola d to the ground
state.
The processes described in the final part of this section may be of
importance for every optically active centre which can easily change its
valency, as will be clear from the examples presented.
Let us now turn to phenomena which are due to interaction between
centres.
(2.12)
Energy transfer 37
o R
- --0
---r--S* ----,r--A*
H
E----------- ~
-~--S A
------+ E
Fig. 2.11 Energy transfer from S to A. R is the SA separation, H is the SA
interaction. The hatched area in the lower picture presents the spectral overlap. See
also text.
Here the integral represents the spectral overlap. The distance dependence of
PSA depends on the interaction mechanism.
If the interaction is of the exchange type, PSA decreases exponentially with
the S-A distance (RSA)' because the wavefunctions do so too. If the
interaction is of the multipolar type, PSA decreases as Rs; where n depends
on the type of interaction. For electric dipole-dipole interaction, for
example, n = 6 (see Chapter 1).
It should be realized that a high value of PSA does not imply automati-
cally that transfer will occur. The excited state of S, i.e. S*, has other ways to
decay, i.e. radiatively (P r ) and/or nonradiatively (Por)' Transfer occurs if
PSA>Pr+Por' The distance for which the transfer rate equals the internal
decay rate is called the critical distance Re. For exchange-mediated transfer
Re is not much larger than 7-8 A, determined by wavefunction overlap. For
transfer by multipolar interactions Re can be much larger, i.e. 50--100 A if
favourable spectral overlap of allowed S*-+S and A-+A* transitions occurs.
An example of energy transfer can be found in the classic lamp phosphor
Ca5(p04h(F, CI):Sb3+, Mn2+ [28]. Short wavelength ultraviolet excitation
excites only the Sb 3 + ion which yields blue emission. Some of these ions,
however, transfer their excitation energy to Mn 2 + which yields yellow
emission. In this way white emission occurs. The emission colour can be
varied by varying the Sb3+:Mn 2 + ratio. The Sb3+ -Mn2+ transfer occurs by
exchange; its critical radius is 12 A.
38 Luminescent centres in insulators
20 _ _ _ _ 50
,
----50
o
10
Fig. 2.12 Cross-relaxation for two Eu 3 + ions (left-hand side) and for two Tb 3 +
ions (right-hand side). The arrow indicates the amount of energy which is trans-
ferred from one ion to the other. As a consequence the higher-level emission is
quenched.
Energy transfer 39
site), the luminescence efficiency of that composition will be low. This
phenomenon is called concentration quenching. This type of quenching will
not occur at low concentrations, because then the average distance between
the S ions is so large that the migration is hampered and the killers are not
reached.
Energy transfer is often studied by measuring the time dependence of the
S emission intensity after a short excitation pulse. We will discuss now a
couple of expressions often used in the literature. Simultaneously they
illustrate the several regimes of energy migration which have been observed.
Here energy migration is defined as a large number of subsequent energy
transfer steps between identical centres.
Consider first a crystal with donors and acceptors and assume that only
one-step energy transfer from S to A is possible. The S and A ions will be
distributed at random. An excited donor can interact with all unexcited
acceptors and it is necessary to account for the distribution in SA separa-
tions. This problem has been treated by Inokuti and Hirayama [29].
They obtained the following expression for the decay of S in the presence
of A:
(2.15)
(2.16)
where y = 2RjL with Rc the critical distance and L an effective Bohr radius.
Note that I(t) is not an exponential in the presence of A. In this treatment
only SA transfer is considered and SS transfer is assumed not to occur. In
the absence of A, the S species decays exponentially according to
I(t)=I(O) ex p ( - :J (2.17)
The first term on the right-hand side corresponds to processes which bring
the species n back to the ground state: YR is the radiative probability, Xn is
the transfer rate to acceptors and Ln' Wnn , gives the transfer rate from species
n to other donor species n'. The second term describes the reverse process.
For simplicity back-transfer from the acceptors is neglected.
The energy difference between ground and excited state, En" will vary
from donor to donor owing to perturbations from impurities, strains,
etc. This yields the inhomogeneous line broadening observed under broad-
band excitation. There are two techniques to follow the excitation energy
migration in the donor system, i.e. fluorescence line narrowing (FLN)
and the time evolution of the donor luminescence in the presence of
acceptors.
In FLN a pulsed, narrow band light source (a laser) excites those donors
whose resonance frequencies span a small part of the inhomogeneous line.
After the pulse, the luminescence evolves as shown schematically in Fig.
2.13. Broadband luminescence arises due to energy transfer to donors which
were not excited directly. The decay of the narrow component yields
information on the microscopic transfer process.
A well-known example is the case of Lao.sPro.2F 3 [31]. In Fig. 2.14 we
give the time evolution of the emission of the 3P o--+ep 6)1 transition on the
Pr 3 + ion. Excitation is at 12 cm -1 higher energy than the line centre. Note
(a)--=-----=------
(b) _-"'---_ _ _ _ __
6v(cm-')
Fig. 2.14 Time-resolved emission spectra for the 3PO-.eH6h luminescence in
La o.S Pr O.2 F 3. Excitation is 12 cm -1 above the line centre. T= 14 K. (a) Immediately
after the excitation pulse, (b) 0.811S after the pulse, and (c) 311S after the pulse.
(Constructed after data by Huber et al. [31].)
42 Luminescent centres in insulators
fIt)
0.1
0.01
0.001 "-_"'--"'---'--_....L-_........_---'-_ _
o 4 8 t
Fig. 2.15 Schematic plot of f(t) versus t: curve a, donor-donor transfer absent
(equations (2.15) and (2.16; curve b, rapid transfer (equation (2.21; curve c,
intermediate between curves a and b (e.g. equation (2.22.
exp [ -3 n
~ 3/2
CA
(C )1/2
t
(1 + 10.87x+ 15.50X 2)3/4]
1 +8.743x (2.22)
2.4.2 The rare earth ions Eu3+ (4f 6 ), Gd3+ (4f7) and Tb3+ (4f 8 )
Absorption spectra of the rare earth ions consist of sharp and weak lines
and, at higher energy, of broad band(s). Figure 2.17 gives an example. The
sharp lines are parity-forbidden transitions within the 4f n configuration, and
the broad bands are allowed transitions from the 4f n to excited configur-
ations. These are either 4fn -15d or 4f n + 1L -1 (charge transfer, L = ligand).
This situation bears an analogy with the Cr 3+ ion, and shows again the
influence of parabolae offset on spectral band shape, since AQ ~O for
intraconfigurational and AQ =/: 0 for interconfigurational transitions.
The emission of the three ions under discussion consists usually of
sharp lines due to transitions from the excited level to the ground levels:
5Do- 7F J for Eu3+ in the red, 6P7/2-+8S for Gd3+ in the ultraviolet, and
5D4 -+ 7FJ for Tb 3 + in the green. However, under special conditions other
emissions appear, for example 5D 2- 7 F J for Eu3+ in the blue, 5D3-7FJ for
Tb3+ in the blue, and for Gd3+ even vacuum ultraviolet (187 nm). Higher
level emission is only possible if (i) the vibrational frequencies in the lattice
are not too high, and (ii) cross-relaxation is absent owing to a low
concentration.
46 Luminescent centres in insulators
20 30 40
10 cm- 1
3
Fig. 2.17 Schematic absorption spectrum of the Eu 3 + (4f 6 ) ion in an oxide. The
broad band on the high-energy side is due to the charge-transfer transition. The lines
are due to transitions within the 4f 6 configuration.
The latter point is immediately clear from our discussion above. The
former will become clear from equation (2.11). The higher CO max is, the larger
is N. Indeed, the higher level emissions mentioned here do not occur at all in
hosts such as borates and phosphates.
If the vibrational frequencies become exceptionally high, as in aqueous
solutions (co max ~ 3500 cm -1), the quantum efficiency of even the lowest level
emission decreases. Only in Gd3+ (AE ~ 32 000 cm -1) does the radiative
rate still exceed the nonradiative rate by many orders of magnitude, but in
Tb3+ (AE ~ 15 000 cm - 1) and Eu 3+ (AE ~ 12 000 cm - 1) a drastic decrease
of the quantum efficiency occurs, whereas the other rare earth ions practi-
cally do not emit in aqueous solution. If, however, heavy water is used
(D 20, Vmax ~ 2200 cm -1) the efficiencies go up again. For solids these effects
can be studied in NaLa(S04h H 20 where the rare earth site is coordinated
to one H 20 molecule only. The q values are 100% for Gd3+, 70% for Tb3+,
10% for Eu3+ and 1% for Sm3+ and Dy3+ [7,8].
In many compounds of Eu3+, Gd3+ and Tb3+ energy migration has been
observed [6-8]. An example was given above for Gd3+. The Gd3+ ion
transfers efficiently to Eu3+ and Tb3+, and Tb 3+ to Eu3+ which has the
lowest emitting level of the three. The latter transfer explains why some
Tb 3+ compounds emit in the red: the Tb 3+ ions are excited, and the
excitation energy migrates over the Tb 3+ sublattice until it is trapped by a
Eu 3+ impurity from where emission occurs.
Finally we want to discuss what happens on excitation into the broad
absorption band. As an example we consider the charge transfer CT band of
Eu3+ [37]. In the commercially applied lamp phosphor Y203:Eu3+ excita-
tion into the CT state results in relaxation to the minimum of the CT
parabola, followed by a transition to the higher parabolae of the 4f 6
Some case studies 47
CT CT
\ \
\ \ E
\ \
\ E
\.
"- /
5D 5D
(a) (b)
Fig. 2.18 Configuration coordinate diagram for the Eu 3 + ion in (a) an yttrium host
lattice and (b) a calcium host lattice. In the latter the charge transfer state (CT) feeds
the 7F ground levels directly which quenches the luminescene due to emission
transitions between the 5D and 7F levels.
'" 1 eV) and structureless (see Fig. 2.4{c)), and the Stokes shift is very large
(1.5-2.0 eV).
It is interesting to note that in CaW0 4 there is no concentration
quenching although the tungstate groups are neighbours in the lattice.
However, the large reorganization of the excited state, as observed from the
large Stokes shift, brings the excited state out of resonance with the ground
state, i.e. the spectral overlap between emission and absorption vanishes, so
that tungstate-tungstate transfer becomes impossible: although there is
probably enough interaction, there is no resonance (compare equation
(2.12)).
This is not necessarily always the case. In YV0 4 :Eu3+ and Ba2MgW06:
U 6 + the dO complex ions (VOl- and WO~- respectively) have a relatively
small Stokes shift ('" 1 eV) and at room temperature (but not at low
temperatures) the thermal activation is high enough to allow energy migra-
tion over the sublattice of the complex ions. Finally this energy is trapped by
the activator (Eu3+ and U 6 + respectively) from where emission occurs. The
composition YV0 4 :Eu 3 + has been for some time a cathode ray and a lamp
phosphor [40]. At room temperature the undoped host lattices YV0 4 and
Ba2MgW06 are for the greater part concentration quenched: the migrating
excitation energy is trapped by quenching centres.
Let us return to CaW0 4 Why are isomorphous SrW0 4 and BaW0 4
considerably less luminescent than Ca W0 4 ? The answer to this problem is
in principle easily found. The ionic radius of the alkaline earth ion increases
from calcium to barium. Therefore expansion of the excited state is easier in
the softer BaW0 4 than in CaW0 4 . Again small changes in AQ have drastic
influences on the luminescence efficiency (i.e. on the rates of the nonradiative
processes).
It would be incorrect to think that the exact nature of the optical
transitions in such classic materials is known. The reason for this is that the
excited state, with an electron in the d orbital of the central ion, and a hole
spread over the ligands, gives rise to a large number of excited levels. Two
facts seems to be clear nowadays. First, the emission originates from a triplet
state [39], so that it is spin forbidden and slow (the excited state of the
vanadate group, for example, lives about 1 ms). Second, the excited state is
strongly Jahn-Teller distorted as shown recently by van der Waals' group
(see for example ref. 41). This implies that the simple configurational
coordinate used so far is certainly incorrect. Other coordinates have to be
taken into account too.
FREE ION
Cs 3 Bi2Brg Bi 4 Ge 3 0'2
Fig. 2.19 Schematic representation of luminescent centres with different character-
istics: Elf' electron-lattice relaxation; J, band broadening. See also text.
shows that there are still more, but these phenomena are discussed in other
chapters in this book.
REFERENCES
1. Blasse, G. (1989) Chern. Mater., 1, 294.
2. Becquerel, E. (1867) La Lurniniere - ses Causes et ses Effets, Didot, Paris.
3. Imbusch, G.F. (1988) In Lasers, Spectroscopy and New Ideas (eds W.M. Yen and
M.D. Levenson, Springer, Berlin, p. 248.
4. DiBartolo, B. (ed.) (1987) Spectroscopy of Solid-state Laser-type Materials,
Plenum, New York.
5. Henderson, B. and Imbusch, G.F. (1989) Optical Spectroscopy of Inorganic
Solids, Clarendon, Oxford.
6. DiBartolo, B. (ed.) (1984) Energy Transfer Processes in Condensed Matter,
Plenum, New York.
7. Blasse, G. (1988) Prog. Solid State Chern., 18, 79.
8. Blasse, G. (1990) Adv. Inorg. Chern., 35, 319.
9. Holstein, T., Lyo, S.K. and Orbach, R. (1981) In Laser Spectroscopy of Solids (eds
W.M. Yen and P.M. Selzer), Springer, Berlin, p. 39.
to. Weber, M.J. (1981) Laser Spectroscopy of Solids (eds W.M. Yen and P.M. Salzer),
Springer, Berlin, p. 189.
11. Glidel, H.U. (1990) Chern. Phys. Lett., 175,262.
12. DiBartolo, B. (ed.) (1980) Radiationless Processes, Plenum, New York.
13. DiBartolo, B. (ed.) (1991) Advances in Nonradiative Processes in Solids, Plenum,
New York.
14. Struck, C.W. and Fonger, W.H. (1975) J. Lurnin., 10, 1.
15. Struck, C.W. (1991) In Advances in Nonradiative Processes in Solids (ed. B.
DiBartolo), Plenum, New York.
16. Englman, R. and Barnett, B. (1970) J. Lurnin., 3, 37, 55.
17. Donnelly, CJ., Healy S.M., Glynn, TJ., Imbusch, G.F. and Morgan, G.P. (1988)
J. Lurnin., 42, 119.
18. Bleijenberg, K.c. and Blasse, G. (1979) J. Solid State Chern., 28, 303.
19. Blasse, G., Dirksen, G.J., Sabbatini, N. and Perathoner, S. (1987) Inorg. Chirn.
Acta, 133, 167.
References 51
20. van Dijk, 1M.F. and Schuurmans, M.F.H. (1983) J. Chem. Phys., 78,5317.
21. Balzani, v., Bolletta, F., Gandolfi, M.T. and Maestri, M. (1978) Top. Curro Chem.,
75, 1.
22. Balzani, V., Sabbatini, N. and Scandola, F. (1986) Chem. Rev., 86, 319.
23. Sabbatini, N. and Balzani, V. (1985) J. Less-Common Met., 112, 381.
24. Pedrini, e., Rogemond, F. and McClure, D.S. (1986) J. Appl. Phys., 59, 1196.
25. Moine, B., Courtois, B. and Pedrini, e. (1989) J. Phys. (Paris), 50, 2105.
26. Biasse, G., Schipper, W.J. and Hamelink, 1.1. (1992) Inorg. Chim. Acta., 159, 77.
27. Hamilton, D.S., Gayen, S.K., Pogatshnik, G.J. and Ghen, R.D. (1989) Phys. Rev.
B,39,8807.
28. Soules, T.F., Bateman, R.L., Hewes, R.A. and Kreidler, E.R. (1973) Phys. Rev. B,
7, 1657.
29. Inokuti, M. and Hirayama, F. (1965) J. Chem. Phys., 43, 1978.
30. Huber, D.L. (1981) In Laser Spectroscopy of Solids (eds W.M. Yen and P.M.
Selzer), Springer, Berlin, p. 83.
31. Huber, D.L., Hamilton, D.S. and Barnett, B. (1977) Phys. Rev. B, 16, 4642.
32. Burshtein, A.I. (1972) Sov. Phys. J ETP, 35, 882.
33. Yokota, M. and Tanimoto, I. (1967) J. Phys. Soc. Jpn., 22, 779.
34. Chien, R.L., Berg, 1M., McClure, D.S., Rabinowitz, P. and Perry, B.N. (1986) J.
Chem. Phys., 84,4168.
35. Imbusch, G.F., Glynn, T.J. and Morgan G.P. (1990) J. Lumin., 45, 63.
36. Biasse, G. (1990) Mater. Chem. Phys., 25, 393.
37. Biasse, G. (1992) Mater. Chem. Phys., 31, 3.
38. Kroger, F.A. (1948) Some Aspects of the Luminescence of Solids, Elsevier,
Amsterdam.
39. Biasse, G. (1980) Struct. Bonding, 42, 1.
40. Levine, A.K. and Palilla, F.e. (1964) Appl. Phys. Lett., 5, 118.
41. Barendswaard, W. and van der Waals, J.H. (1986) Mol. Phys., 59, 337.
42. Ranfagni, A., Mugnai, D., Bacci, M., Viliani, G. and Fontana, M.P. (1983) Adv.
Phys., 32, 823.
43. Biasse, G. (1987) Mater. Chem. Phys., 16, 201.
3
Luminescence spectroscopy
U. w. Pohl and H.-E. Gumlich
3.1 INTRODUCTION
Energy levels within the band gap of semiconductors play an important role
in both the physics and the technology of semiconductors. They control
mobility and lifetime of free carriers and give rise to both transitions
connected with luminescence emission and radiationless transitions. While
luminescence results from radiating relaxation, a considerable amount of the
relaxation usually occurs within nonradiative processes.
Luminescence spectroscopy is a very helpful tool to unravel the some-
times complicated electronic structure of localized centres and the pro-
cesses of energy transfer among different centres in the semiconductors.
Concerning visible luminescence, the II-VI compound semiconductors
are of special interest because their spectra cover the whole visible and
infrared range, while the 111-V cover a part of the visible and infrared
spectrum and the elementary semiconductors such as silicon the infra-
red range only.
Luminescence emission can be excited in various ways. The photo-
luminescence (PL) is excited by photons, cathodoluminescence (CL) by
cathode rays, electroluminescence (EL) by electric fields and thermolumines-
cence (TL) by thermal release of energy stored in previously excited solids.
This chapter predominantly refers to photoluminescence spectroscopy. A
brief survey of different luminescence processes in semiconductors (section
3.2) is followed by a description of the experimental set-up which is
necessary for their excitation and detection (section 3.3). Finally some case
studies demonstrate applications from luminescence and excitation spectros-
copy in current research (section 3.4).
53
54 Luminescence spectroscopy
3.2.1 Excitons
Excitons are localized excited electronic states in a nonmetallic crystalline
lattice. The theory of excitons developed in the 1930s is well established by
now [1]. Two models are applied depending on the degree oflocalization of
the excitation energy. The Frenkel exciton [2] occurs if the energy is
essentially localized on one ion within the crystal as realized in the alkali
halides. The opposite extreme of the Wannier-Mott exciton [3] is found in
materials with a considerable degree of covalent bonding as in II-VI, III-V
or element semiconductors. In this case the excitation energy is extended
over several unit cells and can be regarded as an interacting electron-hole
pair with a small binding energy. The effective mass theory gives a good
description of this model in analogy to the hydrogen atom. The properties of
the crystal are represented by effective masses of the electron m:
and the
hole mt and by the dielectric constant e. The model leads to hydrogen-like
energy series of the exciton with quantum number n:
f.1*e 4 1 fj 2 k2
E =E - - - -+---..,-
n 9 2e 2 h 2 n2 2(m: + m:)
with the reduced mass 1/f.1* = 11m: + 11m:. The ionization energy is repre-
sented by the gap energy E g As the valence band of a semiconductor can be
split by spin-orbit interaction or electric fields originating from crystal fields
or strains several gap energy values E~, E: can appear and thus also
different corresponding exciton series. In addition direct and indirect transi-
tions have to be considered. The last term in the formula represents the
kinetic energy which leads to a broadening of the free exciton lines.
Excitons may form complexes by a bond to defects as donors, acceptors
or dislocations [4, 5]. The exciton transition energy in these cases is lowered
by the binding energy which increases approximately linearly with increas-
ing ionization energy of the defect [6]. For an additional offset factor called
the central-cell correction see ref. 7. As the kinetic energy of the bound
exciton complex is zero the optical transitions lead to sharp lines.
Under intensive excitation of the crystal, excitons start to interact with
each other by scattering processes and by forming biexcitons. Formation
and decay of these excitonic molecules give rise to various nonlinear optical
phenomena [8, 9]. At an extreme increase of the excitation intensity and
thus of the exciton density a threshold value called the Mott density is
reached where the Coulomb bond between individual electrons and holes is
broken and a plasma state is formed.
The general features of excitonic systems described above provide much
information on crystals by the study of exciton transitions. By recording the
free exciton luminescence the gap energy E g , its temperature dependence,
Information deduced from luminescence spectra 55
EA and ED being the ionization energy of the donor and acceptor respective-
ly; additional van der Waals and s-p exchange corrections are discussed in
ref. 11. As donors and acceptors occupy definite sites within the crystal
lattice the distribution of different pairs with distances R j is determined by
lattice statistics. The validity of this basic approach was demonstrated first
by highly resolved spectra of zero-phonon lines due to donor-acceptor pairs
in GaP where an unambiguous assignment of pair lines to the impurity
distances R j has been given [13].
Even if the zero-phonon lines are not resolved the series of pronounced
phonon-assisted transitions dominated by LO phonons can be detected. The
ratio of the phonon peaks with numbers n of one series is given by a Poisson
distribution In= Ion!/N n with the number N of emitted phonons of the order
56 Luminescence spectroscopy
--
'o.!r.~~~~~~~~~~~~~~~~~~~~~~~~
~ t== - - - +-~ ~_ -=- ~_
~ .200 watt mercury_lamp - - --
_ .-
. m-_'-
g
oo
t;;
"'E-
:1--I----lI--.r.--~--H-_I__-H-I---I---~ r-- -l-
l~.O
:.~_
.
tg --
.0
3 .0
2.
1111 --
1'-
\:
--
-b~ ~ .:::
- -->:) -'0
---=f----:.~:j:t:t=t~
150 watt xenon lamp
-- -- m-~
-'- -.--+~
.--
!-
U- j
.
~.
--
--l&'--M - - IV . \ - - ~;:;;7
I-
1: ~::
~ 0.05
~ ~:: ~-L ~h~t.~~~~~~~~~~:~~-~~=~~.~..~.-~-~-~-~~~~~~
'" I--- ~50 watt deuterium lamp
e
~ .02
O.OIS
. !!._o.~ _
::; 006
0.006
_ _ ~
0.005
00< II\.
000 l'
O.O(n$
0.002 ~=:j==!:=t=:t==~=+:s::*~~=f4=4=+=+=+=~
1_
0.00' L200-...L-300..L-L-.,.,l,oo:--..-L-----;500..!:--L--600:;!;;-L-.LJ----;;!,00~---L..--..1600~--L---;;""';;;;'
Wavelength (nm)
Fig. 3.2 Spectra of various lamps used as broadband radiation sources (Catalog
Oriel).
Table 3.2 Data for cryogenic liquids at normal pressure (1013 mbar): boiling point
T boll , liquid density Q, heat capacity cp at T boll volume ratio of gas (at 273 K) and
liquid (at T boil ) and latent heat of vaporization Iv
3He 4He H2 Ne N2 Ar
The study of the effect of electric and magnetic fields on the optical
properties of solids helps in understanding the nature of electronic and
vibronic states. The application of hydrostatic pressure provides an efficient
tool to introduce high electric fields preserving the symmetry of the crystal.
Uniaxial pressure, on the other hand, introduces an electric field with a
unique axis which reduces the crystal symmetry. The same is true for the
application of magnetic fields. In any case, the application of an external
constraint changes the electronic and vibronic properties of the crystal and
of localized centres in the samples.
Hydrostatic pressure essentially changes the chemical bond and conse-
quently band structure, phonon modes and, at sufficient strength, even the
crystal structure. Thus zincblende structured A-B compounds (such as the
Zn and Cd chalcogenes) are known to undergo a phase transition into an
NaCl structure in a medium pressure regime and a metallic (mostly
orthorhombic) structure at high pressure [27]. These changes are closely
connected to the ionicity of the chemical bond.
Uniaxial stress (Stark effect) and a magnetic field (Zeeman effect) lift
partially or fully the degeneracy of electronic levels. In particular, in the
study of zero-phonon lines related to localized defect centres the symmetry
of the involved electronic states may be deduced from the resulting splitting
pattern. The parameters to be determined are the number of components for
the external fields along some particular crystalline axis (in cubic crystals
along [100], [110] and [111]), the energy shift of each component, their
relative intensity and their polarization (n or 0'). The comparison of these
experimental findings with calculations based on group theory allows the
assignments of electronic states to energy term levels of well-defined symme-
try [18]. In the following some experimental details of pressure and Zeeman
set-ups are considered.
For the application of high hydrostatic pressure the gasketed diamond
anvil cell has proved to be the most suitable technique [28]. It can be used for
optical measurements ranging from the ultraviolet to the infrared (apart from
a range with diamond absorption between 4 Jlm and 6 Jlm) with pressures up
to 1.7 X 1011 Pa (1.7 Mbar) within a cell size of typically 0.1 mm height and
0.2 mm diameter. The cell is composed of two opposed transparent diamond
anvils with parallel faces. The anvils are pressed against a metal platelet with a
centred hole of the cell diameter (Inconel gasket). This pressure chamber is
filled with a fluid acting as a pressure-transmitting agent, the sample and some
tiny pieces of ruby [29]. The ruby is used as a pressure gauge by recording the
66 Luminescence spectroscopy
1. The sensitivity, usually given in amperes of the electric signal current per
watt of the incident light power. The detector may also be characterized by
its quantum efficiency, defined as the ratio of signal electrons to incident
photons.
2. The range of the wavelengths in which the detector is sensitive.
3. The signal-to-noise ratio (S + N)/N. Usually the noise of the detector is
characterized by an equivalent of the incident light which would cause the
same noise level as the detector does (noise-equivalent power (NEP)). The
reciprocal of the NEP is designated as the detectivity D.
4. The upper limit ofthe frequency of the detector which is determined by the
lowest build-up time of the signal triggered by an ideal rectangular light
pulse.
5. The linear range of the detector where the signal is proportional to the
incident radiation intensity.
100
10
Wavelength (IJm)
Fig. 3.3 Spectral response and quantum efficiency of several photocathode ma-
terials [31].
where again the number of electrons is increased. The typical gain at each
dynode is given by a factor of 4. After typically ten dynodes the photo-
current is enhanced by a factor of 4 10 ~ 10 7 . Apart from the dynode
configuration the time resolution is given by the load resistance and the
capacitance of the anode and is typically of the order of 10- 9 s.
In the case of extremely weak light intensities photomultipliers are used in
the photon counting mode. In this case voltage pulses due to the registration
of single quanta are measured instead of an integrated photocurrent. In a
discriminator photoelectron pulses may be separated from smaller pulses
created by thermionic emission and from larger pulses originating from
cosmic radiation. Compared with the analogue registration the method of
photon counting has several advantages.
1. Fluctuations due to the statistical amplification by the dynodes do not
play any role because photoelectron pulses are selected by discrimina-
tion. Also, pulses triggered by cosmic or radioactive radiation near the
tube can nearly be eliminated.
2. The leakage current at the photomultiplier tube will not be counted.
3. The counting rate can be digitally treated, facilitating the computing of
the spectra.
The upper limit of the counting rate depends essentially on both the dead
time of the tube and the time resolution of the discriminator and the counter.
As a consequence of this limit, which is of the order of 10- 8 s, sequences of
statistically distributed pulses occurring at a rate of 10 MHz can be handled.
Since the dark current of photomultiplier tubes decreases to a certain
extent with decreasing temperature the signal-to-noise ratio can be en-
hanced by cooling. In most cases, cooling down to - 40 DC is sufficient,
which can easily be achieved by Peltier elements.
A drastically improved real-time response in luminesence spectroscopy is
attained by optical multichannel analysers (OMAs). These detectors can be
divided into two classes depending on the method by which the stored
spectrum is read out [32]. In the vidicon TV camera technology the charge
pattern on the target of the vidicon is read out by an electron beam. The
other group uses a solid-state array of photodiodes and shift registers for the
read-out procedure. OMA systems are particularly useful for studying
short-time optical phenomena because the whole spectrum can be stored
after a single laser pulse regardless of its pulse duration. To reduce noise
many spectra can be integrated before read-out. OMA systems are thus
equivalent to up to 104 photomultiplier tubes working in parallel.
Often OMA systems are equipped with image intensifiers such as the
micro channel plate (MCP). The MCP consists of a photocathode coated
on a fibre optic face plate which is mounted in close proximity to a
mosaic of electron channel multipliers. These multipliers are hollow glass
fibres of 10-40 Ilm diameter with an internal resistive coating. The fibres are
72 Luminescence spectroscopy
The single photon counting technique is especially useful for recording the
transient of a very weak luminescence in the nanosecond timescale. The
principle of this method is based on the fact that the transient of lumines-
cence after pulse excitation reflects the number of emitted photons which is
proportional to the probability of detecting photons. The luminescence must
be so weak (or attenuated so strongly) that not more than one single photon
can be detected after a single pulse excitation. For many laser shots the
delay time between the pulse excitation and the registration of a single
photon is measured and subsequently an abundance of different delay times
is counted. In the electronic equipment this is achieved by a time-to-
amplitude converter (TAC) creating a pulse height that is proportional to a
delay time between a 'start' and a 'stop' signal (during this time a capacitor
is charged by a constant current supply). While the start pulse is provided by
the laser during the laser shot the stop pulse is created from the detection of
the luminescence photon. The voltage of the T AC is read out and collected
in a multichannel analyser operating in a pulse height analysis mode. Thus
the collected pulse height distribution which is delivered after a large
number of laser shots scans the decay of the weak luminescence signal. It
should be emphasized that the measured transient is falsified by 'pile-up
errors' if more than one single photon is allowed to reach the detector after
one laser pulse. While this technique is well known with photomultiplier
detection its use has also been described in the near infrared using gated
avalanche photodiodes by Levine et ai. [33].
Wavelength (nm)
444 443 442
12 ZnSe/GaAs
T=1.6K
I; X1h
I I
0
a
~
li.I;. I;. 12c 12d 12
Ii)
."!:
c:
::J
b
.c
~ 0
->
-
'in
c:
Q)
c:
c
O~-r ____~____~____~__~
2.790 2.795 2.800 2.805
Energy (eV)
Fig. 3.4 Luminescence under band-band excitation (a), excitation of the 12 line (b)
and reflection (c) of a ZnSe/GaAs epilayer in the excitonic energy range [36].
could be clarified (CI resulting from the substrate cleaning procedure before
growth, Al from residual impurities of the source materials).
A similar picture of strain probing as obtained from the measurements of
different samples can be obtained with a single sample by resonant Raman
scattering if the laser energy is tuned [37]. The variation of the laser energy
versus the energy of the Stokes lines resulting from a resonant enhancement
at the 12i transitions shows an identical dependence to that for the excitation
line shifts versus the 12 detection energy shown in Fig. 3.5. This phenomenon
has been explained by a resonant amplification of the Raman signal only by
those (DO, X) complexes which have suitable 12i energy levels with respect to
the irradiated laser energy owing to their local strain. While the lumines-
cence and excitation spectra provide superimposed signals reporting the
average strain of a broad region in a layer the resonant Raman scattering
therefore allows one to scan the depth profile of the strain in one layer by
tuning the laser.
76 Luminescence spectroscopy
b-E(meV)
ZnSe
-5
E(eV)
Fig. 3.5 Line positions of the 12i excitation resonances of the 12 emission line in
differently strained ZnSe/GaAs layers [36].
~:I----: FE
- 0
::
:c
'"
'">
"c: 0c;;_
~
~
"'c:" Co
E
lattice
~'"
o misfit strain
"
-10
Ih
-15 I
I ZnSe/GaAs
__
Ih JI TG'OWlh = 593 K
ThP = 1.6K
0.1 d, do 10 d(~m)
Layer thickness
Fig. 3.6 Splitting and energy shifts of the heavy hole (hh) and light hole (lh) valence
bands in ZnSe/GaAs epilayers in dependence of the ZnSe layer thickness. The
dashed lines represent the contributions of lattice-misfit and thermally induced strain
[38].
Tm=15K 00 I~
ZnSe:N
12
?:
.,
iii
c:
a,
E.
~
~
.E
...J
D.. I~
4400
~
4450
I~-LO
O2
4000
t 03
04
5000 6000 7000
Wavelength (A)
junction laser with ZnSe has been performed with an N-doped MBE
structure [47].
The luminescence intensity provides additional information. An effective
doping can be read from a high ratio of the intensity owing to bound exciton
lines to the deep level centre luminescence. However, a reduced ratio might
be observed at high doping, indicating the formation of associates. On the
other hand, from a high intensity ratio of the free exciton lines versus the
bound exciton lines a high purity of the sample can be concluded.
Strong bound exciton lines indicating an effective doping have especially
been found in lattice-matched ZnSSe/GaAs layers. It should be noted that
this feature was observed with p doping [48] as well as with n doping [49].
In both cases the strong ratio of the bound exciton lines with respect to the
deep 'self-activated' luminescence observed in the optical spectra was
accompanied by low electrical resistivities.
ell:citation
emission
::I:tJT.L~:: ::~}.[~::
c ~
I
~
I
C B CI / CI / B
Fig. 3.9 Four different lattice sites of Mn2+ in polymorphic Zn8 differing in the
stacking layers above and below the impurity. Points and circles represent Zn2+ and
8 2 - ions, respectively, small circles denote the Mn2+ impurity ion [55].
a total
(")
(")
N
N
en
T N
'So N co
(")
N
N
co
ot
N
N
ot
ot
N
N (")
Ltl
N
C axial 1 N
d axial :2
rec.:18006 cm-'
.E
11'1
0
~ co
>.
CI
"-
III
C
III
C
0
'iii
UI 11'1
'f 0
-
0
III co
emission lines appears in the form of a series, the limit of which is represented
by the single Mn2+ emission zero-phonon line. A convincing illustration of
site selection can be seen in the excitation spectra recorded in these emission
lines labelled Pj, P j and P k in Fig. 3.12. Because of spin-orbit coupling two
ZnS:Mn
6 A ,(S)-+4 T,(G)
Rec.: P,
x3
Rec.: Pk
::J
x4
'"
~
'iii
c:
Ql
C
Rec.: P,
Rec.: Sem
19660 19700
Wavenumber (cm ')
Fig. 3.12 Zero-phonon excitation spectra recorded at the 'single' Mn2+ emission
line (Sem) and at the additional emission lines Pj' Pk and Pl' Arrows denote positions
of excitation lines connected to the selected emission line [60].
Some case studies 87
sharp lines and one broad line are expected in the Mn 2 + zero-phonon
excitation of the 4T 2(G} level. Despite the strong superposition of the lines
from the dominating Mn 2 + 'singles', selective registration allows the
detection of the much less abundant centres. The relative intensities of their
emission lines with respect to the single Mn 2 + emission increase with Mn
doping up to a limit of about xMn~3 X 10- 3 This finding is in accordance
with calculations of the probabilities of Mn 2 + pairs with different mutual
interimpurity distances. For a given interaction range the probability for
each kind of small cluster can be expressed by a polynomial whose powers
and coefficients have been calculated for various cubic-structured lattices
[19]. The application to substitutional Mn 2 + ions on a Zn f.c.c. sublattice of
a zincblende crystal with an interaction reaching the seventh-nearest neigh-
bour confirmed a maximum in the doping-dependent probability below
X~ 10- 2 followed by a decrease at stronger doping (this effect is related to
the formation of larger clusters).
From the series behaviour and the cluster statistics an assignment of the
Mn2+ pair clusters emitting the lines Pi (not shown in Fig. 3.12), Pi' P k and
PI to pair distances ranging from the third- to the sixth-nearest lattice site
has been inferred. At smaller distances the fine structure changes drastically
owing to the formation of the d lo electron system reported earlier [61].
Wavelength
+---
TO
ZnS:Mn
T=5K
-
17600 17700 17800 cm- 1
Energy
Fig. 3.13 Gated emission spectra of ZnO.99Mno.OlS, Spectra 1 and 2 differ only by
the time gate set to 0.2-5.0 ms and 0-0.2 ms, respectively. Spectra 2 and 3 appear at
different excitation energies [61].
T=100 K
x=0.41
80.00 v.x =21650cm- 1
(i)
.~
c:
::J 60.00
.ci
-
~
>
40.00
-
iii
c:
Q)
c:
20.00
Fig. 3.14 Time resolved emission spectra of Znl-xMnxS (x=O.4) at T= 100 K for
different time gates (ttft2). Excitation occurs into the 4E(G} level of the single Mn2+
ion. The curves are normalized.
ZnS:Mn
BII [111] B= 5T
Fig. 3.15 Zeeman emission spectra of the single Mn 2 + emission Sem and the
selectively excited lines P j and L 5. Energy scale refers to the L5 spectrum, the spectra
of P j and Sem are shifted by - 23 em -1 and -40 cm -1, respectively. Marked
postions denote transitions to Ms states of the 6A 1(S) ground state [64].
Wavelength (nm)
1225 1223 1221
ANv, AN v AN III ANII
I I 1570 lJev l
ZnS
B=5T
B11[111]
~ T=1.8K
.iii
...c:c:
Q)
1.012 1.014
Energy (eV)
Fig. 3.16 Zeeman emission spectrum of Fe3+ in ZnS. Transitions into different Ms
states are labelled AN(-AN v( [65].
to a tetrahedral axis into two levels and one into four levels. A configuration
of the detection system parallel to the magnetic field (Faraday configuration)
leads to polarization selection rules with (j +, (j - and 'It polarized lines from
which the symmetry of the initial and final state can be deduced [66]. The
results of these experiments are plotted in Fig. 3.17. The great number of
transitions in the magnetic field originates from excitation lines starting
from various thermally occupied levels of the split 6 Al (S) ground state. The
numbers of the notation of the Zeeman levels in Fig. 3.17 refer to these Ms
levels, the lowest-lying level being denoted by O. Lower-case letters denote
excited states. As transitions starting from all six ground state levels have
been found the application of the selection rules enabled the assignment of
not only the Zeeman levels of the excited 4E(G) state but also those of the
6 A 1 (S) ground state. The experimental assignments have been confirmed by
theoretical perturbation calculations [67].
The action of an external electric field is introduced by the application of
either hydrostatic or uniaxial pressure on the studied sample; see section
3.3.3. The application of uniaxial stress involves a change of the site
symmetry of a studied impurity. The reduction of the symmetry differs,
however, from the magnetic case because the electric field does not affect
directly the spin of the studied centre. Both hydrostatic and uniaxial
pressure experiments can thus be used to complement the results of Zeeman
measurements.
Some case studies 93
00
21255
21250 eO
dO
21245 e11
'i
E
.e rS
> 21240
Ol
Q)
C o d2
ra
Q)
o 0 91
21235
~~ ____+-------h1
r1 o
h2
92
09 2
o h2
21230 o 93
o h3
94
Icp
21225
o h5 rcp
linear
21220+0----~1----~2-----3r----4~---+5~
REFERENCES
1. Cho, K. (ed.) (1979) Excitons, Topics in Current Physics, Vol. 14, Springer, New
York.
2. Frenkel, 1.1. (1936) Phys. Z. Sowjet Union, 9, 158.
3. Wannier, G.H. (1937) Phys. Rev., 52, 191.
4. Lampert, M.A. (1958) Phys. Rev. Lett., 1, 450.
5. Dean, PJ. and Herbert, D.C. (1979) In Excitons (ed. K. Cho), Topics in Current
Physics, Vol. 14, Springer, New York, p. 55.
6. Haynes, 1.R. (1960) Phys. Rev. Lett., 4, 361.
7. Halsted, R.E. (1967) In Physics and Chemistry of II-VI Compounds (eds M. Aven
and 1.S. Prener) North-Holland, Amsterdam.
8. Haug, H. (ed.) (1988) Optical Nonlinearities and Instabilities in Semiconductors,
Academic Press, New York.
9. Klingshirn, C. (1990) Semicond. Sci. Technol., 5, 457.
10. Ungier, W. Suffczynski, M. and Adamowski, 1. (1981) Phys. Rev. B, 24, 2109.
11. Dean, P.l. (1973) In Progress in Solid State Chemistry (eds lO. McCalvin and G.
Somorjai), Vol. 8, Pergamon, New York, p. 1.
12. Williams, F. (1968) Phys. Status Solidi 25, 493.
13. Thomas, D.G., Gershenzon, M. and Trumbore, F.A. (1964) Phys. Rev., 133,
A269.
14. Monemar, B. (1990) In Impurities, Defects and Diffusion in Semiconductors: Bulk
and Layered Structures (eds D.l. Wolford, 1. BernhoIc and E.E. Haller), MRS
Symposium Proceedings, Vol. 163, Materials Research Society, Pittsburgh, PA.
15. Imbusch, G.F. and Kopelman, R. (1981) In Laser Spectroscopy of Solids (eds
W.M. Yen and P.M. Selzer), Topics in Applied Physics, Vol. 49, Springer, New
York.
16. Curie, D. (1963) Luminescence in Crystals, Wiley, New York.
17. Sugano, S., Tanabe, Y. and Kamimura, H. (1970) Multiplets of Metal Ions in
Solids, Academic Press, New York.
18. Griffith, 1.S. (1971) The Theory of Transition Metal Ions, Cambridge University
Press, Cambridge.
19. Pohl, U.W. and Busse, W. (1989) J. Chern. Phys., 90, 6877.
20. Curie, D. and Prener, 1.S. (1967) In Physics and Chemistry of II-VI Compounds
(eds M. Aven and lS. Prener), North-Holland, Amsterdam.
21. Forster, T. (1949) Z. Naturforsch. A, 4, 321.
22. Dexter, D.L. (1953) J. Chern. Phys., 21, 836.
23. Hamilton, T.D.S., Munro, I.H. and Walker, G. (1978) In Luminescence Spectro-
scopy (ed. M.D. Lumb), Academic Press, New York.
References 95
24. Henderson, B. and Imbusch, G.F. (1989) Optical Spectroscopy of Inorganic
Solids, Oxford University Press, New York.
25. Shimoda, K. (1984) Introduction to Laser Physics, Springer Series in Optical
Sciences, Vol. 44, Springer, New York.
26. Mollenauer, L.F. and White, J.e. (1987) Tunable Lasers, Topics in Applied
Physics, Vol. 59, Springer, New York.
27. Phillips, J.e. (1971) Phys. Rev. Lett., 27, 1197.
28. Martinez, G. (1980) In Handbook of Semiconductors (ed. M. Balkanski), Vol. 2,
North-Holland, Amsterdam, p. 184.
29. Piermarini, G.J., Block, S. and Barnett, J.D. (1973) J. Appl. Phys., 44, 5377.
30. Demtroder, W. (1977) Grundlagen und Techniken der Laserspektroskopie, Spring-
er, Berlin.
31. Keyes, R.J. (ed.) (1977) Optical and Infrared Detectors, Topics in Applied Physics,
Vol. 19, Springer, New York.
32. Chang, R.K., Long, M.B., Cardona, M. and Giintherodt, G. (1982) Light
Scattering in Solids II, Topics in Applied Physics, Vol. 50, Springer, New
York.
33. Levine, B.F. and Bethea, e.G. (1984) Appl. Phys. Lett., 44, 558, 581, 649.
34. Gutowski, J., Kudlek, G. and Presser, N. (1990) Phys. Status Solidi A, 120, 11.
35. Kleiman, J., Park, R.M. and Quadri, S.B. (1987) J. Appl. Phys., 61, 2067.
36. Kudlek, G., Presser, N., Pohl, U.W., Gutowski, 1., Lilja, J., Kuusisto, E., Imai, K.,
Pessa, M., Hingerl, K. and Sitter, H. (1992) J. Cryst. Growth, 117, 309.
37. Kudlek, G., Presser, N. and Gutowski, J. (1991) Semicond. Sci. Technol., 6, A83.
38. Kudlek, G. and Gutowski, 1. (1992) J. Lumin., 52, 55.
39. Gumlich, H.-E. (1988) In Growth and Optical Properties of Wide-Gap II-VI
Low-Dimensional Semiconductors (eds T.e. McGill, e.M. Sotomayor-Torres and
W. Gebhardt), NATO ASI Series, Vol. 200, Plenum, New York.
40. Mandel, G. (1964) Phys. Rev., 134, A1073.
41. Mandel, G., Morehead, F.F. and Wagner, P.R. (1964) Phys. Rev., 136, A826.
42. Neumark, G.F. (1989) Phys. Rev. Lett., 62, 1800.
43. Laks, D.B., Van de Walle, e.G., Neumark, G.F. and Pantelides, S.T. (1991) Phys.
Rev. Lett., 66, 648.
44. Goetz, G., Po hi, U.W. and Schulz, H.-1. (1992) J. Phys. C, 4, 8253.
45. Kukimoto, H. (1988) In Growth and Optical Properties of Wide-Gap II-VI
Low-Dimensional Semiconductors (eds T.e. McGill, e.M. Sotomayor-Torres and
W. Gebhardt), NATO ASI Series, Vol. 200, Plenum, New York.
46. Yoshikawa, A., Muto, S., Yamaga, S. and Kasai, H. (1988) Jpn. J. Appl. Phys., 27,
992.
47. DePuydt, J.M., Haase, M.A., Qui, J. and Cheng, H. (1992) J. Cryst. Growth, 117,
1078.
48. Suemune, I., Yamada, K., Masato, H., Kanda, T., Kan, Y. and Yamanishi, M.
(1988) Jpn. J. Appl. Phys., 27, L2195.
49. Fujita, S., Terada, K., Sakamoto, S. and Fujita, S. (1989) J. Cryst. Growth, 94,102.
50. Motegi, N. and Shionoya, S. (1973) J. Lumin., 8, 1.
51. Szabo, A. (1970) Phys. Rev. Lett., 25, 924.
52. Langer, D. and Ibuki, S. (1965) Phys. Rev., 138, A809.
53. Neumann, E. (1971) Thesis D83, Technical University, Berlin.
54. Lambert, B., Buch, T. and Geoffroy, A. (1973) Phys. Rev. B, 8, 863.
55. Pohl, U.W., Gumlich, H.-E. and Busse, W. (1984) Phys. Status Solidi B, 125, 773.
56. Busse, W., Gumlich, H.-E., Geoffroy, A. and Parrot, R. (1979) Phys. Status Solidi
B,93,591.
57. Parrot, R. and Blanchard C. (1972) Phys. Rev. B, 6, 3992.
96 Luminescence spectroscopy
58. Pohl, U.W., Ostermeier, A., Busse, W. and Gumlich, H.-E. (1990) Phys. Rev. B,
42,5751.
59. Busse, W., Gumlich, H.-E., Knaak, W. and Schulze, J. (1980) J. Phys. Soc. Jpn.,
49 (Suppl. A), 581.
60. Pohl, V.W. and Gumlich, H.-E. (1989) Phys. Rev. B, 40, 1194.
61. Benecke, C., Busse, W., Gumlich, H.-E. and Pohl, V.W. (1985) Phys. Status Solidi
B, 128, 701.
62. Benecke, C. and Gumlich, H.-E. (1992) In The Luminesence of Wide Band Gap
JI-Mn-VI Semimagnetic Semiconductors in Diluted Magnetic Systems (ed. M.
Jain), World Scientific, London.
63. Benecke, C., Busse, W., Gumlich, H.-E. and Moros, H.-1. (1987) Phys. Status
Solidi B, 142, 301.
64. Pohl, V.W. and Gumlich, H.-E. (1990) J. Cryst. Growth, 101, 512.
65. Hoffman, A., Heitz, R. and Broser, I. (1990) Phys. Rev. B, 41, 5806.
66. Litzenburger, B., Mahrt, 1., Pohl, U.W. and Gumlich, H.-E. (1993) To be
published.
67. Parrot, R., Diarra, M. and Boulanger, D. (1991) Personal communication.
68. Parrot, R., Naud, C., Gendron, F., Park, C. and Lemercier, M. (1986) J. Chem.
Phys., 85, 4932.
4
One-photon rare earth optical
transitions: recent theoretical
developments
G.W. Burdick and M.e. Downer
97
98 One-photon rare earth optical transitions
1
_ L [(i IE 1 Dln)(nIE2 ' Dlf) + (iIE2 ' Dln)(nIE1 'Dlf)] 12 (4.1)
n h(w n - wd h(w n - (2)
where E j ' D represent electric dipole operators, h(w n -Wj) denote the
energies of intermediate states n above the single photon energy hWj, and
the sum is taken over all intermediate states n of opposite parity to the
ground configuration 4fN.
In the one-photon theory of Judd [2] and Ofelt [3], an expression
analogous to formula (4.1) is used, called the 'forced' electric dipole, where
the second-order perturbation of the interaction Hamiltonian is replaced by
a time-independent perturbation involving the noncentrosymmetric part of
the crystal field, H CF ' That is, admixtures of opposite parity configurations
into the initial and final state wavefunctions allow the odd-parity electric
dipole to connect two states of the same parity. Thus, we obtain the
corresponding one-photon transition probability,
1
_ L [(iIHcFln)(nIE' Dlf) + (iIE' Dln)(nIHcFlf)] 12 (4.2)
n hWni hWn!
where hWni and hWn! denote the energies of intermediate states n above
initial state i and final state f respectively.
Actual computation of (4.2) for one-photon and (4.1) for two-photon
transitions is impractical because of the large number of intermediate levels
which must be summed over. For example, the lowest opposite-parity
excited configuration of Tb3+ (4f 7 5d) has 3006 SLJ levels [14]. In order to
get around this problem, the 'closure approximation' was independently
introduced by Judd [2] and Ofelt [3], in which each excited configuration
is regarded as being completely degenerate. This closure approximation is
reasonable for trivalent lanthanide ions owing to the large energy gap
between excited configuration levels and the ground and final state 4fN levels.
General theory of one-photon absorption 99
For example, in Pr3+, observed 4f2 levels lie within 23000 cm -1 of the
ground state [15] while energy levels of the lowest excited configuration,
4f5d, range from 61000cm- 1 to about nOOOcm- 1 [14,16]. Thus, the
energy denominator h(Wni -w) in formula (4.1) for two-photon excitation
ranges from 50000 to 70000 cm -1 . For one-photon excitation, the calcula-
tion is somewhat more complicated, since there are two different energy
denominators hWni and hWnJ. However, the closure approximation assumes
that the matrix elements in the two numerators of formula (4.2) are identical,
to within a phase factor, and we can replace the two denominators by a
single denominator using the identity
-1+ - =
1 2( 1 1 -w-2 -)
) ( 1+ (4.3)
hWni hWnJ hWni - hw/2 4 WniWnJ
(4.5)
n= 1 kodd q=-k
Since the M J components are generally mixed by the crystal field and
often not resolved, it is customary to define the total probability, or line
strength, of a transition as the summed intensity over all components M J
and Mj of the ground and final states. The line strength is anisotropic
owing to the coupling of the electric field to the crystal field. Since the
magnitudes of the coefficients BJk) for k odd are not known with any
degree of accuracy, this polarization dependence is difficult to predict.
Therefore, in order to simplify the comparison of theory and experiment,
this complicated line strength anisotropy is averaged out by using un-
polarized light or samples with randomly oriented rare earth ions, such as
an aqueous solution or powder. Thus, we sum over all polarizations q, and
equation (4.4) reduces to
p= L vT~(iIIU())lIf)2 (4.6)
),=2.4.6
This expression is the major result of the Judd-Ofelt theory, and is the usual
form used for analysis of data.
Reduced matrix elements U().) have been tabulated between all levels
within the fN configuration for LS coupling by Nielson and Koster [17], and
between the ground state and low-lying excited fN states for intermediate
coupling by Carnall et al. [18]. The T). however, are not determined from
first principles because of the difficulty of computing odd-parity crystal field
coefficients and radial integrals. Instead, these 'Judd-Ofelt parameters' are
left as phenomenological parameters, determined by fitting experimental
oscillator strengths. In so far as the closure approximation and other
simplifications made by the Judd-Ofelt theory are valid, the T). are constant
for all transitions of a particular rare earth ion in a given host crystal or
solution, allowing all observed transitions to be fitted by a single set of
parameters.
General theory of one-photon absorption 101
where the first sum represents the standard second-order term and the
second and third sums represent new third- and fourth-order terms respect-
ively. Two-photon excitation from a single beam has been assumed for
simplicity.
Setting H equal to the spin-orbit interaction Hso, Judd and Pooler [8]
showed that the third-order term explained experimental observations of
the 8S 7 / 2 --.6PJ transitions in Gd 3 +. Downer and Bivas [10] explained
observed two-photon absorption of the 8S7/2 --.6IJ transitions in Gd3+ by
use of third-order terms with H equal to the crystal-field interaction H eF ,
and fourth-order terms including both spin-orbit and crystal-field interac-
tions. These higher-order terms seem to be particularly important for
transitions which violate the selection rules of the lower-order theory.
Thus, the spin-orbit interaction is particularly important for transitions
which violate the I1S = selection rule, and the crystal-field interaction is
important when the I1L, 11)::;; 2 selection rule is violated. A complete
review of the importance of these higher-order terms is given by Downer
[5].
Other third-order contributions involving electron correlation have been
introduced [20, 21] in two-photon absorption calculations. In these calcu-
lations, only those terms which scale directly with the second-order V(2)
term have been calculated [22], since Smentek-Mielczarek and Hess [23]
have concluded that those third-order terms which are dependent on the
particular transition are small compared with the other terms examined,
and thus can be ignored. Since third-order terms involving electron corre-
lation do not affect any of the second-order selection rules, we would not
expect these terms to have the large selectively important effects found in
the spin-orbit and crystal-field interaction terms. Third-order electron
correlation has been shown to have a significant impact on absolute line
strengths, but its impact on relative transition probabilities is considered to
be negligible.
General theory of one-photon absorption 103
4.1.3 Third-order contributions to one-photon absorption
Each of the third-order contributions discussed above for two-photon
absorption has an analogous contribution in one-photon absorption.
Explicitly adding the third-order contribution, we rewrite expression (4.2)
as
_ I [(iIHcFln)(nIE.DIf) + (iIE.Dln)(nIHcFlf)]
1
n hWni hWnf
(4.10)
4.2.1 Derivation
The line strength for a one-photon intra-IN transition from initial state i to
final state f is
M~Mj 1(i'M'f)
- L{LI';/([i]IEDln)(nIHcFIU])
n
f)
where (i IMI is the standard magnetic dipole term, the first term in braces
gives the standard forced electric dipole contribution, and the last term in
Derivation and calculation of third-order terms 105
braces gives the new electric dipole contribution. The LJ kj denote energies of
intermediate state k above the initial state i or final state f Standard forced
electric dipole terms arise in the first order of time-dependent perturbation
theory. While the even-parity magnetic dipole operator directly links pure [N
states, the electric dipole term requires a first-order static perturbation of
initial and final states by the non centro symmetric crystal field. Consequent-
ly, this term is designated as the 'second-order' contribution. For spin-
forbidden transitions, static perturbations by the spin-orbit interaction must
also be included. This point has been emphasized by placing [i] and [f] in
brackets to denote that intermediate coupled wavefunctions must be used
for AS> 0 transitions.
The final term in this expression arises in the second order of time-
dependent perturbation theory, and introduces an extra energy denomina-
tor along with the new interaction H. In general, H represents the com-
plete Hamiltonian, but in this case we restrict our attention to the part
H = Hso, which is the most important contribution for spin-forbidden
transitions. The first-order odd-parity static crystal-field perturbation of
initial and final states is required in this term as well, and is shown
explicitly. Consequently, this term is designated as the 'third-order' contri-
bution. When H = Hso, initial and final states which differ in spin by one
unit can be linked directly even in the Russell-Saunders limit. In order to
emphasize this point, i and f are written without brackets to denote LS
coupled wavefunctions. Because of this direct linkage, it is possible for
third-order terms to rival or exceed lower-order terms in magnitude des-
pite the extra energy denominator.
In order to evaluate the second- and third-order terms for specific
transitions, we employ the same approximations as in the original theory [2,
3]. Since the intermediate states belong to opposite-parity configurations
which lie far above both i and f in energy, we follow the conventional
approximation of Judd and Ofelt that the LJ kj are constant for all k within a
particular configuration, allowing sums over each excited configuration to
be evaluated by the closure approximation. Individual operators then unite
into compound operators acting solely between levels of configuration [N.
After these compound operators are recoupled into a convenient spherical
tensor form, their matrix elements can be computed. Details of this operator
recoupling are given in the appendix of ref. 36 using the powerful techniques
of second quantization [37]. Resulting operators, acting between initial state
i and final state f of [N, with intermediate states belonging to [N - 11', are the
second-order term
2[1] [1'] ([
o
1 ~) Ei? L (~
0 kodd
k 1')
0 0
L [t]
teven
1/2 {1
[
~ ~} (E(1) B(k)(t) U(t)
(4.12)
106 One-photon rare earth optical transitions
2[/][1'](1
o
1 I')E,-?
0 0
L
kodd 0
(I k
1')(/{/}1/2
0 0 teven
L {1I k
It'}
+2[/][1'](1
o
1 I')E/-:?
0 0 kodd
L (I k
1') (dl'}1/2
0 0 0
(4.13)
where the notation {/}=/(/+ 1)(2/+ 1)/2, [/]=21+1, has been used for
brevity. The (I are spin-orbit coupling coefficients for nl electrons. Spheri-
cal Racah tensors [38] U(t) and W(lA)t result from the recoupling of
the electric dipole, crystal-field and spin-orbit operators. E(1) denotes the
electric field vector and B(k) is a tensor, the qth component of which
equals (-l)q B':q, where the B: are the ordinary crystal-field para-
meters.
For lanthanides, 1=3, and the only possible values of l' are 2 and 4,
corresponding to d- and g-type excited configurations respectively. The
k = 3, 5 components provide the only important contributions to the
noncentrosymmetric crystal field; the special case of k = 1 will be discussed
later. Substituting these values into expressions (4.12) and (4.13) and
evaluating [39] the required 3-j, 6-j and 9-j symbols, we obtain the following
explicit expressions for the second-order term,
(4.14)
4.2.2 Tensors
As illustrated by expressions (4.14H4.l7), in the standard second-order
theory, intra-fN transitions can be connected only through the second-,
fourth- and sixth-rank tensors U(2), U(4) and U(6). When third-order
terms are included, other even-rank U(/) W(l1)O and W(1.l.)1 terms are
added. Reduced matrix elements ([S L J ] II U(/) II [S' Lj,]) for intermediate
coupled ions throughout the trivalent lanthanide series have been com-
piled by Carnall et al. for ions in aqueous solution [40-43] and in LaF 3
[18]. Relevant third-order eLJl1 W(l1)OIIS'Lj,) and eLJl1 W(1.l.)/IIS'Lj,)
reduced matrix elements have been extensively tabulated by Burdick
[36].
Keeping track of appropriate multipliers, we can add third-order tensors
of the same rank to each other to form single second-, fourth-, and
sixth-rank operators Of), O~4), and O~6). When these composite third-
order terms are added to the corresponding second-order tensors and the
sums squared, we obtain combined second- and third-order operators which
are acted on by phenomenologically fitted parameters T(2), T(4), and T(6) to
yield total line strength predictions. Summing over M J and Mj, we obtain
108 One-photon rare earth optical transitions
the simplified final equation for the electric dipole contribution to the line
strength P:
(a) k= 1 terms
The k= 1 terms arising in expressions (4.12) and (4.13) are possible only if
linear crystal-field terms B(I) are allowed. Although it can be argued that
linear terms imply nonequilibrium between the rare earth ion and its
complex, Judd [44] has pointed out that electrons of the rare earth ion may
produce an electric field at the nucleus which exactly cancels that coming
from the ligands. Consequently, the nucleus remains stationary while the
electrons develop an asymmetric charge distribution which reflects the
presence of ql) terms in the crystal-field Hamiltonian. Thus, in crystalline
sites possessing symmetries which allow a linear crystal-field term (C n , Cm
and Cs), these k = 1 terms may playa significant role.
Allowing k= 1 in expressions (4.12) and (4.13), we get the following new
second-order terms,
+ 6(fs-)1/2 E;;/(E(I) B(I)(2) U(2)
(4.19)
+i(~)1/2 E;I(E(I) B(I)(2) U(2)
(4.20)
for respectively d and g orbital excited configurations. Notice that, except
for an extra factor of 2, these terms are mathematically equivalent to those
used for two-photon absorption. That is, if we replace the B(1) in expressions
(4.19) and (4.20) with E(1) and divide by 2, we obtain the second-order and
third-order spin-orbit operators necessary for two-photon transition calcu-
lations [5].
An examination of the 7Fo+-+500 transition of Eu 3+ and Sm2+ provides
the best indication of the importance of these linear crystal-field terms in
general, and the W(11)O term in particular. Since neither U(O) nor any other
second-order operator can provide a connection between these two levels,
this transition is allowed in second order only through J-J mixing, known
to be very weak for lanthanide ions. The third-order W(l1)O term, in
contrast, provides a direct linkage between two states with J = O. In most
environments, this transition is absent or extremely weak, which can be
adequately explained by means of J-J mixing. However, strong 7FO+-+500
transition lines have been observed in Eu 3+ - and Sm2+ -doped crystals with
ionic substitution sites possessing symmetries allowing a linear crystal-
field term. To give a few examples, Kiss and Weakliem [45] and Alam and
di Bartolo [46] observed that the 50 0-+ 7FO emission line in Sm2+ : BaCIF
(site symmetry C 4v ) was comparable in intensity with the neighbouring
allowed 500-+ 7F 2 line. Blasse and Bril [47] and Nieuwpoort and Blasse
[48] studied the intensity of this emission line in Eu3+ in a variety of
crystals. They found that, in Eu3+ :Sr2 Ti0 4 (C 4v site symmetry) and
Eu 3+ :Ba3Gd2 WO g (C 3v site symmetry), 50 0-+ 7FO had 50% of the intensity
of 5DO -+ 7F 2, and showed strong intensities in several other crystals with
C2v , C 4v and C. substitution site symmetries. In their phenomenological
analysis of energy levels in Sm2+: SrCIF, Grenet et al. [49] invoked a
dominant D[(ll)OA1] term, analogous to the W(l1)O term discussed above,
which was not significantly altered by inclusion of J-J mixing. A more
recent study of the fluorescence of Eu3+ -doped oxyapatites [50] found the
500-+ 7F 0 transition to dominate. In these selected cases, J-J mixing
appears to be inadequate to explain either the magnitude of the transition
intensity or hypersensitivity to the crystalline environment. They seem to
provide strong evidence for the existence and importance of linear crystal-
field terms. Preliminary calculations [51] have been done in an attempt to
quantify magnitudes of the linear crystal field necessary to explain these
observations.
110 One-photon rare earth optical transitions
where U~).) are the second-order g orbital terms, scaled so that U().) in
expression (4.14) can be replaced by the sum U~).) + U~).). Calculation is more
complicated for fourth-rank terms, since the ratio between crystal-field
terms (E(1) B(3)(4) and (E(1) B(5)r) must be known. Considering three cases,
(a) the first term dominates, (b) the two terms are equal, and (c) the second
term dominates, we obtain respectively
3 (4flr3In'g)
11 (4flr3In'd)
U(4) = Edc (4flrln'g) (4)X 15 (4flr 3In'g)+(4flr 5In'g) (4.23)
g Egc (4flrln'd) U d ~ 14 (4flr 3In'd)+(4flr 5In'd)
(4f Ir 5 1n'g)
6 (4f1 r5In'd)
Since the magnitudes of these crystal-field terms will not change for different
transitions from the same ion site, the effect of these g orbital contributions
can be completely masked by an appropriate scaling of the T().).
Radial integrals (4f11In'l') are included in the T().) parameters and thus
normally are not calculated. However, from the Raman asymmetry data of
Becker [55], the value of (Edr/Egc)(4flrln'g)/(4flrln'd) can be determined
rigorously, and is found to be 0.91 for ErP0 4 and 1.09 for TmP0 4 . Since
these ratios are close to unity, it is reasonable to make the further
assumption that (4fllln'd)=(4fllln'g) for k=3,5. Thus, setting all ratios of
energy denominators and radial integrals in expressions (4.21~(4.23) equal
to unity, we find that inclusion of these g orbital terms necessitates a 92%
decrease in the parameter 1(2), anywhere from a 38% to 98% decrease in the
parameter 1(4), and a 14% decrease in the parameter 1(6)'
Parameter revisions of these dimensions will strongly affect any calcula-
tions which attempt to relate the magnitude of the Judd-Ofelt parameters to
the underlying structure of the crystal field, such as those of Krupke [54]
and Alfrey et al. [56]. Recent calculations of Smentek-Mielczarek [57] tend
to discount the importance of g orbital contributions in explaining Raman
asymmetry data. However, as the above calculations show, even small g
orbital contributions will yield significant modification of the Judd-Ofelt
parameters.
112 One-photon rare earth optical transitions
If these odd-rank tensors are important, we would have to add two other
parameters T(3) and 1(5) into our calculations, increasing the number of
fitted parameters from three to five. We have found, however, that these
odd-rank terms are always smaller than the corresponding even-rank terms,
typically by several orders of magnitude. Thus, for these odd-rank terms to
be at all important, magnitudes of the fitted parameters T(3) and T(5) would
have to be very much larger than of the parameters T(2), T(4) and T(6). That
is, crystal-field terms in (E(1) B(3)(3) or (E(1) B(5)<5) would have to be more
than a hundred times the magnitude of the corresponding (E(1)B(3)(2),
(E(1) B(3)(4), (E(1) B(5)(4) and (E(1) B(5)(6) terms, which is extremely unlikely.
4.2.4 Conclusions
These third-order computations can determine unambiguously just how
important the third-order terms are relative to conventional second-order
terms. Since the ratios of third- to second-order contributions do not depend
on the radial integrals or fitted parameters 1().), their relative importance or
unimportance for specific transitions is an unambiguous output of the
calculation.
Through appropriate modification of fitted parameters T().), line strength
calculations which include third-order terms can be refitted to experiment,
allowing us to determine whether inclusion of third-order terms can better
explain the experimental data than can second-order terms alone.
(a) Nd3+
Table 4.1 gives the second- and third-order line strength contributions for
the various transitions from the ground state 419/2 for Nd3+ in units of
P x 108 , where P is the probability of transition. The first column specifies
the particular transition. The second and third columns give second- and
third-order contributions respectively, and the fourth column gives the
percentage ratio between the third-order contribution,
(4.26)
where the 1(;.) are given in the bottom row of this table for the fitting when
both second- and third-order terms are included, i is the ground state 419/2,
and the final state f is listed in the first column. The last three columns of
this table compare total experimental line strengths, in the fifth column, with
line strengths calculated using only second-order terms and including
third-order terms, in the sixth and seventh columns respectively. The
seventh column is a summation of the values presented in the second and
third columns. Transitions which are not experimentally resolved have been
grouped together in the last three columns.
Since the third-order contribution (4.25) includes cross-terms between
second- and third-order matrix elements, as well as squares of third-order
matrix elements, the third-order contribution can actually be negative, as is
the case for a number of transitions listed in this table. Third-order
contributions for spin-allowed quartet to quartet transitions presented in
this table have systematically small positive values ranging from 5% to 13%
114 One-photon rare earth optical transitions
Table 4.1 Calculated second- and third-order contributions together with experi-
mental and calculated line strengths for Nd3+ transitions from ground state 41 9/2 ,
given in P x 108
(b) Sm 3 +
Second- and third-order electric dipole and magnetic dipole contributions
for transitions of Sm 3+ are tabulated in Table 4.2. The first column gives the
transition from the ground state 6H s/2. The second column of this table
presents the calculated magnetic dipole contribution to the oscillator
strengths in units of P x 108 As shown in this table, only the 6Hs/2-+4GS/2,
4F S/2 and 4H7/2 transitions have a significant magnetic dipole contribution,
with a zero or negligible contribution in all other observed transitions. The
third and fourth columns present the second- and third-order electric dipole
contributions to the transition oscillator strengths respectively, also in units
of P x 108 , and the fifth column gives the ratio of the third- to second-order
electric dipole contributions.
Third-order contributions to the spin-allowed sextet to sextet transitions
are all consistently between 5% and 12% of the corresponding second-order
contributions. In contrast, third-order contributions to the spin-forbidden
sextet to quartet transitions have some of the largest relative third-order
contributions of any trivalent lanthanide, often being greater than 50% of
the corresponding second-order contributions. For three transitions,
6Hs/2-+4F3/2, 4H9/2 and 4Hll/2' the third-order contribution is actually
greater than the second-order contribution, with respective contributions of
127%, 160% and 121 %. Experimental observations of the latter two
transitions are hidden by the close-by dominant 6H s/2-+ 4D7(2 transition.
However, the 6H s/2-+ 4F 3/2 transition is clearly resolved experimentally. The
large third-order contribution results in significant improvement in the
fitting of this transition, as illustrated in the last three columns of Table 4.2.
The experimental oscillator strength for this transition is 0.90 x 10- 8 , while
Table 4.2 Calculated magnetic dipole and second- and third-order electric dipole contributions
together with experimental and calculated line strengths for Sm 3 + transitions from ground state
6H s/2 , given in P x 10 8
6F 1 / 2 22 3 11.5% 26 26 26
6H 1S / 2 1.5 0.1 9.0%
6F 3 / 2 89 8 9.3% 97 99 98
6F s/2 166 15 8.9% 170 184 181
6F 7 / 2 280 21 7.5% 274 300 301
6F 9 / 2 186 12 6.3% 197 198 198
6F 11/2 30 2 5.6% 31 32 32
4G S / 2 2.3 1.1 0.5 50.4% 3.0 3.5 3.9
4F3/2 0.10 0.13 126.8% 0.90 0.10 0.23
4G 7 / 2 5.8 -1.1 -19.5% 5.2 6.2 4.7
41 9/2 3.2 1.3 42.0% 153 54 67
4M 1S / 2 35 9 24.7%
4111/2 13 6 50.7%
41 13 / 2 32 11 34.5% 68 34 44
4FS/2 0.6 0.5 -0.3 -56.7% 3.0 1.6 0.8
4M17/2 6.5 1.8 27.0% 39 13 13
4G 9/2 5.4 -1.1 -19.8%
41 1S12 0.3 0.1 38.7%
6. 4P S12 51 6 11.5% 56 57
4L 13/2 29 5 17.4% 43 31 34
4F7/2 2.9 -0.7 -23.8% 384 380 381
6P 3/2 339 34 10.1%
4Kll/2 4.6 1.0 21.0%
4L 1S12 8.5 2.7 32.2% 42 11 13
4G l1/2 1.6 0.3 18.3%
4Dl/2 0.05 -0.05 -96.7% 106 126 126
6P 7/2 114 8 6.9%
4L17/2 0.3 0.1 26.2%
4K 13/2 2.7 0.1 3.2%
4F9/2 0.9 0.1 14.6%
4D3/2 56 -2 -4.5% 123 103 90
(4D,6P)SI2 38 -1 -3.6%
4H7/2 0.9 2.0 1.3 63.2% 6.0 2.9 4.2
4H9/2 1.5 2.4 159.9% 80 67 68
4D7/2 61 0 0.0%
4Hll/2 1.0 1.3 121.3%
T(2) T(4) T(6)
(c) Eu3+
Second- and third-order electric dipole contributions to transitions from the
ground state 7F 0 of Eu 3 + are given in Table 4.3 in the same format as Table
4.1. As shown in Table 4.3, third-order terms in ground state transitions of
Table 4.3 Calculated second- and third-order contributions together with experi-
mental and calculated line strengths for Eu 3 + transitions from ground state 7FO'
given in P x 10 8
Eu3+ are quite significant, with average contributions greater than one-third
the magnitude, either positive or negative, of the corresponding second-
order terms. Large negative third-order contributions cancel out 54% and
39% of corresponding second order terms for the 7FO-+5D2 and 514
transitions respectively. Also, large third-order contributions add 77%, 61 %
and 53% to respective second-order terms for the 7FO-+5G4, 5G 6 and 5F4
transitions.
The first transition given in Table 4.3 is the 7F 0 -+ 5D 2 transition. Since
this is the only transition used in fitting the empirical parameter T(2), an
exact fit between calculation and experiment can be guaranteed. Third-order
results for this transition are interesting because the third-order term is more
than 50% of the magnitude of the second-order term and of opposite sign.
Thus, the Judd-Ofelt parameter 1(2) must be more than doubled from its
value of 1.88 x 10 - 9 cm to 4.08 x 10 - 9 cm, as shown at the bottom of this
table.
(d) Gd 3 +
Second- and third-order electric dipole and magnetic dipole contributions
for ground state transitions of Gd3+ are tabulated in Table 4.4 with the
same format as for Table 4.2. The first three transitions given in this table
are for the 8S 7/2-+6PJ transitions. For the 8S 7/2-+6P 7/2, 6P 5/2 transitions,
third-order terms are 60% and 113 % of the magnitude of the corresponding
second-order terms. However, third-order contributions have little effect on
predicted intensities for these two lines since the magnetic dipole term
dominates the line strengths. The 8S7/2 -+ 6p 3/2 transition, by contrast, has
no magnetic dipole term, so the large third-order term dominates. The
oscillator strength measured by Carnall et al. [19] for this transition is an
order of magnitude greater than predicted by second-order theory. How-
ever, inclusion of the large third-order term, 948% of the magnitude of the
second-order contribution, accounts for the strength of this previously
unexplained transition.
Relative to the corresponding second-order contributions, overall third-
order contributions for Gd3+ transitions are the greatest of any lanthanide
ion, with contribution percentages ranging from 60% to 948% for
8S 7/2-+6PJ transitions, as discussed above. This fact could have been
expected for two reasons. First, all observed transitions for Gd3+, along
with Eu3+ and Tb3+, are spin forbidden in LS coupling, a fact which
accounts for the reduction of the second-order matrix elements for these
three ions compared to the other trivalent lanthanides. The second reason
comes from a unique characteristic of half-filled shells [60], namely that all
e
diagonal reduced matrix elements s + 1 LII U(k) 11 2S + 1 L) vanish. This latter
fact tends to reduce the value of intermediate coupled reduced matrix
e
elements ([2S + 1 L] II U(k) II s ' + 1 L']) connecting states of differing spin, since
Table 4.4 Calculated magnetic dipole and second- and third-order electric dipole contributions
together with experimental and calculated line strengths for Gd3+ transitions from ground state
BS 7/2, given in P x lOB
the initial and final states will share common LS coupled admixtures,
making second-order contributions unusually small.
(J (J
T(2) T(4) T(6) T(2) T(4) T(6)
(x JO- 9 em) (P x J08) (x JO- 9 em) (P x J08)
where separate energy denominators Elli and En' are calculated explicitly in
C(v). Since this scaling factor reduces the range of energy denominators
which must be considered degenerate, it should decrease the amount of error
closure introduces into the calculation.
For observed transitions across the lanthanide series, C(v) is usually
within 10% of unity, and almost always within 20%. One significant
exception is the Nd3+ ion, where scaling factors greater than 40% are found
for selected observed transitions. When this scaling factor is introduced into
the fitting calculations for Nd3+, the Judd-Ofelt parameters must be
significantly revised in the second-order fitting as well as the fit with
third-order terms included. However, with this revision, standard deviations
of fits to experiment are found to be 4% worse than the corresponding
fittings done without scaling factors [36].
We conclude that, for most trivalent ions, the approximation of closure
over the ground configuration does not introduce significant error into the
References 129
calculations. In the case of Nd 3 +, where the closure approximation would
seem to introduce the greatest error, removing closure over the ground
configuration actually increases the error between fit and experiment,
suggesting that there might be other, as-yet unrecognized physics which is
included in the closure approximation.
Since removal of closure in the ground configuration does not have a
significant effect on fitted line strengths, we can suppose that closure of the
excited configurations will not produce a significant calculational error
either. However, explicit line strength calculations which do not invoke
closure of either the ground or the excited configurations will be necessary
before this truly can be known.
ACKNOWLEDGEMENTS
The authors wish to thank J.e. Gacon and M. Kibler for helpful discussion
during the writing of this chapter. Financial support by the Robert A. Welch
Foundation (Grant F-I038) and the National Science Foundation (Grant
DMR-8858388) is gratefully acknowledged. We also acknowledge the support
of the Ministere de l'Education Nationale (France) for one author (G.W.B.).
REFERENCES
1. van Vleck, J.H. (1937) J. Phys. Chem., 41, 67-80.
2. Judd, B.R. (1962) Phys. Rev., 127, 750-761.
3. Ofelt, G.S. (1962) J. Chem. Phys., 37, 511-520.
4. Peacock, R.D. (1975) Struct. Bonding, 22, 83-122.
5. Downer, M.e. (1989) In Laser Spectroscopy of Solids II (ed. W.M. Yen),
Springer, Berlin, pp. 29-75.
6. Axe, J.D., Jr. (1964) Phys. Rev., 136, A42-A45.
7. Dagenais, M., Downer, M., Neumann, R. and Bloembergen, N. (1981) Phys. Rev.
Lett., 46, 561-565.
8. Judd, B.R. and Pooler, D.R. (1982) J. Phys. C, 15, 591-598.
9. Downer, M.C., Bivas, A. and Bloembergen, N. (1982) Opt. Commun., 41,335-340.
10. Downer, M.e. and Bivas, A. (1983) Phys. Rev. B, 28, 3677-3696.
11. Downer, M.e., Cordero-Montalvo, e.D. and Crosswhite, H. (1983) Phys. Rev. B,
28, 4931-4943.
12. Kitai, A.H. (1992) In Visible Luminescence - Solid State Materials & Applications
(ed. A.H. Kitai), Chapman & Hall, London, Chapter 1.
13. Goppert-Mayer, M. (1931) Ann. Phyz. (Leipzig), 9, 273-294.
14. Dieke, G.H. and Crosswhite, H.M. (1963) Appl. Opt., 2, 675-686.
15. Carnall, W.T., Fields, P.R. and Wybourne, B.G. (1965) J. Chem. Phys., 42,
3797-3806.
16. Brewer, L. (1971) J. Opt. Soc. Am., 61, 1666-1682.
17. Nielson, e.W. and Koster, G.F. (1964) Spectroscopic Coefficients for the pN, dN
and fN Configurations, MIT Press, Cambridge, MA.
18. Camall, W.T., Crosswhite, H. and Crosswhite, H.M. (1977) Energy Level
Structure and Transition Probabilities of Trivalent Lanthanides in LaF 3, Argonne
National Laboratory Report.
130 One-photon rare earth optical transitions
19. Carnall, W.T., Fields, P.R and Rajnak, K. (1968) J. Chern. Phys., 49, 4412-4423.
20. Jankowski, K. and Smentek-Mielczarek, L. (1987) Mol. Phys., 60,1211-1219.
21. Smentek-Mielczarek, L. and Hess, B.A., Jr. (1987) Phys. Rev. B,36, 1811-1817.
22. Smentek-Mielczarek, L. and Hess, B.A., Jr. (1989) J. Chern. Phys., 90, 1390-1395.
23. Smentek-Mielczarek, L. and Hess, B.A., Jr. (1987) J. Chern. Phys., 87, 3509-3512.
24. Smentek-Mielczarek, L. and Hess, B.A., Jr. (1988) J. Chern. Phys., 89,703-710.
25. Smentek-Mielczarek, L. and Hess, B.A., Jr. (1989) Phys. Rev. B, 39, 517-522.
26. Jorgensen, e.K. and Judd, B.R. (1964) Mol. Phys., 8, 281-290.
27. Mason, S.F., Peacock, R.D. and Stewart, B. (1975) Mol. Phys., 30, 1829-1841.
28. Judd, B.R (1979) J. Chern. Phys., 70, 4830-4833.
29. Reid, M.F. and Richardson, F.S. (1983) J. Chern. Phys., 79, 5735-5742.
30. Reid, M.F., Dallara, J.J. and Richardson, F.S. (1983) J. Chern. Phys., 79,
5743-5751.
31. Reid, M.F. (1987) J. Chern. Phys., 87, 6388-6392.
32. Poon, U.M. and Newman, DJ. (1984) J. Phys. C., 17,4319-4325.
33. Henrie, D.E., Fellows, R.L. and Choppin, G.R. (1976) Coord. Chern. Rev., 18,
199-224.
34. Peacock, RD. (1977) Mol. Phys., 33, 1239-1246.
35. Reid, M.F. and Ng, B. (1989) Mol. Phys., 67, 407-415.
36. Burdick, G.W. (1991) Ph.D. Thesis, University of Texas, Austin.
37. Judd, B.R (1967) Second Quantization and Atornic Spectroscopy, Johns Hopkins,
Baltimore, MD.
38. Judd, B.R (1960) Operator Techniques in Atornic Spectroscopy, McGraw-Hill,
New York.
39. Rotenberg, M., Bivins, R, Metropolis, N. and Wooton, J.K., Jr. (1959) The 3-j
and 6-j syrnbols, Technology Press, Cambridge, MA.
40. Carnall, W.T., Fields, P.R and Rajnak, K. (1968) J. Chern. Phys., 49, 4424-4442.
41. Carnall, W.T., Fields, P.R. and Rajnak, K. (1968) J. Chern. Phys., 49, 4443-4446.
42. Carnall, W.T., Fields, P.R and Rajnak, K. (1968) J. Chern. Phys. 49, 4447-4449.
43. Carnall, W.T., Fields, P.R and Rajnak, K. (1968) J. Chern. Phys., 49, 4450-4455.
44. Judd, B.R. (1966) J. Chern. Phys., 44, 839-840.
45. Kiss, ZJ. and Weakliem, H.A. (1965) Phys. Rev. Lett., 15, 457-460.
46. Alam, A.S.M.M. and di Bartolo, B. (1967) J. Chern. Phys., 47, 3790-3801.
47. Blasse, G. and Bri!, A. (1966) Philips Res. Rep., 21, 368-378.
48. Nieuwpoort, W.e. and Blasse, G. (1966) Solid State Cornrnun., 4, 227-229.
49. Grenet, G., Kibler, M., Gros, A., Souillat, J.e. and Gacon, 1.e. (1980) Phys. Rev.
B, 22, 5052-5067.
50. Piriou, B., Fahmi, D., Dexpert-Ghys, 1., Taitai, A. and Lacout, 1.L. (1987) J.
Lurnin., 39, 97-103.
51. Burdick, G.W. and Downer, M.e. (1991) Eur. J. Solid State Inorg. Chern., 28,
217-220.
52. Holsa,1. and Porcher, P. (1985) J. Less-Cornrnon Met., 112, 121-125.
53. Becker, P.e., Edelstein, N., Judd, B.R, Leavitt, Re. and Lister, G.M.S. (1985) J.
Phys. C, 18, Ll063-Ll066.
54. Krupke, W.F. (1966) Phys. Rev., 145, 325-337.
55. Becker, P.e. (1986) Ph.D. Thesis, Lawrence Berkeley Laboratory, University of
California, Berkeley.
56. Alfrey, A.J., Stafsudd, O.M., Dunn, B. and Yang, D.L. (1988) J. Chern. Phys., 88,
707-716.
57. Smentek-Mielczarek, L. (1989) Phys. Rev. B, 40, 6499-6504.
58. Downer, M.e., Burdick, G.W. and Sardar, D.K. (1988) J. Chern. Phys., 89,
1787-1797.
References 131
59. Burdick, G.W., Downer, M.e. and Sardar, D.K. (1989) J. Chem. Phys., 91,
1511-1520.
60. Judd, B.R. (1988) In Handbook on the Physics and Chemistry of Rare Earths, Vol.
11, (eds K.A. Gschneidner, Jr, and L. Eyring), North-Holland, New York, pp.
81-195.
61. Gacon, le., Jacquier, B., Marcerou, J.F., Bouazaoui, M. and Kibler, M. (1990) J.
Lumin., 45, 162-164.
62. Kibler, M. and Gacon, J.e. (1989) Croat. Chem. Acta, 62, 783-797.
63. Kibler, M. (1991) In Symmetry and Structural Properties of Condensed Matter
(eds W. Florek, T. Lulek and M. Mucha), World Scientific, Singapore, pp.
237-250.
64. Kibler, M. (1992) In Excited States of Transition Elements (eds W. Strek, W.
Ryba-Romanowski and B. Jezowska-Trzebiatowska), World Scientific, Sin-
gapore, pp. 139-148.
65. Sztucki, J., Daoud, M. and Kibler, M. (1992) Phys. Rev. B, 45, 2023-2028.
66. Gacon, le., Marcerou, J.F., Bouazaoui, M., Jacquier, B. and Kibler, M. (1989)
Phys. Rev. B, 40, 2070-2075.
67. Gacon, le., Bouazaoui, M., Jacquier, B., Kibler, M., Boatner, L.A. and Ab-
raham, M.M. (1991) Eur. J. Solid State Inorg. Chem., 28, 113-116.
5
Thin film electroluminescence
G.O. Muller
5.1.1 History
As easily imaginable, phosphorescence or afterglow was the first observation
of luminescence. People used to observe light from objects in nature as some
kind of re-emission or reflection in daylight or from incandescent or high
temperature sources, emitting thermal radiation. A cold object emitting in
the dark was completely different from this usual experience and of course
found great attention as early as 1603, when Vincenzo Cascariolo detected
the Bologna stones [1].
It was not until Destriau [2J reported on the influences of electric fields
on a luminescing material that electroluminescence (EL) was born. Later,
powder electroluminescence was something very common as a research and
development object over the years. The advent of semiconductor physics,
comprising single crystal growth, doping and p--n junctions brought along
the search for injection (electro-) luminescence. The concept of p--n junc-
tions, emitting minority carriers into the adjacent region of a crystal with a
different type of conductivity, where they recombine radiatively (in part), can
be modified. It has to be changed for materials which are hard to dope
ambivalently. Rather sophisticated structures have been invented and inves-
tigated to overcome these difficulties. The blue diode and the blue diode
laser, however, emerged from concentrated efforts to produce p-type con-
ductivity in ZnSe [3].
Thin film electroluminescence, after becoming successful, exhibited many
roots. Clearly identifiable is the effort at Bell Laboratories (see for instance
ref. 4, and papers cited therein) which was focused on phosphors deposited
in thin film form on anodized Ta. It had been anodized to prevent electric
breakdown when driven with voltages which would give rise to it otherwise.
133
134 Thin film electro luminescence
The paper by Chen and Krupka [4] marked the end of these activities at
that time, as the work was strongly believed not to be useful. The first
activities on thin film electroluminescence by VIa senko and Jaremko [5]
should also be mentioned. Some years later, Inoguchi and coworkers [6, 7]
gave thin film electroluminescence, now referred to in this way, a comeback.
They obtained excellent properties, and, what was the most impressive
feature, very long lifetimes. It was some years until the first results published
by the IBM group [8] convinced a broader scientific and engineering
community that thin film EL could be a solution for high field electro-
luminescence in general.
Less than 10 years after Inoguchi's first paper, his company, Sharp, went
on to industrial production of monochrome, yellow EL displays of more
than postcard size and a complexity of 256 by 512 picture elements (pixels).
In 1991 both at Sharp (Japan) and Planar (USA) large-scale production of
up to 19 inch diagonal displays with 1024 x 768 pixels of high contrast and
'crisp' aesthetic appearance was established, and the race for multicolour
and even full-colour displays continues.
Metal-_ _ _w::'!~
Insulator
inner shell transition) gives rise to the emission. Clearly, the applica-
tion of voltage to any electroluminescent device must cause just this,
either generating electron-hole pairs (if one kind of carrier is present,
only minorities are needed), or exciting luminescence centres. As we
discuss EL, the excitation energy comes from the applied field and dis-
sipative current must flow. Excluding in this chapter the 'trick' of injec-
tion of minority carriers, the carriers forming the dissipative current
must be energetic enough to excite either the e-h pair or the centre. This
excitation is the Coulombic interaction of the energetic carrier with the
one to be excited, which we shall call 'Coulomb impact' or simply
'impact'.
The usual electrical conduction in semiconductors (or metals) does not
supply any kinetic energy comparable with the thermal energy kT to the
carriers - they stay 'cool'. In high fields, however, when the energy gained
from the field between two interactions with cooling media - phonons,
defects or other elementary excitations, or in general the 'lattice' - exceeds
the energy dissipated in this interaction, the carrier becomes 'hot'. Obviously
only hot carriers can excite (noninjection) EL.
To be more specific, as we are referring to visible emission mainly, the
excitation energy (in general at least as high as the energy of the emitted
photon) is of the order of 2.5 eV. So, in order to observe EL, there must be
carriers with about 2.5 eV of kinetic energy. That is far beyond the regime
usually addressed, if in semiconductor physics 'hot electrons' are identified.
Nobody until now has mentioned the possibility of electron energies of
25000 K in a material at room temperature.
However, as the balance between energy gain from the field and energy
loss to the lattice is a statistical process, it might happen that some of the
electrons are 'lucky' in gaining much higher momenta in the direction of the
field than the average, and the energy distribution is no longer a thermal
one, not even to a temperature higher than the lattice temperature. Run-
away electrons occur.
To rely on runaways, occurring eventually with a concentration of 10- 6 ,
would limit the efficiency to less than 10- 6 , as all the other cool carriers add
to the energy input the same amount, but nothing to the output.
Situations in which the fraction of hot electrons dominates are known,
however; for example, the avalanche breakdown regime. There electrons (or
holes) reach high enough energies to multiply, usually more than sufficient
to excite the dopants. So one possibility for efficient EL is operation under
avalanche conditions. Long life, however, does not seem to be guaranteed
for a device breaking down electrically.
One of the above-mentioned important properties, the threshold be-
haviour, however, would be realized in this avalanching device. As avalanch-
ing itself is a threshold process, a steep B(V) characteristic would be
assured.
Basics of thin film electro luminescence 137
5.2 BASICS OF THIN FILM ELECTROLUMINESCENCE
As outlined above, but understood only after the first devices were develop-
ed and waiting for production, noninjection or high field EL only has the
chance to be efficient if the field is high enough to generate plenty of hot
electrons.
Even an avalanching structure, however, does not necessarily luminesce
efficiently; suitable dopings have to be found which can readily be incorpor-
ated in such a way as to conserve their radiative efficiency.
The following section will therefore be split into these two ways of looking
at electroluminescence: electroluminescence and electro luminescence.
5
ZnS
Insul. Semiconductor I Insul
S'S
CaS
Tunnel
em ission
excita tion
Mn 2
Tb3 +
Ce3+
Eu 2
Fig. 5.2 The 'simple' model of a TFEL device, showing the basic processes.
ZnS:: Mn
.,
OJ
~
o
H.
>
Fig. 5.3 Time behaviour of dissipative 'current' and 'brightness' under the action of
the applied 'voltage'.
To give a last important formula, the energy dissipated in the S film in one
pulse is related to the (outer) threshold voltage U th by
I
s
~M
Fig. 5.4 The so-called back-to-back Zener model of the TFELD.
140 Thin film electro luminescence
160 400
v
80 200
1.6 ..
F
0.8
o Time, f.lS
Fig. 5.5 (a) Simultaneously measured waveforms of applied voltage V and dissi-
pative current density D; (b) voltage across the active film, expressed in normalized
form as average field F.
Basics of thin film electro luminescence 141
160
.A
80
a 1
Field, mV/cm 2
Fig. 5.6 A plot of the dissipative current D (from Fig. 5.5(a)) versus the average
field F gives the (normalized) current-voltage characteristic of the active film D(F)
[10,11].
Until now in this section luminescence or light emission has not been
mentioned. As it is the goal of the whole chapter, we shall concentrate on
this question separately (see section 5.2.l(c)). As outlined above, excitation of
centres can be by Coulomb impact or by recombination of the e-h pairs
generated during avalanching. The latter could proceed either directly, or
via excitons (free or bound), via trapping at suitable defects or impurities;
also, in any of those processes a photon could be generated or the energy
quantum-mechanically transferred to another transition, which decays radi-
atively or nonradiatively.
4
I
,,
N(E), a.u.
J\'I
ZnSe
ZnS
/\\
2 ~
~
\
'. \
~
a 0
./- I -R
/
/
I
~
I
a 2 4 6 8
E, eV
10 12 14
Fig. 5.7 Energy distributions of the electrons emitted into vacuum through the
thinnest gold film which could be made at present; the inner EEDs evolve by adding
the differences between the work function of gold and the electron affinity of the
active material ZnS or ZnSe. Note that the a.u. scales are not the same for the two
materials. (From ref. 20.)
is well-known from powder EL (e.g. see [22]). The mechanism is simply the
field ionization of the activators, which gives rise to another chance for
nonradiative recombination channels to catch the minority carrier.
So in all doping considerations for EL one has to be careful with the 'field
stability' of the wanted transition.
To begin with a very new, negative experience, we shall discuss one recent
attempt to use the well-behaving ZnS, if doped with Mn, for blue emission
by doping Te into it. It has been known for a long time that ZnS:Te emits in
the UV if lightly doped, in the blue if the Te concentration increases, and
eventually shifts to the green, if the concentration is raised further. Te,
iso-electronic to S, binds by short-range forces to excitons, which decay
highly radiatively, especially if bound to a Te pair by about 600 meV (the
blue line). As an exciton is a neutral entity, the risk of field ionization from a
centre binding it at 0.6 eV should be much less than, for example, for the
hole on the Ag acceptor.
After overcoming the problem of incorporating Te into ZnS properly in a
low temperature preparation (film deposition) process, the result was devas-
tating: well-photoluminescing samples showed no EL in the blue.
Also, looking for photoluminescence under applied voltages (photo-
electroluminescence (PEL)), even below the threshold for tunnel emission,
quenching became obvious (Fig. 5.8). Whether quenched directly by the field
or by impact of energetic carriers would have to be determined, but the
result is certain: no blue EL from ZnS:Te [23].
15000
photo- (PL)/electroluminescence (EL)
1-PL
10000
:J 2-EL
co 3,4-PEL
i-
'iii
c
Q)
E 5000
ZnS:Mn
1.6
O.B
o Time,l..Is 100
Fig. 5.9 Brightness B is roughly the time integral of the dissipative current D in a
low doped ZnS:Mn sample, as the decay time of luminescence is long compared
with the exciting current flow.
146 Thin film electroluminescence
0.12 0.3.
- - CaS: Eu
0.04
~t i;"
::
---SrS: Ce
. o. 1
1/
.':
....... ::. __ ..
a Time, IJS 100
Fig. 5.10 In contrast to ZnS:Mn, shown in Fig. 5.9, in the AES doped with 5d-+4f
lanthanides, exhibiting short decays, brightness B follows closely the exciting
current D.
Brightness, cd/m 2
3 r-------~-----------------,
10
ZnS:Mn 150
50Hz
100
50
230 260
Vpeak' V
Fig. 5.11 Brightness versus the peak voltage Vpeak of a bipolar pulse voltage of
50 Hz, on logarithmic and linear scales.
(b) Dopants
The selection of dopants has been discussed before in section 5.2.l(c). It is
mainly the field quenching problem that governs decision making. However,
the solubility of the dopant in the host is of some importance too.
Consider the case of rare earths in ZnS. The solubility of most of them is in
the parts per million range. In low temperature deposition processes - electron
beam evaporation and d. sputtering have been mostly used - it is of course
possible to incorporate in some way Ln3+ ions and/or LnCl 3 molecules into
the film of ZnS, although whether into the lattice of the growing crystal grains
or to what other sites is questionable. However, even if we assume that the
incorporation on cation sites at least to some degree is achieved, the question
arises as to on what time scale the system, which is not in thermodynamic
equilibrium (TDE), travels under the action of energy input towards it. Also,
TDE is a phase mixture of ZnS doped to the solubility limit with the rare earth
and other phases containing more Ln, e.g. LnZZnS4.
The very big problem for TFEL, aiming at other or all colours, is the rather
limited choice of dopants able to dissolve in one of the known suitable hosts
and emitting a wanted colour.
The alkaline earth chalcogenides came into the play just from these
considerations - their solubility for lanthanides is orders of magnitude higher.
5.3.1 Efficiency
A check of the understanding of the basic physics - and the most important
issue for any application - is the modelling of efficiency. Efficiency will
Values and margins 149
In the case of the most useful materials the refractive index is about ns = 2.3,
and n out =0.1. Light guiding in the highly refracting material is responsible
for this dramatic loss of usefulness. (Interference by the necessary (insulator)
or additional films may change the angular distribution (differently for
150 Thin film electro luminescence
different wavelengths), but cannot improve the net outcoupling.) With these
results inserted, we have
n=O.lsNnlum
To measure nl um for ZnS:Mn is no problem. The radiative lifetime trad is
well known [28]. Also, from a comparison of the decay of brightness after a
sufficiently short excitation, b(t), with the purely radiative decay time trad ,
the efficiency nl um follows directly [29] and without any need for absolute
light measurements, which are tedious:
nl um = I
Loo b(t) dt Loo exp( - t/t rad ) dt
The only assumption made is that the starting conditions of the real decay
and the fictitious purely radiative decay are the same; this is true if the
excitation period is short compared with the time in which a decay of for
example 5% (to 95%) of the initial brightness occurs.
N has to be determined by a depth profiling, for instance SIMS, analysis,
to be sure about homogeneous doping. Also, by measuring n for a sample
for which one is sure that the outcoupling efficiency nOU! = 0.1, the cross-
section s can be determined, and holds for all samples with just this dopant.
This investigation has been conducted for a variety of preparation
methods, and the results are summarized in Fig. 5.12 [30,31]. Figure 5.l2(a)
shows the measured values of n and Fig. 5.l2(b) the nl um values, measured in
the way described; Fig. 5.12(c) resulted from the multiplication of the nl um
values by sN, with s known from a previous determination, to give an inner
efficiency nin!. Dividing the experimental values n by nin! of Fig. 5.12(c), we
obtain the nOU! values displayed in Fig. 5.12(d). Consistency is seen from the
fact that 0.1 is a lower margin for nOU! (except two samples at high dopant
concentrations, in which case the uncertainties are at a maximum).
Higher values of nOU! than 0.1 have to be explained, however. The classic
optics formula holds for a situation of a perfectly plane surface of the
emitting material. In fact the surface of the polycrystalline film in general
will consist of growth faces inclined against the average surface plane.
Therefore scattering will occur in addition to the refraction and total
reflection, adding to nOU! = 0.1. In fact, on probing surface roughness for
instance by surface plasmon coupling [32], a qualitative correlation is found
between nOU! > 0.1 and the expected dip in plasmon reflection [33]. A more
quantitative comparison of the values from Fig. 5.l2(d) with independently
determined values from brightness profiles [30] did not go beyond qualita-
tive agreement.
n.lm/W
lD'~~-----------------r--------~
I: a '".~
r- x
+
+
.. +
+ +
+
EBE1,HT
+ ElSE 1, L T
THE1
a THE:!
~ ALII:
X RFS
D.l 1 lD
(a) Mn-eoneentration.1020 em-'
nl um
1
+
~ X
D.l~ ..... . ..................................
+ .
I- +
+
(b)
... x
lD ~ .................. ~ ... .d otlI~" O'.;l< ...... :
+ +
+
1 ~.
(e)
nout
1 I:
~ ........................................... .
I- a ~ ~
JC +
D.l ~ ..... . + X. +
'\0 :.. '.' :
I-
D.l 1 lD
(d) Mn-coneentration.10Z"cm-
Fig. 5.12 (a) The EL efficiency n as a function of the Mn concentration measured
on samples made by different deposition methods: EBE, electron beam evaporation;
HT, LT, high, low substrate temperature; THE, thermal evaporation; RFS, radio
frequency sputtering. The efficiency has been decomposed into (b) nl um , (c) ninl and (d)
noul' The symbols are the same in all parts ofthe figure, showing clearly that there are
improvements possible for all of them [30].
152 Thin film electro luminescence
More difficult to analyse certainly are these partial efficiencies in AESs,
especially the luminescence efficiency nlum , as not only are the radiative
decay times of the prominent dopants not known yet, but the decays are
much faster (in the nanosecond range), so that measuring is much more
complicated [34].
The physics behind the measured nlum(N) is 'concentration quenching'.
Assuming some defects (impurities or native defects) to act as killers of
luminescence by giving rise to nonradiative transitions, there will be more
chance for an excitation to reach these killers if the centres are highly
concentrated, so that by transfer from the excited via unexcited ones the
killer can be reached during one radiative lifetime. Also, as the probability
for quantum-mechanical transfer goes steeply up with decreasing distance R,
e.g. for dipole-dipole interaction as R- 6 , this means that there exists a
certain concentration where quenching starts suddenly. Of course it will
depend on the killer concentration too, but if the latter is low, the dopant
concentration dominates the behaviour. As known from the classical papers
by Dexter [35], the coefficient in the transfer probability can be approxi-
mated by the spectral overlap between absorption and emission spectra of
the transferring centres, with of course the oscillator strengths determining
the absolute value. So transfer should be more likely for the 5d -+4f dopants
Eu2+ and Ce3+, used mainly in the AESs. However, the radiative lifetimes
scale down in the same way, so that predictions about the concentration
quenching are not simple from hand-waving arguments.
Even if the concentration dependence of nlum is known, the optimum
dopant concentration N opl depends on the absolute value of the excitation
cross-section s, as is evident from the inner efficiency ninl of Fig. 5.12(c), and
manifested for convenience schematically in Fig. 5.13 again.
Concluding, the maximum efficiency attainable in electroluminescence
critically depends on the concentration quenching behaviour of the chosen
dopant in a given host, and its excitation cross-section for the energetic
electrons in that host lattice. This is true for the inner efficiency ninl. A bonus
pertaining to the real outer efficiency can be achieved by introducing
scattering into the optical system, e.g. by surface roughness.
To give a numerical example, which is not far from reality (ZnS:Mn), we
assume, that at a dopant concentration N = 2 X 1020 em - 3 the luminescence
efficiency nlum =0.4; the cross-section ofthis dopant we set to 4 x 10- 16 em 2
This gives us an inner efficiency ninl = 0.4 x 4 x 2 X 104 em -1 = 3.2 ~m -1.
This means that, on its way through the active film of say 1 ~m thickness,
each electron gives rise to the generation of 3.2 photons.
The energy efficiency we obtain by accounting for the photon energy of
2.1 eV, and the fact that the electron travels in a field of 1.5 x 106 V em -1,
which results in 3.2 x 2.1 eV/(e x 1.5 x 106 x 10- 4 V)=4.48 X 10- 2 or 4.5%.
Knowing that the luminous equivalent in the ZnS:Mn emission region is
590 1m W- 1 , the inner luminous efficiency is 4.5% of that, or 26.5 1m W- 1 ,
Values and margins 153
.1 20
m 2
Fig. 5.13 Schematic of the factorization of the efficiency: the shaded area of n)um
indicates the variations obtained experimentally using the different technologies; the
limiting curves of ninl reflect this variation.
or, what is really meaningful for the user, there is an (outer) efficiency of
10.5 1m W - 1, if nOUI = 0.4, which is not unrealistic.
Another way of speaking is the following. A factor of 0.4 is lost be-
cause of the luminescence efficiency; at the assumed concentration and
cross-section, the average impact length is (SN)-1 = 12.5 X 10- 6 cm- 1,
but the electron gains enough energy to excite (2.1 eV) on la =
2.1/1.5 X 10 6 V cm -1 = 1.4 X 10- 6 cm; this accounts for a loss factor of
1.4/12.5=0.112, and these two multiply to 0.112 x 0.4=0.0448. In words, at
the optimum dopant concentration the luminescence efficiency is relatively
high (40%), but this concentration is so small that only 11 % of the hot
electrons are used for excitation, and the others miss their targets.
So what is wanted is a dopant which has a higher cross-section, or shows
concentration quenching not earlier than N'=I/s1 a =1.8xl0 21 cm- 3 . As
there is no general expression available for the concentration quenching
nlum(N), no optimum condition can be generally formulated.
In practice, however, things are even more complicated. An example again
is ZnS:Mn. In this case a rather strong nonlinear interaction in the dopant
system occurs. Incidentally, Mn shows cross-relaxation in the sense of a
transfer of excitation energy from an excited Mn to an excited neighbouring
Mn. The last one is excited to the second excited state, which is degenerate
with the interband continuum of the host, and so two excitations are lost by
reabsorption of this 4.3 eV photon. The transfer probability of this process is
some orders of magnitude larger than the one for transfer from an excited to
an unexcited Mn 2 + (the transition from the first to the second excited state
is dipole allowed), so it is a rather effective nonlinear loss, which shows up
154 Thin film electro luminescence
already at moderate excitation levels. As is obvious from the description, it
will depend on the dopant concentration itself, and is less pronounced far
from the optimum doping [36].
5.3.2 Brightness
From a user's point of view the efficiency of a display may be only relevant if
it is so low that heating becomes a limitation. If this is not the case, mainly
the brightness (and contrast) will be an essential specification. As is evident
from the understanding of the mechanism, brightness is roughly propor-
tional to the drive frequency, and to the transferred charge Q. The latter
depends on the drive voltage, or more accurately on the voltage above
threshold. It has already been mentioned that driving a matrix display
usually means that a select pulse is applied to a row at a time, and
simultaneously information pulses of opposite polarity (in order to add at
the pixel) are supplied to all columns that should light when they cross the
selected row. As the latter voltage is limited by the capability of the column
driver, very often in the literature the brightness at some voltage above
threshold (mostly at 30 or 40 V above) is specified. This is important for the
designer of drive circuitry, but almost irrelevant for the physics. In section
5.2.1(a) the charge-voltage characteristic was discussed, and from that it is
very obvious that the brightness B(U th +40V)=nQ(Uth +40V) depends not
only on the efficiency n but strongly on the capacitance ratio C;/( C j + C.),
and on C j itself too. So the capacitances are important design criteria.
Talking about values and margins, for display purposes at mains fre-
quency (50 or 60Hz) about lOOcdm- 2 are wanted, but 60cdm- 2 can be
sold, if the contrast is good [37]. So we have to discuss the limits imposed
by physics on the charge and/or the voltage (independent of the availability
of driver ICs).
The avalanching of the active film clamps the voltage across it, and all
amplitude above threshold stresses the insulator films. So ultimately their
breakdown field limits the brightness. Looking closer, it turns out [38] that
the storable charge on the dielectric films - EoEjF b - is the margin of the
transferrable charge (the E are the permittivities of vacuum (subscript 0) and
insulator (subscript i), and Fb is its breakdown field strength). Maximum
safe values of 6 x 10 - 6 A s cm - 2 have been reported.
So, if there is a principal limitation in Q, the brightness can be easily
increased by raising the drive frequency. 30000 cd m - 2 can be reached safely
by devices driven at 8 kHz [37]. This is not possible, of course, if a complex
display has to be addressed, as even only 200 rows would decrease the select
pulse width to 0.6 J,1S, and with the high capacitive load of a 200-line display
(of reasonable size) this is impossible to achieve. However, applications of
electroluminescence are imaginable for which those brightness levels are
mandatory.
Acknowledgement 155
5.4 OUTLOOK
Thin film electroluminescence is still far from having reached a textbook
level of comprehension. The mechanism appears to be understood on the
basis of a linearized concept, not taking into account, for instance, the
consequences of avalanching for inhomogeneities in carrier concentrations,
space charge distributions and recombination destroying inhomogeneous
space charge in every half-period of the drive voltage. Progress in this
direction might be expected from the evaluation of time-dependent electron
energy distributions, measured in vacuum emission.
Almost nothing is known about the energetic location of the dopant
ground and excited states with respect to the energy bands of the host
crystals. Strong correlation effects have to be expected. It is this feature
which makes any consideration of interactions between a dopant's electrons
and free carriers in the bands very complicated. Especially in the case of the
alkaline earth cha1cogenides, the liberation of electrons from the excited
state of Ce 3 + and Eu 2 + has been discussed [25, 26] which could give rise to
more complicated temperature dependences of (effective) impact excitation
cross-sections of these dopants.
One of the most challenging deficiencies of relevant knowledge concerns
the palette of materials available for EL from the viewpoint of loss-free
acceleration. This special high field transport mode up to now has only been
established in lIB and IIA cha1cogenides, but it would be surprising if
additional materials do not join this class.
Concerning the luminescence processes, the concentration quenching,
discussed mainly in connection with laser materials, enters the limelight
again. A thorough understanding, by simulation for instance, of the complex
interplay of various transfer processes, linear and nonlinear, is urgently
needed. It is not unlikely that, similarly to lamp phosphors, sensitizers could
bring about improvements of the use of available electronic impact power.
As mentioned above, the balance between excitation and quenching by field
and/or impact has in no instance been thoroughly investigated. Also, the
partition of relevant processes into laminates of the active film [39, 40] either
differently doped or composed of different materials has only been touched on.
Summing up, one can state that the first steps in the understanding of thin
film electroluminescence have been taken, but the whole field is not far
beyond infancy - and this in view of a tremendous application potential,
ranking from very simple indicators, displays for computers, monochrome
and colour, up to high definition television.
ACKNOWLEDGEMENT
Thorough and stimulating discussions with Dr R. Mach are gratefully
acknowledged.
156 Thin film electroluminescence
REFERENCES
1. Queisser, H.J. (1981) J. Lumin., 24, 3-10.
2. Destriau, G. (1936) J. Chim. Phys., 33, 587-625.
3. DePuydt, 1.M., Haase, M.A., Qiu, 1. and Cheng, H. (1992) In Proc. 5th Int. II-VI
Con!, Tamano, 1991, in J. Cryst. Growth, 117, 1087.
4. Chen, Y.S. and Krupka, D.C. (1972) J. Appl. Phys., 43, 4089.
5. Vlasenko, N.A. and Jaremko, A.M. (1965) Opt. Spektrosk., 8, 39.
6. Inoguchi, T., Takeda, M., Kakihara,Y., Nakata, Y. and Yoshida, M. (1974) In
SID 74 Digest, pp. 86-87.
7. Inoguchi, T. and Mito, S. (1977) In Topics in Applied Physics, 17, Springer,
Heidelberg, p. 222.
8. Riihle, W., Marello, V. and Onton, A. (1979) J. Lumin., 18--19, 729-738.
9. Russ, MJ. and Kennedy, D.I. (1967) J. Electrochem. Soc., 114, 1066.
10. Miiller, G.O., Mach, R, Selle, B. and Schulz, G. (1988) Phys. Status Solidi A, 110,
657-669.
11. Mach, R. and Miiller, G.O. (1991) Semicond. Sci. Techno/., 6, 305-323.
12. Mach, Rand Miiller, G.O. (1982) Phys. Status Solidi A, 69, 11-66.
13. Mach, R. and Miiller, G.O. (1984) Phys. Status Solidi A, 81, 609.
14. Mach, Rand Miiller, G.O. (1989) In Proc. 4th Int. Con! on II-VI Compounds,
1989, in J. Cryst. Growth, 91, 967.
15. Fitting, HJ. and Czarnovski, A. (1986) Phys. Status Solidi A, 93, 385.
16. Brennan, K. (1988) J. Appl. Phys., 64(8), 4024-4030.
17. Allen, J. (1984) In Proc. ICL 1984, in J. Lumin., 31-32, 665-670.
18. Miiller, G.O. and Mach, R. (1986) In Proc. 20th ICPS, 1986, pp. 1587-1590.
19. Miiller, G.O., Mach, R., Halden, E. and Fitting, H.-J. (1990) Proc. 20th ICPS,
World Scientific Press, Singapore, pp. 2510-2513.
20. Miiller, G.O., Mach, R., Reinsperger, G.-v. and Schulz, G. (1991) In Proc. IDRC
91, San Diego, CA, IEEE, Piscataway, New Jersey, p. 17.
21. Miiller, G.O., Mach, R, Reinsperger, G.v., Halden, E. and Schulz, G. (1992) In
Proc. 5th II-VI Con!, Tamano, 1991, in J. Cryst. Growth, 117,948.
22. Gumlich, H.J. (1970) Energy transport in Electroluminescence and Electro-
photoluminescence of II-VI Devices, Sammlung Vieweg.
23. Mach, R., Reinsperger, G.-V., Miiller, G.O., Selle, B. and Matzkeit, G. (1992) In
Proc. 5th Int. II-VI Co/if., Tamano, 1991, in J. Cryst. Growth, 117, 1002.
24. Mach, R, Gericke, W., Treptow, H. and Ludwig, W. (1978) Phys. Status Solidi A,
49,667.
25. Ando, M. and Ono, Y.A. (1991) J. App/. Phys., 69(10), 7225-7230.
26. Yoshiyama, H., Sohn, S.H., Tanaka, S. and Kobayashi, H. (1989) In Electro-
luminescence, Springer Proceedings in Physics, Vol. 38, Springer, Berlin, pp.
48-55.
27. Minami, T., Miyata, T., Takata, S. and Fukuda, I. (1992) In SID Int. Symp.,
Boston, MA, May 17-22, 1992.
28. Busse, W., Gumlich, H.E., Meissner, B. and Theis, D. (1976) J. Lumin., 12-13,
693.
29. Benalloul, P., Benoit, 1., Mach, R., Miiller, G.O. and Reinsperger, G.-v. (1990) J.
Cryst. Growth, 101, 989.
30. Mach, R, Miiller, G.O., Schrottke, L., Benalloul, P. and Benoit, 1. (1990) Acta
Poly tech. Scand., Appl. Phys. Ser., 170, 161.
31. Mach, R, Miiller, G.O., Reinsperger, G.-v., Benalloul, P., Benoit, 1. and
Ohnishi, H. (1990) In SID 90 Digest, pp. 238-241.
32. Harbeke, G. (1985) In Polycrystalline Semiconductors, Springer Series on Solid
State Sciences, Vol. 57, Springer, Berlin, pp. 156-169.
References 157
33. Mach, R., Schrottke, L., Milller, G.O., Reetz, R., Krause, E. and Hildisch, H.
(1988) J. Lumin., 40-41, 779-781.
34. Hilttl, B., Rinck, W., Milller, G.O. and Mach, R. (1982) Proc. Workshop on EL
1992, El Paso, Cinco Puntos Press, EI Paso, Texas, USA (1990) p. 123.
35. Dexter, D.L. (1953) J. Chern. Phys., 21, 836.
36. Milller, G.O., Neugebauer, J., Mach, R. and Reinsperger, G.-v. (1988) J. Cryst.
Growth 86, 890-894.
37. Mach, R. (1993) In Solid State Luminescence (ed. A.H. Kitai), Chapman & Hall,
London, Chapter 7.
38. Howard, W.E. (1977) IEEE Trans. Electron Devices, 24, 903-908.
39. Suyama, T., Okamoto, K. and Hamakawa, Y. (1982) Appl. Phys. Lett., 41, 462.
40. Herrmann, R., Milller, G.O. and Selle, B. (1990) Acta Poly tech. Scand., Appl.
Phys. Ser., 170, 171-179.
6
Powder electroluminescence
Surjit S. Chadha
6.1 INTRODUCTION
This chapter will survey the phenomenon of electroluminescence (EL) in
powder phosphors. Electroluminescence is luminescence produced by the
application of an electric field to a dielectric phosphor; it is also termed
electroftuorescence. Reviews of the AC powder electroluminescence (AC-
PEL) phenomenon as discovered by Destriau [1] in 1936 and of the DC
powder electroluminescence (DCPEL) as promoted by Vecht [2] will be
presented. As far as ACPEL is concerned, the period of 1936-1963 was
reviewed by Ivey [3] and Henisch [4]. Therefore, in this chapter, apart from
historical references and comparisons with the DCPEL devices, the review
of the ACPEL cells will be limited to any progress since 1963 noting major
investigations such as Fischer's [5] and a few others.
The organization of this chapter is in 11 sections. Sections 6.2 and 6.3
consist of historical perspectives of both AC and DCPEL structures re-
spectively. Section 6.4 describes the powder phosphors, their preparation
techniques, the role of the luminescent centres in zinc sulphide (ZnS) with
special reference to copper (Cu) and manganese (Mn) and the Cu coating
process that is essential for DCPEL devices. The possible presence of
copper sulphide phases on the Cu-coated ZnS phosphor is also discussed
in section 6.4. Section 6.5 covers the fabrication of ACPEL and DCPEL
devices. Section 6.6 discusses the forming process that is unique to DCPEL
devices. The possible excitation and injection mechanisms occurring in
ZnS phosphors are covered in section 6.7 with the optoelectrical responses
of both AC and DC excited powder EL devices in section 6.8. The driving
schemes employed for the AC and DC structures are described in section
6.9 and the maintenance characteristics in section 6.10. Finally section 6.11
illustrates some of the typical usage of both AC and DC powder EL
devices.
We begin with a brief history of the ACPEL cells.
159
160 Powder electroluminescence
6.2 ALTERNATING CURRENT ELECTROLUMINESCENCE
IN POWDER PHOSPORS: HISTORICAL
As stated above, Destriau [1] discovered EL in Zn:SCu powder phosphors
in 1936. The structure of his original EL cell is shown in Fig. 6.1.
Figure 6.2 shows a modified version of the original Destriau cell. Al-
though very high voltage (15 kV) AC excitation was used, the luminance was
extremely poor. Thus, even by 1950, Leverenz [6] questioned whether the
Oil
mercury
ZnS
Salt water
Mica
Metal
Fig. 6.2 A schematic diagram of a modified Destriau cell (no transparent conduct-
ing electrodes available then).
AC electro luminescence in powder phosphors 161
light emission process was in fact due to the excitation of the ZnS:Cu
powder phosphor by AC fields. According to him, the phosphor may in fact
be excited by the UV light emitted by the electrical breakdown of the gases
in the fairly porous powder cells. Consequently the effect may be merely
photoluminescence (PL).
The real impetus for EL displays was provided by the arrival of the
transparent conducting electrodes (TCE) based on tin oxide (Sn02) in the
late 1940s. All devices until then had been rather fragile and as a result
precluded industrial interest.
The 1950s saw massive academic and industrial research effort being
devoted to the 'source de lumiere du futur' as Destriau had claimed. The
many attractive features claimed for this new possible light source included
flatness, low power dissipation, multicolour emission and the lack of
catastrophic failure. Among possibilities suggested. for these devices, a TV
screen was prominent. Unlike the bulky and cumbersome TV screens based
on the cathode ray tube (CRT), these slim, lightweight units could have been
hung on the wall, much like a painting.
The cells developed in the early 1950s were made by using a suitably
doped ZnS powder (5-20Ilm grains) suspended in a dielectric (binder),
sandwiched between two electrodes, one of which was transparent, and
supported by a substrate. The construction is essentially that of a flat-plate
capacitor in which the phosphor is the dielectric medium. The structure has
changed very little over this time (Figs 6.3 and 6.4). The substrate may be
glass or flexible plastic or may be metallic, in which case the top electrode
must be transparent. The dielectric binder may be organic, e.g. cyanoethyl
cellulose with a plastic or glass substrate or a low melting glass with a metal
substrate. Usually a secondary white dielectric layer is used as extra
protection against catastrophic breakdown and to reflect the light emitted
through the transparent layer.
Protective glaze
Tin oxide
ZnS:Cu in enamel
Reflector layer
Grip coat
Metal electrode
Plastic
ZsS:Cu in high
dielectric constant
binder
Aluminium foil
black inslating
resin
EL-monograin layer
fl.l.l .l.lX
on thermoplastic
.. ~.~
. : ~., resin, optionally
dyed with
W fluorescent
pigment
vacuum-deposited
AI 20 3 insulator
5000 A thick
Fig. 6.5 Use of TFT drive for ACPEL cells as proposed by Fischer [5].
Phosphor
Layer
Glass
",-~---
Substrate
Fig. 6.7 (a) Cubic (sphalerite) and (b) hexagonal (wurtzite) modifications of zinc
sulphide.
Table 6.1 shows some of the binary and ternary systems that have been
investigated over the years together with the activator(s) used for ACPEL.
Some alkaline earth based phosphors that have been found to be useful for
DCPEL are given in Table 6.2. The maximum luminance under continuous
and pulse DC condition is also given together with the power efficiency
(section 6.4.3(f).
the CRT phosphors were modified for the manufacture of the ACPEL
phosphors. For DCPEL, with its inherent requirement that particles should
be small and of uniform size, newer preparative techniques had to be
developed. Most luminescent and optoelectronic II-VI materials are pre-
pared by the addition of controlled amounts of specific impurities (dopants),
called 'activators' and 'co-activators', to a high purity II-VI compound.
In order to prepare high purity starting materials, one of the following
methods could be used.
1. Direct reaction of the group II and VI elements.
2. Precipitation from aqueous group II salt solutions by gaseous hydrogen
sulphide, selenide or telluride. Another method uses the reaction between
group II salt in aqueous medium with, for example, thiourea or
thioacetamide that provides the sulphur. The reaction yields high purity
ZnS.
3. Thermal decomposition of suitable organometallic compounds such as
zinc or cadmium diethyldithiocarbamates.
Although totally feasible, the direct reaction of pure elemental Sand Zn,
for example, is not recommended, since the reaction is difficult to control
and is explosive especially if the reagents are in the form of finely divided
powders. The thermal decomposition of organometallic compounds leads to
the production of toxic organic residues and traces of carbon in the product.
As a consequence, this method is also not used for the production of powder
II-VI compounds (however, this method has been used successfully for the
deposition of thin films of ZnS [93]). Therefore the precipitation methods
are the most widely used for the preparation of powder II-VI phosphors.
Some of these precipitation methods will now be described in more detail.
~~-----------------.-----------------.
I
50 ................................................................................1.................................................................................. ..
~
!
40 ................................................................................1...................................................................................
g
3. ------------ -----+--------------- i
.~ 20 ................................................................................ !. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.3 10 ...................................................................................;I...............................................................................
.
!
O+-____~~~~~~~i----~~~~~~~
!
10 100 1000
Log,o Milling time (mins)
Fig. 6.9 The reduction in ACPEL output at identical drive conditions for ZnS:Cu
phosphors milled over a range of time.
% 50
0.1 1 10 100
Particle size ().1m)
Fig. 6.11 The particle size distribution of DCPEL phosphor showing a very
narrow spread.
/
voltage and constant current respectively.
7
1
- 005:
0.3% Mn ./
,
:- 10.1
-
:::J
c.
:::J
o
/
Q)
g 10.2
co
c
E
:::J
...J
10.3 L..-1...-.L.-J'---1---L---L--'---'---'
o 40 80 120 160
CI content (p.p.m.)
Fig. 6.12 The effect of Cion EL output at constant voltage [29].
In ZnS, the most important luminescent centres are due to Mn and Cu. In
DC PEL, Mn emits a bright yellow-orange peak centred around 585 nm.
The Cu centre is only efficient under AC excitation, emitting either green or
blue, the colour being dependent on [CuJ and the preparative techniques
used. Both of these luminescent centres are discussed below.
10
0.3% Mn
0.05% Cu
--
()JA cm- 2 )
- ..............
-:';:~,
: :I
C.
: :I
00.1
Q)
()
C
<1l
C
E
::I
...J
.,
o
o 40
CI content (p.p.m.)
80 120
""
160
100r---------------~----------------------~
so
o~----~----~----~------~----~----~~
200 300 400 SOO 600 700 800
Wavelength (nm)
Fig. 6.14 Typical emission spectra of the blue emitting ZnS:Cu ACPEL phosphor
(conditions: 60 Hz, 200 V, and 400 Hz, 200 V).
loor-------------------~~------------------~
so
o~----~----~~--~
600
~
700
~~
800
Wavelength (nm)
Fig. 6.15 Typical emission spectra of the green emitting ZnS:Cu ACPEL phosphor
(conditions: 60 Hz, 200 V, and 400 Hz, 200 V).
o - Vacuum level
Conduction band
5
Valence band
20
25
Fig. 6.16 Energy levels of the Mn2+ ion.
interaction with other previously excited impurity systems. Since the excited
state responsible for emission is only 2.1 eV above the ground state, and the
band gap of ZnS is 3.7 eV, the activator system in ZnS:Mn is stable against
field ionization in applied fields as great as 10 7 V cm - 1.
In ZnS, because of its 'characteristic' nature, Mn2+ produces only one
emission band with its peak centred at "" 585 nm (Fig. 6.17). (Characteristic
activators luminesce by means of electronic transitions within closed in-
terelectronic shells. Ionization is not involved. This should be contrasted
with fundamental activators such as Cu. These give rise to new energy levels
between the conduction band and the valence band and are capable of being
ionized.) This is very close to the maximum sensitivity of the human eye. The
modes of excitation or the presence of any co-activator has little effect on the
emission spectrum of the Mn2+ centres. In fact, as Fig. 6.18 shows, changes
in the host lattice, CaS:Mn [101], produce only minor shifts in the emission
peak wavelength.
Mn is very soluble in ZnS. Nitsche [102] published a phase diagram for
the ZnS: MnS system. Three points of interest are as follows.
16
12
-<
~
r:::
o
o
-as .
....
.......
r:::
.gjo
/
ew 4
o /"
,,
CaS:Mn,Sc,Mg
E ::::I
>-E
.~ Ox
-
en co
5ir::: _E 50
.- 0
Q)
.~
co u
5i
-
QiQ;
0::0.
-::J
C.
~ 10.1
(I)
u
c
ca
c
'E
::J
....I
1
0.045% eu
....--.-
400 v
-::J
C.
S 10.1
o
~
c
ca
c
'E
/'/-
/.
/' 300 v
::J
....I
-E
......., 5
Q)
E
..;::;
>-
ctI
~ 2
o
Skolnick [67] reported that the decay times at [Mn] > 0.01 % were
non-exponential and became increasingly so for higher currents. In other
words, the decay curve was composed of two or possibly more components.
The luminance intensity, I(r), could be best described by the equation
(6.2)
where tl and t2 are the decay time constants of the isolated Mn2+
luminescent centres and Mn 2 + pairs respectively. The averaged dominant
decay times of Mn 2 + are given in Table 6.3.
In summary, Mn 2 + is the most efficient EL activator in ZnS with a spectrum
output that is very close to the maximum sensitivity of the human eye and is
almost unaffected by changes in the host lattice. However, despite its high
efficiency, it suffers from saturation effects due to quenching by Mn 2 + pairs.
This quenching places limits on achieving extremely high brightness devices.
Mn concentration '["
(%) (ms)
0.25% Mn
:::J 1
t...
:::J
Q.
\.
~ 10-1
Q)
(,)
c:
co
c:
e 10-2
:::J
...J
10-3 '--......L.._'--......L.._'--......
o 0.1 0.2
Cu (%)
Fig. 6.22 The effect of [Cu].a. on the luminance characteristics at constant current
[29].
10
...
:::J
Q.
~ 10-1
Q)
(,)
c:
co
c:
e 10-2
:::J
...J
>
u
c:
Q)
.(3
~ 0.2
0.1 -+-----r-----.------,-----,.-----,r---~
0.00 0.01 0.02 0.03 0.04 0.05 0.06
Lattice copper concentration %[w/w]
Fig. 6.24 The effect of [CU]lat on the power efficiency of DCPEL devices [108].
Cu++ SOURCE
Zn++ Zn++
2Cu+
+
~ ,-. --~~ -1---' --------- ~l~~ -~.- ----
ZnS BASE
Fig. 6.25 The exchange reaction between the ZnS surface and the Cu2+ ions in the
solution.
For ZnS powder phosphors, however, Smith [111] found that the
replacement reaction was only possible when excess surface Zn was present.
Thus phosphors fired in a nitrogen-sulphur atmosphere were very difficult
to coat with Cu whereas phosphors fired in excess Zn, were coated easily
with Cu. ZnS powder phosphor is normally fired in air and, owing to the
greater volatility of S and its high reactivity with oxygen, the surface is
expected to be Zn rich (because of the presence of only a limited amount of
oxygen in a closed crucible). Some ZnO is also formed and this is
subsequently removed by a dilute acetic acid wash. Smith gave the following
reaction equations after discovering that the amount of Cu taken up by the
phosphor almost equalled the amount of Zn found in the solution:
ZnO+2Cu 2+.-. Zn 2+ +2Cu+ (6.5)
The cuprous ions then react with the ZnS to give cuprous sulphide:
ZnS+2Cu+.-. Zn 2+ +CU2S (6.6)
Overall, the reaction can be written as:
Zno + ZnS + 2Cu 2+ .-. 2Zn2 + + CU2S (6.7)
Powder phosphors 183
126.9C 12.5C
2.0,."...-----~--__,_--------____,
E
c.
c.
Q)
c5 1.5
C.
::::>
CD
c.
c.
<31.0
o
Cl
o
......J
O.5+----+----+----+-----i------if...-_--I
0.0025 0.0030 0.0035
Temperature (11K)
Fig. 6.26 Copper uptake variation as a result of the coating temperature [108].
the thickness of the CU2-xS layer is nearly 0.2 J.l.m. Hence characterization
using most modern surface and bulk analytical techniques does not pose a
problem.
Some studies have been carried out to establish the nature of the CU2 - xS-
coated surface layer. Kelly [112] found distinct peaks in the resistance plots
of Cu-coated devices as a function of temperature (Fig. 6.27). These also
coincided with the well-known CU2-xS phase transition temperatures.
However, no discernible peaks were obtained using differential thermogra-
vometric analysis (DT A), indicating an absence of bulk phase transforma-
tions, since only the phase change transformations occurring within a
CU2-xS precipitate were involved. The amount of material involved would
be too small to be measurable by DT A but the electrical conductivity was
affected appreciably.
Resistance Resistance
103 ohms 103 ohms
20 200
10
J
II 100
20 40 60 80 100 120
Heat Treat Temperature cC
Fig. 6.27 Resistance changes as a function of temperature for C-coated ZnS:Mn,Cu
powder layers [112].
220
~ l-ta+".
32
~
.,I<~
~ 200 30
!'J~
....
~
.
180 28
:::J
~
Conductance J.A ~
..
~ 160 26
1- -1--) ~ "' : ~
0 -
82 C ~ Q)
C)
E 140 24 <:
M '"c
Q)
'0
Q) 120 ~ rr~ 22 E
85 C
~
C) ~
'"
U
<:
100 ~ I
1 20 c
It)
I
~~ r-- Q)
::J
"0 64 C Luminance ~
<: 80 ~ 18
V ~~
0
--
U
60 ....-"'i 16
"
40
. ~ 14
U)e-4
~ 8.0e5 SSC38(1/R)
<:
It) " SSC39 (1/ R)
U o SSC145(1 R)
-5 6.0<:.5
<:
o
U
4.0e5
2.0e.5 ~~~~~~---"'~-~-r-~~~~~....,...-~~
20 40 60 80 100 120 140
Temperature (0C)
Fig. 6.29 The conductance- temperature characteristics of the excess Cu-coated
devices (SSC38 and SSC 39) compared with an un aged device (SSC 145) [108].
studies (Fig. 6.29), the possible presence of CU2-xS phases was indicated in
devices that were fresh and for those where the Cu coating was carried out
at elevated temperatures [108]. The conductivity versus temperature plot
shows a decrease in conductance above 90 C in all cases. This may reflect
some Cu phase change or merely loss of adsorbed water. The role of the
adsorbed species will be mentioned later. Devices fabricated using Cu-
186 Powder electro luminescence
Fig. 6.30 Photolithographically etched ITO- (or SnOr) coated glass substrate.
Fabrication of powder EL structures 187
Phosphor layer
( ~ 25)Jm thick)
=~~_ AI evaporated
back electrode
Scribed phosphor/rear
electrode layer
Fig. 6.33 Scribing of the excess AI/ZnS:Mn,Cu layers to yield electrically separated
emitting pixels.
connections are made between the Al cathode and the contact pad lead-ins
using electrically conductive paint (Ag dag).
Whereas ACPEL cells emit light on the first application of an electric
field, DCPEL devices require a forming process. This leads to the creation of
a very narrow ('" 1 /-lm), highly resistive, layer adjacent to the TCE (anode).
188 Powder electroluminescence
The light emission occurs in this formed region. In the next section, the
forming process, which is peculiar to the DCPEL devices only, will be
described.
=
in Fig. 6.34.
T=
AI cathode
1
-------AI,Ov film
25 m I
I
I
1
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
Phosphor
I'Y"
ITO anode
Glass substrate
~~BaCklayer
Cu coated
ZnS:Mn,Cu
powder phosphor
- - Formed layer
50
Threshold voltage
40 (A1 2 0 3 breakdown)
~ Luminance
E 30 (ft - L)
+-'
c
....
OJ
.... 20
u 10
:::J Forming
voltage (V f)
0
0 10 20 30
Applied voltage (V)
Fig. 6.36 A typical forming trace for a DC PEL cell (0.2 cm 2 area).
binders can lead to devices that are very stable but dim and others lead to
devices that are initially very bright but exhibit poor maintenance. Normally
DCPEL devices are encapsulated with a desiccant placed in the backcap.
The cell structure after forming is shown in Fig. 6.37.
The first description of the forming process observed in Cu-coated
ZnS:Mn,Cu DCPEL devices was given by Vecht et al. [29] in 1969. They
stated that, 'when a voltage is applied to a Cu coated phosphor, during the first
minute or so (depending on the initial voltage) a marked drop in current is
observed associated with the onset of light emission'. The forming process was
said to be power dependent requiring about 1-2 W cm- 2. It was suggested
that the conduction in the DCPEL devices was via a non-stoichiometric
phase of CU2 - xS.
190 Powder electro luminescence
- - - A I cathode
T~ ____
25 IJm I I I I I I I I I I I I I
BaCklayer
Emitting
Cathode region Anode
+
After
forming ---t
E 101 (200 v) Before
E forming
1 (10 v)
~
u i
10-1 f \
! 1
co
0'0
~
!!l 10-
2
f-
,,1
.0
\ '~
"0 O~j/
> I"CT
10-3
o 4 8 12
Distance (mm)
Fig. 6.38 Potential profile in a gap cell [29].
10
200 300 400 500
Wavelength (mil)
Fig. 6.39 Short-circuit current versus wavelength [77].
o
4000 4500 5000 5500 6000 6500 7000
Wavelength (A)
Fig. 6.40 PL spectra of unaged device (spectrum a), formed device (spectrum b) and
aged device for 700 h (spectrum c) [89].
As shown, the unformed device has two peaks (curve a), centred at
460.0 nm and 516.3 nm, which are associated with Cu (blue and green)
centres and one at 584.0 nm, which is the well-known Mn peak [29]. Curve
b shows that the peak due to the green Cu centres is greatly reduced whilst
the blue centre peak is almost absent. Curve c shows that the Cu green band
has been now substantially reduced. Interestingly, the Mn peak is also
reduced and some other, unidentified, peaks have appeared. However,
Abdalla et al. did not consider the possibility that the luminescence from the
Cu centres may be reduced by some other mechanism such as formation of
non-radiative centres, e.g. Cu! ions. The Cu ion pairs may loose their
energy non-radiatively in a similar fashion to the Mn-Mn pair centres. The
194 Powder electro luminescence
Voltage
(V)
Fig. 6.41 Uncorrected I-V characteristics for a standard test device of area
0.2 cm 2 Curve a was taken prior to forming. The remaining curves were taken after
forming to the following: b, 20 V; c, 30 V; d, 40 V; e, 50 V; f, 60 V; g, 70 V; h, 80 V;
i,90V.
current (A)
I m
100 abc de f h i
80
60
40
20
-50 -25
Voltage
-125 -100 -75 ,Ij 25 50 75 100 125
(V)
20
40
60
80
h gfedcba 100
Fig. 6.42 I-V curves obtained after subtraction of the curve obtained prior to
forming. The curves are derived from those taken at the following voltages: a, 20 V;
~~~~~~~~~~w~Lm~&w~~~~
across the formed layer is constant and is independent of the current passed.
Indeed, the characteristics deduced for higher forming voltages indicated
that the voltage across the layer decreased as the current increased. This
indicates that there is the possibility of negative differential resistance (NOR)
in the ZnS:Mn,Cu powder devices under normal drive conditions.
Alder et al. [64] also found that the reciprocal capacitance was a linear
function of the formed voltage for both the normal (i.e. transparent electrode
as positive) and the inverse voltages. This is shown in Fig. 6.43.
The results of their forming studies led Alder et al. [63] to conclude that,
according to their idealized model (Cu migration), there was 'no reason for
the forming process to stop at any particular value of applied voltage'.
Using surface analytical techniques, Alexander et al. [116, 117] elucidated
that initially (before forming) the Cu on the surface was present as Cu + ions
associated with sulphur, probably as CU2S. The storage of Cu-coated
phosphor in air had the effect of oxidizing some sulphide to sulphate. Auger
spectroscopy revealed that Cu depletion had occurred on the surface of ZnS
after the forming process. Electron probe microanalysis (EPMA) indicated
that the total [Cu] had not altered in the formed region. This was confirmed
in both gap cells and in through cells. The results further support the work
of Vecht et al. [29].
Alexander et al. also found that certain gaseous ambients affected the
forming and the degradation rates of the powder devices. Devices were said
196 Powder electro luminescence
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
o 10 20 30 40 50 60 70 80 90
Breakdown voltage (volts)
odi -2.6
...J
0
-2.8 +-...,--.--.--.-..........-+----.--.---.---.--+..........---.--.--.-...--i
2.0 2.5 3.0 3.5
Log,o time (5)
Fig. 6.44 Changes in the forming rates as a result of device exposure to wet and dry
air [57].
surface of the powder particles, and two have involved chemical changes of
the surface layer of the Cll, namely a CU2-xS phase change and an
oxidation model involving formation of the oxides and then sulphates of Cu
and Zn respectively. The possibility exists, of course, that forming involves
combination of two or more of these basic models. As noted earlier,
evidence has only been presented for the migration of Cu into the ZnS
lattice. No evidence exists for the migration of Cu over the surface of the
phosphor particle although it may be a considerably easier and therefore a
faster process and Beatty's [115] reservations about the likelihood of Cu
migration into an already saturated ZnS lattice. No evidence exists for
migration of Cu towards the cathode and as the results of Alexander et al.
indicate, the total [Cu] (surface+lattice per particle) is the same in the
unformed and formed regions. Evidence for any phase change has not been
conclusive.
198 Powder electro luminescence
It appears from the available data that the forming process involves an
initial migration of the Cu ions into the ZnS particles followed by an
oxidation process involving surface moisture (",0.3% by weight). This
adsorbed moisture is extremely difficult to remove.
After the DCPEL devices have been formed up to 70 V, they are given a
long-life treatment which involves the replacement of the dry air in the
backcap by an inert gas such as Ar or N 2 . Normally this treatment is carried
out at 105 DC as recommended by Werring [114]. The high temperature
appears to enhance the useful life of the DCPEL devices, probably by
removing some the adsorbed moisture.
The excitation mechanisms responsible for the luminescence process in
powder EL devices are discussed next.
( a) Field emission
Electrons from localized levels in the band gap are released by the local high
field. This process is very similar to the Zener effect of a reversed biased
diode where a very high field is produced in the depletion region. The
emission current (1) is related to the applied voltage (V) by
Ioc V 2 exp( -constant/V) (6.8)
Piper and Williams [120] and Taylor [121] calculated that the luminance
B obtained is related to the applied voltage, V, by
B=aV" exp( -bVO. 5 ) (6.9)
where a and b are constants and n=O, 2.
I =Io[exp(eV/kT)-lJ (6.13)
where k is Boltzmann's constant.
The large band gap of 3.7 eV in ZnS leads to negligible production of
thermally generated carriers at room temperature (Le. lois very small).
In both cases the excitation of the luminescent centres may occur by two
possible mechanisms. Firstly the Cu + centre is excited by the impact of
electrons or trapping of a hole and then the luminescent Mn 2+ centre is
excited by a resonance energy transfer mechanism through the excited Cu +
centre [107]. Secondly, the luminescent Mn 2+ is directly excited by the
impact of electron or trapping of a hole. In either case the characteristic
emission of an Mn2+ centre is obtained.
d. . . . . brightness at voltage V
(6.15)
IscnmmatIon ratIO = b' h '/2
ng tness at tj
With the DR being only around (3-5):1, these displays are not easy to
multiplex and normally useful alphanumeric and graphical displays would
not be feasible. The only way to overcome this problem is to incorporate a
TFT drive into the AC powder panel structure (Fig. 6.S) as described by
Fischer [S] and attempted by Brody et al. [125]. Unfortunately, this was a
very expensive solution at the time owing to the very low TFT yields and
never fully pursued. In addition, the stable AC thin film structures reported
by Inoguchi et al. [126] from Sharp in 1974 also further diminished interest
in AC powder for alphanumeric and graphical display purposes.
202 Powder electro luminescence
600.-__________________~----------------_,
+ ............................................................................................................................................,.r............................
::J
,
:S 400 -t ......................................................................................!..................................:.-/-..............................................
:;
C.
:; ; ....................................................................................; .............. ,1)(..................................................................
o
Q)
u
c:
co
c: 200
'E
::J
....J
.............................................................:;;'.""'.............+................................................................................. . 1
o
o 100 200
Applied voltage (V)
1000~-----------,------------~-----------.
~ 100 ....................................................t....................................................1".....................................................
....::J
C.
:;
o
Q)
1
100 1000 10000 100000
Frequency (Hz)
1\
:;
1\
c.
:;
o ~
\ -
Q)
(.)
c
co
c
._"\
E 1
::J
...J
.1
\ ~
o 10 20 30 40 50 60 70 80 90 1 00
ZnS:Cu phosphor layer thickness (~m)
Fig. 6.47 The effect of phosphor layer thickness on the luminance of a typical
ACPEL cell.
'"'"c: 101
Q)
.E
OJ
.;::
CD
1~L---------~~~____~~
m ro ~ ~ ~ 00 W M 00 ~
Volts
I
CirCUit
t~H,PG"
Y, -- - - - - - - - - - - - - - - - - - - H,PG
II x, !I
--+i~
Counter
I,,I
circuits
I
I
I,
, EL matrix II Vert.
5ynch.
5i9
! Ym -- -+1 H"PG
r- - -- ---- ---
i
i
Ir--L-'---I.....---...L...-----L..,
i Delay Line f4t-----~_ ___'
L___~:a~~~~~c..u_it _ _ _ .:
Speaker
Fig. 6.49 The block diagram for the drive circuitry of a ACPEL panel for TV
applications [19].
few papers describing the use of AC drive for the running of DC PEL panels
have also been published with emphasis being placed on apparently en-
hanced stability as a consequence.
Panels with true alphanumeric and graphical display capabilities require
fairly high voltage drivers. The rest of this section will be devoted to a
description of driving circuitry for the DCPEL panels starting with a brief
description of the methods of addressing x-y matrix panels. This will be
followed by the first reported EL TV from Matsushita Electric Industrial
Company to the present displays available from Phosphor Products Com-
pany and Cherry Electrical.
All flat panels employ an x-y matrix structure as shown in Fig. 6.50.
The two sets of electrodes are placed orthogonally so that any single pixel
may be selected by sending pulses along the column and row lines. It
should be recalled that the standard pixel-by-pixel scanning methods used
in conventional CRT receivers is not a viable addressing technique for
matrix panels of any type as this method would result in very low duty
cycle and hence very low display brightness. The preferred method is
therefore sequential line at a time. This method not only simplifies the
system configuration by eliminating the need for horizontal scanning but
also increases the brightness of the display. Even then a display with 640 x
200 pixels driven at 240 Hz only allows 12.5 Jls on time per pixel per frame.
206 Powder electroluminescence
Row
Vo ltage
Fig. 6.50 Schematic diagram showing the selection of a pixel in any x-y matrix
panel.
DCPEL was also investigated for TV purposes by the workers from the
Matsushita Electric Company [78]. The manufacture of a 13 in diagonal
display with 50 176 pixels was attempted. The display had 150 TV lines and
emitted 10 ft-L. It appears that a contrast ratio of 20:1 and a grey scale
capability of 16 steps were achieved but the maintenance was poor.
However, this display consumed 150 W compared with the 100 W consumed
by the AC variety.
The drive systems for a range of display sizes ranging from 32 to 1250
characters were reported by Mears et al. [61] in 1973 (Fig. 6.51). In a
sense, they were similar to those reported by Yoshiyama et al [19] since
only discrete components were used (although integrated circuit drivers
with the necessary high voltage output had been designed and these
included the necessary control logic). With the available logic controls, the
maximum input rate was only 300 Hz. This would be unacceptable since to
sequentially assemble a line of information (300 bits) and then to write to
the display would reduce the duty cycle to a very low value. The display
would then be very dim and show flicker. The problem could be overcome
by the inclusion of a memory-mapped RAM with a byte representing one
character.
Very similar to the ACPEL cells, the DCPEL devices also suffer from a
gradual increase in the resistance of the light emitting formed region (section
6.10). As a consequence, when driven by a constant voltage drive, typical
large area multiplexed panels decay by 30% over 10000 h. This is still more
than the average lifetime required for most laptop and portable computer
displays (it is worth remembering that for a computer used for 8 h per day,
i.e. nearly 2000 h per annum, a 10 000 h display still has a useful life of 5
years).
Maintenance of powder EL devices 207
Line
counter
Display
Control
Unit
Fig. 6.51 An early drive system for the DCPEL devices [61].
VS DCEL
HS
DATA Interface IC
CLOCK--~~-'r-~~~~
Display
PGM current
PINS
Ref ..
voltage
639
Level 20 Current
Column
Shifter Lines (REF)
L..--T-~1-~ driver
201C'S
Row 0 Row 199
,--......L.......L.----.
DATA
Row
CLOCK Driver
+ 12V & IC'S
Fig. 6.52 The constant current drive circuit used by Cherry Electrical [74].
100
....:::I
....0.
g~
Q) I 50
u
c:"O
E
ctlU
c: ~
'E
:::I
.-..J
(b)
5000 10000
Time (hrs)
Fig. 6.53 Maintenance curves of ACPEL cells with (curve a) and without (curve b)
an inverter.
Maintenance of powder ELdevices 209
been proposed for these devices. Roberts [128] invoked ionic diffusion in
the electric field as a possible cause though mentioned that moisture
caused discoloration of the phosphor. Further evidence for the ionic
diffusion model came from the work of Jaffe [90] who showed that by
replacing the halide co-activator in ZnS:Cu,X phosphors from Cl- to Br-
decreased the rate of decay. He thus believed that the slower migration of
the Br- compared with Cl- was responsible for the improvements ob-
served.
The most important work regarding the maintenance of ACPEL cells
was carried out by Fischer [5]. He showed that the decay in brightness
was related to the blunting of microscopic tips of CU2-xS inclusions in the
ZnS phosphor (Fig. 6.54).
Fig. 6.54 CU2-xS inclusions shaped like a double-comet in the ZnS:Cu,Cl powder
phosphor [5].
These CU2 _xS tips decorated imperfection lines in the ZnS and could be
blunted by the outward migration of the Cu ions amongst others. In a
well-executed experiment, Fischer showed that initially EL was observed in
a double-comet line shape. After some time, the double-comet lines changed
into just two light-emitting points, just at the positions where the comet
heads had been before.
Some deterioration was believed to be photon assisted because blue
emitting phosphors deteriorated faster than orange emitting ones - the host
lattice being identical. Lehmann [11] believed that it was the S vacancies
(Vs) in the ZnS phosphors which were a prime cause of the deterioration.
Formation of Vs is expected since these phosphors are normally fired at
elevated temperatures making loss of S highly probable. Lehmann went on
to make the 'hypermaintenance phosphor' which consisted of phosphor fired
in an S atmosphere.
210 Powder electro luminescence
6.10.2 Maintenance of DCPEL panels
Earlier on it was noted that the initial creation of the formed region in
DCPEL devices probably occurred as a result of Cu migration, although the
gaseous ambient also may be involved. The above section on drive circuitry
also mentioned the luminance drop that is always observed in DC devices
operated at a constant voltage. The initial rate of decay is high, followed by
a more gradual decrease. The decay is not exponential, but it has become
conventional for the maintenance of a device to be defined as the time for
the luminance at a given voltage to reach half the initial value. The current is
also found to decrease simultaneously. In this section, the pertinent litera-
ture on the maintenance characteristics will be presented.
Probably the earliest mention of the degradation phenomenon in DC PEL
devices was by Kazankin et al. [21] in 1959. Their devices had very poor
maintenance, lasting a few tens of minutes. Kazankin et al. [21] attributed
the decay in brightness and current to polarization effects since they
assumed that their cells were condensers. In support of their hypothesis, the
authors reported that under the application of the electric field, the crystals
(particles?) actually align in the direction of the applied field and formed
little bridges between the plates of the condenser.
Vecht et al. [32] found that the luminance decay had been significantly
slowed by the encapsulation of the EL devices. In fact, the devices had not
reached their half-life after hundreds of hours. This result clearly points
towards the role of some constituent of the ambient atmosphere in the
luminesence deterioration process. No mention was made of desiccant to
preserve a dry atmosphere in this paper. However, a subsequent paper
reported the use of dry nitrogen and a desiccant [33]. The current flowing
through the device also decreased. Thus essentially the efficiency of the
devices was found to remain stable. It was suggested that the maintenance of
the device could be improved considerably by using a series resistor and by
further improvements to the encapsulation technique. These therefore for-
med the first two methods for increasing the maintenance of the powder
devices.
Although the half-life was improved from a few minutes to over 900h,
Vecht et al. [31] had recognized that devices must last for 5000-10000 h to
be useful.
Figure 6.55 shows the maintenance curves of a typical device containing
dry air and of a device containing a desiccant (magnesium perchlorate) and
an inert gas.
The first electrical measurements on life-tested devices indicated that the
impedance of devices was increasing with operating time [33]. Following
this, the use of constant power to achieve a pseudo-stability in luminance
was suggested by Ellis et al. [34]. It was shown that running a device at
constant power resulted in luminance remaining essentially unchanged for
Maintenance of powder ELdevices 211
100
:::J,
...
~
:; \. Improved drying
...
I
Q.
--
::J
0 10
CD
tJ
c:
(1)
c:
E
::J
....J
1
20 40 60 80 100 120 140 160
-Hours~
Fig. 6.55 Life curve, showing the effect of dry nitrogen atmosphere. 'Through'
luminescence cell at 100 V [34].
0.2
:::J
t 0.1
Efficiency >-
tJ
c:
CD
CD
tJ
50 u
c:
(1)
if
CD
c: 20
E 0
Ci
::J 0
....J ....J
0
Ci
0
....J o 0 1 2 3
Time (hrs) x 1000
Fig. 6.56 Brightness and efficiency versus operating time, at 100 V [36].
Mears et al. [61] reported that between 30 and 100 ft-L were achieved
with devices pulsed at 0.5% duty cycle with good operating performance.
Figure 6.57 shows the maintenance curve of these pulsed devices.
According to Abdalla et al. [85], the continued migration of Cu was the
culprit for the loss of brightness. In addition, it was felt that the Cu-Mn
resonance energy transfer process may become less efficient on degradation,
again because of continued Cu migration.
In a subsequent publication Abdalla et al. [86] suggested the use of an AC
drive to stabilize DC formed devices. The results of a device formed and run
212 Powder electroluminescence
50
::::J
~
:;
B-
:J
o
OJ
u b
c:
co
c:
"E
:J
...J
10
10
Hours of operation
Fig. 6.57 Stability of a 100 V device driven at 185 V, 3 IlS pulse width, 0.7% duty
cycle (curve a) and 160 V, 121ls, 0.5% duty cycle (curve b) [61].
at 70 V DC were given. Initially, this device emitted 9.8 ft-L and the
luminance dropped to less than 0.1 ft-L after 200 h. Replacement of the 70 V
DC drive with 81 V (rms, 800 Hz) AC drive increased the luminance to
9.8 ft-L again which then dropped to 8.4 ft-L after another 450 h.
Until 1980, most of the authors had referred to device degradation in
terms of either loss of luminance with time or time to half-life or having to
increase the voltage to keep constant brightness level. It was left to Alder et
al. [24] to rectify this situation and to treat EL cells in the same way as Si
devices. They measured pulsed I-V characteristics before, during and after
forming to certain voltages (section 6.6). At each stage, the capacitance of the
devices was also measured. In addition, devices were characterized during
aging. Three possible types of device failure modes were diagnosed.
1. Further forming. As shown in Fig. 6.58, further forming is a parallel shift
in the I-V characteristics normally accompanied by an increasing
efficiency. This failure mode is ascribed to the widening of the formed
120
<100
oS
c: 80
~
:; 60
"
:J
~ 40
.,
il: 20
Fig. 6.58 Parallel shift in I-V curves indicates further forming (VF= Vo at To and
VF = VI at TI)'
Maintenance of powder EL devices 213
layer that normally results in more Mn 2 + being excited. Thus although
the brightness may be expected to decrease because of lowered current,
the increase in the quantum efficiency often partially offsets this loss (Fig.
6.59). The sharp threshold in the I-V characteristics led to the panels
that failed by further forming being referred to as 'hard' devices.
2. Softening. As shown in Fig. 6.60, a significant leakage of current prior to
the 'threshold voltage' leading to inefficient devices was termed softening.
This failure mode was the result of parallel leakage paths, probably on
the surface of phosphor particles in the formed region. The B-1 curve is
therefore depressed (Fig. 6.61) and the quantum efficiency is reduced.
T=T,
...
:::J
Co
'5
o Significant increase in
Q)
u luminance efficiency exhibited by
c:
III
c: panels which further form
"E
:::J
...J
120
;{100
.s
1: 80
~
B60
:::J
~4O
a:::
20
o
o 20 40 60 80 100 120 140 160
Applied voltage (V)
Fig. 6.60 Softening in the I-V characteristics leads to significant current passage
before the threshold voltage point (VF= VO at To and VF = Vl at Td.
214 Powder electro luminescence
o
o 20 40 60 80 100 120 140
Pulse current (mA)
120
;?100
E
~
80
E
~
::; 60
u
:;,
<1l 40
~
<1l
a::: 20
o
o 20 40 60 80 100 120 140 160
Applied voltage (V)
Fig. 6.62 Large increase in the series resistance is termed load line instability
(where formed voltage VF = Vo at To and VI at Td.
Maintenance of powder EL devices 215
voltage to maintain constant luminance. According to Alder et al. [64J, this
failure mode was successfully suppressed by a heat treatment process at
105C [114]. The cure for load line change requires limiting high processing
or running temperatures. The only failure mode for which the cause was not
known was softening. However, it was stated that, by resubjecting the 'soft'
devices to continuous DC (i.e. as during initial forming), they could be
restored to 'hard', i.e. efficient, devices.
Vecht et al. [53J reported the use of thin film (TF) interlayers between the
TCE and the ZnS:Mn,Cu. These TF interlayers of ZnS, SiO, Y203 and
polymeric materials were found not only to reduce the initial forming
powders but also to eliminate softening. The obvious implication of this
result is that ionic migration from the glass may be responsible for soft
devices. The effects of the gas ambient were discussed and the choice of the
correct molecular sieve to eliminate the residual gaseous impurities was said
to be a dominant factor in obtaining improved maintenance. This was one
of the first publications to note the possible ~ole of a negatively charged
species (probably Vs) in the maintenance characteristics.
According to Fujii and Hoshina [79J, DCPEL devices emitting 30--60 ft-L
(lOllS pulse width and 0.5% duty cycle and 120 V) for over 3000 h were
possible. This was achieved by using an encapsulation process that was
completely different from all others in the literature. The devices were heated
to 105C under a vacuum of 10- 3 Torr for 3 h. The devices were then kept
under a vacuum of 10- 6 Torr for 24 h to remove any adsorbed gases and
water. Finally they were sealed in a glass capsule with molecular sieve and
an atmosphere of dry argon.
The optoelectrical characteristics were reinterpreted by Alexander et al.
[116, 117]. They examined the standard technique of using linear plots of
both the I-V and the B-1 characteristics initiated by Alder et al. [24J in
1980. By plotting log I versus V and log I versus log B, they found another
failure mode. Figure 6.63 reproduces the plots of Alexander et al. This new
failure mode pointed towards an increasing barrier height for injection.
According to the authors, there were three important features of the
physics of the DC PEL powder device:
,;7
,, '"
"
" co
Cl
.2 / " Cl
.2
" /
" /
I
/
(a)
V log I
'"
---
Cl /
/ '" '" CO
Cl
.2 / .2
f
'/
~
~
(b)
V log I
CO
Cl
.2 Cl
o
'I
I
f
f
(c)
V log I
".-
"...
,-
I I"
f ,/
f
,
.r
Cl
I co ,,-
.2 Cl /
f
f .2 I
I
I
/
I"
(d)
V log I
Fig. 6.63 Schematic changes in injection and efficiency plots for different degrada-
tion modes: (a) further forming; (b) load line change; (c) softening; (d) effective barrier
height change or decrease in injection site density.
Maintenance of powder EL devices 217
Although no evidence was found for any of these mechanisms actually
operating in the DCPEL devices, all involved transferring an electron across
an energy barrier by field-assisted thermal activation or tunnelling. The
following relationship was obtained for current (I) and field (E):
I =Io(E) exp{ -e[H -F(E)]/kT} (6.17)
where H is the energy barrier for electron injection and F(E) is the
field-dependent reduction in barrier height. This is a variant of the Schottky
thermionic emission expression. The authors chose to put F(E) = A V/d
where A was a constant and d the formed layer thickness. Thus a log I
versus V plot (injection plot) was linear with the slope dependent on the
thickness of the formed layer and the intercept related to the barrier height,
H. Equation (6.16) was used for the luminance, B, i.e.
B=KIn
The slope of log B versus log I (efficiency plot) was said to show the photon
production mechanism and the intercept a measure of the quantum efficiency.
The object was to prove that there was no threshold in both the I-V and the
B-1 characteristics as advocated by Alder et al. [24]. According to Alexander
et al., the same mechanism was operating at both low and high electrical fields
in both the current and the luminance characteristics. The existence of a
threshold voltage has been argued for a long time, beginning with Destriau in
1947 [129J, Thornton in 1960 [10J, Vlasenko and Yaremko in 1965 [130J and
so on. In addition, just as any increase in the formed layer thickness would
have been shown up by a change in the slope of the characteristics, any
parallel shift in the I-V characteristics would have indicated a change in the
barrier height for electron injection (H in equation (6.17)) (Fig. 6.64).
2 40V
55 V
70 V
- 85 V
...
c:
~
::J
U
go O+-~~~'---+---T-~~-+--
....J
250
Applied voltage (V)
-1
Fig. 6.64 Parallel changes in the I-V characteristics are indicative of changes in the
barrier height for electron injection [116].
218 Powder electroluminescence
Since Alexander et al. found that, during forming, the I-V characteristics
exhibited a parallel shift, they concluded that, after the initial forming had
occurred, the thickness of the formed region does not increase. Thus either
the barrier height for electron injection was increasing or there was a
reduction in the injection site density (see Fig. 6.65).
Finally, the use of constant current drive by Cherry Electricals (section
69.2) has led to the appearance of very stable devices. The maintenance
curve of their displays is shown in Fig. 6.66.
The lack of stoichiometry leading to poor maintenance has been inves-
tigated recently [57]. This research produced very stable devices (6%
degradation in light output in 10 000 h). The devices were treated with H 2 S
gas. Although a similar model to that of Lehmann [11] for the ACPEL cells
can be proposed by invoking the reduction of Vs, the actual mechanism is
probably more subtle in the DCPEL case. We have proposed that Vs
vacancies lead to trapping centres and that the trap density affects the
conductivity of the ZnS. For example, O 2 trapped at an S site would reduce
the surface conductivity since it is an acceptor centre. Similarly, H2
adsorbed at the same site would lead to enhanced conductivity since it will
act as a donor centre. The H 2 S gas treated devices are inefficient and would
be traditionally regarded as 'soft'. This more conductive surface may provide
for a more stable injection interface by a reduction in the barrier height to
injection. The improved maintenance may also reflect the greater stability of
the conductive CU2-xS surface layer. Excess S on the surface may inhibit Cu
migration. In a recent paper, it was stated that a decrease in the number of S
vacancies leads to lower clamping field Ez , which results in lower efficiency
in ZnS thin films [131]. The situation regarding reduced power efficiency in
DCPEL panels appears to be the same although it is slightly more difficult
to be precise in this case. This is because the thickness of the formed region,
::J 2
~
Q)
u
C
<II
c
"E
..:1
0
Ci
0
-.J
-1 2
Log,o current
-1
Fig. 6.65 Injection and efficiency plots for a DCPEL device formed in air in the
range 40-85 V [116].
Application areas 219
:r-l:
120.-----~.----~.----~.----~----~----~-----r
t I : 1.8
1.6 ....
:I-:::-:---:-~[-:--:-=:-~~~!~~;~.-:- :-:- ~1-;-___":__"_ ~_
....
co
1 --=-=-r_:
___ ..
_ 1.4 ~
....
IJ)
Q;
"0 1.2 ~
-
> IJ)
.J ~~~~~_------~---~~--~---~--i---------- c
Q)
E
Qi
x ! ! 0.8 ::J
0.4
0.2
0 I i 0
0 2 4 6 8 12 14
Hours
(Thousands)
Fig. 6.67 A photograph showing typical shapes and sizes of ACPEL lamps.
6.12 CONCLUSIONS
The history and developments of both AC and DC powder EL structures
have been reviewed. The bulk of the effort on ACPEL cells was in the period
1950-1963 and most of this was reviewed by Ivey [3] and Henisch [4]. The
situation has not changed much since then for the ACPEL cells, though they
are more efficient and stable than the lamps of the early 1960s. Obviously
the manufacturing processes have improved considerably since then as well.
However, their poor DR has ensured that they will not be considered for
alphanumeric and graphical applications. As a result, most of this review has
been on DCPEL devices although, where applicable, comparative reviews of
Fig. 6.68 A prototype car dashboard DCPEL display.
Fig. 6.69 A DCPEL display for alphanumeric and graphical applications, e.g.
laptop computers. (Courtesy PPC.)
222 Powder electro luminescence
Fig. 6.70 A DCPEL display with touch switch capability for dealer room applica-
tions. (Courtesy PPc.)
ACKNOWLEDGMENTS
The author is indebted to many friends and colleagues in the displays
industry. Particular thanks for supplying photographs and data go to Mr Jo
Mayo and Mr Eric Olsen (Cherry Electrical), Mr Ray Ellis (Phosphor
Products Company), Mr Fred Tagg (Craft Data) and Dr Paul Beatty
(formerly Smiths Industries). He would like to express his sincere thanks to
Mr David Smith (University of Greenwich) for his helpful criticism. Finally,
he is indebted to Professor Aron Vecht for allowing him to 'dabble' in EL
and other display systems during the last 12 years.
REFERENCES
1. Destriau, G. (1936) J. Chim. Phys., 33, 589.
2. Vecht, A. (1984) SID Semin. Notes, vol. 1, section 1, p. 2.
3. Ivey, H.F. (1966) Advances in Electronics and Electron Physics, Suppl. 1, Electro-
luminescence and Related Effects, Academic Press, New York.
4. Henisch, H.K. (1962) Electroluminescence, Pergamon, Oxford.
5. Fischer, A.G. (1971) J. Electrochem. Soc., 118, 139c.
6. Leverenz, H.W. (1950) An Introduction to Luminescence of Solids, Wiley, New
York, p. 61.
7. Thornton, W.A. (1957) J. Appl. Phys., 28(2), 262.
8. Thornton, W.A. (1958) Bull. Am. Phys. Soc., 3, 233.
9. Thornton, W.A. (1959) Phys. Rev., 113, 1187.
10. Thornton, W.A. (1960) J. Electrochem. Soc., 107, 895.
11. Lehmann, W. (1966) J. Electrochem. Soc., 113, 40.
12. Larach, S. and Shrader, R.E. (1957) J. Phys. Chem. Solids, 3, 159.
13. Zalm, P., Diemer, G. and Klasens, H.A. (1954) Philips Res. Rep., 9, 81.
14. Zalm, P. (1956) Philips Res. Rep., 11,353-399,417-451.
15. Vlasenko, N.A. and Popkov, I.A. (1960) Opt. Spectrosc., 8, 39.
16. Bowtell, J.N. and Bate, H.C. (1956) Proc. IRE, 44, 697.
17. Homer, H.M., Rulon, R.M. and Butler, K.H. (1953) J. Electrochem. Soc., 100,
566.
18. Busca, G. and Cortese, C. (1971) Lett. Nuovo Cimento, 2(5), 261.
224 Powder electro luminescence
19. Yoshiyama, M., Kawarada, H. and Sato, T. (1970) Proc. IEEE Int. Computer
Group Conj, pp. 261-269.
20. Nicoll, F.H. and Kazan, B. (1955) Proc. IRE, 43, 1012.
21. Kazankin, D.N., Pekerman, F.M. and Petoshina, L.M. (1959) Opt. Spectrosc., 7,
458-459.
22. Favorin, V.N., Kozina, G.S. and Tikhonova, L.K. (1959) Opt. Spectrosc., 7,
420-421.
23. Favorin, Y.N. and Kozina, G.S. (1961) Opt. Spectrosc., 10, 43.
24. Alder, c.J. Cattell, A.F., Dexter, K., Kirton, J. and Skolnick, M.S. (1980) SID
Biennial Display Research Conf. Digest, Cherry Hill Meeting, p. 1.
25. Vecht, A and Ellis, R (1966) Nature (London), 210, 1251.
26. Vecht, A and Ellis, R (1966) Proc. Int. Conf. on Luminescence, 1966, pp.
1986-1991.
27. Vecht, A, Werring, N.J. and Smith, P.J.F. (1968) Br. J. Appl. Phys., Ser. 2, 1, 134.
28. Vecht, A. (1968) Proc. 4th Oxford Lecture Course on Electronic Components,
October 1968, vol. 9, p. 1153.
29. Vecht, A, Werring, N.J. Ellis, R and Smith, P.J.F. (1969) Br. J. Appl. Phys., Ser.
2,2,953.
30. Vecht A. (1969) Spectrum, 7, 66.
31. Vecht, A., Werring, N.J., Ellis, R. and Smith, P.J.F. (1970) J. Electrochem. Soc.,
117(1), 134.
32. Vecht, A. and Werring, N.J. (1970) J. Phys.: Appl. Phys., 3(2), 105.
33. Vecht, A., Werring, N.J., Ellis, R. and Smith, P.J.F. (1970) J. Phys. D, 3,165-166.
34. Ellis, R, Werring, N.J., Vecht, A., Smith, P.J.F. and Williamson, J.H. (1971)
Conf. on Displays, Loughborough, September 1971, pp. 205-208.
35. Waite, M.S. and Vecht, A. (1971) Appl. Phys. Lett., 19(11),471.
36. Werring, N.J., Ellis, R, Smith, P.J.F., Vecht, A. and Mears, A (1972) SID Symp.
1972, p. 170.
37. Vecht, A (1973) J. Vac. Sci. Technol., 10(5), 795.
38. Vecht, A. (1973) J. Lumin., 7, 213.
39. Vecht, A (1974) J. Phys. (Paris), Colloq. C3, Suppl. 4, 35, 173.
40. Waite, M.S. and Vecht, A. (1974) J. Electrochem. Soc., 121, 109.
41. May, D. and Vecht, A. (1975) J. Phys. C, 8, LS05-L509.
42. Vecht, A. (1976) Electrochem. Soc. Extended Abstr., 76(2), 563.
43. May, D. and Vecht, A (1976) Thin Solid Films, 36, 137-140.
44. Vecht, A., Higton, M.H. and Nicholas, D.M. (1976) Electrochem. Soc. Extended
Abstr., 76(2), 574.
45. Vecht, A., Mayo, J. and Higton, M.H. (1977) SID Int. Symp. Digest 1977, p. 88.
46. Higton, M., Vecht, A. and Mayo, J. (1978) SID Int. Symp. Digest 1978, pp.
136-137.
47. Werring, N.J. (1978) Electron, pp. 1337-8.
48. Vecht, A., Higton, M., Mayo, J. and Blackmore, J. (1980) SID Int. Symp. Digest
1980, p. 110.
49. Vecht, A (1981) SID Int. Symp. Digest 1981, p. 32.
50. Vecht, A, Waite, M., Higton, M.H. and Ellis, R (1981) J. Lumin., 24-25,
917-920.
51. Smith, P.J.F. and Werring, N.J. (1981) Proc 1st European Displays Conf., VDE
Verlag, Berlin, pp. 149-151.
52. May, D. (1981) Electronic Displays '81, Conf. Proc., London, Vol. 1,1, pp. 82-92.
53. Vecht, A, Chadha, S., Hayes, R. and Ellis, R (1982) SID Int. Symp. Digest 1982,
p.16.
54. Vecht, A. (1982) J. Cryst. Growth, 59, 81.
References 225
55. Chadha, S.S. Haynes, C.V. and Vecht, A. (1988) SID Int. Symp. Digest 1988, pp.
35-38.
56. Mayo, 1., Hutchinson, P., Hayes, R and Ellis, R. (1984) SID Int. Symp. Digest
1984, p. 22.
57. Chadha, S.S. and Vecht, A. (1988) Electroluminescence, Springer Proceedings in
Physics, Vol. 38, Springer, Berlin, pp. 337-341.
58. Mayo, J.W., Hutchinson, P., Hinsley, J.L., Alexander, P.W. and Davies, M.
(1986) SID Int. Symp. Digest 1986, Paper 17.4, pp. 313-315.
59. Kirton, 1., Lloyd, K., Sarginson, R, Smith, P. and Freer, W. (1971) 5th Annual
Conf. on Solid State Devices, University of Lancaster, September 14-16, 1971,
Paper 5.8, p. 60.
60. Mears, A.L., Parker, P. and Sarginson, RW. (1972) Solid State Devices 1972,
Paper C5.8, p. 224.
61. Mears, A.L., Parker, 1., Sarginson, R.W. and Ellis, R. (1973) SID Int. Symp.
Digest 1973, p. 30.
62. Mears, A.L. and Sarginson, RW. (1976) SID Int. Symp. Digest 1976, p. 10.
63. Alder, C.J., Cattell, A.F., Dexter, K., Dixon, M. and Kirton, J. (1980) Electron.
Lett., 16, 571.
64.Alder, c.J., Cattell, A.F., Dexter, K., Dixon, M., Kirton, 1. and Skolnick, M.S.
(1981) IEEE Trans. Electron Devices, 28(6), 680-688.
65. Kirton, 1., (1981) Proc. 1st European Displays Conf., VDE Verlag, Berlin, p. 144.
66. Kirton, J. (1981) In Handbook on Semiconductors (ed. C. Hilsum), Vol. 4,
North-Holland, Amsterdam, p. 665.
67. Skolnick, M.S. (1981) J. Phys. D, 14, 301.
68. Dean, P.1. (1981) J. Lumin., 23, 17-53.
69. Wickenden, D.K. (1980) Phys. Technol., 11,211-217.
70. Shepherd, B. (1981) Displays, 2, 338.
71. May, D. and Chadha, S.S. (1982) Conf. on II-VI Compounds, University of
Bradford, February 1982.
72. Fickiet, J.M. and Plumb, J.L. (1973) J. Electrochem. Soc., Solid-state Science and
Technol., 120(9), 1238.
73. Hanak, J.1. and Yocom, P.N. (1973) Final Report Contract No. DAAB07-
69-c-0290 for us Army Electronics Command AD-766741.
74. Glaser, D. and Kupsky G.A. (1986) SID Int. Symp. Digest 1987, Paper 16.2, p.
292.
75. Mayo, 1. and Channing, D. (1990) Inform. Disp., 6(2) 6-8.
76. Yoshiyama, M., Kawarada, H. and Sato, T. (1971) 17, 670-676.
77. Yamamoto, Rand Yamazoe, H. (1973) J. Appl. Phys., 44(7), 3191-3192.
78. Kawarada, H. and Ohshina, N. (1973) Proc. IEEE, 61(7), 907-915.
79. Fujii, H. and Hoshima, T. (1983) Japan Displays '83, Paper 4.6, p. 96.
80. Luo, X. and Zhou, L.X. (1981) Extended Abstracts from the May 1981 Elec-
trochem. Soc. Meeting, Abstract No. 173, pp. 431-433.
81. Xu, S., Chou, L. and Luo, X. (1982) SID Int. Symp. Digest 1982, Paper 4.2, pp.
18-19.
82. Chou, L., Lou, X. and Xu, S. (1982) Extended Abstracts, Electrochem. Meeting
Soc. 1982, Abstract No. 519, p. 847.
83. Wu, B., Chern, 1., Hong, X. and Liu, F. (1984) J. Lumin., 31-32, 951-953.
84. Zhongyi, Y., Weidong, H., Jingpu, L., Nanggang, Z., Wei, Z. and Zhiquan, L.
(1983) Japan Displays '83, Paper 3.6, p. 586.
85. Abdalla, M.I., Godin, A. and Noblanc, 1.P. (1979) J. Lumin., 18-19, 743-748.
86. Abdalla, M.I., Brenac, A. and Noblanc, 1.P. (1980) SID Biennial Display Research
Conf. Digest, Cherry Hill Meeting, pp. 174-176.
226 Powder electroluminescence
87. Pecile, D., Brunei, c., Thomas, l and Godin, A. (1982) Extended Abstracts,
Electrochem. Soc. Meeting 1982, Abstract No. 172, pp. 428-430.
88. Abdalla, M.I., Godin, A., Brenac, A. and Noblanc, lP. (1981) IEEE Trans.
Electron Devices, 28(6), 689-693.
89. Destriau, G. (1937) J. Chim. Phys., 34, 327.
90. Jaffe, P.M. (1961) J. Electrochem. Soc., 108, 711.
91. Gobrecht, H. and Gumlich, H.E. (1956) J. Phys. Radium, 17, 754.
92. Gobrecht, H., Gumlich, H.E., Nelkowaki, H. and Langar, D.Z. (1957) Z. Phys.,
149,504.
93. Vecht, A., Smith, D.W., Chadha, S.S., Morton, D.C. and Miller, M.R. (1990)
Electrochem. Soc. Extended Abstracts, October 1990, No. 625, pp. 896-897.
94. Brown, R.A. (1968) Electrochem. Technol., 6(7-8), 246-251.
95. Willard, H.H. (1956) Anal. Chem., 22(11), 1372-1374.
96. Jackson, P.A. (1967) British Patent 1,071,062 (Cl. C09K) (June).
97. Cusano, D.A. (1962) In Luminescence of Organic and Inorganic Materials (eds
H.P. Kallman and G.M. Spruch), Wiley, New York.
98. Aven, M. and Parodi, J.A. (1960) J. Phys. Chem. Solids, 13, p. 56.
99. Nickerson, J., Goldberg, P. and Baird, D.H. (1963) J. Electrochem. Soc., 110(12),
1228-1230.
100. Gushurov, G. and Banks, E. (1967) J. Electrochem. Soc., 114(11), 1143-1148.
101. Kawamura, T., Hironori, T. and Kazuhiko, T. (1989) J. Electrochem. Soc.,
136(4), 1229-1234.
102. Nitsche, R. (1971) J. Cryst. Growth, 9, 241.
103. Juze, R., Rabrean, A. and Paschier, G. (1956) Z. Anorg. Allg. Chem., 285, 61.
104. Tornqvist, R. (1983) J. Appl. Phys., 54, 4110.
105. Kitai, A.H. (1988) J. Lumin., 39, 227.
106. Miyata, T., Nakagawa, S., Hirayama, H. and Hasegawa H. (1970) Jpn. J. Appl.
Phys., 9, 615-624.
107. Goldberg, P. and Nickerson, J.W. (1963) J. Appl. Phys., 34, 1601-1608.
108. Chadha, S.S. (1981) Unpublished results.
109. Thornton, W.A. (1959) J. Appl. Phys., 30, 123-124.
110. Gaudin, A.M., Fuershenan, D.W. and Mao, G.W. (1959) Min. Eng., (April), 11,
431-436.
111. Smith, P.J.F. (1973) M.SC. Thesis, Thames Polytechnic.
112. Kelly, J. (1979) M.Sc. Thesis, Thames Polytechnic.
113. Hartmann, H., Mach, R. and Selle, B. (1982) Curro Top. Mater. Sci., 9, 224.
114. Werring, N.J. (1973) Ph.D. Thesis, Thames Polytechnic.
115. Beatty, P.H.l (1970) M.Sc. Thesis, Thames Polytechnic.
116. Alexander, P.W., Sherhod, C. and Stowell, M.J. (1988) J. Phys. D, 21, 1627-
1634.
117. Alexander, P.W., Sherhod, C. and Stowell, M.l (1988) J. Phys. D, 21, 1635-
1641.
118. Beatty, P.H.J. (1974) Ph.D. Thesis, BruneI University.
119. Cabrera, N. and Mott N.F. (1949) Rep. Prog. Phys., 12, 163.
120. Piper, W.W. and Williams, F.E. (1955) Br. J. Appl. Phys. (Suppl.), 6, 39.
121. Taylor, J.B. (1954) Br. J. Appl. Phys., 45.
122. Zallen, R., Eriksen, W.T. and Ahlburg, H. (1960) J. Electrochem. Soc., 107,
288-95.
123. Chadha, S.S., Smith D.W., Vecht, A. and Williams, lL. (1991) SID Technical
Digest, Paper 6.6, pp. 80-83.
124. Waite, M.S. and Thomas, A. (1980) J. Electrochem. Soc., 127(2), 509.
125. Brody, T.P. and Yu, K.K. (1975) IEEE Trans. Electron Devices, 22,739.
References 227
126. Inoguchi, T., Takeda, M., Kakihara, Y., Nakata, Y. and Yoshida, M. (1974) SID
Int. Symp. Digest 1974, p. 84.
127. Faria, S. (1988) J. Electrochem. Soc., 135(10), 2627.
128. Roberts, S. (1957) J. App/. Phys., 28(2), 245.
129. Destriau, G. (1947) Phi/os. Mag., 38, 700.
130. Vlasenko, N.A. and Yaremko, A.M. (1965) Opt. Spectrosc., 15, 263~266.
131. Hirabayashi, K., Shibata, T. and Kozawaguchi, H. (1988) IDRC 1988, San
Diego, CA, p. 42.
7
Thin film electroluminescence
devices
R. Mach
7.1 INTRODUCTION
Thin film electroluminescence devices (TFELDs) are discussed in the litera-
ture at present mostly under the heading 'flat panels'. It will be shown that
this is by no means comprehensive; other applications are feasible and
upcoming. Nevertheless, the TFEL Display is on the marketplace, bright
enough for us to predict a bright future for it.
We are going to discuss in what follows the device engineering aspects.
Concerning the underlying physics, the reader is referred to other chapters
of this book (Chs 1, 2 and 5).
The most exciting feature of any TFELD is that it is all-solid-state (Fig.
7.1). The structure is simple. One substrate - glass, in general - is covered
by a stack of thin films of dielectric and semiconducting or phosphor
materials, which are sandwiched between electrodes, one of which is
Insulator -...../-:Jr::='-J"",,,,'-'
Semicondu rTnr --,-
Glass
Fig. 7.1 Schematic of the physical structure of the thin film EL display.
229
230 Thin film electroluminescence devices
7.2 DEVICES
7.2.1 Displays
TFEL displays are matrix organized, in a manner shown by Fig. 7.1, and
addressed as shown in Fig. 7.2. The thin film stack of an insulator -
semiconductor (phosphor) - insulator (lSI) sequence is deposited onto a first
system of (bottom) electrode stripes, running parallel to each other. On top
of it a second set of electrode stripes, perpendicular to the first one, is
deposited. A picture element (pixel) is defined by the crossing of top and
bottom electrodes. If voltages of different polarities with respect to a common
Devices 231
f!
ltn . . .
r~
n
012
time
.
Fig. 7.2 The addressing scheme, a line at a time, mostly used in flat panels.
(ground) level are applied to the electrodes, they add across the pixel. This
is the basis of addressing or driving. The most common way to drive any of
the matrix-organized flat panels is a line at a time [1]. In contrast to cathode
ray tubes (CRTs), in which the addressing electron beam scans the pixels in
a row sequentially (and the rows in a frame scan), in a TFEL display all
pixels in a row are addressed simultaneously. Consequently the dwell time
per pixel is independent of the number of columns; it is only determined by
the number of lines, m, as the frame time divided by m.
As is well known from liquid crystal (LC) displays, problems might arise
with such a simple concept of addressing if the response of the active
material is not prompt, and no real threshold behaviour is encountered. If
for instance the response is to the effective voltage, as for LCs, and is rather
soft, almost linear, the multiplex rate, or number of addressable lines, is low.
In the TFEL case, however, the response is instantaneous and steep, so that
1000 multiplexable lines are easily reached. If the frame time is 20 ms, a 20 f.lS
dwell time for each pixel imposes no problems on the luminescence,
although possibly on the driver circuitry.
The steepness of the response of TFEL devices to an applied voltage is
shown in Fig. 7.3. The brightness-voltage characteristic B(V) is shown with
a suppressed zero of the voltage scale, to make it meaningful. Four orders
of magnitude in brightness are bridged by 20 V, or 10% of the applied
voltage. Fortunately enough, in spite of this threshold behaviour, no
bistability is encountered. Grey scale, or the selection of defined brightness,
is achievable by amplitude selection. To evidence this, the B(V) curve of Fig.
7.3 is redrawn in Fig. 7.4 on a linear brightness scale: putting a select pulse
of 200 Von the row, information pulse amplitudes between 10 and 50 V can
give grey scaling.
232 Thin film electro luminescence devices
ZnS:Mn
100 Hz
250
Fig. 7.3 Brightness-voltage characteristic of a typical ZnS:Mn TFEL sample under
100Hz bipolar pulse drive, showing the extreme steepness of response.
600
ZnS:Mn
300 100 Hz
250
Fig. 7.4 The brightness-voltage characteristic of Fig. 7.3, redrawn on a linear
brightness scale, to demonstrate the easy grey scaling by voltage modulation.
other flat panels. However, drivers are available at about 5 and 10 cents for
one column and row respectively, an appreciable cost item. There has been
much discussion about reducing the voltage needed for operation. The
feeling behind it was the belief '300 V transistors are expensive'. Certainly
such a transistor uses more area on an Si chip, which counts in large-scale
production. There are, however, some trade-offs which need to be consider-
ed. The first is that for a given efficiency the light output can be achieved by
a certain electric power consumed; this can be supplied either by a high
voltage at low current or vice versa. The high current version, however,
would impose insurmountable problems with the transparent electrodes.
They are at their limit even at 200 V with diagonal dimensions of 14 in. On
the other hand, reducing the voltage requirement means that the capacitive
load is increased. This is quite obvious: the field needed in the active film is
set by physics; reducing the thickness of the active film reduces the voltage
but also the brightness for a given transferred charge. Either the voltage
must be increased to increase the transferred charge, or the frequency;
increasing the use of the applied voltage by improving the capacitive
division ratio between dielectrics and active films increases the total
capacitance. In both cases the capacitive currents increase and so do the
losses on the transparent electrode resistance. Inhomogeneities of brightness
arise from these voltage drops.
To give some figures, a typical, well-designed structure has a total
capacitance of about 10nFcm- 2 . The capacity seen by a row or column
driver depends on the details of the driving scheme; for one obvious solution
the calculations have been done by Miller and Kelley [2]. One result,
surprising at first sight, is that the column load is at a maximum when half
of the pixels are lit.
Little has been written in connection with Fig. 7.1 about the location of
the transparent electrodes. They are shown to sit on the substrate glass. This
is the conventional design. Recently in the discussion of colour the advan-
tage of this placing has been questioned. At this point, however, the
discussion of the insufficient conductivity of indium tin oxide (ITO) or
ZnO:Alleads us to mention the tendency to invert the classical structure by
depositing the transparent electrode on top (Fig. 7.5). Below the lSI stack a
thick ITO stripe brings about appreciable 'climbing' of the next films over
the edges of these stripes, giving rise to inhomogeneities and field concen-
trations on the edges, etc. The metal electrodes on glass with only 50 nm
thickness are more convenient. On top of the MISI, on the other hand,
thicker ITO does no harm. Also, even reinforcements by very narrow
metallic lines are more easily made on top.
Available from large-scale industrial production at present is a large
variety of monochrome TFELDs, based on ZnS:Mn as the active material.
Table 7.1 lists the main dimensions, pixel numbers and arrangements, and
the brightnesses of the commercial products. The largest TFELD available
234 Thin film electroluminescence devices
Fig. 7.5 Inverted structure of a TFELD; the transparent electrode (e.g. ITO) on top
of the stack has many advantages.
(a)
(b)
Fig. 7.6 Photographs of the top TFEL products of 1991: (a) a 19 in diagonal
1024 x 864 pixel display of Planar, USA; (b) a 640 x 400 pixel grey scale display made
by Sharp, Japan.
236 Thin film electro luminescence devices
exchangeability with a CRT, which was one goal earlier, may be a poor way
of viewing TFEL.
The extreme simplicity of the TFELD with only one glass (or other
substrate) and no filling as for plasma panels or LCDs avoids the need of
sealing seams around the active area, and lends itself to 'tiling' of a large
display from smaller modules. Of course, a problem with the electric
connections arises if one tries to go beyond four quarters which is simple, as
Fig. 7.7 illustrates. The difficulties arising from the optical discontinuity of
the meeting edges can be overcome, as shown on small samples [3].
~ ..
~
RGB-modul.l I
,:--
RGB-modul.3
/J
L-
~~ .. r::-'""7' .
select =
=
register =
=
3
1
- ~ I-- V
select :::
:::: 4
register
2
=
:-
_V V
... --0 o~J
i###
I RGB-modul.2 I RGB-modul.4
1--
Fig. 7.7 Tiling of large displays from four quarters is easy, as only two edges are
needed for contacts, and no sealing seams are necessary [3].
brightness 30000 cd m 2
leading edge (10% to 90%) 0.1 ms
trailing edge (90% to 10%) 0.3 ms
ripple <7%
spatial brightness variation < 5%
ZnS:Mn
8 kHz
B,
10' cd m-
Fig. 7.8 Rather extreme brightness values can be reached if the driving frequency is
increased and design is carefully tailored to the specifications [5].
Addressing was with 8 kHz pulse bursts between 4 and 10 ms, but the
repetition frequency was not higher than 10 Hz. Some thousand hours of
useful life were achieved under these conditions, which can be termed 'stress
aging'. The doping concentration had to be carefully chosen, and a special
design of the film stack was necessary.
Certainly in future the potentialities of this TFEL application will be
much more exploited.
238 Thin film electroluminescence devices
Dielectrics
Fig. 7.9 The light guided in the thin film stack along the substrate plane can be
emitted from an edge, addressing photo printers a line at a time [6].
The main idea in this concept is the fact that light guiding along the plane
of the film stack gives access to a much deeper (thicker) zone of the active
film, increasing the light output relative to that from the plane of a display
by factors of some thousands. While the emitting depth of the display is the
thickness of the active film (about 1 Jlm), the length of electrode stripes is the
relevant dimension in the case of the EE. However, severe problems were
encountered when attempts were made to achieve this. It turned out that, in
the usual ZnS:Mn-based devices, so successful for displays, the attenuation
length for the light travelling along the film plane is in the order of some
hundred microns only. Also, this is not because of the scattering out of the
plane, mentioned in ref. 1, but because of absorption. The reason for the
absorption is not really understood, but it seems to be tied to the manganese
centre itself.
As well as the principal advantages of no moving parts, greater expected
lifetimes, solid state design and ease of adjustment, the edge emitter offers
higher resolution (> 1000 lines in - 1), grey shading, and rather high speed,
with the achievable values for the latter not yet assessed or published. On
first glance the single elements (stripes) seem to need a driver each. An
Devices 239
of electrons
a: collected electrons
o
5 I
t
::::l
en
Z Au ...... ..
" ............... - -
. . ' ...... . g
Fig. 7.10 Schematic of the hot electron cold cathode, and the arrangement to
measure EEDs [9, 10].
240 Thin film electro luminescence devices
40mV, ....... j .. .
i(t}
I'
n Ivac
U!
20mV
/div .
~
!not!
trig'd
~
-160mV--~--~__~~__~__~__~__~~__~
o time, !-Is 100
Fig. 7.11 Comparing the time behaviours of the dissipative current i(t) and the
vacuum current I vac with the retarding grid tuned to the voltage value given near the
curve.
nothing more than a d.c. voltage, of say 20 kV, between the thin electrode of
the HECC and the top A1 coating of a conventional CRT (colour) screen.
The pixels of the HECC should have a 1:1 correspondence to the pixels of
the screen. Usual a-line-at-a-time addressing of the HECC gives the screen
phosphor in each pixel one line time to be excited, or at least the HECC
emission time (Fig. 7.11), which is much longer than the time the electron
beam 'rests' on each pixel during its line scan. The price which has to be paid
is that of the driving circuitry, more expensive than the deflection coils and
(hot) cathodes of the CRT. First estimates, however, indicate that the
brightness levels could be raised [10].
The application of the HECC was addressed in two other papers almost
at the same time [11, 12].
Thinking of simpler applications first, one very attractive feature of the
HECC is its ability to be of arbitrary geometry: a line focus, which is quite
difficult to achieve from the usual electron optics, is no problem at all.
Miniature electron guns become possible, consisting of nothing more than
the patterned HECC and something to be bombarded a suitable distance
from it. Two lines perpendicular to each other, exciting a commercial colour
TV screen, are shown in Fig. 7.12. It is evident from this first trial that such a
resolution is no problem to achieve.
Monochrome and colour 241
Fig. 7.12 A conventional colour TV screen excited by electrons emitted from two
HECC lines, perpendicular to each other, to demonstrate the application potential.
7.3.1 Monochrome
(a) ZnS:Mn
As mentioned earlier, ZnS:Mn is the material first used for TFEL and on
which the commercial production of TFELD at the end of 1991 was solely
242 Thin film electroluminescence devices
based. The reason is historical, but were this material not as forgiving as it
is, TFEL probably would not have reached this stage of development.
ZnS is easily vacuum deposited without too much decomposition. It is not
able to deviate appreciably from stoichiometry, and it is completely miscible
with MnS. The latter means that no clustering or precipitation of the dopant
is to be expected and was never reported. ZnS:Mn crystallizes at low
temperatures even more readily than undoped ZnS. There are two crystallo-
graphic phases of ZnS known, the hexagonal wurtzite and the cubic
zincblende. Both incorporate Mn 2+ on cation sites. The first excited stage of
the dopant, about 2.4 eV above the ground state, is crystal field split and
highly metastable, as the radiative transition to the ground state is dipole
forbidden. It has a radiative lifetime depending on the crystal structure [13]
of about 1.5 ms. The emitted broad band is amber or yellow-orange and the
spectrum is given in Fig. 7.13. Concentration quenching sets in at about
10 20 cm - 3, which is due to increasing excitation travelling in the dopant
system by quantum-mechanical transfer, reaching eventually a killer centre.
This concentration quenching is obvious from the faster decay, which is no
longer exponential [14,15].
intensity, a.u.
100r--------------..------------.
50
500 700nm
(b) ZnS:Ln
Ln stands for the lanthanides or rare earths. They have been tested for quite
some time, as they are the only possibility of making ZnS emit at other
colours than the yellow of Mn 2 +. Other transition metals are not known to
emit in the visible, if doped into ZnS. The big risk with all the rare earths in
ZnS is their small solubility, said to be some parts per million. Segregation
or clustering is likely to occur, even if it is possible to incorporate
concentrations above this limit in deposition processes far from the ther-
modynamical equilibrium (TDE).
Nevertheless, good efficiency and brightness values have been reported for
ZnS:Tb3+ or, to indicate how it was prepared, usually ZnS:TbF 3. Its green
emission in narrow lines of the (forbidden) 4f-+4f transitions (Fig. 7.14) is
attractive for display applications.
intensity, a.u.
100.-----------~--------------~
ZnS:Tb
50 -
500 700nm
<= (2 )
=>
300K
~
~
::J SrS : Ce thin film ::J
-
tV
'-' cti
'-'
u
>- 420nm >-
c:
-
CD 'iii
c:
'u
:e
CD
CD
c:
c:
=>
c:
-
0 0
'Z:; 'iii
tV Ul
420nm
300K
The term 'full colour' is used if there are at least 64 grey shades possible in
each of the basic colours, In what follows we shall use the term 'colour'
display or devices as meaning either multi- or full colour.
Two different concepts are followed in achieving colour TFEL devices:
I, patterned phosphor, or
2, colour by filtering.
246 Thin film electroluminescence devices
Electrodes
--=. .=',.,=. ========::::;::-"'" . . '
"""-.. _.----=::=====::::::;'......
,~,7
Insulator
Patterned
phosphor
,~,7
1,4I'~:Ph"
,~,7
Patterned
Colour
Filters
Fig. 7.17 Exploded view of a colour TFELD using the colour by white or
patterned filter concept.
The inverted structure with ITO (or other transparent) electrodes on top
of the lSI stack, patterned with the filters used as the etch mask for the ITO
(self-aligning, parallax-free system), was proposed recently. It is illustrated in
Fig. 7.18. Testing of the device can be done before the filters are deposited;
screen printing in a flow process of organic filters with high colour
saturation may be used before wet etching.
Very recently multicolour displays were demonstrated by Planar [34] and
by Sharp [35]. Both used colour filtering at least partially. The Planar
demonstrator is shown in Fig. 7.19. It uses a further method of filter
arrangement. The red part of the Mn emission, which is red shifted
compared with the spectra shown in Fig. 7.13 if the Mn concentration is
high, was used for a standard red using an inorganic Cd(S,Se) film below the
ITO on glass, acting as a short wave cut-off. Green emisson, however, was
obtained from ZnS:Tb 3 + made as mentioned previously by inhomogeneous
doping in ALE. So this in fact is a patterned phosphor with filters.
The Sharp example uses filtering from the broad emission spectra of
ZnS:Mn of green and red in an inverted structure, with organic filters (Fig.
7.20) on a second glass, which is put face to face on the inverted structure's
ITO stripes. The visual impression of the example is very good, mainly
because of very good brightness, which seems to be the result of a novel
CVD process [36]; black stripes along the edges of the various colour stripes
increased the contrast. A filtered white phosphor device was introduced by
248 Thin film electro luminescence devices
R G B
"',, '/'~ '~ITO,'. . ' . . ." "
Glass
Fig. 7.18 Inverted structure with parallax-free filters on top, which can be used as
an etch mask for the transparent electrode stripes. The white or broadband emitter is
shown as composed of two differently doped layers in one host.
Aluminum row
Top insulator
1 st insulator
ITO Column
ALE Barrier
Red filter
Tanaka et al. [32]. The white emission was achieved by double doping of
SrS with Ce 3 + (blue-green) and Eu 2 + (red). The separated colours are
shown together with the 'white' in Fig. 7.21. Rather poor efficiency and
brightness were achieved as transfer from Ce 3 + to Eu 2 + limited the
concentrations of the dopants to a rather low use of the hot electrons [1].
Presumably it is much better to separate the two dopants into layers of the
same film, as shown already in Fig. 7.18. This was tested by Tanaka too, but
dismissed because of colour changes with drive.
Summarizing the section on concepts for colour displays, it appears to be
of primary importance to find out whether the efficiency and/or brightness
of a broad band emission can be made so high that filtering becomes
feasible. The subtractive process of filtering spoils the effective efficiency of
Monochrome and colour 249
10 0 (a)
........
0~
50
c:
0
iii
e
f/)
f/)
c: 0
...
~ (b)
........
::J
~
...>
.iii
....c:
CIl
~
...J
W
for example mixed white by a factor of about 2.5; on the other hand, at least
11m W- 1 should be maintained to avoid excessive heating. There is no
question that the colour-by-filtering concept guarantees much higher pro-
duction yields and lower costs and is most preferable for larger sizes.
White SrS:Ce,K,Eu
U) Blue
.~
c:
~
J:i
.!
...>-
iii Green
...c:
c:
Q)
Red
convincing argument of Mita et al. is that vacancies are avoided with this
active atmosphere.
It was mentioned above, but should be stressed again, that the transfer
process mentioned first by Tanaka et al. in the 1988 SID paper can be
avoided by a physical separation of the two dopants into two or more layers
of the SrS emitter film. A change of the colour coordinates of the white with
voltage is an argument against the application as a white display, not
against colour by filtering. In the latter case different nonlinear brightness-
voltage characteristics result for each colour, but this would be the case
anyway.
C=0.01
0.03
....>-
.iii
c:
....Q)c:
(b)
SrS:Eu(abs.) SrS:Ce,K SrS:Eu
,--!'''
: ",
~---~:-::.--~~
, ,
,, '
,,
,
!
,,
,
-'
Fig. 7.22 (a) The total emission is shifted strongly to the red with increasing dopant
concentration as a result of energy transfer from the Ce 3 + ion in SrS to the Eu 2 +,
because of (b) strong overlap in their emission and absorption bands [32].
0.1 10
Mn-concentration, 1020 cm- J
Fig. 7.23 Dependence of the efficiency of ZnS:Mn TFEL on the Mn concentration
for different deposition techniques [16].
turns out that the best in the total efficiency is not the best in all partials,
which means that it could be improved [16].
However, even more important appears the conclusion that there is no
technology in principle inferior to others. Each deposition process, EBE,
thermal evaporation, d. sputtering and CVD, can be optimized and used to
make highly efficient TFELD. One of the most important parameters in the
process is the substrate temperature, influencing the crystallinity of the
active film. The surface roughness caused by the well-developed grain
surfaces 'penetrating' a geometrical mean surface plane, increases the out-
coupling by scattering, but obviously nlum is increased as well. It is not clear
whether this is caused by a segregation of heavy metals into the grain
boundaries, as tentatively considered by Benalloul et al. [15].
An overview of the various techniques of preparation was given by Ono
[38] at the Helsinki Workshop on EL.
Very recently Mikami et al. [36] proposed a modified CVD process for
the deposition of ZnS:Mn and ZnS:Tb by a so-called HCI bypass flow
process. While Mikami et al. claimed 4lm W - 1 for the yellow emission, in
an obvious application of this method to the Sharp multicolour display [35]
surprising brightnesses were achieved (at 60 Hz): in the filtered green,
80 cd m - 2; in the filtered red, 30 cd m - 2.
Some more standardization appears in the material for the insulator films.
Mostly d.-sputtered Si(O,N) is used; only the Planar division in Finland
uses ALE-grown (AI,Ti) oxide. As will be discussed in section 7.5, the main
influence of the dielectrics is in degradation. Also, the crucial point of the
TFEL structure is the dielectric-semiconductor interface. A constant doping
assumed, any asymmetry in the lightwave emitted under low frequency (e.g.
60 Hz) can be traced to the inadequacy of (at least) one interface. Emission of
Technology and materials 253
the electrons at too Iowan electric field causes a fraction of the charge to be
transferred at energies at which no excitation of luminescent centres is
possible.
As well as efficiency and/or brightness, one further feature is important for
a display product, and this is price, or as a basis for the price, the production
costs. Decisive in this respect is throughput or manufacturing time, automa-
tion of the processes, and production yield. All of these and other factors will
be very dependent on the choice of technology and/or materials. Some of the
materials are much more easy to handle than others; some processes are
better to automate.
In the case where the necessary specifications can be met by virtually all
deposition processes, as for ZnS:Mn [16], costs will become decisive for any
judgement made on new production facilities in the future. The first period
of decision making on availability aspects of production equipment and
training aspects has now ended for ZnS:Mn. It will start again for colour
TFEL.
Materials and technology aspects for colour TFEL at present are rather
diverse. As outlined above, the materials favoured are the good hosts for
rare earths, which embody the loss-free acceleration of electrons at fields
which can be withstood by common dielectrics. These are the alkaline earth
sulphides and selenides.
Numerous techniques have been tried. The main problem is to obtain
stoichiometric, oxygen-free material from SrS and related compounds,
which are difficult to handle. This is all the more true for attempts to
prepare thin films on low temperature substrates. Good crystallinity and
high luminescence efficiency are still to be reported.
Generally speaking, the deposition technique used in most cases for the
alkaline earth sulphides or selenides (AESs) is EBE with coevaporation of
sulphur and deposition at the highest possible temperatures withstood by
the substrate glass, usually 500 0c. It has been developed to some degree of
sophistication at the Tottori University in Japan by Tanaka [39] and
Kobayashi and the latest results [40] have been mentioned before. Further
Japanese groups successful in the field are those of Abe et al. [41] and
Okamoto et al. [27].
CVD and AESs has some tradition. The RCA group published the first
results as early as 1987 [42, 43]. One of the outstanding results of these
efforts was the statement of appreciably different mechanisms of electro-
luminescence in ZnS:Mn and AES.
Following the way prescribed by Suntola [37] for ZnS:Mn, atomic layer
deposition (ALE) of AES was developed in Finland, mainly by Niinisto [44]
and his group at the Helsinki University of Technology. One of the main
concerns in this, as in other CVD processes, is the oxygen content of the
starting and/or doping materials. It appears detrimental to incorporate even
traces of oxygen. On the other hand, most of the usual starting organo
254 Thin .film electroluminescence devices
60
'j
ZnS-:i',
. :{' ~ .
" +
I I
I ,
, I
80 I~' :
[,
r
J. :
o
o 1 2
(a) Field, MV/cm
mV~--~~~~~~~~--~~
o 100
(b) Time, ~s
Fig. 7.24 (a) Current density versus average field in the active film, exhibiting
striking similarities of ZnS, SrS, and CaS, even quantitatively [49]. (b) In samples of
CaS prepared by Ohnishi et al. [50] by r.f. sputtering, trailing edge emission is
marginal, !,!specially at higher transferred charges [51].
7.5 DEGRADATION
For each practical device, degradation is an important aspect. The lifetime
of CRTs has been raised over the last decade to values well above 5000h,
Degradation 257
and certainly all displays will be measured by comparing them with this
value. Since, to our knowledge, no reports on degradation by the manufac-
turers or independent organizations have been published, we will concen-
trate on degradation mechanisms studied on test samples, in general without
any complex matrix organization of a multitude of pixels.
Again the ZnS:Mn system is the best known, and all other colour
materials are much less investigated.
7.5.1 ZnS:Mn
The most elaborate investigation [58] started from the concept of develop-
ing methods which allowed, on the basis of aging or forced aging of an
ensemble of 'identical' samples, one to follow the changes of some important
parameters. These chosen parameters were
1. the voltage division ratio below threshold Cd( C i + Cs),
2. the threshold voltage, and
3. the low light level value of the outer efficiency no.
The total pixel capacitance C was measured and used together with the
voltage division ratio to determine C i and Cs.
'Identical' samples were those cut from the same large substrate or several
substrates manufactured in the same run, i.e. made under identical condi-
tions. Most of the samples were made in a sequence of EBE depositions
without breaking the vacuum, and consisted of one or two layers of Al2 0 3
and Zr02, stabilized by Ca or Y, for each of the insulator films. The whole
lSI stack was deposited on ITO and contacted from the top by AI. ITO and
Al were wet etched after photolithography. Measurements of the parameters
mentioned previously were made (without destruction of the samples)
intermittently while driving the samples at constant transferred charge Q per
pulse at different frequencies between 50 Hz and 5 kHz.
Up to about 2 kHz, depending slightly on Q, the data scaled nicely with
the sum of the transferred charge, a first scaling law for forced aging. About
this frequency or at very high Q, the temperature in the samples obviously
increased, leading to additional effects, which were not followed.
One first surprising result was the changing of Ci mainly in the down-
wards direction, as illustrated by Fig. 7.25 [58]. This meant that either the
permittivity of the dielectric decreased or its thickness increased. It was
found that this behaviour was tied to non stoichiometric oxides containing a
surplus of oxygen. In extreme cases a segregation of oxygen into micro-
bubbles of gas could be observed during operation, as well as the oxidation
of very thin Al films up to the stage of visual vanishing. The last process can
really account for the growing thickness, and surely the dielectric constant
should decrease when ionic conductivity is destroyed. Careful preparation
of stoichiometric insulator films prevented this degradation mechanism.
258 Thin film electroluminescence devices
CIA pF mm- 2
';'0
-50
I
~.
C;/A pF mm- 2
-120
~100
I
~.6 a
~.
r-0.5 1
Fig. 7.25 Changes of the total capacitance C/A, the insulator capacitance Cd A, the
voltage division ratio ot = C;/( C j + Cs), and the threshold field E, with the time under
forced aging at 5kHz, 10- 6 Ascm- 2 [58,59].
7.6 CONCLUSIONS
Thin film electroluminescent devices are well established with respect to
technology, materials, and understanding of the physics (at least in a
linearized model) as far as ZnS:Mn is concerned. Some $10 8 market share in
the tough competition of the LCD-dominated flat panel market of 1991
260 Thin film electro luminescence devices
proves that. Production yields of much better than 75% have been reported
[61]. Multicolour displays based on ZnS:Mn are likely to go into large-scale
production in 1992, and the first steps toward full-colour displays have been
taken in several laboratories. As well as the excellent appearance with
respect to viewing angle, contrast and brightness, the simplicity of produc-
tion in flow-line equipment without the need of costly handling processes
provides a challenge to the older flat panel technologies.
A bright future can be forecast for bright and soon full-colour TFEL
devices.
ACKNOWLEDGEMENT
Gratefully acknowledged are numerous discussions, and the coordination of
the contents of the chapters in this volume with G.O. Miiller.
REFERENCES
1. Miiller, G.O. (1993) In Solid State Luminescence (ed. A.H. Kitai), Chapman &
Hall, London, Chapter 5.
2. Miller, M.R. and Kelley, T.G. (1989) In Electroluminescence, Springer Proceed-
ings in Physics, Vol. 38, Springer, Heidelberg, pp. 259-263.
3. Miiller, G.O., Mach, R., Reinsperger, G.-v. and Schrottke, L. (1991) SID 91
Tech. Digest, p. 567.
4. Uchiike, H. (1989) Electroluminescence, Springer Proceedings in Physics, Vol. 38,
Springer, Heidelberg, pp. 238-245.
5. Miiller, G.O. and Mach, R (1993) Phys. Stat. Sol. Submitted.
6. Kun, Z.K., Leksell, D. and Brandt, G.B. (1992) Proc. 5th Int. Conf on II-VI
Compounds. Tamano. 1991, J. Cryst. Growth, 117, p. 987.
7. LekseIl, D., Kun, Z.K., Asars, J.A., Phillips, N.J., Brandt, G.B., Stringer, J.T. and
Matty, T.e. (1990) 1st Laser Printer Conf.. Colorado Springs. 1990.
8. Miiller, G.O., Mach, R, Halden, E. and Fitting, H.-J. (1990) Proc. XX ICPS,
Thessaloniki. 1990, World Scientific, Singapore, p. 2510.
9. Miiller, G.O., Mach, R., Reinsperger, G.-U. and Schulz, G. (1991) Proc. Int.
Display Research Conf. 1991. San Diego. CA, p. 17.
10. Miiller, G.O., Mach, R, Reinsperger, G.-v., Halden, E. and Schulz, G. (1992)
Proc. 5th Int. II-VI Conf. Tamano.1991, J. Cryst. Growth, 117, p. 948.
11. Kitai, A., Dalacu, N. and Zhizheng, H. (1991) SID 91 Technical Digest, p. 440.
12. Okamoto, S., Nakazawa, E. and Suzuki, T. (1991) Jpn. J. Appl. Phys., 30 (7B),
L1321-L1323.
13. Busse, W., Gumlich, H.-E., Geoffroy, A. and Parrot, R. (1979) Phys. Status Solidi
B, 93, 591-596.
14. Hurd, J.M. and King, e.N. (1979) J. Electron. Mater., 8, 879.
15. Benalloul, P., Benoit, J., Geoffroy, A., Mach, R, Miiller, G.O. and Reinsperger,
G.-v. (1990) J. Cryst. Growth, 101, 989.
16. Mach, R, Miiller, G.O., Reinsperger, G.-v., Benalloul, P., Benoit, J. and
Ohnishi, H. (1990) SID 90 Digest, pp. 238-241.
17. Khomshenko, V.S., Konenec, YF., Vlasenko, N.A., Mach, R, Reinsperger,
G.-v., Selle, B. and Reetz, R (1990) J. Cryst. Growth, 101, 994-998.
References 261
18. Ohnishi, H., Yamasaki, Y. and Iwase, R. (1987) Proc. Soc. In! Disp. 28(4),
345-350.
19. Hiirkonen, G., Hiirkonen, K. and Tornqvist, R. (1990) SID 90 Digest, pp.
232-235.
20. Tanaka, S., Morimoto, S., Samada, K., Kobayashi, H., Zhang, Z. and Jiang, X.
(1992) In Proc. 5th Int. II-VI Con!. Tamano. 1991, J. Cryst. Growth, 117,
997.
21. Heimbrodt, W. and Goede, O. (1986) Phys. Status Solidi B, 135, 795.
22. Mach, R., Reinsperger, G.-D., Miiller, G.O., Selle, B. and Matzkeit, G. (1992),
Proc. 5th Int. II-VI Con!. Tamano. 1991, J. Cryst. Growth, 117, p. 1002.
23. Oseto, S., Kageyama, Y., Takahashi, M., Deguchi, H., Kameyama, K. and
Fujimura, I. (1989) In Electroluminescence, Springer Proceedings in Physics, Vol.
38, Springer, Heidelberg, pp. 191-194.
24. Miiller, G.O. Mach, R. and Ohnishi, H. (1990) In Eurodisplay 90. Proc. 10th Int.
Display Res. Conf.. Amsterdam, pp. 88-91.
25. Ohnishi, H. and Okuda, T. (1989) SID 89 Digest, pp. 317-320.
26. Tanaka, S., Morita, H., Yamada, K. and Kobayashi, H. (1991) IDRC 91, pp.
137-140.
27. Okamoto, S., Nakazawa, E. and Tsuchiya, Y. (1989) In Electroluminescence,
Springer Proceedings in Physics, Vol. 38, Springer, Heidelberg, pp. 195-202.
28. Barrow, W.A., Coovert, R.E., Krug, CN. and Zruchkovski, MJ. (1988) SID 88
Digest, pp. 284-286.
29. Tuenge, R.T. (1989) Electroluminescence, Springer Proceedings in Physics, Vol.
38, Springer, Heidelberg, pp. 132-138.
30. Mita, J., Gao, Q.Z., Kobayashi, M. and Kawamura, K. (1991) SID 91 Digest, pp.
290-293.
31. Konishi, T., Fujita, Y., Minamoto, M., Yamamoto, S. and Tanaka, Y. (1991) SID
91 Digest, pp. 271-274.
32. Tanaka, S., Yoshiyama, H., Nishiuva, 1. and Ohshio, S. (1988) SID 88 Digest, pp.
293-206.
33. Tanaka, S., Kawakami, H., Nakamura, K. and Kobayashi, H. (1989) SID 89
Digest, pp. 321-324.
34. Laakso, C, Khormaei, R. and King, C. (1991) IDRC 91. San Diego 1991, pp.
43-44.
35. Okibayashi, K., Ogura, T., Terada, K .. Taniguchi, K., Samashita, T., Yosheda,
M. and Nakajima, S. (1991) SID 91 Digest, pp. 275-278.
36. Mikami, A., Terada, K., Tanaka, K., Taniguchi, K., Yoshida, M. and Nakajima,
S. (1989) SID 89 Digest, pp. 309-312.
37. Suntola, T. (1990) Proc. 1st Int. Symp. on Atomic layer Epitaxy, Helsinki, in Acta
Poly tech. Scand., Chem. Technol. Metall. Ser., 195, 93-106.
38. Ono, Y.A. (1990) Acta Poly tech. Scand. Appl. Phys. Ser., 170,41-48.
39. Tanaka, S. (1990) Proc. 4th Int. Conf. on II-VI Compounds 1989, in J. Cryst.
Growth, 101, 958.
40. Tanaka, S., Yoshiyama, H., Nakamura, K., Wada, S. Morita, H. and Kobayashi,
H. (1991) Jpn. J. Appl. Phys., 3O(6A), Ll021-Ll024.
41. Abe, Y., Onisawa, K., Tamura, K., Nakayama, T., Honazono, M. and Ono, Y.A.
(1989) Springer Proceedings in Physics, 38, p. 199. Proc. 4th Int. Workshop on
Electroluminescence, Tottori, Japan, 1988. Springer, Berlin.
42. Crandall, R.s. (1987) Appl. Phys. Lett., SO, 641.
43. Crandall, R.S. (1987) Appl. Phys. Lett., 50(10), 551-553.
44. Leppanen, M., Leskelii, M., Niinisto, L., Nykiinen, E., Soininen, P. and Tiitta, M.
(1991) SID 91 Digest, pp. 282-285.
262 Thin film electro luminescence devices
45. Nire, T., Watanabe, T., Tsurumaki, T., Miyakoshi, A. and Tanda, S. (1989),
Springer Proceedings in Physics; 38, p. 218. Proc. 4th Int. Workshop on
Electroluminescence, Tottori, Japan, 1988. Springer, Berlin pp. 218-223.
46. Nakanishi, Y., Suzuki, T., Tatsuoka, H., Kuwabara, H., Fujiyasu, H., Fukuda, Y
and Shimaoka, G. (1990) In Proc. 4th II-VI Conj., Berlin, 1989.
47. Mauch, R.H., Velthaus, K.O., Bilger, G. and Schwek, H.W. (1992) Proc. 5th Int.
II-VI Conj.. Tamano. 1991, l. Cryst. Growth, 117, p. 964.
48. Mach, R. (1990) Acta Poly tech. Scand. Appl. Phys. Ser., 170, 85-98.
49. Muller, G.O., Mach, R., Selle, B. and Ohnishi, H. (1990) l. Cryst. Growth, 101,
999.
50. Ohnishi, H., Iwase, R. and Yamasaki, Y. (1988) SID 88 Digest, pp. 289-292.
51. Mach, R., Ohnishi, H. and Muller, G.O. (1989) In Electroluminescence, Springer
Proceedings in Physics, Vol. 38, Springer, Heidelberg, pp. 176-179.
52. Ando, M, and Ono, Y.A. (1990) l. Appl. Phys., 68(7), 3578-3583.
53. Ando, M. and Ono, YA. (1991) l. Appl. Phys., 69(10), 7225-7232.
54. Takahashi, K., Shibuya, K. and Kondo, K. (1991) Proc. 5th Int. II-VI Con!,
Tamano, 1991, l. Cryst. Growth, 117,979.
55. Hoshina, T. (1980) l. Phys. Soc. lpn., 48(4), 1261-1268.
56. Rennie, 1., Nakazawa, E. and Koda, T. (1990) Jpn. J. Appl. Phys., 29 509.
57. Rennie, 1., Takenchi, K., Kaneko, Y and Koda, T. (1991) J. Lumin., 48--49,
787-791.
58. Mach, R. and Muller, G.O. (1989) In Electroluminescence, Springer Proceedings
in Physics, Vol. 38, Springer, Heidelberg, pp. 264-272.
59. Muller, G.O., Mach, R., Reetz, R. and Reinsperger, G.-v. (1988) SID 88 Digest,
pp.23-26.
60. Tanaka, K., Mikami, A., Ogura, T., Taniguchi, K., Yoshida, M. and Nakajima, S.
(1986) Appl. Phys. Lett., 48(25),1730-1732.
61. King (1990)
62. Onisawa, K.O., Abe, Y., Nakayama, T. and Hanazono, M. (1991) Jpn. l. Appl.
Phys., 30, 314-319.
8
Light emitting diodes:
materials growth and properties
S.P. DenBaars
8.1 INTRODUCTION
The major subject of this book is 'high field' electroluminescence: however,
an equally important class of luminescence is 'low field' injection electro-
luminescence found in light emitting diodes (LEDs). In contrast to the
localized luminescence centres and impact excitation of free carriers in 'high
field' luminescence, excitation in 'low field' injection luminescence occurs by
the injection of minority carriers in semiconductor materials. Current
research interests focus on the development of bright LED sources for full
colour displays. Recently, significant progress in the epitaxial growth of
compound semiconductor materials has been made. This has created new
applications for high brightness LEDs in the outdoor display market. The
purpose of this chapter is to review the physics of LEDs and materials
selection, and to describe the epitaxial growth technologies employed in the
manufacturing of LEDs.
263
264 Light emitting diodes
n-TYPE p-TYPE
Acceptor
Level
Junction
Region
Fig. 8.1 Schematic representation of a p-n junction in a light emitting diode under
forward bias.
spectrum exist. The solid state nature of the device makes it a high reliability
product choice. Average lifetime of an LED is of the order of tens of years
[1]. Some of the commercial applications of LEDs include visible displays,
plastic fibreoptic communications, opto"couplers, motion control, bar-code
scanners, LED printers and even automotive brake lights.
>.
~
Q)
c:
W
c:
e
~
jjj
Momentum
Direct
Fig. 8.2 Energy band diagram of direct and indirect bandgap semiconductors.
Electron energy is plotted versus momentum, k. In the direct semiconductor
radiative transitions are allowed because of conservation of momentum in which
only vertical transitions are allowed. In an indirect semiconductor the electrons in
the conduction band minimum possess a different momentum from the holes and
diagonal transitions are forbidden [1].
Nitrogen and zinc-oxygen pairs are the two most important types of
iso-electronic impurities in indirect bandgap GaP. The energy levels of the
iso-electronic traps lie below the conduction band and therefore the emitted
photon will have an energy approximately equal to the bandgap energy
minus the binding energy of the centre.
where .1no is the initial excess carrier concentration, and !r is the recombina-
tion lifetime. The radiative recombination rate is defined with the excess
carrier density and recombination lifetime as [3]
.1n
Rr = -
!r
The internal radiative quantum efficiency is defined as the ratio of the rate of
radiative recombination '1i = Rr/(Rr+ Rnr) to the total recombination rate.
Therefore, the internal quantum efficiency can be expressed in terms of the
radiative and non-radiative lifetimes:
1/!r
'1. = ---'---
I 1/!r + 1/!nr 1 + !r/!nr
By keeping !r/!nr small a maximum internal quantum efficiency can be
obtained. This is done by making !nr long compared to !r. Growing
materials with low defect densities and eliminating impurities that cause
non-radiative deep levels are certain material improvements that increase
the quantum efficiency. Using the principle of detailed balance [4] in which
at equilibrium the rate of optical generation of electron-hole pairs is equal
to their rate of radiative recombination, the spontaneous recombination rate
for band-to-band emission can be expressed as
Rsp=Brnp
where Br (cm 3 S-1) is the radiative recombination coefficient. Under non-
equilibrium conditions additional carriers are introduced into the material
and the spontaneous recombination rate becomes
Rsp = Br(no + .1n)(po + .1p)
where no and Po are the equilibrium electron and hole concentrations
respectively. Charge neutrality requires
.1p=.1n
and the spontaneous recombination rate becomes
Rsp=Br[nopo +.1n (Po + no) + .1n 2 ]
Therefore the spontaneous recombination rate is a combination of the
equilibrium recombination rate R~p and the recombination rate caused by
the excess injected carriers R~p:
Rsp = R~p + R~p
Physics of LEDs: injection luminescence 267
The equilibrium recombination rate can be expressed 10 terms of the
intrinsic material doping nj:
R~p = Br(nj)2 = BrnoPo
At high injection levels, the excess carrier density will substantially exceed
the background concentration:
An>po+no
so that the average carrier lifetime can be expressed as
1
Tr~ Br An
Since interfaces occur on each side of the active region the interface
recombination velocity is multiplied by 2.
where '1inj is the injection efficiency, '11 is the light generation efficiency, and
'10 is the optical or light extraction efficiency. Since light generation is
predominantly generated on the p side of the diode for GaP [5] and AIGaAs
[6] a high electron injection efficiency is typically desired. This dictates that a
high injection efficiency will be achieved if the electron current dominates
over the hole current. Injection efficiency is defined as the ratio of the
electron current J e to the total forward current J f and can be expressed as
Je Je
'1inj= J +J = J
e h f
From p-n junction theory [4] we can define the injection efficiency in terms
of relevant materials properties:
DnnLp
'1' . = -------'---
In) DnnLp + DppLn
where Dn and Dp are the diffusion coefficients for electrons and holes,
respectively. Lp and Ln are the minority carrier diffusion lengths, and nand p
are the net electron and hole concentrations on the two sides of the junction.
At thermal equilibrium the net electron and hole concentration can be
expressed with the doping level of donors N D and acceptors N A:
n=nf/N A
Physics of LEDs: injection luminescence 269
and
p=nflND
Therefore to obtain a high injection efficiency we want to adjust the doping
such that N D > N A. However, to optimize the optical efficiency the material
must be kept transparent to light generated at the junction. Therefore free
carrier absorption must be minimized by lowering the doping level in the
heavily doped n-type material to below 10 18 cm -3 [6].
AI Contact
~ffUSed p-Type Region
./
n-Type GaAs
....- - Substrate
Au Contact
Fig. 8.3 Construction of a typical homojunction GaAsP LED with a diffused
junction. The p-n junction is formed by a Zn diffusion into the n-type GaAsP
epitaxiallayer. (After Steranka et al. [7].)
270 Light emitting diodes
n side, higher efficiency devices are made by eliminating the minority carrier
injection into the n-type material. By creating a heterostructure at the p-n
interface a valence band discontinuity is formed which provides a hole
potential barrier. By adding an additional heterostructure, confinement at the
other side of the active layer increases because the second heterointerface
prevents the injected electrons from diffusing out of the narrow-gap layer.
Figure 8.4 illustrates the energy band diagram for a p-N heterojunction under
no bias and forward bias. (Capital 'N' signifies the larger energy gap material.)
The conduction band discontinuity A.Ec and the valence band discontinuity
A.Ev produce a barrier to electron and hole flow respectively. The performance
advantages of the double-heterostructure absorbing substrate (DHAS) red
AIGaAs LEDs in comparison with single-heterostructure absorbing substrate
(SHAS) AIGaAs and GaAsP:N LEDs is shown in Fig. 8.5.
No Bias
uui:~l
.......
rev
t......t. c
---Ev
p n
Fig. 8.4 Energy band diagram for a p-N heterojunction under no bias and forward
bias. The conduction band discontinuity flEe and the valence band discontinuity flEv
produce a barrier to electron and hole flow respectively. (After KresseI [3].)
-
Q)
'(3
6.0 DH-AS
;;::::
Q)
-
.g
Q)
E
4.0
~ _____=-SH
... - - - - GaAsP:N
-
0
0 2.0
..c::
a..
0.0
0.1 1.0 10.0 100.0
Forward current (rnA)
Fig. 8.5 Performance (1m A-I) of double-heterostructure absorbing substrate red
AlGaAs LEDs in comparison with that of single-heterostructure AlGaAs and
GaAsP:N LEDs. (After Steranka et al. [7].)
and
PnO =nr /no
The electron-to-hole injection ratio now becomes
Je DnLp nr no
Jh DpLn nr Po
From Fermi-Dirac statistics the electron-hole density product is a function
of the materials energy gap:
np=nr =NcNy exp( -Eg/kT)
where N c and Ny are the conduction and valence band effective density of
states respectively. Therefore the electron-to-hole injection ratio now be-
comes a function of the energy gap difference:
- - - -no-N~N~
J e-_ -DnLp expu (AE /k T )
J h DpLn Po N~N~ g
4
ZnSo
3 r-
>-
~ AlP 0
ZnSe o
o CdS ZnTe
go GaP 0
AIAso 0
1>0
;... 2 I-
CdSe
foil 00 CdTe
~ GaAso A1Sb 0
~ 0
Si 0 InP
1 r-
Ge o oGaSb
InAs o
InSb 8,. Sn
I I I I I
0
2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9
Bond length (A)
Fig. 8.6 Bandgap energy versus bond length for various semiconductors. Note the
general increase in bandgap as the bond length decreases.
Materials selection 273
100
--
~
~
I
I Red, Yellow,
co
--
and
~
( J) DH AIGaAs/AIGaAs I Green AlinGaP
c 10
Q) ~-------
DH AIGaAs/GaAs
E Red Autobmobile Taillight
--------------------
:::J
.......
Q) SH AIGaAs/GaAs
u Red, Yellow,
c Green GaAsP:N
co
E
(;
't:
Q) Blue SiC
a..
I
0.1
1970 1975 1980 1985 1990 1995 2000
Time
Fig. 8.7 Evolution of LED performance (1m W- I ) for the commercially important
materials. (Courtesy Hewlett-Packard Optoelectronics Division.)
levels exceeded 31m W -1. Table 8.1 summarizes the commercially import-
ant visible LED materials currently achievable.
3 25
a...
I ......
" ~ ............
'"
;e r
!!..... 2.5 20
c:
3
==
w
E
::J
2 .
~---;',--..--./
15
s
0
c:
fI)
m
C
0
CII
::J
1.5
I ...... 10
::2:
C')"
iii'
::J
ca
E fl ...... (')
'<
I
~
x
CD
J ....... 5
w 0.5 W
o o
560 570 580 590 600 610 620 630
Wavelength (nm)
Fig. 8.9 External quantum efficiency and luminous efficacy of the newly developed
AlGaInP high brightness LEDs as a function of emission wavelength. (Data from
Kou et al. [14].)
Energy
GaP
Conduction Band
Green Red
Valence Band
Momentum ..
Fig. 8.10 Plot of momentum versus bandgap for indirect GaP materials showing
the iso-electronic trapping levels corresponding to red and green luminescence.
rr:=:::~~~~..-
..-
PH3 + AsH3 + H2 + dopant
H2 +HCI
. . - H2
Fig. 8.11 A vertical multiwafer hydride based vapour phase epitaxy reactor for
growing GaAs,-xP x. (After Burd et al. [21].)
deposition zone,
2GaCI(g)+ H 2(g) + HI - x)AS 4 (g)
T-750C
+!xP 4 (g) , ' 2GaAs 1 - X P X (s)+2HCI(g)
Inspection of the deposition equation indicates that the composition of the
deposited film can be controlled by adjusting the pressure of PH 3 relative to
AsH 3. The typical red GaAsP LED is grown on a GaAs(lOO) oriented
substrate by grading the alloy composition in the first layer from X = 0.0 to
X = 0.4. This buffer layer reduces the stress and defects caused by depositing
the epitaxial layer on a substrate with different lattice constant and thermal
expansion coefficient [22].
Another type of VPE system, halide VPE, uses AsCh and PCh sources in
place of the more toxic hydride sources. The governing chemical reactions
are thought to be as follows:
2PCh(g)+3H2(g)~! P 4 (g) + 6HCI(g)
2AsCI 3(g)+ 3H2(g)~! As 4 (g)+ 6HCI(g)
2HCI 3(g) + 2Ga(l) ~ 2GaCI(g) + H2(g)
Since extra HCI is being generated by the column V halide it is important to
introduce the AsCI 3, or PCh, downstream from the Ga reservoir to avoid
unwanted side-reactions. One advantage of VPE over other growth tech-
nologies, such as LPE or MOCVD, is the ability to etch the surface in situ,
thereby producing a pristine substrate. This is accomplished by eliminating
the hydrogen flow such that the deposition reaction transforms into the
following etching reaction:
2GaCI(g)+ ! As 4 (g) ~2GaAs(s) + Ch(g)
At typical growth temperatures the reaction proceeds to the left, causing a
removal of material from the substrate.
282 Light emitting diodes
In the past few years, MOCD has evolved into a promising technique for
production of visible light emitting opto-electronic devices. Recently, very
high brightness yellow and orange emitting LEDs have been demonstrated
in the AIGaInP materials system [14, 16]. High purity materials and
monolayer abrupt heterojunctions have been demonstrated by this tech-
nique. Because of the economy and simplicity of the technology, MOCVD is
the most viable technique for scale-up to large area epitaxial growth for
commercial applications. MOCVD is a non-equilibrium vapour phase
growth technique that is typically operated at pressures near atmospheric
pressure. In this technique a metal alkyl, such as trimethylgallium (TMGa),
is mixed in the vapour phase with a hydride, such as arsine (AsH 3), and the
reactants are simultaneously injected into the growth chamber where they
thermally decompose to form single-crystal layers.
The MOCVD technique originated from the early research of Manasevit
in 1968 [23] who demonstrated that triethylgallium (TEGa) and arsine
deposited single-crystal GaAs pyrolytically in an open tube reactor.
Manasevit and co-workers [24-26] subsequently expanded the use of this
technique for the growth of GaAs1_yPy, GaAs1_ySby, and AI-containing
compounds. Composition and growth rate are controlled by precisely
controlling the flow rate and dilution of the various components of the gas
stream. The source materials are more often than not pyrophoric and
extremely toxic. This is a special concern when it comes to gaseous sources,
such as the group V elements that are traditionally supplied from their
hydrides: arsine (AsH3) and phosphine (PH 3). Both these sources present a
serious safety problem, since they must be stored in compressed gas
cylinders, which allows for the possibility of catastrophic release. However, a
trend has emerged toward using new group V organometallic liquids, such
as t-butylphosphine (TBP) and t-butylarsine (TBA), which have, until
recently, not been commonly available in sufficient purity for quality
semiconductor growth. The group III sources are almost universally or-
ganometallic liquids, such as trimethylgallium (TMGa) and trimethyl-
aluminum (TMAI). The liquid sources are stored in bubblers through which
a hydrogen carrier gas flows. The carrier gas will saturate with vapour from
the liquid and transport the vapour to the substrate. The substrate rests on a
block of graphite called a susceptor that can be heated by an r.f. coil,
electrically, or radiantly by an infrared lamp heater.
The most common reactor chamber designs are the vertical and horizon-
tal configurations. In the vertical design, reactants are injected through
the top. Typically, the substrate is held flat on a rotating silicon carbide
coated graphite susceptor that is perpendicular to the gas flow direction
as shown in Fig. 8.12. Heating is accomplished by r.f. induction, and
temperature monitoring is accomplished by either an infrared pyrometer or
Growth technologies 283
MFC's (Computer Contro lied)
Vent Flush H2
Main H2
Disilane Computer
Controlled
DEZn
Differential Pressure _ _H r - -
Transducer:~
Susceptor
The exact chemical pathways operative in the MOCVD process are not yet
clearly understood. The basic reaction most commonly used to describe the
growth of the compound semiconductor GaAs by MOCVD is as follows:
Ga(CH3h(v) + AsH3(v)-+GaAs(s) + 3CH 4 (v)
This balanced expression ignores the facts that the specific reaction path and
reactive species are largely unknown. A more likely reaction pathway
leading to growth of the GaAs epitaxial layers involves the homogeneous
decomposition of TMGa as was reported in an earlier study [28, 29]:
Ga(CH 3h(v)+1 H2(v)-+Ga(v)+ 3CH 4 (v)
Arsine is thought to decompose heterogeneously to yield AS 4 and AS 2 at
normal growth temperatures:
+
~ Surface
c:
E kinetic
---
.3 It'limited
...I!!
CD
.1
...~
..r:: Ea = 19 kcal/mol
e
C)
Ea = 22 kcal/mol
.01
0.9 1.~ 1.3 1.5
1000/T(K)
Fig. 8.13 MOCVD growth rate versus reciprocal temperature for epitaxial layers
on (100) and (111) Ga oriented GaAs substrates. (After Reep et al. [30].)
J _ DGaPTMGa
Ga- ob RT
From the above equation we can see that control of the growth rate is easily
obtained by controlling the number of moles of metal alkyls injected into the
reactor. Precise control of the growth rate is then achieved by use of
accurate electronic mass flow controllers that control the flow through the
metal alkyl bubbler. In addition, a uniform boundary layer thickness Ob
ensures uniform deposition.
In the temperature regime 300-550 e, the kinetics of surface reactions
dominates the growth. Reep and Ghandi [30] theorized that the growth
process is dominated by adsorption and decomposition of arsine at the
growing interface. This view is supported by the fact that the activation energy
for the heterogeneous decomposition of arsine on a single-crystal GaAs(l00)
surface of 18 kcal mol- 1 [28] is approximately equal to the activation energy
of 19 kcalmol- 1 observed for growth on the GaAs(l00) orientation. How-
ever, the presence ofTMGa still plays an important role since the growth rate
is strongly influenced by the partial pressure of TMGa in the reactor.
286 Light emitting diodes
~ 0.8 ImPLl
c:
0
~
..;::;
en 0.6
0
a.
E 0.4
0
u
:2
(5 0.2
(f)
0.0
0.0 0.2 0.4 0.6 0.8 1.0
Mole fraction (AI)/[(AI) + (Ga)]
Fig. 8.14 Compositional control in MOCVD is achieved by linear adjustment of
the molar fraction of the constituents.
This equation illustrates one of the greatest assets of the MOCVD growth
process: the distribution coefficient is essentially unity for the column III
elements. This makes the growth of AI-In compounds feasible, which is in
contrast to liquid phase epitaxy where the growth of AI-In compounds is
unachievable because of the large liquid distribution coefficient [31]. For the
growth of AIGaInP, we can write a balanced MOCVD expression as:
Fused - Silica
FumaceTube
Graphite Barrel
Pull
Graphite
.......... ,~Ii~~~H2
Growth Seed
Fig. 8.15 Schematic representation of a multilayer LPE apparatus employing the
sliding graphite boat technique. (After Casey and Panish [9].)
( a) Liquid phase epitaxy - growth rates for step cooling and uniform cooling
There are two very distinct ways in which growth from a liquid solution
can be initiated. The first method involves bringing a substrate into
contact with a melt at equilibrium and then slowly cooling the melt and
substrate. This method is commonly called uniform cooling (UC). The
second method, step cooling (SC), involves the saturation of the melt at
some temperature T hot at which point the temperature is reduced to Teold,
such that the melt becomes supersaturated and the substrate is brought
into contact at the same temperature and serves as a source of nucleation
sites for growth to be initiated. In both these techniques, growth at the
substrate sets up a compositional gradient in the melt by locally depleting
the melt of the more dilute of its constituents. It is the diffusion through
this expanding depletion layer in the melt that determines the growth
rate.
For illustration, we will consider the growth of GaAs from a Ga-rich melt.
It is assumed that the concentration of As is the determining factor, since it
is the more dilute constituent of the melt. The solution will apply in general
to any growth from a melt where a single constituent is the limiting factor
for growth. The amount of solute per unit area M t leaving the solution at the
crystal surface in time t can be obtained by integration of Fick's first law of
diffusion:
M t= ft D[dC~s(Z' t)] dt
o dz z=o
where D is the diffusivity, and C~s is the concentration of As in the liquid.
z=O is then assigned to the substrate surface. The resulting layer thickness,
d, is given by
d= M t
C~s
Growth technologies 289
where C~s is the concentration of As in the solid. Hsieh [32] evaluated M,
by solving the one-dimensional diffusion equation
D d2C~s(z, t) = dC~s(z, t)
dz 2 dt
under the boundary conditions for uniform cooling of C~s(z, 0) = C~s(Thot)
and C~s(O, t)= C~s(Thot)-(R/m)t, where m is given from the slope of the
liquidus curves as
The result is
d= ~ ~ (!!.)1/2 t3/2
3 C~sm 1t
The boundary conditions for pure step cooling are C~s(z, 0) = C~s(Thot) and
C~s(O, t) = C~s(Tcool). The solution is
The result for a combination of step and uniform cooling in the LPE growth
of III-V semiconductor is given by the sum of the two results above:
8.5 CONCLUSIONS
Significant progress has been made in the epitaxial growth and properties of
visible LEDs in recent years. Displays based on LEDs continue to compete
with thin film EL devices. Notable progress has been made in the develop-
ment of AIGaInP and AIGaAs efficient emitters. Efficiencies of these
materials have achieved levels high enough for large panel outdoor display
applications. An efficient blue emitter is required for full colour displays
based on LEDs to become viable. ZnSe, GaN and SiC are all being
developed as sources for blue LEDs. Currently, GaAsP and GaP are still the
dominant materials used in the marketplace. Newly developed materials
such as AIGaAs and AIGaInP will replace the existing LED materials as
production costs are reduced.
REFERENCES
1. Craford, M.G. (1977) IEEE Trans. Electron Devices, 24, 935.
2. Henry, c.H. (1976) J. Luminescence, 12-13, 47.
3. Kressel, H. (1977) Semiconductor Laser and Heterojunction LEDs, Academic
Press, New York.
4. Shockley, W. (1950) Electrons and Holes in Semiconductors, Van Nostrand, New
York.
5. Bergh, A.A. and Dean, P.I. (1976) Light Emitting Diodes, Oxford University
Press, London.
6. Varon, 1., Mahieu, M., Vandenberg, P., Boissy, M. and Lebailly, I. (1981) IEEE
Trans. Electron. Devices, 4, 416.
7. Steranka, F.N., DeFevere, D.c., Camras, M.D., Tu, C.W., McElfresh, D.K., Rudaz,
S.L., Cook, L.W. and Synder, W.L. (1988) Hewlett-Packard J., (August), 84.
8. Capasso, F. (1987) J. Appl. Phys., 24, 319.
9. Casey, H.C. and Panish, M. (1978) Heterostructure Lasers, Academic Press, New
York.
References 291
10. Tsai, M.J. (1984) Proceedings of the 11 th IntI. Symp. on Gallium Arsenide and
Related Compounds, Institute of Physics Publishing, Bristol, p. 26.
11. Cook, L.W., Camras, M.D., Rudaz, S.L., and Steranka, F.M. (1987) Proceedings
of the 14th Inti. Symp. on GaAs and Related Compounds, Institute of Physics
Conf. Series, 91, Institute of Physics Publishing, Bristol, p. 777.
12. Kellert, F. and Moon, RM. (1986) J. Electron. Mater., 15, 13.
13. Ikeda, M., Nakano, K., Mori, Y., Kaneko, K. and Watenabe, N. (1986) Appl.
Phys. Lett., 48, 89.
14. Kuo, e.P., Fletcher, RM., Ostenowski, T.D., Lardizabal, M.C., Craford, M.G.
and Robbins, V.M. (1990) Appl. Phys. Lett., 57, 2937.
15. Bour, D.P., and Shealy, J.R. (1988) IEEE J. Quant. Electron. 21, 1856.
16. Sugawara, H., Ishikawa, M. and Hatakoshi, G. (1991) Appl. Phys. Lett., 58, 1010.
17. Campbell, J.e., Holonak, N., Craford, M.G. and Keune, D.L. (1974) J. Appl.
Phys., 45, 4543.
18. Hase, M.A., Qiu, J., DePuydt, J.M. and Cheng, H. (1991) Appl. Phys. Lett., 59,
1272.
19. Akasaki, I., Amano, H., Hiramatsu, K., Koide, N. and Sawaki, N. (1987)
Proceedings of the 14th IntI. Symp. on GaAs and Related Compounds, Institute of
Physics Conf. Series, 91, Institute of Physics Publishing, Bristol, p. 633.
20. Amano, H.K., Kito, M., Hiramatsu, K., Akasaki, I. (1989) Jpn. J. Appl. Phys., 28,
L2112.
21. Burd, J.w. (1969) Trans. Metall. Soc. AIME, 245, 571.
22. Abrahams, M.S. (1969) J. Mater. Sci., 4, 223.
23. Manasevit, H.M. (1968) Appl. Phys. Lett., 12, 156.
24. Manasevit, H.M. and Simpson, W.I. (1969) J. Electrochem. Soc., 116, 1725.
25. Manasevit, H.M., (1971) J. Electrochem. Soc., 118, 647.
26. Manasevit, H.M. and Hess, K.L. (1979) J. Electrochem. Soc., 126,2031.
27. Giling, L.1. (1982) J. Phys. (Paris), Colloq. C5, 235.
28. DenBaars, S.P., Maa, B.Y., Dapkus, P.D., Danner, A.D. and Lee, H.e. (1986) J.
Cryst. Growth, 77, 188.
29. Nishizawa, J., Abe, H. and Kurabayashi, T. (1985) J. Electrochem. Soc., 132,
1197.
30. Reep, D.H. and Ghandi, S.K. (1983) J. Electrochem. Soc., 130,675.
31. Yuan, J.S., Hsu, e.e., Cohen, RM. and Stringfellow, G.B., (1985) J. Appl. Phys.,
57, 1380.
32. Hsieh, J.J. (1974) J. Cryst. Growth, 20, 49.
9
Atomic layer epitaxy of phosphor
thin films
Brian W. Sanders
9.1 INTRODUCTION
293
294 Atomic layer epitaxy of phosphor thin films
deposition methods that result in atomic layer growth of a film. This has
resulted in some confusion in accessing the literature.
9.1.4 Terminology
The terms successive ion adsorption and reaction [18], molecular layer
epitaxy [20], digital layer epitaxy [21], electrochemical atomic layer epitaxy
[19], flow-rate modulation epitaxy (FME) [22], and modulated molecular
beam epitaxy [23] have all appeared in the literature to describe thin film
deposition methods relying on the alternating exposure of reactants to a
substrate resulting in film growth of one monolayer per cycle. The prepon-
derance of these terms has become confusing especially when, often as not,
the films resulting from the growth techniques are polycrystalline and
therefore do not fit the term epitaxy.
The techniques of flow-rate modulation epitaxy and modulated molecular
beam epitaxy are a different class of technique. They do result in a net
growth of one atomic layer per cycle of a material, but they do not rely on
self-limiting reactions at the surface of the growing film. They rely on simply
limiting the amount of reactant supplied to the substrate at each step to only
that which is necessary for monolayer formation. In this way a purging step
is not required for the film growth. Although these deposition techniques do
296 Atomic layer epitaxy of phosphor thin films
result in good control over film growth, they will not be discussed in this
chapter. The reader is referred to the references if further information is
desired. For those techniques that rely on the historical definition of atomic
layer epitaxy, a new naming system has recently been suggested.
Most of the names mentioned above contain the term epitaxy, which is
usually used to describe the growth of single-crystal films. This has resulted
in some confusion as many films prepared by ALE are polycrystalline or
amorphous. Since the term is meant to describe the way in which the film
grows, not the crystallinity, it has been suggested by Leskela [14] that the
term atomic layer deposition (ALD) be adopted for the production of
polycrystalline films by this method of deposition. A suitable prefix can be
used to illustrate the primary deposition method. Using this terminology,
chemical vapour atomic layer deposition (CV ALD) would be used for the
deposition of a polycrystalline material through the exposure of a substrate
to alternating streams of chemical vapours. Conversely, molecular beam
atomic layer epitaxy (MBALE) refers to the production of single crystal thin
films by exposing a substrate to alternating elemental molecular beams. The
suggestion seems reasonable. If it is widely adopted it should make the
literature on this subject much more consistent and access less confusing.
This terminology will be adhered to by the author from now on when a
specific deposition technique is being explained. The more general terms
ALE and ALD will be used when the specific growth method does not
matter in the discussion.
IIIIIIIIII
1
IIIIIIIIII
2
IIIIIIIIII
@@@@@@@@@
Fig. 9.1 Schematic of ALD using elemental starting materials. (After Suntola and
Hyvarinen [24].)
Figure 9.2 shows an energy diagram for the absorption of an atom to the
surface of a solid. Excluding the possibility of condensation, below some
temperature Tl only physisorption occurs. Thin film growth is influenced
greatly by the substrate temperature and time between reactant pulses as
only a weak association occurs between the surface and any adsorbed
298 Atomic layer epitaxy of phosphor thin films
>
...CD
C)
C
CD
I
.;;
C
....oCD
0..
Chemisorption
atoms. The whole monolayer would desorb given enough time. Between the
temperatures Tl and T2 actual ALD occurs as the molecules or atoms will
adsorb but not desorb from the surface (chemisorption). Additional mono-
layers do not form owing to the differences in bond energies between like
and unlike elements. Above T2 the growth is again dependent on tempera-
ture and time between pulses as atoms are desorbing from the surface. ALD
only occurs when chemisorption is the sole process occurring. To attain this,
substrate temperature must be adjusted to within some optimum range.
Extending this model further to cover compound reactants results in Fig.
9.3. This figure describes processes that compete with ideal ALD when
compound starting materials are used. If the temperature is too low,
desorption and/or incomplete decomposition of the reactants occurs and an
uncontrolled growth results (region B). Condensation of the reactants may
occur, which results in a higher than ideal growth rate (region A). If the
temperature is too high, then re-evaporation of elements from the surface
may occur (region D) or, for compound reactants, undesirable decomposi-
tion or side reactions may result (region C). ALD is expected to occur only
over a limited temperature range [7].
This temperature range must be high enough to prevent condensation and
to overcome the activation energy to chemisorption, but low enough to
prevent unwanted reactions or re-evaporation. Thus, the ratio between the
vaporization temperature of source elements and the ZnS film determines
the width of the ALD 'window'. It is therefore expected that using com-
pounds with low vaporization temperatures, instead of elemental source
Theory of ALE 299
!II 2
!II
!II G)
!II C
G).,lIi:
.i .!:!
.!:! -E
Ideal ALE
-E :D 1
iij r; or ALD
~o
o C
<C 0
~
Temperature
Fig. 9.3 Schematic showing the relationship between ideal ALD growth and
competing processes with temperature when compound reactants are used. (After
Suntola [7].)
materials, will increase the temperature range over which ideal growth will
occur.
~~~~I/~~
0~ 0+00 70~0 70~
0
-
0
0-
0
0
_0
(1)
0- 1
~~I/~~~L/
0000Gr0
~070 15
cfJ 00 cfJ 00
~0cfJ 0
3
'~~L / ,
0000000 ,! L/
Fig. 9.4 Ideal ALD from compound reactants (after Suntola and Hyvarinen [24]).
Also shown are two possible branches that still could result in ideal monolayer per
cycle deposition.
zinc [15] and hydrogen sulphide does not always produce one monolayer per
cycle growth. The side groups on the zinc compounds take up space on the
surface of the monolayer and do not let a completed monolayer form. Thus,
less than one monolayer per cycle is observed. In the case of dimethyl zinc, only
about 80% of a monolayer forms during the film growth. When zinc chloride is
used, values as low as 33% of ideal ALD have been observed [14].
A theoretical treatment of this behaviour has been published by Pak-
kanen et al. [25] (it has also appeared in a review by Suntola [41]). The
authors use Hartree-Fock calculations for determining the thermodynamics
of absorption of dimethyl zinc on the (111) planes of zinc sulphide for
growth in the [111] direction (the direction most ZnS films deposited by
ALD grow). They discovered that, if molecules adsorb singly on the surface,
then only one-third of the surface may be covered with the zinc compound
at any time. The large chlorine atoms prevent the zinc atoms from getting
close enough to one another for a complete layer to form. Their calculations
also show that it is energetically favourable for the zinc chloride to chain
across the surface. When this occurs, 50% or even 100% of an ideal
monolayer can form from the zinc atoms. A combination of the above
reaction schemes can be used to explain almost any fraction of ideal growth
encountered. This model also might be applicable to the use of dimethyl
zinc, as a methyl group is approximately the same size as a chlorine atom.
Equipment used to deposit thin films by ALD 301
Substrate
InniC-' Iitehambe,
ttt
Reactants To Pump
t
or
-
Atmosphere
Substrate
To Pumping
System
- I
I
6:+:-8
I
I I I
I I
I I
I I
Shutters
WWW
Effusion
Cells
Fig. 9.5 Schematic of CVALD (top) and MBALE (bottom) systems. (After Suntola
and Hyvarinen [24].)
302 Atomic layer epitaxy of phosphor thin films
This type of system has also been used at ambient pressures [15J using
dimethyl zinc and hydrogen sulphide as reactants. The reactant gases in this
case have a very high vapour pressure, so high in fact that the dimethyl zinc
source had to be cooled to 273 K to lower its vapour pressure. A disadvan-
tage of this CVD application is the extremely reactive nature of the
organometallic compound. Here, the authors passed the exhaust gases
through a high temperature furnace to crack any unreacted organometallic
compound.
The bottom diagram of Fig. 9.5 is a simplified schematic of a MBE system
that could be used for ALD. A series of computer-controlled shutters allow
the necessary alternating exposure of the reactants to the substrate to occur.
A unique application of the ALE technique has been made by Gregory et
al. [19]. They have successfully grown CdTe by electrochemical ALE
(ECALE). Although this material has not found use as a phosphor, this
variant on a well-known deposition process deserves to be covered here
since many other II-VI compounds of interest to the phosphor field have
been deposited using conventional electrodeposition. A chamber similar to
the one used by these authors is pictured in Fig. 9.6. To effect ALE, these
authors change solutions repetitively during the experiment. Each solution
only contains one element necessary for film deposition. The solution
exchange is accomplished by blowing spent solution out the bottom of the
chamber using compressed nitrogen. The nitrogen flow is then shut off and
the second solution is fed into the electrode chamber through pinholes with
the use of capillary action. When this solution has been reacted it is blown
out with nitrogen and the first solution is then drawn into the electrode
chamber. ALD is possible in this system because of the phenomenon of
Electrode 0.001 in
Gap
Fig. 9.6 Schematic of a deposition chamber used for ECALE. (After Gregory et al.
[19].)
Compounds deposited by atomic layer deposition 303
Reactants
Fig. 9.7 Schematic of ALE deposition systems for solution (left) (after Nicolau
[26]) and chemical vapour (right) (after Bedair et al. [27]) deposition techniques.
The last type of apparatus (pictured on the right-hand side of Fig. 9.7) to be
discussed is a variant of the CVD-type system presented earlier. Instead of
valving the source gases the substrate is rotated in and out of two or more
reactant gas streams. This system has been used in the past to produce films of
GaAs using streams of trimethyl gallium and arsine as reactant materials [28].
alkaline earth sulphides. GaAs has been the most widely studied III-V
compound studied to date but some of its alloys with aluminium have also
been grown. ALD techniques have also produced many high quality oxides
useful for luminescent displays. It is also not surprising that, given the
wealth of compounds grown, complete electroluminescent devices have been
grown inside ALE systems without exposure to atmosphere.
Given the number of compounds grown, an all-encompassing summary of
the characteristics of each compound grown cannot be made. Most of the
compounds useful as phosphors that have been grown by ALD will be
covered here. However, because of the vast number of starting materials
available, and the number of research groups working on thin film growth
by ALD, only a small fraction of their results can be covered owing to space
limitations. It is hoped that the following summary will at least give a
flavour for the diversity of films produced and the ingenuity that researchers
have used to adapt conventional thin film growth techniques to produce
ALD of a material.
A word of caution should be issued at this point. Many of the compounds
mentioned in this chapter are toxic to humans in even small quantities.
Some materials are also very reactive with air or water, creating enough heat
to start a fire or releasing noxious and toxic fumes. For these reasons, the
reader is advised that he or she should obtain the necessary information on
the safe handling and use of these compounds before any experimentation
begins.
(a) ZnS
Historically ZnS was the first compound deposited by ALE. It has been one
of the most widely studied thin film materials ever since (of materials grown
by ALD, that is). Since this is a book on luminescent materials, it is
fortunate that so much is known about growing this compound by ALD
since ZnS-based thin film phosphors are one of the most promising
materials available for display of information.
ZnS has been grown from elemental as well as compound reactants.
Compound metal sources have included ZnCI 2 , Zn(CH 3 COOHh, dimethyl
and diethyl zinc, and some furan adducts of zinc to name a few. Usually, the
sulphur adduct of choice has been H 2 S.
The very first implementation of ALE was the growth of ZnS from
elemental Zn and S. The material was of superior crystal quality despite
being grown at relatively low temperatures (593-643 K) [1].
Growth from elements is simpler than from compounds since only the
desired elements are used in the growth avoiding sometimes complicated
and unwanted side reactions. However, because of equipment design, or if
Compounds deposited by atomic layer deposition 305
by this technique is the entrapped voids that are revealed in TEM images
[18].
(c) ZnSe
Yoshikawa et al. [36] have grown ZnSe using MBALE using dimethyl zinc
and hydrogen selenide as reactants. Ideal monolayer per cycle growth was
restricted to a substrate temperature range of 473-548 K. They also noted a
dependence of the growth rate on the dimethyl zinc flow rate with less than
monolayer per cycle deposition at low flow rates but a limit of one
monolayer per cycle at higher flow rates. This indicates that not enough
time was allowed at the low flow rates for a monolayer of Zn to form.
Yoshikawa et al. [36] also produced films by CV ALE with the same
reactants as given above. In this case it was found that ideal ALE-type
growth could only be accomplished with a hydrogen selenide flow rate at
least twice that of the MBALE growth and if a much longer pulse duration
was used. This increased the time of deposition significantly.
Despite the fact that films prepared by each technique were grown in an
ALE mode there were still marked differences in the films. The MBALE
films were much smoother in appearance than the CV ALE films. The
MBALE films also showed superior photoluminescence spectra with more
intense and better-defined peaks.
(e) ZnO
Sanders and Kitai [37J have grown zinc oxide by ALCVD. The process used
dimethyl zinc and the trace oxygen or water contained in the nitrogen purge
gas. Slightly less than one monolayer growth per cycle was exhibited for
growth temperatures ranging from 473 to 573 K. Rutherford back scattering
analysis indicated that good stoichiometric films resulted from this tech-
nique. Despite a very fine-grained material (grain size < 300 A), reasonable
resistivities (0.005 n cm) were obtained. Growth was primarily in the (0002)
direction. These films would be suitable as transparent conductors in a
variety of thin film luminescent applications.
found in the films under Auger analysis despite the high content of these
elements in the whole complex. X-ray diffraction patterns of the films
revealed that the three films grew in three different orientations. CaS grew
primarily along the (220) direction, SrS grew (111) and BaS grew along (200).
These different growth directions may be a result of the apparently
complicated growth mechanism proposed for thin film formation. Mass
spectrometric studies show that these complexes maintain their integrity in
the gas phase, and thus molecules adsorbing on the surface of a growing film
are large. They are so large, in fact, that it is impossible to generate a
complete monolayer in one growth cycle. Only 0.3-0.4 A per cycle are
achieved. This means that, after the first cycle, the incoming molecules see
several nonidentical sites in which to adsorb. Thus, different growth direc-
tions will occur according to the most energetically favourable adsorption
si te filling.
(a) GaAs
Growth of GaAs by CV ALD techniques has rapidly expanded since the
original paper presented by Bedair et al. in 1985 [27]. Most researchers use
trimethyl gallium and arsine with a hydrogen carrier gas when depositing
films by this technique. The dominant chamber used is similar to the one
pictured in Fig. 9.5. The substrate is rotated alternately between two or
more gas streams. Using such a system ideal ALE growth is routinely
obtained for a wide range of growth temperatures and flow rates [40].
When diethyl gallium chloride is used as the group III source a wide
deposition window is also present. Ideal monolayer growth is maintained
through a temperature range of 673-873 K for over 3000 cycles. The system
also maintained ideal ALE growth over a wide range of exposure times [40].
(b) AIGaAs
A report on the growth of this compound by CV ALE using trimethyl
gallium, trimethyl aluminium and arsine as source materials has been
Compounds deposited by atomic layer deposition 309
(a) Al 2 0 3
Thin film aluminium oxide was one of the first oxides prepared by ALD [5].
Thin films which were pinhole free suitable for use as ion barrier or dielectric
and passivation layers in electroluminescent devices were produced. The
reactants used were AICl 3 and H 2 0. Amorphous layers were produced
despite a substrate temperature of 720 K.
(c) MgO
High quality MgO thin films deposited on (111) silicon substrates by
CV ALD are formed by a hydrolysis surface reaction of Mg(C 2 Hs hand
H 2 0. The growth orientation ofMgO changes from (111) to (100) when the
310 Atomic layer epitaxy of phosphor thin films
temperature of the silicon substrate changes from 600C to 900 0c. The
growth orientation difference of MgO grown by CV ALD is rationalized in
terms of the surface reaction in CVALD [17].
9.5 SUMMARY
The technique of ALE has been applied successfully to the deposition of
both single-crystal and polycrystalline thin films of II-VI and III-V
semiconductors as well as a few oxide dielectrics. The technique is easily
adapted to most conventional deposition methods and has found wide-
spread use for the production of thin films.
Because of the nature of the ALE technique, large substrate areas can be
covered homogeneously without the need for substrate rotation. Perfect
layer-by-layer growth has been realized by this technique, making it an ideal
method for producing superlattice structures.
Perhaps the most severe restriction of the technique is its low growth rate
when compared with other methods of film deposition. The high crystal
quality of films resulting from this growth method can more than compen-
sate for the longer deposition times. In fact, using this technique, structures
that cannot be grown conventionally are possible. Atomic layer deposition
techniques have become, and probably always will be, a valuable tool for
thin film phosphor growth.
REFERENCES
1. Suntoia, T. and Antson, J. (1977) US Patent 4,058,430.
2. Suntoia, T., Antson, M.J., Pakkaia, A. and Lindfors, S. (1980) SID 80 Digest,
p.108.
3. Suntoia, T., Pakkaia, A. and Lindfors, S. (1983) US Patent 4,413,022.
4. Herman, M.A. (1991) Vacuum, 42(1-2), 61-66.
5. Goodman, C.H. and Pessa, M.V. (1986) J. Appl. Phys., 60(3), R65-R81.
References 311
6. Peng, Y. (1990) Wuli, 19(12), 733-738.
7. Suntola, T. (1989) Acta Poly tech. Scand., Ser. Electr. Eng., 64, 242.
8. Niinisto, L. (ed.) (1990) Acta Poly tech. Scand., Chern. Metall. Ser., 195.
9. Wantanabe, H., Mizutani, T. and Usui, A. (1990) In Very High Speed Integrated
Circuits: Heterostructure, (ed. T. Ikama), Semiconductors and Semimetals, Vol. 30,
Academic Press, San Diego, CA, pp. 1-52.
10. Sun to la, T. and Simpson, M. (eds) (1990) Atomic Layer Epitaxy, Blackie,
Glasgow.
11. Tanninen, v.P., Oikkonen, M. and Tuomi, T. (1981) Phys. Status Solidi A, 67(2),
573-583.
12. Lahtinen, lA., Lu, A. and Tuomi, T. (1985) J. Appl. Phys., 58(5), 1851-1853.
13. Wu, Y., Toyoda, T. and Kawakami, Y. (1990) Jpn. J. Appl. Phys., Part 2, 29(5),
L727-L730.
14. Leskela, M. (1990) Acta Poly tech. Scand., Chern. Metall. Ser., 195, 67-80.
15. Hunter, A. and Kitai, A.H. (1988) J. Cryst. Growth, 91,111-118.
16. Tammenmaa, M., Antson, H., Asplund, M., Hiltunen, L., Leskela, M. and
Niinisto, L. (1987) J. Cryst. Growth, 84, 151-154.
17. Huang, R. and Kitai, A.H. (1992) App/. Phys. Lett., 61, 1450-2.
18. Nicolau, Y.F., Dupuy, M. and Brunei, M. (1990) J. Electrochem. Soc., 137(9),
2915-2924.
19. Gregory, B.W., Suggs, D.W. and Stickney, lL. (1991) J. Electrochem. Soc., 138(5),
1279-1284.
20. Nishizawa, J., Abbe, H. and Kurabayashi, T. (1985) J. Electrochem. Soc., 132(5),
1197-1200.
21. Usui, A. and Wantanabe, H. (1987) Proc. Electrochem. Soc., 87(8), 766-775.
22. Kobayashi, N., Makimot, T., Yamauchi, Y. and Horikoshi, Y. (1990) Acta
Poly tech. Scand., Chern, Metall. Ser., 195, 139-145.
23. Gerard, J.M., Marzin, lY. and Jusserand, B. (1991) J. Cryst. Growth, 111,
205-209.
24. Suntola, T. and Hyvarinen, 1 (1985) Annu. Rev. Mater. Sci., 15, 177-195.
25. Pakkanen, T.A., Nevalainen, v., Lindblad, M. and Makkonen, P. (1987) Surf
Sci., 188(3), 456-474.
26. Nicolau, Y.F. (1985) Appl. Surf Sci., 22-23, 1061.
27. Bedair, S.M., Tischler, M.A., Katsuyama, T. and EI-Masry, N.A. (1985) App/.
Phys. Lett., 47(1), 51-53.
28. Bedair, S.M. (1990) Acta Poly tech. Scand., Chern. Techno/. Metall. Ser., 195,
17-37.
29. Pes sa, M., Makela, R. and Suntola, T. (1981) Appl. Phys. Lett., 38(3), 131-132.
30. Oikkonen, M., Blomberg, M., Tuomi, T. and Tammenmaa, M. (1985) Thin Solid
Films, 124, 317-321.
31. Dryburgh, P.M. (1988) J. Cryst. Growth, 87(4), 397.
32. Tornqvist, R., Antson, l, Skarp, J. and Tanninen, V.P. (1983) IEEE Trans.
Electron Devices, 30(5), 468-471.
33. Sutela, T. (1984) Displays, 5(2), 73-78.
34. Tammenmaa, M., Koskinen, T., Hiltunen, L., Leskela, M. and Niinisto, L. (1984)
VTT Symp. Ser., 54, 18-21.
35. Tammenmaa, M., Leskela, M., Niinisto, L. and Koskinen, T. (1986) J. Less-
Common Met., 126, 209-214.
36. Yoshikawa, A., Okamoto, T., Yasuda, H., Yamaga, S. and Kasai, H. (1989) J.
Cryst. Growth, 101, 86-90. .
37. Sanders, B.W. and Kitai, A.H. (1991) Proc. SPIE-International Society of Optical
Engineering, 1398,81-87.
312 Atomic layer epitaxy of phosphor thin films
38. Ohnishi, H. (1989) Annu. Rev. Mater. Sci., 19, 83-101.
39. Dapkus, P.D., Maa, B.Y., Chen, Q., Jeong, W.G. and DenBaars, S.P. (1990) Acta
Poly tech. Scand., Chern. Technol. Metall. Ser., 195, 39-54.
40. Tischler, M.A. and Bedair, S.M. (1990) In Atomic Layer Epitaxy (eds T. Suntola
and M. Simpson), Blackie, Glasgow, pp. 110--154.
41. Suntola, T. (1989) Mater. Sci. Rep., 4, 261-312.
10
Lamp phosphors
T.E. Peters, R.G. Pappalardo and R.B. Hunt, Jr
313
314 Lamp phosphors
Ionization
10
> Q)
c:
>-
....
C)
Q) 5
c:
Q)
Q)
>
Q) 185 nm
....J
o~--~~-------------------------
6e So)
Fig. 10.1 Lowest electronic energy levels for neutral mercury.
550
0.7
560
0.6
500 570
G
>- 580
t 0.5
590
0.4
4000 "'.
tJ.- ~ 600-
8
~,3000-2P90 T~610
~490 10000 ~ ,~, 15~0~7"" 620
r 5000 1000
0.3
~q~o ~~ 6.50
I~
~~ 8000-
Absolute
temperature 670
(K)
0.2 '0
480
0.1 B"
470
460'" 450
~400
... 0.5
....
CI)
e0
CI)
0.4
a
...=
CI)
0.3
Q.
....
....
<II
0.2
as
~ 1\ : .,"''\....
J\
0.1
-j\ --'
~~ ~,
I
o r-..J ~
350 400 450 500 550 600 650 700 750
Wavelength (nanometers)
Fig. 10.3 SPDs for two fluorescent lamps with identical color points on the BBL,
at 4100K.
Bril and de Jager-Veenis [9J only report conservatively that the quantum
yield is bracketed between 40% and 60%. With this proviso in mind it can
only be stated that the quantum yield for 254 nm excitation in the traditional
lamp phosphors is in the 60-70% range. In the case of the novel rare-
earth-activated phosphors the quantum efficiency moves into the 80-90%
range, and finally for Y203: Eu the quantum efficiency may reach '" 95%.
In the context of lamp manufacturing the phosphor quantum yield is a
useful, but not a final, indicator of phosphor performance in lamps. As
discussed in later sections, both the processing of the phosphor during lamp
fabrication and exposure of the phosphor to the plasma during lamp
operation may seriously lower its conversion efficiency. From a device
standpoint the basic question in comparing the lamp performance for
phosphors emitting in different spectral regions is, having selected a lamp
type, namely a specific plasma, how many (visible) photons does one obtain
from the lamp by using a specific phosphor, or a specific blend of phos-
phors? The lamp output is normally available in the form of the SPD for
100 h of operation, for reasons explained in section 10.1.8. It is a simple
matter to convert this information in a photon-per-nanometer format, and
hence to arrive at the total photon content of the SPD. One could then
derive an absolute value of the 'lamp' quantum yield for the phosphor, if the
number of UV photons emitted from the plasma was also reliably known.
This unfortunately is not the case at present. Still, it will be a useful exercise
to rank lamp performance according to relative lamp quantum yield, as is
done in Table 10.1.
In this table Y203: Eu stands out for its superlative performance, followed
by the Ce-Tb-activated hex a-aluminate. These phosphors will be discussed
Table 10.1 Relative lamp quantum yield of commercial phosphors, as derived from
the SPDs of 40 W lamps at 100 h of operation (SPD from 380 nm to 760 nm)
......
CII
CII
0.15
S
0
ar:: A
...
V\
0.1
CII
g,
......
In
1\
0.05
~
.."
~v- J ~
o
300 400 500 600 700 800
Wavelength (nanometers)
Fig. 10.4 SPD of a Sylvania 'cool white' fluorescent lamp (40 W 1.5 in diameter).
322 Lamp phosphors
CaHP0 4 3.0
CaC0 3 1.205
CaF2 0.425
NH4Ci 0.180
MnC0 3 0.089
Sb 2 0 3 0.038
CdO 0.06
Total Ca 4.630
Total Ca+Cd+Mn+Sb 4.855
Total F+Cl 1.035
Phosphors for low-pressure Hg plasmas 323
Making the reasonable assumption that Cd, Sb, and Mn substitute for Ca
in the apatite lattice, and that Cl substitutes for F, it would appear that the
reagent blend described in Table 10.2 could not yield a stoichiometric
halophosphate. In particular the blend is cation deficient and has an excess
of halide.
In effect, early workers found that phosphors prepared from blends
having precisely correct cation:phosphate and halide:phosphate mole ratios
gave rise to poorly luminescent materials.
This provided the impetus for studies that employed differential thermal
analysis (DTA), thermogravimetric analysis (TGA), electron paramagnetic
resonance (EPR), X-ray diffraction, optical spectroscopy, and total chemical
analysis to follow all of the processes occurring during phosphor formation
[14-19]. It was possible to identify precisely the temperatures at which the
raw materials thermally decomposed or reacted, what transient phases were
formed, which phases acted as carriers for Sb and Mn, how the furnace
atmosphere modified the processes, at what point apatite was obtained, and
what undesirable phases might remain in the product.
Some of the more pertinent results of these studies are worthy of further
comment. The CaF AP can be prepared at temperatures as low as 700-
800C, but optimum crystallinity and brightness requires temperatures of
1000-1200C. The synthesis of CaFAP is a dynamic process, wherein the
raw materials undergo thermal decomposition or react to form new phases
that are themselves later consumed in forming the phosphor. Any change in
the raw materials, heating rate, furnace atmosphere, batch size, powder bed
depth or furnace loading will modify the characteristics of the phosphor
obtained. To maintain chromaticity and brightness within specifications in
industrial-scale syntheses, formulae have been developed that actually adjust
blend composition to compensate for raw-material changes [20].
It was also learned that a substantial part of the antimony, typically
about half of the added amount, is lost during the firing process, together
with some of the halide and cadmium. The antimony is probably volatil-
ized in the form of Sb 20 3 or SbCh. The use of cation-deficient blends
helps to minimize contamination of the phosphor by minor phases detri-
mental to phosphor brightness. The compounds Ca(Sb0 3h and ~
Ca3(P04h are particularly objectionable, while Ca2P207 appears to be
relatively harmless. Ca(Sb0 3h robs the apatite phase of needed antimony,
thereby reducing its UV absorption, and at the same time introduces a
parasitic UV absorption of its own. The ~-Ca3(P04h:Mn phase reduces
brightness because of a visible absorption (pink body color) caused by
oxidation of Mn [15]. Formation of the orthophosphate is influenced by
the actual reagent mix, and is more likely to be observed in fluoroch-
loroapatite phosphors that have been exposed to air or moisture during the
high-temperature synthesis. In effect, an inert atmosphere is needed for the
halophosphate synthesis, since oxidation of even a small fraction of the
324 Lamp phosphors
Post-synthesis processing
On completion of the CaFClAP:Sb,Mn firing, the phosphor is obtained in the
form of a sintered cake. This cake is reduced to a powder and incorporated
into a paint-like suspension capable of producing a smooth, uniform, adherent
coating on the interior of the fluorescent-lamp envelope. Butler [3] provides a
detailed description of the early attempts at lamp coating. Initially, the desired
coating characteristics were achieved by ball milling the phosphor until
particles greater than 30l1m were eliminated. However, ball milling also
increased the overall particle size distribution of the phosphor, creating many
'fines' which lowered lamp brightness, because of their reduced UV absorption
and decreased visible transmission. A phosphor with a narrow particle size
distribution was required, with neither the large particles that reduced coating
quality and adherence nor the fine particles that decreased brightness.
Milling the phosphor has other disadvantages. The particles are fractured,
exposing Mn and Sb to oxidation, reduction, and/or interaction with the Hg
plasma, all of which is harmful to lamp brightness and maintenance. Early
studies, though, showed that a sequential acid-base wash could remove
these reactive ions [21].
The major lamp manufacturers have devised their own methods for
producing halophosphates with the particle size distribution required to
provide high-quality phosphor layers of optimum brightness and lumen
maintenance. These phosphor finishing processes have been tailored to
specific phosphor suspension chemistries and lamp fabrication methods, and
do not produce equivalent results when partially transferred to a different
lamp manufacturing environment.
Although the details of these methods are proprietary, some general
informaton is available. It is now well known that the particle size and
morphology of the CaHP0 4 are reflected in the finished phosphor. Thus,
the first step in the control of particle size distribution of the phosphor is to
control the particle size of this critical component.
Various chemical washing procedures have also been developed to
eliminate surface Mn, Sb and minor non-apatitic phases from the phosphor.
The more successful of these use chelating agents such as EDTA [22] and
DTPA [23] to bind the soluble ions so they are not re-deposited onto the
phosphor. These chemical washes are conveniently sequenced with wet
milling and elutriation or decantation techniques for particle size control.
(a) (b)
Fig. 10.5 (a) Diagram of the fluoroapatite structure. Each large circle represents
the group of atoms shown below. These circles form planes perpendicular to the
c-axis. The Ca(I) ions are below the plane of the large circles and the dashed
rhombus outlines a unit cell with F- ions at the vertices. (b) Diagram illustrating
the orientation of the halogen chain relative to the Ca(IJ) sites in the apatite
structure [24]. Reprinted by permission of the publisher, The Electrochemical
Society, Inc.
The site (or sites) occupied by Sb3+ in the apatite host is still being
debated. According to the most widely accepted model [28], antimony
enters the Ca(II) site and is charge compensated by an oxygen replacing an
adjacent halogen. The Sb-O pairing explains the insensitivity of the Sb 3 +
emission to changes in the halogen (Cl or F). Davis et al. [29] confirmed the
model by demonstrating that Sb 3 + emission shifts to 400 nm when the Sb-O
pairing is precluded, and that this high-energy emission can be converted
into the normal band (Amax '" 48 nm) by re-establishing the Sb-O pairing.
Soules et al. [30] subsequently calculated the energy levels for the Sb3+
centers in fluorapatite by means of a molecular orbital model and compared
their predictions with the available spectral information. This validated the
model of Sb 3 + in the Ca(II) site, and of charge compensation via the Sb-O
pairing.
Unfortunately the direct methods used to probe the location and environ-
ment of the Mn 2 + ion in fluoroapatite are not applicable to the diamagnetic
Sb 3 + ion. However, Mossbauer spectroscopy was recently used to track
antimony during the apatite synthesis process, through the transient
Ca(Sb0 3 h phase, and the subsequent incorporation into the apatite lattice
[31]. The spectra were not sufficiently resolved to extract all of the
Mossbauer parameters, but a variation in the isomer shift values with the
CI:F ratio was observed, thereby providing evidence that Sb was incorpor-
ated into the halogen environment of the lattice. This again suggested Sb 3 +
occupancy of the Ca(II) site.
In another Mossbauer study Mishra et al. [32] reached quite a different
conclusion, reporting that they found strong support for the model [33]
wherein a (P0 4 )3- group is replaced by an (Sb0 3)3- group. This interpreta-
tion of the Mossbauer results, however, is strongly dependent on the
calculations of several models representing different Sb 3 + surroundings, and
the models selected were considered to be incomplete [34].
Perhaps the strongest direct evidence for the Sb 3 + location derives from a
recent X-ray diffraction study by DeBoer et al. [35]. By using both Reitveld
refinements and electron density difference maps in the analysis of data
obtained from undoped and strongly doped (2.2% by weight Sb) Ca
fluoroapatite, it was possible to conclude that Sb 3 + was present in a Ca(II)
site, and that a missing or displaced fluoride ion was probably associated
with the substitution. (Similar measurements on a sample containing 3% by
weight Sb, that was non-typical in several other respects, provided a
different and less acceptable result.)
It has been noted [34] that Sb3+ is not a spherical ion, but has a lone pair
of electrons approximately occupying the volume of an oxygen ligand [35].
Occupancy of the adjacent F- site by this lone pair could explain both the
missing fluorine and the insensitivity of the Sb emission-excitation to
changes in halide type, but it still leaves unexplained how charge compensa-
tion is achieved.
Phosphors for low-pressure Hg plasmas 327
In further speculations on the X-ray results DeBoer et al. [35] pointed out
that the lone pair of electrons was also responsible for the one-sided
coordination of ligands in some 79 known crystal structures containing Sb 3+.
This type of one-sided coordination would also be expected in halophos-
phates and would require something like the (Sb0 3 )3 - model of Mishra et al.
[32], but with Sb 3+ residing on a Ca(II) site rather than a pS + site.
Somehow this picture must be reconciled with that obtained by Moran et
at. [36] from a study of Sb-activated CaF AP using the technique of
magic-angle spinning-nuclear magnetic resonance (MAS-NMR). In this
study the MAS-NMR of 19F and 31p were used to derive quantative
information about the number of fluorides affected by each Sb3+, as well as
information on the spatial proximity of the affected fluorides. From this it
was possible to conclude that Sb 3+ must occupy a phosphorus site.
80 \ \
~Ex i+- Em
.~ 60
=
In
~ 40
20
o
200
~ 300
t-\.A J lA. L.l
600
11
700
A
Wavelength
Fig. 10.6 Excitation and emission spectra for Y203:Eu: Ex, excitation for 611 nm
Eu 3+ emission; Em, Eu3+ emission for excitation at 253.7 nm.
Phosphors for low-pressure Hg plasmas 329
broad absorption band that starts at - 290 nm and peaks at - 230 nm. This
absorption is attributed to charge transfer transitions involving Eu3+ and
nearby oxygen ions.
From the charge transfer states the excitation energy relaxes to the 5Dj
excited states of the 4f 6 configuration of Eu 3 +. Line emission is normally
observed from the 5D 2 , 5D h and 5DO levels to the 7F; ground term levels.
For typical commercial phosphors, less than 1% of the emitted photons
originate from the 5D 1 excited state, and fewer still from 5D 2 The dominant
emission is from the lowest excited state 5D o, particularly in the 5Do- 7F 2
hypersensitive transition at 611 nm, with a lifetime of -1 ms. As shown in
Fig. 10.7, two inequivalent cation sites are available to Eu3+, a C 2 site and
the centrosymmetric C 3i site. At room temperature the bulk of the emission
proceeds from Eu3+ centers in the C 2 site, as a result of energy transfer
between the two sites. At low temperatures the narrow linewidth (- 1 A at
4 K) of the intra-4f6 transitions can be exploited to follow a bewildering
array of energy transfer pathways [38].
o yttrium
Oxygen
Fig. 10.7 Cation sites in Y203' Oxygen-deficient fluorite structure with oxygen
vacancies (joined by dot-dashed line) along a face diagonal for C2 sites, and along a
body diagonal for C3i sites.
The quantum efficiency of this phosphor, within a few per cent of unity, is
the highest of all lighting phosphors, and remains constant over a wide
range of Eu concentrations from 1 to 20 mol. %. The brightness maintenance
in lamps also exceeds that of other phosphors. At 254 nm the phosphor has
a relatively high reflectance (25%), so in blends its ability to capture the
available UV radiation is inferior to that of other blend components. As a
result this phosphor is not only the most expensive component of the blend,
but it is also the dominant blend fraction by weight.
t
Spinel
Me2+.Me3+
Block
!
Mirror Plane .....~":O-r"1r
Fig. 10.8 The magnetoplumbite structure of SrAI 12 0 19 , as a combination of spinel
blocks and ion conduction layers. The spinel block is composed only of AI3+ and
0 2 - ions, but will accommodate Ga3+ or Mg2+ substituting for AI3+. Larger
cations (Sr2+ ,La3+ ,Ce3+ ,Tb3+) are located in the conduction layer which is a mirror
symmetry plane. Only half the unit cell is shown, the complete cell being obtained by
reflection on either mirror plane.
----. A (nanometers)
Fig. 10.9 Relative excitation spectrum (Ex) of the Tb 3+ emission and diffuse
reflection spectrum (R) of (Ceo.67Tbo.33)MgAll1019' Also indicated are the excita-
tion spectrum Ex( x 20) for direct emission by Tb 3+, amplified 20 times, and the
Ce3+ emission (Em) in CeMgAI l1 0 19 [41]. Reprinted by permission of the pub-
lisher, The Electrochemical Society, Inc.
100
80
Em
>-
.,!:: 60
=
'"CII
...d 40
20
o
300 400
A \.A J\ 600 700
500
Wavelength (nanometers)
Fig. 10.10 Emission from the 5D4 level of Tb3+ in (Ce,Tb)MgAll1019 (254nm
excitation).
\ ..l ,, 'J \
,
50 Ex , I
,, I
I
,
o __---'-____.,.""'____.I,.......;"'---L----'
~
ance of various types of cool white lamps (1.5 in diameter) is depicted in Fig.
10.12, together with their operating current and loading characteristics in
watts per foot of lamp length. The lamps with the higher power inputs have
poorer maintenance than the standard F40 lamps because the overall
intensity of UV radiation impinging on the phosphor(s) and on the lamp
envelope has increased, and because the intensity of the 185 nm Hg emission
has increased relatively to the 254 nm component. While all the lamp types
maintain respectable levels of light output for very long times, it is evident that
an improvement in the lumen maintenance of the higher-loaded lamps is
desirable. Maintenance is of course also dependent on the type of phosphor
employed in the lamp and on the type of glass used in bulb fabrication.
=
'"
E
:s
...:I
60
~
......
S
=
8
40
""
GI
Q.,
20
0
0 4000 8000 12,000
Hours Burned
Fig. 10.12 Lamp output brightness as a function of current loading in commercial
fluorescent-lamp types.
Germicidal lamps are designed to emit in the uve region, and consist of
a phosphorless fluorescent tube with an envelope transparent to the
253.7 nm radiation. Absorption of this radiation disrupts the DNA mole-
cules, and so these lamps are used to kill airborne viruses, bacteria and
fungi. Such lamps can be also be used in water purification systems. uve
radiation is of course equally damaging to human tissues and must be used
with caution.
120
:. . Hg
100
I
80
:>.
:!:l
rn
1:1 60
CD
B 40
20
0
200 250 300 350 400 450
Wavelength (nanometers)
Fig. 10.13 Excitation spectra (Ex) for peak emission and emission spectra (Em) for
254 nm excitation for phosphors used in skin-tanning lamps: spectra A, SrB 60 lO :Pb;
spectra B, Sro.3Mgo.63CeO.61All1018.3; spectra C, BaSi 2 0s:Pb.
emission band extends slightly into the UVB region for adequate
melanogenesis. The phosphor exhibits chronically poor maintenance in
lamps, especially at the high current loadings of the tanning lamps. This
requires frequent re-Iamping at the tanning salons to ensure consistent
radiation dosing. The BaSi 2 0 s :Pb 2 + phosphor is also normally present in
the 'long-wavelength UV' mineral lamps (black lights).
As the art of tanning improves, custom blends of this phosphor with other
UV emitters are being introduced in lamps to control the degree of tanning
in users of different racial stock. The UV -emitting phosphors commonly
found in these blends are cerium-activated aluminates, such as SrAI 12 0 19 :Ce.
A borate phosphor with a narrow (35 nm) emission band peaking at 300 nm,
namely SrB 6 0 1o :Pb2+, is used as an intense source of UVB radiation.
Many industrial processes requiring accelerated bleaching and fading of
materials also employ lamps containing UV -emitting phosphors and blends
thereof.
the photosensitive element on the copier drum. The phosphor should also
show appreciable emission in the spectral region absorbed by the document
ink.
Finally, the europium-activated strontium pyrophosphate, Sr 2 P 20 7 : Eu 2 + ,
is contained in lamps used to make architectural blueprints in the Diazo
white printing process.
22mmOD
I
~ ~-~~~~~~-~::::::~~ ---
~~~~~~~~~~~~
c
Fig. 10.14 Schematic of a high-pressure mercury vapor (HPMV) source: A, cath-
ode; B, exhaust tubulation; C, molybdenum foil seal. Reproduced with permission,
MIT Press.
Wavelength (om)
Fig. 10.15 SPD from an HPMV lamp. Reproduced with permission, MIT Press.
REFERENCES
1. Waymouth, IF. (1978) Electrical Discharge Lamps, MIT Press, Cambridge, MA.
2. Wyszecky, G. and Stiles, W.S. (1967) Color Science. Concepts and Methods,
Quantitative Data and Formulas, Wiley, New York.
References 347
3. Butler, K.H. (1980) Fluorescent Lamp Phosphors, Pennsylvania State University
Press.
4. Blasse, G. (1987) Mater. Chem. Phys., 16, 201-236.
5. Brattain, H.G. (1987) In Chemical Analysis (ed. J.D. Winefordner) Vol. 77, Wiley,
New York.
6. Pappalardo, R.G. (1987) In Spectroscopy of Solid-state Laser-type Materials (ed.
B. Di Bartolo), Plenum, New York, pp. 397-429.
7. Welker, T. (1991) J. Lumin., 48-49, 49-56.
8. Pappalardo, R.G. and Peters, T.E. (1984) J. Lumin., 31-32, 284-286.
9. Bril, A. and de Jager-Veenis, A.W. (1976) J. Electrochem. Soc., 123, 396-398.
10. McKeag, A.H. and Ranby, P.W. (1949) US Patent 2,448,733.
11. Jenkins, H.G., McKeag, A.H. and Ranby, P.W. (1949) J. Electrochem. Soc., 96,
1-12.
12. Chenot, C.F. (1990) Personal communication.
13. Kreidler, E.R. (1967) Thesis in Ceramics 68-3549, Pennsylvania State University.
14. Wanmaker, W.L., Hoekstra, A.H. and Tak, M.G.A. (1955) Philips Res. Rep., 10,
11-38.
15. Rabatin, IG. and Gillooly, G.R. (1964) J. Electrochem. Soc., 111, 542-546.
16. Parodi, J.A. (1967) J. Electrochem. Soc., 114, 370-377.
17. Rabatin, J.G., Gillooly, G.R. and Hunter, lW. (1967) J. Electrochem. Soc., 114,
956-959.
18. Hoekstra, A.H. (1967) PhD Thesis, Technical University of Eindhoven.
19. Wachtel, A. and Ryan, F.M. (1973) J. Electrochem. Soc., 120, 693-695.
20. Chenot, C.F. (1986) US Patent 4,613,453.
21. Homer, H.H., Butler K.H. and Bergin, MJ. (1956) Illumin. Eng., 51, 214.
22. Martyny, W.C. (1962) US Patent 3,047,512.
23. Graff, E.A. (1972) J. Electrochem. Soc., 119, 118-120.
24. Flaherty, J.M. (1981) J. Electrochem. Soc., 128, 131-139.
25. Johnson, P.D. (1961) J. Electrochem. Soc., lOS, 159.
26. Ryan, F.M., Ohlmann, R.c., Murphy, 1, Mazelsky, R., Wagner, G.R. and
Warren, R.W. (1971) Phys. Rev. B, 2, 2341-2352.
27. Suitch, P.R., LaCout, J.L., Hewat, A. and Young, R.A. (1985) Acta Crystallogr. B,
41, 173-179.
28. Ouweltjes, lL. (1952) Philips Tech. Rev., 13, 346-351.
29. Davis, T.S., Kreidler, E.R., Parodi, lA. and Soules, T.F. (1971) J. Lumin., 4,
48-62.
30. Soules, T.F., Davis, T.S. and Kreidler, E.R. (1971) J. Chem. Phys., 55,1056-1064.
31. Fraknoy-Koros, V., Gelencser, P., Levay, B. and Vertes, A. (1975) J. Lumin., 49,
467-474.
32. Mishra, K.C., Patton, RJ., Dale, E.A. and Das, T.P. (1987) Phys. Rev. B, 35,
1512-1520.
33. Dale E.A. and Berkowitz, J.K. (1986) 170th Electrochemical Society Meeting,
October 1986, Abstract 708.
34. Oomen, E.W.J.L., Smit, W.M.A. and Blasse, G. (1988) Mater. Chem. Phys., 19,
357-368.
35. DeBoer, B.G., Sakthivel, A., Cagel, lR. and Young, R.A. (1991) Acta Crystallogr.
B, 47,683-692.
36. Moran, L.B., Berkowitz, lK. and Yesinowski, J.P. (1992) Phys. Rev. B, 45,
5347-60.
37. Verstegen, J.M.P.J., Radielovic, D. and Vrenken, L.E. (1975) J. IES, 4, 90-98.
38. Pappalardo, R.G. and R.B. Hunt, R.B., Jr (1985) J. Electrochem. Soc., 132,
721-730.
348 Lamp phosphors
39. Iyi, N., Takekawa, S. and Kimura, S. (1989) J. Solid State Chem., 83, 8-19.
40. Sommerdijk, J.L. and Verstegen, J.M.PJ. (1974) J. Lumin., 9, 415-419.
41. Verstegen, J.M.PJ. (1974) J. Electrochem. Soc., 121, 1623-1628.
42. Martyny, W.e. (1961) US Patent 2.987,414.
43. Repsher, R.W. (1971) US Patent 3,585,207.
44. Beaumont, H.H. and Friedman, A.I. (1969) US Patent 3,424,605.
45. Mooney, RW. and Gungle, W.e. (1956) Illumin. Eng., 51, 793.
46. Lowry, E.F. and Major, E.F. (1949) Illumin. Eng., 44, 98.
47. Lowry, E.F. (1952) Illumin. Eng., 47, 639.
48. Vrenken, L.E., de Vette, T.H. and van der Wolf, R.W. (1964) J. Illumin. Eng., 59,
59-66.
49. Onorato, P.I.K. and Alexander, M.N. (1983) Personal communication.
50. Ryan F.M., Ohlmann, R.e., Murphy, J., Mazelsky R., Wagner, G.R. and
Warren, R.W. (1971) Phys. Rev. B, 2, 2341-2352.
51. Singleton, J.H. and Such ow, L. (1963) J. Electrochem. Soc., 110, 36-41.
52. Narita, K. and Tsuda, N. (1975) Bull. Chem. Soc. Jpn., 48, 2047-2051.
53. Apple, E.F. (1963) J. Electrochem. Soc., 110, 374-380.
54. Aia, M.A. and Poss, S.M. (1960) US Patent 2,965,786.
55. Warren, R.W., Ryan, F.M., Hopkins, R.H. and Van Broekhoven, J. (1975) J.
Electrochem. Soc., 122, 752-760.
56. Ryan, F.M. (1981) J. Lumin., 24-25, 827-834.
57. Lehmann, W., Ryan, F.M., Manocha, A.S. and McAllister, W.A. (1983) J.
Electrochem. Soc., 130,171-174.
58. Hammer, E.F. and Kaduk, E.E. (1975) US Patent 3,886,396.
59. Taubner, F.R. and Roy, J.A. (1987) US Patent 4,639,637.
60. Flaherty, J.M. (1984) US Patent 4,459,507.
61. Flaherty, J.M. (1986) US Patent 4,607,191.
62. Sigai, A.G. and Flaherty, J.M. (1987) US Patent 4,670,688.
63. Sigai, A.G. (1986) US Patent 4,585,673.
64. Sigai, A.G. (1987) US Patent 4,710,674.
65. Peters, T.E. and Hunt, R.B. (1989) US Patent 4,803,400.
66. Piper, W.w., deLuca, J.A. and Ham, F.S. (1974) J. Lumin. 8, 344-348.
67. Sommerdijk, J.L., Bril, A. and de Jager, A.W. (1974) J. Lumin., 8, 341-343.
68. limas, E.R. and Savikhina, T.I. (1970) J. Lumin., 1-2, 702.
69. Berkowitz, J.K. and Olsen, J.A. (1991) J. Lumin., 50,111-115.
70. Flaherty, J.M. (1986) US Patent 4,633,133.
71. Smets, B.M.J. and Verlijsdonk, J.G. (1986) Mater. Res. Bull., 21, 1305-1310.
11
Phosphors for other applications
G. Blasse
11.1 INTRODUCTION
Other chapters on luminescent materials in this book deal with lamp
phosphors and electroluminescent thin films. There are still many other
applications of luminescent materials. Some of these will be dealt with in
this chapter. The applications in this chapter are to a large extent in the
medical field. The excitation source is for the greater part high-energy
excitation.
First we will deal with X-ray phosphors. There are two types, those used
for direct imaging, and those which store the information obtained. The
latter in particular are at this moment in the focus of interest. Subsequently
we will deal with scintillators, large crystals which are used to detect high
energy particles.
In the two following sections we deal with luminescent compositions
which can be used in laser materials and with rare earth cryptates, an
interesting and exotic class of compounds which is promising for application
in fluorescence immunoassay. A section on cathode ray phosphors and a
conclusion end this chapter.
349
350 Phosphors for other applications
----
I
I
I
X-ray source patient intensifying screen
Fig. 11.1 Schematic representation of a medical radiography system based on the
use of an intensifying screen.
1111111]
F=film
Cassette front
Screen base
S
~~~~~~~~~~Fluorescent
Supercoat layer
Emulsion
'I ,,\
\ i\ I I,'
F ('1\' ',~ : :: - Film base
f' "
s{
upercoat
Fluorescent layer
;-~'(' ..~"'n base
-Casslette back
Fig. 11.2 Section through an X-ray cassette containing a double-coated film (F)
and a pair of intensifying screens (S).
X -ray phosphors 351
emission with high conversion efficiency (blue and green match the spectral
sensitivity of the film).
Surprisingly enough, the successful CaW0 4 does not satisfy these
requirements too well. Its X-ray absorption is relatively low, since only
one of the six constituent atoms shows strong X-ray absorption in the
range 30-80 keY which is of medical importance. Also, its density
(6.06 g em -3) is not very high. Its broadband emission (Fig. 11.3) is hard to
use completely: a blue-sensitive film does not utilize the green part of the
emission. On the other hand, it is an inexpensive and stable material. Most
detrimental, however, is the poor X-ray to light conversion efficiency of
only 5%.
350 500nm
Fig. 11.3 X-ray excited emission spectrum of CaW0 4
300 500nm
Fig. 11.4 X-ray excited emission spectrum of BaFCl:Eu2+. The peak is a 4f1
intraconfigurational transition, the band a 4f 6 5d-4f1 interconfigurational transition.
360 480nm
Fig. 11.5 X-ray excited emission spectrum of LaOBr:Tm3+.
300 t.OOnm
Fig. 11.7 X-ray excited emission spectrum of M'-YTa04' The band is due to
tantalate emission; the line at 312 nm is due to emission from a Gd3+ impurity.
354 Phosphors for other applications
OL---~--~--~--~--~
30 50 70keV
Fig. 11.8 X-ray absorption of layers of CaW0 4 (1), YTa04 (2) and LuTa04 (3).
CB ., .+-
c b 6E
1~
d
a
2-<>-
VB
Fig. 11.9 Energy band model showing the electronic transitions in a storage
phosphor: a, generation of free electrons and holes; b, electron and hole trapping; c,
electron release due to stimulation; d, recombination. Solid circles are electrons, open
circles are holes. Centre 1 presents an electron trap, centre 2 a hole trap.
X-ray storage phosphors 355
promoted from the valence band to the conduction band. In a storage
phosphor a number of the created free charge carriers are trapped in
electron traps and hole traps. The traps are localized energy states in the
bandgap due to impurities or lattice defects. If the trap depth AE is large
compared with kT, the probability for thermal escape from the trap will be
negligibly small and a metastable situation is created.
The stored energy can be released by thermal or optical stimulation. In
the case of thermal stimulation the irradiated phosphor is heated to a
temperature at which the energy barrier AE can be overcome thermally. The
trapped electron (or hole) can escape from the trap and recombine with the
trapped hole (electron). In the case of radiative recombination, luminescence
is observed, which is called thermally stimulated luminescence (TSL).
Under optical stimulation the energy of an incident photon is used to
overcome AE. The luminescence due to optical stimulation is called photo-
stimulated luminescence (PSL). The phenomenon of stimulated lumines-
cence from storage phosphors is not new [12]. Storage phosphors have
found a wide range of applications, e.g. as IR detectors and in the field of
dosimetry [13].
The X-ray imaging system based on a photostimulable storage phosphor
is depicted in Fig. 11.10. The X-ray film is replaced by a storage phosphor
screen as a primary image receptor. The transmitted X-ray photons are
absorbed by the storage phosphor in the screen, where a dose-proportional
amount of energy is stored. The latent image in the screen is read out by
scanning the phosphor screen with a focused He-Ne laser. The red laser
light (A. = 633 nm) stimulates recombination, resulting in photo stimulated
luminescence. The intensity of the photostimulated luminescence is propor-
tional to the X-ray dose. For each spot of the laser beam on the screen the
intensity of the photo stimulated luminescence is measured by a photomul-
tiplier tube and stored in a computer. The X-ray photograph in the
computer can be visualized on a monitor or by making a hard copy.
He-Ne laser
screen
hard copy
I
I
I
I
X-ray patient storage I computer
source phosphor L. photomultiplier
Fig. 11.10 Schematic representation of a medical radiography system based on the
use of a photostimulable storage phosphor.
356 Phosphors for other applications
This new technique for X-ray imaging offers several important advan-
tages in comparison with conventional screen-film radiography [11]. The
response of the systems is linear over at least four decades of the X-ray
dose (10- 2-10 2 mR). This wide dynamic range prevents overexposure and
underexposure. The sensitivity of the system is higher owing to a higher
sensitivity of a photomultiplier tube to light compared with film. The
higher sensitivity enables a reduction of the exposure time. Finally, the
digitized images obtained with this system can be processed by a com-
puter which offers the possibility of digital manipulation and facilitates
archiving.
Apart from the high price, the main disadvantage of the digital
X-ray imaging system is the lower resolution. Because of scattering
of the laser beam the performance of the system is worse than the
performance of the conventional screen-film system. This prohibits applica-
tion in areas where a high resolution is required. The development of a
translucent storage phosphor screen with minimal scattering can solve this
problem.
A good storage phosphor has to meet a number ofrequirements [11].
1. The absorption coefficient for X-rays must be high. This can be achieved
by using compounds with a high density.
2. The amount of energy stored in the phosphor per unit X-ray dose has to
be large to obtain a high sensitivity.
3. A short decay time ( < 10 Ils) of the photostimulated emission is required
for a fast readout.
4. The fading of the information stored in the phosphor has to be slow (the
information must still be present several hours after exposure).
5. Stimulation has to be possible in the red or near infrared. The stimulated
emission has to be located between 300 and 500 nm where the photomul-
tiplier tube has its highest sensitivity.
The photostimulable phosphor used at present in nearly all commercial
digital X-ray imaging systems is BaFBr:Eu 2 +. A physical mechanism for
the PSL in Eu 2 + -activated barium fluorohalides has been proposed by
Takahashi et al. [14]. Optical, EPR and photoconductivity studies indicate
that on X-ray irradiation a fraction of the holes is trapped by Eu2+ ions,
giving Eu 3 +, and a fraction of the electrons in halide vacancies, giving F
centres. Illumination in the F centre absorption band stimulates recom-
bination of the trapped electron with the hole trapped on Eu 2 +, resulting
in Eu 2 + in the excited 4f 6 5d state. The Eu 2 + ion returns to the ground
state radiatively and the characteristic Eu 2 + emission around 390 nm is
observed.
Recent developments concern two topics, i.e. a refinement of the theory
and a search for new and better storage phosphors. These developments
X -ray storage phosphors 357
are reviewed here briefly. It was soon realized that the simple scheme of
Fig. 11.9 is an oversimplification. Studies on the recombination mechanism
suggest that electron-hole recombination in BaFBr:Eu 2+ does not take
place via an electron entering the conduction band, but via tunnelling. The
evidence for this originates from different sources.
1. De Leeuw et al. [15] have measured the decay of the photostimulated
luminescence under continuous stimulation in the temperature region
100-300 K and found it to be temperature independent.
2. Von Seggern et al. [16] have measured the increase of the photo-
stimulated luminescence intensity with increasing X-ray dose. This
increase appears to be linear which points to a tunnelling process.
3. Meijerink and Blasse [17] have measured the thermally stimulated
luminescence. Analysis of these results points also to tunnelling.
An important contribution to the theoretical model formation originates
from the use of vacuum ultraviolet radiation from a synchrotron source
[18]. It was shown that photostimulable centres in BaFBr:Eu2+ can be
created by irradiation into the vacuum ultraviolet region (>6.7 eV), i.e. in
the excitonic and interband region of the host lattice.
The model originating from this work invokes as a first step the relaxation
of a free exciton in the neighbourhood of a lattice distortion induced by
Eu 2+. This leads to an e + VK centre (where e stands for electron, and VK for
the VK hole centre consisting of a Bri molecule on two Br- sites).
Subsequently an off-centre self-trapped exciton is formed, i.e. a nearest-
neighbour F-H pair (where F is the well-known F centre, in which an
electron is trapped at an anion vacancy, and H is the H centre, which can be
regarded as an Xi molecule occupying an X- anion site; the anion
concerned here is Br-). This pair is assumed to be stabilized by the presence
of a substitutional Eu2+ ion which is much smaller than Ba2+. The
photostimulable centre is then a Eu 2+-F-H complex.
Photo stimulation is thought to occur as follows: on excitation the F
centre induces a certain relaxation which destabilizes the H centre and an
excited e + VK centre is formed. The excitation energy is transferred to the
nearby Eu2+ ion and emission follows. This model accounts also for the fact
that practically no Eu 3 + has been observed after X-ray irradiation of
BaFBr:Eu2+ [15-17]. Note that the simple model of Fig. 11.9 requires the
presence of Eu 3 + after irradiation.
Many other X-ray storage phosphors have been proposed. The most
promising among these are BasSi04Br6:Eu2+ and Bas Ge04Br6:Eu2+ [17,
19]. Others are Ba2Bs09Br:Eu2+ [17], Ba40Br6:Eu2+ [20], Y2SiOs:Ce3+
[21] and RbBr:TI + [22].
Since this field is at the moment still developing, it is not possible to
indicate which material will be the choice for application in the future.
358 Phosphors for other applications
11.4 SCINTILLATO RS
Scintillator materials are used to detect high-energy radiation such as y rays.
They are used in, for example, medical diagnostics and in nuclear and
high-energy physics. A spectacular example of the latter is the use of 12000
crystals of Bi4Ge3012 at CERN (Geneva) in order to detect all types of
particles in the field of high-energy physics [23]. Examples of nuclear
diagnostics in the medical field are positron emission tomography (PET),
single-photon emission computed tomography (SPECT), and computed
tomography (CT) [24].
PET is an in vivo tracer technique which makes use of the annihilation of
positive electrons emitted by a radionuclide. This results in two almost
collinear gamma quanta with an energy of 511 keY. It is possible to measure
a three-dimensional distribution of the radioactivity by detecting the y
radiation with a scintillator. Two detectors are necessary to obtain a
three-dimensional picture.
In SPECT a radioactive substance is introduced into the body and its
gamma radiation is measured. A frequently used radionuclide is 99Tc m
(E y = 140.5 keV, tl/2=6h). In PET a radioactive isotope of C, N or 0 is
used. These are the 'biological' elements in the human body, so that
metabolic processes can be followed. These isotopes are not gamma, but
positron, emitters. Positrons (positive electrons) cannot penetrate far into
the tissue (only a few millimetres). At rest the positron annihilates with an
electron yielding two photons with 511 keY energy emitted at '" 180. These
are detected by two scintillators (Fig. 11.11). From the coincidence data,
three-dimensional images can be reconstructed. PET has several advantages
over SPECT, but the detection is very expensive.
:' Sc I
I Sc
Fig. 11.11 Principle of a PET camera. There are two scintillator crystals (Sc). The
source Sp is in the body of the patient. Note that the positron decay results in the
emission of two y quanta under an angle of _180.
i i
400 600nm
Fig. 11.12 X-ray excited emission spectra of crystals for use in CT: spectrum a,
NaI:TI; spectrum b, CsI:TI; spectrum c, ZnW0 4; spectrum d, Bi4Ge3012 (BGO).
5PBa 2+
Fig. 11.13 Energy band model of BaF 2 showing cross-over luminescence. Eg gives
the forbidden zone between the conduction band (CB) and the valence band (2PF-)'
The core band 5PBa2+ lies below the valence band. The excitation CE results in the
fast cross-over luminescence indicated by f. The slow emission s is due to a level in
the forbidden zone.
Fig. 11.14 y ray excited emission spectrum of LaF 3 :Nd3+ (0.03%). In addition to
the Nd3+ emission band at 173 nm there are also emission bands due to the
impurities Pr 3+ and Ce 3+ .
I.
----19/2
Fig. 11.15 Part of the energy level scheme of Nd3+. The laser action of Nd3+-
containing solids is based on the emission transition 4F 3/2 ..... 4111/2 (1.06 ~m). Excita-
tion is from the ground state 419/2 to the excited levels here indicated by X. From the
4111/2 level the system relaxes rapidly to the ground state 41 9 / 2,
362 Phosphors for other applications
whereas the 4111/2 level empties easily to the 419/2 ground level. The largest
lasers in the world use Nd3+ -doped glass rods which can be rather easily
prepared in large and perfect shape. They playa role in the nuclear fusion
experiments in the Lawrence Livermore National Laboratory (USA) [39].
The Nd 3+ - YAG laser is nowadays commercially available and well known
in the field of laser spectroscopy (Y AG = Y3Ais 12). The application of
crystalline material requires the growth of large and optically perfect
crystals.
For more recent developments we turn to the transition metal ions.
Actually the first solid state laser used ruby (A12 0 3:Cr3+ ) and was develop-
ed by Maiman [40]. Figure 11.16 gives the energy level scheme of Cr3+ as a
function of the crystal field strength LI. In ruby LI is large and laser action
occurs in the 2E-+4A2 transition. This is a three-level laser because pumping
occurs in the 4A2-+4T2 and higher transitions. The 2E-+4A2 transition is
sharp, so that the ruby laser is not tunable.
---~!::.
Fig. 11.16 The position of the lower energy levels of the Cr3+ ion as a function of
the crystal field LI. In the weak-field case the emission is the broadband 4T2-+4A2
transition (b) on which a tunable near-infrared laser can be based. In the strong-field
case the emission is the line 2E-+4A2 transition (1) on which a not-tunable three-level
red laser can be based.
In other host lattices the Cr 3+ ion may have a weaker crystal field and the
emission is due to the broadband 4T2-+4A2 transition. Using this transition
a tunable infrared laser can be designed. Suitable host lattices for this
purpose are alexandrite (BeAI 20 4) [41, 42], KZnF3 [43], ScB0 3 [44J, and
GSGG (Gd3Sc2Ga3012) [45].
Also, other transition metal ions can be used to obtain a tunable laser.
Here we mention AI 20 3:Ti3+ [35,46]. Further, the luminescence of colour
centres should be mentioned in this respect. This subject has been reviewed
by Henderson and Imbusch [35]. For further examples the reader is referred
to the general literature cited.
Crystals and glasses for lasers 363
Many efficient luminescent materials cannot be used as a laser material
because of excited state absorption (ESA). Examples are Y3Ais 0 12 :Ce3+
[47], MgAI204:Mn2+ [48], and CaW0 4 [49]. It will be obvious that it is
detrimental for laser action if the emitted radiation is reabsorbed by the
other still-excited ions.
As an example we consider Y3AIs012:Ce in more detail (see also our
earlier chapter in this book [1]). In this crystal laser action is completely
quenched by excited state absorption. In the ground state absorption
spectrum of Y3AIs012:Ce3+(4f1) the lower bands are due to allowed 4f-5d
transitions on the Ce 3+ ion. A broad emission band extends from 500 to
650 nm. The excited state absorption spectrum shows a broad band overlap-
ping this region. It has an oscillator strength of 0.03.
The excited state absorption starts on the lower 5d level and ends on a
level in the conduction band. Therefore photoionization of the Ce 3+ ion
results. The minimum of the conduction band lies only 10000 cm -1 above
the lowest relaxed 5d level of Ce 3+. Therefore the higher part of the ground
state absorption spectrum (A < 300 nm) is due to direct (one-photon) photo-
ionization of the Ce 3+ ion.
In MgAh 04:Mn2+ the situation is simpler. Excited state absorption is
observed in the visible and near-infrared spectral region. The lowest excited
state from which the emission occurs is the 4T 1 level of the d S configuration
of Mn 2+. Excited state absorption empties the emitting level and promotes
the Mn 2+ ion to higher levels of its own d S configuration. This makes laser
action based on Mn2+ luminescence practically impossible. Figure 11.17
gives an energy level scheme and some of the transitions of importance.
~ iESA
M X
Fig. 11.17 Schematic configurational coordinate diagram for Mn2+ after data by
Petermann et al. [48]. The ground state level g is 6 A b the lowest excited state e from
which the emission occurs is 4T 1. Excitation is indicated by X, emission by M. The
broken line ESA indicates excited state absorption from 4T 1 to the level e' (4T 1)'
364 Phosphors for other applications
Fig. 11.18 The [2.2.1] and [2.2.2] cryptands. In these the central metal ion has
been removed for clarity. This is not the case for the [Lncbpy bpy bpy]3+ cryptate.
The CH 2 groups are not indicated in these formulae (cryptand is the name of the
ligand; cryptate is the name of the whole complex).
given in Fig. 11.19 for one cathode ray tube. Because of the large magnifica-
tion used in imaging, the phosphor screen of the cathode ray tube must be
excited with an intense and small electron spot. The maximum excitation
density may be two orders of magnitude larger than in conventional cathode
ray tubes.
"-.3''---- ---+ L
Fig. 11.19 Projection television set (schematic). The lens system L projects the
picture at the faceplate of the tube T on the viewing screen S. The mirror M provides
a folded light path to keep the cabinet depth small.
Under such high loading the efficiency of the phosphor decreases with
increasing excitation density owing to saturation effects. This effect is much
more pronounced for the semiconductor sulphide phosphors than for the
rare-earth-activated oxides. A simple qualitative explanation of this effect
runs as follows. With increasing excitation density the number of excited
activator ions increases until all activator ions are excited. This leads to
saturation of the luminescence output. Rare-earth-activated phosphors
contain usually a few per cent of luminescent ions; in the sulphides, however,
the amount of luminescent centres is far below 1%, so that saturation (or
activator depletion) appears at much lower excitation density.
Let us consider the situation for the different colours separately. For blue
emission ZnS:Ag+ is used in conventional tubes as well as in PTV. Al-
though it suffers from strong saturation, no acceptable alternative has
been found. What compensates partly for the strong saturation is the fact
that the energy conversion efficiency of ZnS:Ag+ for cathode rays is
extremely high.
For red the phosphor Y203:Eu3+ is preferred above Y20 2S:Eu3+ in
PTV tubes. The oxide shows less saturation than the oxysulphide. Also, the
Phosphors for cathode ray tubes 367
temperature dependence of the Eu3+ emission is less in Y203 than in
Y2O 2S. This is important since the temperature of the phosphor in a PTV
tube may increase up to 100 DC.
For green many candidates have been proposed. The conventional
(Zn, Cd)S:Cu + shows an unacceptable saturation in PTV tubes. First an old,
classic luminescent material was proposed, i.e. Zn2Si04:Mn2+ (willemite). It
has a stronger saturation than Tb3+ phosphors. After Gd 2O 2S:Tb 3+ (an
X-ray phosphor, see section 11.2), the phosphors Y2SiO 5 :Tb3+ and Y3(Ga,
AI)s012:Tb3+ were proposed. Unfortunately their chromaticities lie outside
the specifications given for colour television tubes. This can be remedied
using filters. A phosphor such as InB0 3:Tb3+ has excellent properties but
its decay time is too long (7.5 ms). This is due to the fact that the
crystallographic site for Tb 3+ in InB0 3 has inversion symmetry, so that all
electric dipole transitions are forbidden.
Table 11.1 shows some data on these new phosphors. Many of these have
been studied in detail and several processes to explain the saturation have
been proposed. In addition to the model mentioned above, we mention here
increased self-absorption of the emission [55] and interaction between
excited centres [56, 57].
Y2 0 3: Eu 8 0.64 0.91
Y 2 0 2 S:Eu 13 0.40 0.68
(Zn,Cd)S:Cu + 16 0.12 1.00
Zn2Si04:Mn2+ 9 0.63 0.95
Y3(AI,Ga)s 0 12 :Tb 3+ 9 0.82 0.97
Y 2SiOs:Tb3+ 9 0.88 0.94
Gd 2O 2S:Tb 3+ 11 0.51 0.80
InB0 3:Tb3+ 8 0.76 1.00
ZnS:Ag+ 20 0.36 0.89
"'1: energy conversion efficiency in per cent in conventional cathode ray
tube.
b'1;TV: relative efficiency at excitation density characteristic of PTV
tube (the lower this value, the more pronounced is the saturation).
e'1: 00: relative efficiency at 1()() 0c.
~
c:
Q)
'0
i
Q)
>
~
Qj
CI: 0.5
Fig. 11.20 Saturation behaviour of InB0 3 :Tb and ScB0 3 :Tb prepared by the
spray dry technique as compared with InB0 3 :Tb prepared in the standard way.
(From Hintzen [59].)
the standard InB0 3 :Tb3+ (4%) shows a stronger saturation than the sample
made by the spray-drying technique. Note further that the 4% sample has
less saturation than the 1% sample as is to be expected (see above).
Research to improve the quality of luminescent materials for use in
projection television tubes is still going on. A final decision on the optimum
choice is to be expected soon.
REFERENCES
1. BIasse, G. (1992) In Solid State Luminescence (ed. A.H. Kitai), Chapman & Hall,
London, Chapter 2.
2. Brixner, L.H. (1987) Mater. Chem. Phys., 16, 253.
3. Chenot, D.F. (1972) Canadian Patent 896453.
4. Fouassier, c., Laourette, B., Portier, 1. and Hagenmuller, P. (1976) Mater. Res.
Bull., 11, 933.
5. Stevels, A.L.N. and Pingault, F. (1975) Philips Res. Rep., 30, 277.
6. Tecotzky, M. (1968) Electrochemical Soc. Meeting, Boston, MA, May 1968.
7. Brixner, L.H. and Chen, H.Y. (1983) J. Electrochem. Soc., 130, 2435.
8. Lammers, MJ.1. and Blasse, G. (1984) Mater. Res. Bull., 19, 759.
9. Lammers, MJJ. and BIasse, G. (1987) J. Electrochem. Soc., 134, 2068.
10. Lammers, M.J.J., Severin, 1.W. and BIasse, G. (1987) J. Electrochem. Soc., 134,
2356.
11. Sonoda, M., Takano, M., Miyahara, J. and Kato, H. (1983) Radiology, 148, 833.
12. Boyle, R. (1663) Regist. R. Soc., 213.
13. McKeever, S.W.S. (1985) Thermoluminescence oj Solids, Cambridge University
Press, Cambridge.
14. Takahashi, K., Miyahara, 1. and Shibahara, Y. (1985) J. Electrochem. Soc., 132,
1492.
15. de Leeuw, D.M., Kovats, T. and Herko, S.P. (1987) J. Electrochem. Soc., 134, 491.
16. von Seggern, H., Voigt, T., Knupfer, W. and Lange, G. (1988) J. Appl. Phys., 66,
1405.
17. Meijerink, A. and BIasse, G. (1991) J. Phys. D, 24, 626.
18. Ruter, H.H., von Seggern, H., Reiniger, R. and Saile, V. (1990) Phys. Rev. Lett.,
65,2438.
19. Meijerink, A., BIasse, G. and Struye, A. (1989) Mater. Chem. Phys., 21, 261.
20. Schipper, WJ., BIasse, G., Meijer, G. and Schleid, T. (1991) Mater. Chem. Phys.,
30,43.
21. Meijerink, A., Schipper, W.J. and Blasse, G. (1991) J. Phys. D, 24, 997.
22. von Seggern, H., Meijerink, A., Voigt, T. and Winnacker, A. (1989) J. Appl. Phys.,
66,4418.
23. Rogemond, F., Pedrini, c., Moine, B. and Boulon, G. (1985) J. Lumin., 33, 455.
24. Birks, J.B. (1964) The Theory and Practice oj Scintillation Counting, Pergamon,
New York.
References 371
25. Rossner, W. and Grabmaier, B.C (1991) J. Lumin., 48-49, 29.
26. Laval, M., Moszynski, M., Allemand, R., Cormoreche, E., Guinet, P., Odru, P.
and Vacher, 1. (1983) Nucl. Instrum. Methods, 206, 169.
27. Schotanus, P., van Eijk, CW.E. and Hollander, R.W. (1988) Nucl. Instrum.
Methods A, 269, 377.
28. Aleksandrov, Yu.M., Makhov, V.N., Rodnye, P.A., Syreischikova, TI. and
Yakimenko, M.N. (1984) Sov. Phys. Solid State, 26, 1734.
29. Valbis, YaA, Rachko, Z.A. and Yansons, Ya. (1985) JETP Lett., 42,172.
30. Valbis, YaA, Rachko, Z.A. and Yansons, Ya. (1986) Opt. Spectrosc., 60, 679.
31. Schotanus, P., van Eijk, CW.E., Hollander, R.W. and Pijpelink, 1. (1985) Nucl.
Instrum. Methods A, 238, 564.
32. Dorenbos, R., van Eijk, CW.E., Hollander, R.W. and Schotanus, P. (1990) IEEE
Trans. Nucl. Sci., 37, 119.
33. Derenzo, S.E., Moses, W.W., Cahoon, 1.L., Perera, R.CC and Litton, J.E. (1990)
IEEE Trans. Nucl. Sci., 37, 203.
34. Blasse, G. (1990) IEEE Trans. Nucl. Sci., 38, 30.
35. Henderson, B. and Imbusch, G.F. (1989) Optical Spectroscopy oj Inorganic
Solids, Clarendon, Oxford, 1989.
36. Kaminskii, A.A. (1981) Laser Crystals, Springer, Berlin.
37. DiBartolo, B. (1987) Spectroscopy oj Solid-state Laser-type Materials, Plenum,
New York.
38. Atkins, P.W. (1990) Physical Chemistry, 4th edn, Oxford University Press,
Oxford.
39. Weber, MJ. (1976) In Critical Materials Problems in Energy Production, C Stein
(ed.), Academic Press, New York.
40. Maiman, T. (1960) Nature (London), 187,493.
41. Walling, J.C (1982) Laser Focus, 18,45.
42. Walling, J.C, Petersen, O.G., Jenssen, H.P., Morris, F.C and O'Dell, E.W. (1980)
IEEE J. Quantum Electron., 16, 1302.
43. Durr, U. and Brauch, U. (1986) In Tunable Solid State Lasers II (eds A.B.
Budgor, L. Esterowitz and L.G. De Shazer), Springer, Berlin, p. 151.
44. Lai, S.T., Chai, B.H.T, Long, M. and Morris, R. (1986) IEEE J. Quantum
Electron., 22, 1931.
45. Huber, G. and Petermann, K. (1985) in Tunable Solid State Lasers (eds, P.
Hammerling, A.B. Budgor and A. Pinto), Springer, Berlin, p. 11.
46. Moulton, P.F. (1982) Opt. News, (November-December 1982),9.
47. Hamilton, D.S., Gayen, S.K., Pogatshnik, G.J. and Ghen, R.D. (1989) Phys. Rev.
B,39,8807.
48. Petermann, K., Clausen, R., Heumann, E. and Ledig, M. (1989) Opt. Commun.,
70,483.
49. Koepke, Cz. and Lempicki, A. (1991) J. Lumin., 47, 189; (1992), 54, 151.
50. Sabbatini, N., Perathoner, S., Balzano, V., Alpha, B. and Lehn, 1.M. (1987) In
Supramolecular Photochemistry (ed. V. Balzani), Reidel, Dordrecht, p. 187.
51. Sabbatini, N. (1992) Coord. Chem. Rev., to be published. Mater. Chem. Phys., 31,
13.
52. Blasse, G., Dirksen, G.J., Sabbatini, N. and Perathoner, S. (1987) Inorg. Chim.
Acta, 133, 167.
53. Welker, T (1991) J. Lumin., 48-49, 49.
54. Vriens, L., Clarke, J.A. and Spruit, J.H.M. (1989) Philips Tech. Rev., 44, 201.
55. Klaassen, D.B.M. and de Leeuw, D.M. (1984) J. Lumin., 31-32, 687.
56. van der Weg, W.F., Robertson, J.M., Zwicker, W.K. and Popma, ThJ.A. (1981)
J. Lumin., 24-25, 633.
372 Phosphors for other applications
57. van der Weg, W.F., Popma, ThJ.A. and Vink, A.T. (1985) J. Appl. Phys., 57,
5450.
58. Yamamoto, H. and Matsukiyo, H. (1991) J. Lurnin., 48-49, 43.
59. Hintzen, H.TJ.M. (1990) Thesis, University of Utrecht.
60. Biasse, G. (1990) Adv. Inorg. Chern., 35, 319.
61. Hazenkamp, M.F. and Biasse, G. (1990) Chern. Mater., 2, 105.
62. Weller, H., Koch, U., Gutierrez, M. and Henglein, A. (1984) Ber. Bunsenges.
Phys. Chern., 88, 649.
63. Wang, Y. and Herron, N. (1988) J. Phys. Chern., 92, 4988.
64. Herron, N., Wang, Y., Eddy, M.M., Stucky, G.D., Cox, D.E., Moller, K. and
Beiz, T. (1989) J. Arn. Chern. Soc., 111, 530.
Index
373
374 Index
Epitaxy Filtering
atomic layer, see Atomic layer absorption filters 67
epitaxy birefringent filters 67, 69
digital layer 295 interference filters 66-7
flow-rate modulation 295 multicolour thin film
halide vapour phase 281 electroluminescence devices
hydride vapour phase 280-1 245, 246-9
liquid phase 273, 286-9 Fischer's model 200-1
molecular beam 79, 289-90, 293, Flat panels, see Thin film
296, 302 electro luminescence devices
modulated 295 Flickering 319
molecular layer 295 FLN, see Fluorescence line
vapour phase 279-81 narrowing (FLN)
EPMA, see Electron probe Flocculation 335
microanalysis (EPMA) Flow-rate modulation epitaxy 295
EPR, see Electron paramagnetic Fluorescence immunoassay 364--5
resonance (EPR) Fluorescence line narrowing (FLN)
Europium 40
divalent Eu-activated phosphors Forced electric dipole 98
319 Forming process, see Phosphors
luminescence of ions 45-7 Forster's law 89
one-photon transitions 118-19 Free exciton reflection spectra (FE)
Excitation 53 76
impact excitation 313 Frenkel exciton 54
optically active centres 22
see also Excitation sources;
Gadolinium, luminescence of ions
Excitons
45-7
Excitation sources
Gadolinium compounds 44
lamps 62-3
Gadolinium one-photon transitions
lasers 60
119-21
for spectroscopy 60-3
Gallium arsenide (GaAs) deposition
Excited state absorption 363
Excitons 54--5 308
Gallium arsenide phosphide (GaAsP)
biexcitons 54
diode material 273, 278-9
bound exciton luminescence
Gallium phosphide (GaP) diode
spectra (BE) 76
material 277-8
density 54
Germicidal lamps 340-1
free exciton reflection spectra (FE)
Goppert-Mayer formula 98
76
Grating spectrometer 66-8
Frenkel exciton 54
Gumlich, H.-E. 53-96
Mott density 54
transition energy 54
Wannier-Mott exciton 54, 55 Halide vapour phase epitaxy 281
Halophosphate lamp phosphors 321-7
Harmonic perturbation 10-12
Fabry-Perot elation 62 Hayne's rule 79
Faraday configuration 66 Hel bypass flow process 252
Fick's law 284 Heisenberg uncertainty principle 5,
Field quenching, thin films 143--4, 264, 277
145, 146, 148 Helmholtz pair 66
Index 377