Você está na página 1de 32

Advances in Colloid and Interface Science

76]77 1998. 341]372

New perspectives in mercury porosimetry


Carlos A. Leon
y Leon
U
Quantachrome Corporation, 1900 Corporate Dr., Boynton Beach, FL 33426, USA

Abstract
Seventy-six years ago, Washburn pioneered the concept that the structure of porous solids
could be characterized by forcing a non-wetting liquid to penetrate their pores. At that time
Washburn postulated that the minimum pressure P required to force a non-wetting liquid
like mercury to penetrate pores of size R is given by P s KrR, where K is a constant.
Nowadays that very same concept constitutes the backbone of mercury porosimetry, a
technique applied routinely to the characterization of all kinds of solids. Despite its
perceived fundamental and practical limitations, mercury porosimetry will continue to be
regarded as a standard measure of macro- and mesopore size distributions for years to
come. This is so because this time-tested technique is 1. conceptually much simpler, 2.
experimentally much faster, and 3. unique in its ability to evaluate a much wider range of
pore sizes, than any alternative method practised currently e.g. gas sorption, calorimetry,
thermoporometry, etc... Clearly it would be desirable to derive as much structural information as possible from simple mercury porosimetry experiments. Surprisingly, relatively few
attempts have been made in the open literature to extract much information beyond pore
size distributions from mercury porosimetry data. This contribution emphasizes the need to
develop concerted efforts towards expanding the interpretation of mercury porosimetry data
by examining the virtues and flaws of various reported attempts to generate particle size
distributions, inter- and intraparticle porosities, pore tortuosities, permeabilities, throatrpore
ratios, fractal dimensions and compressibilities from mercury intrusion andror extrusion
curves. Q 1998 Published by Elsevier Science B.V. All rights reserved.
Keywords: Mercury porosimetry; Characterization; Pores; Particles

Tel: q1 561 7314999; fax: q1 561 7329888; e-mail: Qchrome@aol.com

0001-8686r98r$19.00 Q 1998 Published by Elsevier Science B.V. All rights reserved.


PII S0001-868698.00052-9

342

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . .
2. Mercury porosimetry as a characterization tool . . .
2.1. Theoretical foundation . . . . . . . . . . . . . . .
2.1.1. Hysteresis phenomena . . . . . . . . . . . .
2.1.2. Studies of ideal pore systems . . . . . . . .
2.2. Experimental approach . . . . . . . . . . . . . . .
2.2.1. Scanning versus stepwise data acquisition
2.2.2. Contact angle measurement . . . . . . . .
2.2.3. Mercury purity . . . . . . . . . . . . . . . . .
2.2.4. Blank corrections . . . . . . . . . . . . . . .
2.3. Range of applicability . . . . . . . . . . . . . . . .
2.3.1. Sample types . . . . . . . . . . . . . . . . . .
2.3.2. Pressure and pore size limits . . . . . . . .
2.4. Interpretation of mercury porosimetry data . .
2.4.1. Particle size distribution . . . . . . . . . . .
2.4.1.1. Mayer]Stowe MS. theory . . . .
2.4.1.2. Smith]Stermer SS. theory . . . .
2.4.2. Inter- and intraparticle porosity . . . . . .
2.4.3. Pore tortuosity . . . . . . . . . . . . . . . . .
2.4.4. Permeability . . . . . . . . . . . . . . . . . .
2.4.5. Throatrpore ratios . . . . . . . . . . . . . .
2.4.6. Fractal dimension . . . . . . . . . . . . . . .
2.4.7. Sample compressibility . . . . . . . . . . . .
3. Concluding remarks . . . . . . . . . . . . . . . . . . . .
Acknowledgements . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

. 342
. 343
. 343
. 344
. 345
. 347
. 348
. 350
. 351
. 352
. 354
. 354
. 354
. 354
. 354
. 355
. 356
. 358
. 361
. 363
. 364
. 367
. 369
. 369
. 370
. 370

1. Introduction
Mercury porosimetry is a well established technique for the characterization of
porous materials w1]3x. This fact is most clearly manifested in documents such as
the International Union of Pure and Applied Chemistrys IUPACs. Recommendations for the Characterization of Porous Solids, in which it is stated that mercury
porosimetry is widely accepted as a standard measure of total pore volume and
pore size distribution in the macro- and mesopore ranges w4x. Moreover, practical
applications of the technique have led to the development of internationally
recognized standard procedures ASTM, DIN. revolving around its use w5]8x. Yet
the full potential of mercury porosimetry as a characterization tool is far from
being uncovered.
The use of mercury porosimetry is much too often limited to the evaluation of
pore volumes or pore size distributions only. Over the years, numerous literature
reports have described theories and procedures designed to obtain additional
information from mercury porosimetry curves. However, judging from the relatively
few pertinent publications and research reports presented at international meet-

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

343

ings w9x, these approaches have not received the attention they might deserve.
Clearly, it would be greatly beneficial to expand the capabilities of a technique that
has already proven to be fast, simple, reliable, broad in scope and, hence, ideally
suited for industrial and research applications alike.
The purpose of this contribution is to call attention to some approaches that in
our experience can or have already turned mercury porosimetry into a much more
powerful characterization tool. Their examination is preceded by an overview of
the theoretical foundations and experimental approaches that define mercury
porosimetry as a technique including a discussion of how certain practical considerations can impact its applicability to the characterization of solid materials in
general. and is aided by selected illustrative examples.
2. Mercury porosimetry as a characterization tool
Mercury porosimetry is by far the most popular method employed for the
characterization of relatively large pores, in particular macropores w10]12x. Compared to alternative pore size characterization methods e.g. gas sorption w1]3x.
mercury porosimetry is based on a simpler principle the Washburn equation}Sec.
2.1., is much faster producing full pore size distributions in minutes compared to
hours in gas sorption tests. and, perhaps more importantly, is unique in that it
covers a very wide range of pore sizes, including large pores ) 0.5 m m. that are
difficult or impossible to probe reliably by other techniques. Yet much like other
techniques, mercury porosimetry analyses yield cumulative raw data from which
differential pore size distributions are numerically derived.
2.1. Theoretical foundation
The principles governing conventional mercury porosimetry calculations have
been described in a number of publications w1]3,10]15x. Almost two centuries ago,
Laplace w16x reasoned that the work required to expand a nonspherical fluid
surface of principal radii of curvature R1 and R 2 is equal to the work done to the
concave side of the surface, and managed to derive the following equilibrium
expression:

D P s g 1rR1 q 1rR 2 .

1.

where g is the fluid surface tension and D P is the pressure across the interface.
Later on Young w17x derived an expression describing the mechanical equilibrium
of liquid drops on surfaces w1,15x which Washburn w18x linked to Laplaces equation
in order to postulate that, for cylindrical pores of equivalent radius R wetted with
a fluid of contact angle u ,

D P s y2g cos u . rR

2.

The latter equation predicts the behaviour of liquids confined in capillaries.


Accordingly, for wetting liquids u - 908, so D P s Pliquid y Pgas - 0 and the liquids

344

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

rise in the capillaries to develop a column whose head pressure compensates for
that pressure difference. Conversely, non-wetting liquids like mercury for which
908 - u - 1808. tend to recede in capillaries and must be forced hydraulically to
fill them.
The Washburn equation clearly provides a simple and convenient relationship
between applied pressure and pore size. Pore size distributions are therefore
generated by monitoring the amount of non-wetting mercury intruded into pores as
a function of increasing applied pressure w1x. The choice of cylindrical pore
geometry was one of mathematical convenience in order to avoid the complexities
of having to deal with mean radii and contact angles in pores of irregular cross
sections. For that reason mercury porosimetry has traditionally treated solids as
porous materials containing a bundle of capillaries of various sizes. Not very much
error is introduced by the presence in solids of interconnected channels, provided
that all pores are equally accessible to the exterior mercury reservoir; that is,
access to pores of a given size D must always be possible through pores of size
G D. If the pores have only smaller entrances, they can only be filled upon
reaching a pressure higher than that required by their actual inner. dimensions.
Once the pressure required to fill the largest entrance was reached, the filling of
the entire pore would ensue. In such case, the calculated pore size distribution
would be biased towards smaller than actual inner. pore sizes. The converse would
be true upon mercury extrusion, and could presumably lead to entrapment and
hysteresis in so-called ink-bottle pores w19,20x. The total content of ink-bottle
pores, when estimated from the end point of the depressurization curve, has been
reported to vary widely from a negligible fraction of total pore volume in a
silica]alumina gel to ) 80% in an activated carbon w15x.
2.1.1. Hysteresis phenomena
Many explanations have been offered to account for the experimental observation that mercury extrusion curves do not overlap intrusion curves w1,3,13,21]28x.
Currently, three explanations appear to be favored by different groups in the
literature: i. the ink-bottle pore assumption w20x, as described in Section 2.1; ii.
network effects, i.e. an extension of the ink-bottle concept which is substantiated
by complex computer simulations w13,23]26x; and iii. a pore potential theory
whereby mercury is not subjected to pore wall interactions during its initial
intrusion but is partly held in pores upon extrusion as a function of wall interactions w1,27,28x. According to the latter mechanism, the work done to expand the
mercury surface area upon initial intrusion consists of two parts: i. the irreversible
work of intrusion, associated with mercury entrapment in potential wells or
ink-bottle pores; and ii. the reversible work of intrusion, which is greater than iii.
the reversible work of extrusion. The latter is quite feasible thermodynamically
w1,15,27,28x and is ascribed to the extrusion of mercury being facilitated by a
decrease in the tensile extrusion contact angle u E relative to the compressive
intrusion contact angle u I . In this context, good quantitative agreement has been
reported w1,27,28x upon comparing the results of pore potential predictions with
contact angle changes. Moreover, it is an experimental fact that hysteresis gaps can

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

345

Fig. 1.

in many cases be completely eliminated within ca. 1%. by modifying the extrusion
contact angle until the extrusion curve and a second intrusion curve overlap each
other w1,25,27,28x. This fact is illustrated for a typical material tested a silica. in
Figs. 1 and 2. Fig. 1a presents standard mercury intrusionrextrusion volumes as a
function of pore sizes calculated using a constant contact angle u I s u E . of 1408.
Fig. 1b illustrates the excellent overlap obtained after shifting the extrusion angle
u E . to 106.58. Fig. 2a,b confirms the excellent overlap in pore size distributions
calculated before and after shifting the extrusion contact angle, respectively.
Exceptions to this general observation have been attributed to materials having
multiple or variable contact angles w1,3x. Interestingly, if contact angle shifts
dominated hysteresis phenomena in mercury porosimetry, attempting to infer pore
shapes from hysteresis curves as done through network models. would be futile.
2.1.2. Studies of ideal pore systems
One way to substantiate the validity of contact angle hysteresis in mercury

346

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

Fig. 1. Mercury intrusionrextrusion curves as a function of pore size for a typical material; a. assuming
u I s u E s 1408; b. after shifting u E to 106.58; Symbols: circles, intrusion; triangles, extrusion. For
clarity, only one out of every 100 data points collected is shown.

porosimetry is through the analysis of ideal pore systems. For instance, a material
with uniformly sized, straight cylindrical pores would not be subjected to ink-bottle
or network effects, and should therefore show no hysteresis upon mercury intrusionrextrusion experiments. On the other hand, hysteresis loops in such ideal pore
systems would be consistent with an unavoidable shift in intrusionrextrusion
contact angles. Over a decade ago, Lowell w29x performed pertinent mercury
porosimetry experiments on Nuclepore brand polycarbonate membrane filters w30x.
These filters, made by bombarding polycarbonate films with laser beams, exhibit
rather uniform straight cylindrical pores perpendicular to the surface, as confirmed
by SEM pictures w31x. Mercury porosimetry experiments confirmed that a clear
intrusionrextrusion hysteresis gap occurred in these ideal materials w29x. In spite of
a possible influence of sample compression on the results w32x, the study of ideal
pore systems could prove to be invaluable to confirm the validity of different

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

347

Fig. 2.

models of hysteresis phenomena. Unfortunately, MCM41s and related materials


appear to be mechanically too weak to withstand exposure to high pressures w33x..
2.2. Experimental approach
Even though Washburn postulated his equation in 1921, it was not until the late
1940s that the usefulness of mercury porosimetry experiments began to be realized.
At around that time Ritter and Drake w34,35x popularized the use of the technique
through the introduction of a porosimeter capable of reaching 10 000 psi. Note: 1
psi s 6.895 = 10y3 MPa.. Nowadays commercial porosimeters can reach 60 000 psi
and perform intrusionrextrusion experiments automatically. The basic components
and the operation of modern mercury porosimeters have been described in detail
elsewhere w1,3x. Among the different experimental approaches available, perhaps
the most controversial issues are the choice of scanning versus stepwise
intrusionrextrusion data acquisition and the choice or determination of contact
angles. These and other experimental aspects are briefly reviewed in turn next.

348

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

Fig. 2. Pore size distributions calculated from the curves shown in Fig. 1; a. assuming u I s u E s 1408;
b. after shifting u E to 106.58. Symbols: circles, intrusion; triangles, extrusion. For clarity, only one out
of every 100 data points collected is shown.

2.2.1. Scanning ersus stepwise data acquisition


Early porosimeters operated exclusively in stepwise fashion, i.e. raising or
lowering the applied pressure in steps. This approach had several drawbacks,
including: a. an inability to hold the pressure constant at any given step; b. the
need for long analysis times; c. the poor resolution obtained as a result of
collecting relatively few data points in order to construct pore size distribution
curves; and d. the possibility of skipping important information in pressure ranges
corresponding to the filling of narrow or multimodal pore sizes. All these drawbacks were eliminated by the introduction of the scanning technique pioneered by
Quantachrome w1,3x. In addition, Quantachrome porosimeters were designed to
balance mercury compression under pressure and its expansion due to compressive
heating of the hydraulic oil w1x. This advantage leads to blank runs whose intrusionrextrusion volumes are within 0.5% of full scale throughout the entire experi-

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

349

mental pressure ranges, thereby making blank run corrections unnecessary see
Section 2.2.4. w1,3x. Surprisingly, the application of the stepwise approach, including
the need for blank corrections, seems to have found a niche in the literature.
Supporters of the stepwise approach often invoke Darcys law w1x to justify qualitatively that the degree of mercury intrusion should be a function of time at any
pressure. If so, stepwise curves could be claimed to approach a more equilibrated
condition than scanning curves. However, the quantitative application of Darcys
law in combination with the well-known Poiseuilles law w1,3x leads, after allowing
for a gradual increase in mercury viscosity with pressure up to 18% between 1 and
60 000 psi w36x., to an expression of the form
t s 2.21 = 10y1 0 q 7.2 = 10y5rD P . L2rD 2 .

3.

where t in seconds. is the time it takes for mercury to flow into cylindrical pores
of length L and diameter D both in microns. as a function of applied pressure
difference D P in psi.. The predictions of the above equation for pore lengths of
100 m m probably an upper limit for most industrial powders. and D values
covering the range probed by mercury porosimetry are illustrated in Fig. 3. Also
shown in Fig. 3 are the intrusion pressure differences D P required to fill pores of
diameter D according to the Washburn equation. As an illustration, note that
pores with D s 3 m m would take about 0.01 s to fill at 10 psi, if they could be
filled at 10 psi; however, the Washburn equation indicates that pores with D s 3
m m could only be filled at 100 psi, and by the time that pressure difference is
reached those pores would be filled in only 0.001 s or so. Fig. 3 shows clearly that
the times required to fill 100 m m-long pores with mercury vary between about 10y5
s for the largest intrudable pores to 1 s for the smallest intrudable pores. These
quantitative predictions confirm that the filling of pores with mercury occurs
almost instantaneously and is not at all limited by viscous effects within the
experimental range of typical mercury porosimeters. The same conclusion was
reached on experimental grounds by others who, e.g. confirmed the insensitivity of
mercury intrusion and extrusion scanning curves to the applied scanning rates w25x.
It is therefore not surprising to find a fair agreement between porosimetry spectra
obtained via scanning and stepwise approaches on Quantachrome porosimeters, as
shown in Fig. 4. Yet the calculated pore size distributions Fig. 5. differ considerably for curves collected at approximately equal analysis times. This is so because
the stepwise approach can only generate tens of data points within the same time
frame and pressure range in which the scanning approach collects thousands.
In spite of all the above arguments, if one still chooses to apply the stepwise
method to generate mercury porosimetry data, it is advisable to minimize the
pressure drop during each pause w5x. This is desirable in order to prevent the curves
from falling too deeply within the hysteresis loop and thus undergo irreversible
changes. Time-related phenomena upon mercury intrusionrextrusion have been
ascribed to a slow and relatively minor mercury vapor transfer mechanism w1,3x, but
in some instruments their magnitude is considerable, and it could be related to
additional mercury intrusionrextrusion induced by heat dissipation effects.

350

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

Fig. 3. Time required to fill 100 m m-long cylindrical pores of diameter D with mercury at 293.15 K
predicted by Darcy-Poiseuille laws..

Nonetheless, many commercial porosimeters allow for an automatic repressurization to the target pressure when the pressure changes. Particularly when samples
with relatively narrow pore size distributions are analyzed, the extent of depressurization and repressurization can affect the precision of the results w5x. It is also
advisable, therefore, to confirm that stepwise data are truly insensitive to the
choice of equilibration parameters by comparing the results of tests performed
using different equilibration parameters. A necessary condition to ensure equilibrium is the insensitivity of the results to the choice of experimental variables.
2.2.2. Contact angle measurement
The exact value of the contact angle between mercury and solid surfaces
depends on many factors, including the solid surface chemistry, cleanliness, and
roughness, as well as the purity of the mercury and whether the mercury is
advancing or retreating on the solid surface. Ritter and Drake originally adopted
1408 as an average contact angle representative of a wide variety of materials
w34,35x. However, in many cases this is a crude approximation see Table 1. that
can lead to errors in pore size estimates of the order of up to 50% w3x. For utmost
accuracy in mercury porosimetry calculations, contact angles should be measured
for each material tested. However, assigning the same contact angle value to
compare the data of chemically similar materials is qualitatively useful w1x.
There are a wide variety of methods available to measure contact angles w1,3x.

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

351

Fig. 4. Comparison of scanning and stepwise curves obtained on a Quantachrome Poremaster analysis
times of ca. 15 min in both cases.. Symbols: triangles, scanning data; circles, stepwise data; open
symbols, intrusion; filled symbols, extrusion.

Preferably scanning curves should be processed using dynamic e.g. advancing w1x.
contact angles, whereas stepwise curves may be more compatible with static or
Sessile-drop. contact angles w3,15x. In either case, every effort should be made to
assure the purity of the mercury used see Section 2.2.3..
2.2.3. Mercury purity
High purity mercury should be employed because its purity affects both contact
angles Section 2.2.2. and surface tension values required for data interpretation. It
is advisable to employ acid washed, dried and distilled preferably doubly- or
triply-distilled. mercury, even though its cost can be relatively high. Recycling
mercury directly after decanting the contents of spent sample cells is not advisable.
Similarly, spent hydraulic oil should not be recycled because it carries particles that

352

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

Fig. 5. Comparison of pore size distributions calculated from the mercury intrusion and extrusion
curves shown in Fig. 4. Symbols: triangles, scanning data; circles, stepwise data; open symbols, intrusion;
filled symbols, extrusion.

contaminate the mercury and change its dielectric and flow properties. Constant
levels of impurities in the mercury employed should not be assumed to lead to
constant shifts in intrusion or extrusion data. However, small levels of impurities
may introduce uncertainties as small as those expected of pure surface tension
values as a function of pore size w42,43x.
2.2.4. Blank corrections
Subtracting an empty cell run from an actual sample analysis is often invoked as
a means to correct primarily for the apparent volume intruded due to compression

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

353

Table 1
Examples of contact angles reported in the literature
Material

Contact angle 8.

Reference

Alkali borosilicate glass


Aluminum oxide
Calcite
Carbon
Cement
Clay minerals
Coal
Glass
Mica
Nickel
Oxide-type surfaces
Paraffin wax
Pyrite
Quartz
Steel
Titanium dioxide
Zinc oxide

153
127,142
146
155, 162
125
139]147
142
135]140
126
130
140
149
146
132]147
154
141, 160
141

w37x
w37, this studyx
w38x
w37,40x
w40x
w37x
w38x
w38x
w38x
w41x
w38x
w39x
w38x
w37x
w39x
w37x
w37x

of mercury, but also of the sample itself, of the sample cell, and of the rest of the
instrument components w3,44x. Properly designed instruments are built so as to
minimize the latter two effects w1x. Accordingly, the compressibility of a sample can
be accurately determined Section 2.4.7. provided that the mercury compression
does not affect the volume in the dilatometer stem significantly.
The variation in mercury compressibility b Hg in psiy1 . with pressure P in psi.
at ambient temperature can be derived from experimental data w45]47x as: b Hg s
2.7735 = 10y7 ]6.5331 = 10y1 3 P. This implies that a typical sample cell filled with
ca. 6 cc of mercury could experience a drop in mercury level within the dilatometer
stem 0.2 cm i.d.. of at most 0.5% of the total stem height at pressures up to 60 000
psi. This is the order of magnitude typically observed at 60 000 psi during blank
runs performed on Quantachrome porosimeters.
In any event, utmost accuracy would be achieved by subtracting from the sample
analysis results those of an analogous test on a nonporous sample of similar bulk
volume and compressibility if realistically possible., in order to correct the results
for compressibilities and temperature changes. This is so because the compressibilities of the various components in the system will augment even if negligibly. the
measured intrusion values while the pressure-induced heating and consequent
expansion of the system reduces the measured volumes. For a given porosimeter
design, either one of these effects could dominate. Hence, the results of blank tests
can be used to assess a given porosimeter in terms of the relative importance of an
apparent intrusion compressibility dominant. or an apparent extrusion heating
dominant. on the sample analyses.

354

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

2.3. Range of applicability


Mercury porosimetry is unique in its ability to complement the pore size ranges
other pore characterization approaches w1]3x. Two aspects that often raise questions are the types of samples that can be analyzed and the pore size ranges that
are accessible to mercury. Both aspects are discussed in this section.
2.3.1. Sample types
In principle, mercury porosimetry is applicable to all kinds of solid materials. In
practice, materials whose structure could compress or even collapse at high
pressures require corrections for compressibility or analyses at relatively low
pressures. On the other hand, certain metals show a tendency to amalgamate
readily with mercury w44x. One could thus investigate mercury amalgamation
mechanisms with, e.g. gold or silver, by monitoring apparent intrusion rates. Less
noble metals show less of a tendency to amalgamate with mercury because they are
protected by a thin surface oxide layer that can delay amalgamation rates sufficiently to allow regular intrusion measurements to be made.
2.3.2. Pressure and pore size limits
Mercury porosimeters measure intrusion and extrusion pressures and volumes,
and calculate pore sizes from measured pressures. Hence, the pore size ranges
covered by standard porosimeters are limited by the pressures they can achieve.
Modern porosimeters allow the commencement of intrusion tests at pressures as
low as ca. 0.5 psi w3x. This initial pressure is required to force mercury to fill the
sample cell. Regardless of the filling angle, due to their own height all samples are
invariably subjected to a mercury head pressure of the order of 0.1 psi. that could
induce some intrusion prior to commencing an experiment w1x. To reduce the latter
possibility, intrusion experiments are often commenced at pressures slightly higher
than 0.5 psi but seldom higher than 1 psi. On the other hand, the upper pressure
limit of commercial porosimeters is set by safety design margins to 60 000 psi.
The estimation of pore size limits from experimental pressure limits, using the
Washburn equation, requires using an assumed or measured contact angle u ; see
Section 2.2.2.. Table 2 lists typical pore size ranges calculated from different
assumed contact angles. From the Washburn equation, it follows that instruments
that work in the same pressure range cannot cover different pore size ranges. This
observation again highlights the importance of paying close attention to the value
of the contact angle used when converting applied pressures to calculated pore
sizes.
2.4. Interpretation of mercury porosimetry data
2.4.1. Particle size distribution
The notion that mercury intrusion curves yield information about pores between
particles has led several researchers to postulate that the same intrusion curves
also contain structural information about the dimensions of the particles them-

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

355

Table 2
Effect of assumed contact angle on calculated pore size range covered by mercury porosimeters
Contact
angle 8.

Ka Wa shburn
m m.psi.

Calculated upper
pore size limit,
m m pressure s 1 psi.

Calculated lower
pore size limit,
m m pressure s 60 000 psi.

1408
1308

213.4
179.1

213.4
179.1

0.00356
0.00298

Pore size s K Wa shburn rpressure.

selves. Two theories in particular have found general acceptance in the literature:
the simple mercury breakthrough theory of Mayer and Stowe w48x and the integral
patchwise approach of Smith and Stermer w49x. Both are described in turn next.
2.4.1.1. Mayer]Stowe (MS) theory. The manner in which mercury penetrates a bed
of uniform spherical particles was examined in detail by Mayer and Stowe w48x, who
postulated that the breakthrough pressure P b required to force mercury to
penetrate the void spaces between packed spheres of diameter D is given by
Pb s

Kg
D

4.

where K is the so-called MS proportionality constant and g is the surface tension


of mercury. The dimensionless MS constant K was shown to be a complex function
of the mercury contact angle u with, and the packing arrangement of, the particles.
Using intraparticle porosity ea as a measure of packing arrangement, it can be
shown that for randomly packed spheres ea s 37.5%. and a typical mercury
contact angle of u s 1408, K s 10.73. In general, K was found to increase with u
and to decrease with ea w50x. On the other hand, independent work confirmed that
the average particle coordination number Nc . can be estimated from w51x
Nc s

p
r Hg
1y
r He

5.

where r Hg and r He are the bulk particle. and helium true. densities of the
material.
The validity of the MS theory has been confirmed experimentally by quantitative
comparison with particle size distributions derived from independent techniques
such as X-ray sedimentation and electron microscopy. The best agreement among
different techniques was found for solids with narrow monomodal particle size
distributions and for materials with relatively few interparticle voids. Particle shape
is generally thought to play a minor role in the results, although dimensionless

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

356

intrusion or extrusion particle shape factors f have occasionally been introduced


in the MS expression as follows:
Pb s

fKg

6.

2.4.1.2. Smith]Stermer (SS) theory. Upon applying Mayer and Stowes approach,
Smith and Stermer observed that even for narrow particle size distributions
mercury intrusion curves generally do not exhibit unique breakthrough pressures
w52x. These deviations were attributed to the fact that particles of a given size do
not pack in separate regions of a powder bed, as was implicitly assumed by others
w48x. Smith and Stermer generalized Mayer and Stowes approach by postulating
w49x that the total volume of mercury Vi intruded into a bed of particles of
different sizes at any given pressure Pi is the sum of volumes intruded between
particles of each size D according to
Vi s

H K P , D. F D. d D

7.

where K Pi , D . is a kernel which describes mercury intrusion between particles


of fixed diameter D and F D . is the particle size distribution function. Using
experimental Vi and Pi values and a generalized kernel function, Smith and
Stermer adopted a numerical approach in order to solve the previous equation for
F D .. Their numerical approach called for the division of the expected particle size
range into discrete intervals within which the distribution function F Dj . for its
average pore size Dj . could be evaluated:
Vi s K Pi , Dj . F Dj . D Dj

8.

If the summation is applied in the interval 1 F j F N, and the equation is solved


for 1 F i F M, evaluating F D . involves solving simultaneously a system of M by
N equations subject to a non-negativity constraint such as Lawson and Hansons
Non-Negative Least Squares NNLS. approach w53x, i.e. making the function
Es

Vexperiment P . y Vtheory P .

dD

9.

converge within a minimum accuracy E. To minimize the oscillatory behavior of


the calculated distribution function, a smoothing term can be added w54x as follows:
Es

Vexperiment P . y Vtheory P .

dD q a

H F D.

dD

10.

where a s 10 d and d is the so-called regularization parameter. To speed up the


convergence of the iterative solution of the above summations, a relaxation

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

357

parameter, l, can also be incorporated by correcting successive iteration values as


follows:
F D kq 1 . s F D k . q lw F9 D kq1 . y F D k .x

11 .

where F9 D kq 1 . is the iteration value before applying the relaxation approach.


Typical l values fall between 0 and 2, with l s 1 implying that the relaxation
approach is not applied.
The particle size distributions predicted using the MS and the SS methods have
already been applied successfully to a wide variety of materials. Fig. 6 shows a
comparison of MS and SS model predictions for a particle size reference material
PSRM 4000. supplied by Quantachrome. The agreement among particle size
distributions estimated from X-ray sedimentation Microscan II., laser diffraction
CILAS 1064., the MS and the SS methods for this reference material is remarkably good.
An example of how these methods can already be used to address problems of
industrial relevance is that of the characterization of carbon blacks. Carbon blacks
are made up of very fine and compact particles formed usually through the
high-temperature pyrolysis of gaseous hydrocarbons. These nanometer-sized particles tend to agglomerate and thus generate a complex structure that makes their
particle size analysis very difficult. Currently the most widely used approach to
characterize particle sizes of carbon blacks is by electron microscopy. However, the
technique is laborious, time consuming, expensive, and it requires technicians with
special skills. Moreover, the results are uncertain because of representative sampling errors and because they depend on the number of particles counted and on
the assumption that 2-dimensional pictures can characterize 3-dimensional morphology. Mercury porosimetry is free of all these undesirable limitations, and can
provide satisfactory results in a few minutes per sample. The validity of the
technique was tested using a series of reference carbon blacks available from
Titan Products. routinely used by ASTM Committee D-24 to monitor the precision
of their standard test methods. Fig. 7 shows the MS particle size distributions
obtained for these carbon blacks, and Table 3 compares their estimated modal
particle diameters with those expected from their typical size ranges determined by
electron microscopy tests. The results are again quite encouraging, considering the
speed and simplicity of the mercury porosimetry technique both of which are
essential in industrial carbon black manufacturing facilities .. Moreover, the mercury porosimetry technique provides additional information such as pore size
distributions and, perhaps more importantly, mercury intrusion surface areas, given
the critical relevance of surface areas especially external or nonmicroporous
surface areas. to the reinforcing ability of different commercial carbon black
grades.
An obvious advantage of the MS method in the previous examples is its higher
resolution compared to that of the highly computationally-demanding SS method
see Fig. 6.. It is noteworthy that the MS and the SS methods do not necessarily
yield identical results. Work is in progress to establish the sensitivity of each

358

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

Fig. 6. Comparison of particle size distribution predicted for a reference material PSRM 4000. using
the MS and the SS methods. Continuous curve, MS method; filled circles, SS method.

method to parameters such as sample type, particle size range considered, multimodal distributions, surface chemistry, etc.
2.4.2. Inter- and intraparticle porosity
Porosity is defined as the percentage of void space in a solid. Total porosity e . is
often evaluated from mercury density r Hg . and helium density r He . values as
follows:

e s 100 1 y

r Hg
r He

12.

This definition accounts for all pores smaller than those filled by mercury ca. 14.5

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

359

Fig. 7. Particle size distributions estimated for series SRB-5 reference carbon blacks using the MS
method. Symbols: crosses, SRBA5; filled circles, SRBB5; x-marks, SRBC5; open triangles, SRBD5; open
squares, SRBE5; dots, SRBF5.

m m in diameter at standard pressure. and at the same time accessible to helium


in diameter., whose total pore volume Vp . is
i.e. open pores greater than ca. 3 A
Vp s

r Hg

r He

13.

In powders, both intrapore spaces between particles. and interpore voids within
particles. contribute to the total porosity. The demarcation between interior pores
and voids between particles is often unclear. In fact, some materials possess
intraparticle and interpores of the same magnitude, in which case making a
distinction between them is quite complicated. But more often than not distinction

360

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

Table 3
Comparison of mean particle sizes of series SRB-5 reference carbon blacks expected from electron
microscopy as cited in ASTM D 1765. and estimated using mercury porosimetry MS method.
Reference
carbon black
grade.

Typical average particle


diameter nm.
cited in ASTM D1765.

Modal particle
diameter nm.
MS method.

SRB A5 N135.
SRB B5 N330.
SRB C5 N220.
SRB D5 N762.
SRB E5 N660.
SRB F5 N683.

11]19
26]30
20]25
61]100
49]60
49]60

19
28
20
51
59
65

is clear because the sizes of interpores greatly exceed those of intrapores. Hence,
bimodal distributions are frequently observed, in which case the boundary could be
defined as the cumulative intruded volume curves inflection point. On the other
hand, it is worth noting that the volume of interpores is related to the packing
characteristics of the sample bed. Therefore, loose beds will yield different interpore volumes than tapped beds, with the latter yielding more reproducible results
than the former ones.
To distinguish among the various types of porosities present in a powdered
sample, the following equations can be used:
Interparticle Porosity % . s 100
Intraparticle Porosity % . s 100

Va

14 .

Vb
Va y V b
Vc y V b

Mercury Intrusion Porosity % . s 100

Va
Vc

15 .
16 .

where Va is the mercury volume intruded at any given pressure, V b is the mercury
volume intruded at a user-defined Intrapore Filling Pressure Limit usually at the
inflection point of the plateau between low and high pressure steps in mercury
intrusion curves., and Vc is the mercury volume intruded at the maximum experimental pressure attained.
Fig. 8 illustrates an instance in which both intra- and interpore spaces appear to
be present on a sample, judging from the shape of the mercury intrusion curve
Fig. 8a.. This kind of curve invariably leads to a seemingly bimodal pore size
distribution Fig. 8b.. In the past, people experienced in mercury porosimetry
would recognize the lower pressure intrusion as being related to intrapore volume
filling. However, they had no direct way of confirming if that was truly the case or
if they were actually dealing with a bimodal distribution of pore sizes. The sample
shown in Fig. 8 is in fact a reference material currently being used for particle size

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

361

Fig. 8.

distribution round-robing tests by ASTM Committee D-32 on Catalysts. Its particle


size distribution, calculated from mercury intrusion data up to its first inflection
point ca. 400 psi., agrees fairly well with the distribution expected from laser
diffraction CILAS 1064. measurements}see Table 4. This indicated that one is
indeed dealing with a monomodal distribution of pore sizes in a material whose
pore and particle size distributions can be easily, quickly and reliably estimated
from a single mercury intrusion experiment.
2.4.3. Pore tortuosity
When modeling the diffusion of fluids in porous solids it is often reported that
the effective or measured. diffusivity Deff differs from the theoretical or bulk
fluid. diffusivity D b by a factor related to the structure of the solid as follows:
Deff s

D b uc

17.

362

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

Fig. 8. Determination of particle size distribution, pore size distribution and porosities of a reference
catalyst via a single mercury intrusion experiment. a. Cumulative intrusion curve only one of each 100
experimental data points shown.; b. Differential intrusion curve only one of each 10 experimental data
points shown..

where uc is the pore volume fraction and t is the tortuosity factor. This effective
tortuosity factor lumps all deviations from straight diffusion paths into a single
dimensionless parameter which usually falls between about 1 and 7, with a value of
2 being associated with nonintersecting cylindrical pores. Being an average value,
the effective tortuosity is thought by some to be relatively insensitive to changes in
pore structure characteristics w39x. Nonetheless, using Ficks first law to describe
fluid diffusion through cylindrical paths, Carniglia derived the following expression
w55x:

tc s 2.23 y 1.13Vco r Hg

18.

where Vco is the total specific pore volume which can be approximated by the
mercury volume intruded at the maximum experimental pressure attained, Vc . and

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

363

Table 4
Comparison of particle size distribution values obtained by laser diffraction CILAS 1064. and mercury
intrusion MS method. for a reference catalyst
Weight %
finer than
diameter listed

Laser diffraction data


CILAS 1064. m m.

Mercury intrusion data


MS method. m m.

10%
50% median.
90%

34.0
60.6
98.4

34.8
58.5
124.2

r Hg is the bulk or particle density of the solid. Carniglia pointed out that this
relationship was generally applicable within 0.05 F Vco r Hg F 0.95, so that if cylindrical pores truly prevailed in a material, its tortuosity should not deviate greatly
from t f 2. However, experimental t values significantly larger than 2.23 led
Carniglia w55x to expand the previous equation into a more generalized form:
tc s 2.23 y 1.13Vco r Hg . 0.92 y .

1q E

19 .

where
ys

D Vi

4
s

di

20 .

and
S s
D Vi s
di s
E s

total surface area;


change in pore volume within a pore size interval i;
average diameter within a pore size interval i; and
pore shape exponent s 1 for cylinders..

Using S as the BET surface area, Carniglia reported fair agreement between
tortuosity factors computed with the above model and calculated from experimental diffusion measurements for a wide variety of metal oxides and catalysts, with
few E values falling below 1.0 w55x.
2.4.4. Permeability
The permeability K of a solid is defined as the ability of the solid to let fluids
travel across it w13x. Various experimental methods are available to evaluate the
permeability of porous solids, but they often produce different results and semiempirical correlations at best. among them. This is so because different approaches
are affected to different degrees by factors such as head pressure, moleculerpore
opening size ratio, and by adsorption. The use of mercury intrusion for permeability determinations can be regarded as qualitatively useful when comparing, e.g.
applications involving gas or liquid permeabilities. A match among the various
permeability estimates is in principle possible if a slip correction is allowed as is

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

364

often done to match the results of liquid and gas permeability estimates w39x. A
fundamental limitation of the mercury-permeability method is that it needs to
assume a model for the pore structure. This model usually adopted is a medium
consisting of a bundle of parallel capillaries w39x, which is widely regarded as an
adequate frame of reference. Permeability, like tortuosity, is affected by bed
packing efficiency; the most consistent results are again obtained on fully tapped
samples or, alternatively, by compressing a powdered sample into a tablet by
applying a compression force known to minimize interparticle porosity effects w39x.
Numerous theoretical attempts have been made to relate the permeability K of a
solid to intrinsic and more readily measurable properties, such as porosity and pore
diameters. One such approach w56x models the flow of fluids across straight
cylindrical channels in a bed of powder by combining Darcys and Poiseuilles laws
w1x to obtain
Ks

f d p2
32

21.

where f is the powder bed porosity and d p is the average mean volume. diameter
of the pores. Correction for nonuniform pore characteristics leads to an expression
of the general form
Ks

f d p2
16t

22.

where t can be taken to represent the effective tortuosity of the pores. If the pores
are assumed to be straight cylindrical capillaries then t s 2.
As an illustration of the usefulness of quantifying mercury permeabilities,
consider the pore size distributions of two different kinds of filter paper shown in
Fig. 9. One filter paper has more and larger pores than the other, and could thus
be expected to be more permeable. But by how much? If pore sizes or pore
volumes did not correlate directly with filter performance, one would perhaps
wonder if tortuosity effects were involved. Yet the tortuosities of these samples,
calculated from the same mercury intrusion experiments, were similar see Table
5.. Table 5 also shows the calculated mercury permeabilities for these samples,
which differ by a factor of ca. 20. These calculated permeabilities are much lower
than those calculated for, e.g. commercial paper towels ) 10 000 nm2 . or laboratory kimwipes ) 1000 nm2 ., as expected. Being more akin to filter applications, it
remains to be confirmed whether simple quantitative permeability estimates such
as those made here correlate better with the performance of filters or of other
materials. Nonetheless, it seems clear that even relative permeability estimates can
provide valuable information about the structure of porous samples in general.
2.4.5. Throat r pore ratios
Several computer simulations and models in the literature have treated porous
solids as a network of empty chambers pores. interconnected by a network of

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

365

Fig. 9. Pore size distribution of various kinds of paper. Symbols: cross, filter 1; filled circle, filter 2.

smaller channels throats. w13,57,58x. Indeed, network models are often applied to
the characterization of solids such as porous rocks and agglomerated particles
w13,58x. One inference from network models is the relationship between mercury
Table 5
Comparison of mercury permeabilities and tortuosities of two industrial filter papers
Filter paper
sample

Modal pore
diameter m m.

Pore
tortuosity

Mercury
permeability nm2 .

1
2

42
2

2.12
2.19

10.51
0.41

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

366

intrusion and extrusion curves and the shape of pores and throats w57x. From a
purely geometric point of view, the shapes of mercury intrusion curves could in
principle be related to pore constrictions throats., while those of mercury extrusion curves could be related to the shapes of cavities beyond the constrictions
pores.. In such a case, the throatrpore size ratio R TP,i would vary as a function of
the void fraction filled with mercury f i . as follows:
R TP ,i s

DI

PE

fE

cos u I

PI

fI

cos u E

/ / /
DE

23.
i

where D, P, f and u are the throat or pore size, the applied pressure, the pore
shape factor and the mercury contact angle, with the subscripts I and E referring to
intrusion and extrusion curves, respectively, and

fi s

Va

24 .

Vc

with Va and Vc defined previously see Section 2.4.2.. If the shape factors are
assumed to be the same for throats and pores as is often done implicitly w57x., and
the intrusion and extrusion contact angles are assumed to be constant, R TP is
simply given by the ratio of extrusion to intrusion pressures, PE rPI , at any given
f i . The variations in PE rPI ratios with f i have been correlated with pore shape
characteristics of a variety of agglomerated microparticles w57x. On the other hand,
if the intrusion and extrusion contact angles are allowed to vary w1x, the relative
contributions of geometric and surface chemical effects on the shapes of mercury
porosimetry curves can be assessed upon examination of the resulting D IrD E . i
versus f i curves.
For reference purposes, a characteristic throatrpore ratio, R TPC , can be defined
as
R TPC s

Dmax,I
Dmax,E

25.

where Dmax,I and Dmax,E are the modal throat and pore sizes, respectively.
An example of the kind of information that can be extracted from variations in
PorerThroat ratios is shown in Fig. 10. As pointed out by Conner et al. w57x, if
mercury intrusionrextrusion hysteresis occurred solely due to differences in fixed
intrusion and extrusion contact angles Section 2.1.1., plots like Fig. 10 could be
expected assuming f I and f E to be constant. to yield horizontal lines throughout
f i . Fig. 10 shows that, although many samples pharmaceutical excipients and
papers being the cases in point. exhibit fairly horizontal lines, there are exceptions
to such behaviour, and that these exceptions are not necessarily related to the type
of material tested. It was suggested for agglomerates w57x that the ratios are related
to packing characteristics and to the shapes of their corresponding void spaces.
Accordingly, as plotted in Fig. 10 f i s D IrD E , in order to work with ratios

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

367

Fig. 10. Variation in throatrpore ratios of selected materials. Symbols: x-marks, pharmaceutical
excipient 1; open triangles, pharmaceutical excipient 2; open squares, paper 1; crosses, paper 2.

between 0 and 1., low ratios would be indicative of voids resembling those between
rods, and high ratios would resemble more those expected of voids between plates.
Similarly, slopes in these plots would suggest changes in packing arrangements
andror pore shapes.
2.4.6. Fractal dimension
The fractal dimension D of a solid is a parameter that characterizes the degree
of roughness of its surface w59x. Perfectly flat surfaces expose areas that can be
calculated as a function of a characteristic dimension, e.g. 4p R 2 for a nonporous
sphere of radius R. Any surface roughness or porosity would increase the spheres
surface area up to an extreme point in which the sphere would be so porous that its

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

368

entire volume could be occupied by pore walls. At this hypothetical point, the
surface area would be proportional to the volume of the sphere, i.e. 4r3p R 3. Real
solids expose areas which, following fractal arguments w59x, are proportional to R D ,
with their fractal dimension D ranging between 2 for flat surfaces and about 3 for
extremely rough surfaces.
For pore wall surfaces in general it was shown w60,61x that the pore size
distribution function, ydVrd r, could be expressed as
y

dV
dr

s k 1 r 2yD .

26.

where k 1 is a proportionality constant, r is the pore radius, and D is the fractal


dimension. It follows from the Washburn equation that fractal dimensions can be
derived from mercury porosimetry data according to
dV
dP

s k 2 P Dy4.

27.

where k 2 is another proportionality constant and P is the applied pressure. Taking


logarithms on both sides of the expression yields
log

dV

/
dP

s log k 2 . q D y 4 . log P .

28 .

Hence, D values can be derived from the slope of log dVrd P . versus log P . plots.
Smooth surfaces D s 2. would present slopes close to y2, whereas rougher
surfaces for which D approaches 3. would present slopes approaching y1.
Inspection of typical mercury porosimetry curves reveals that constant slopes in the
appropriate range are often encountered at pressure regions which coincide with
the filling or emptying of pores of distinct sizes. Hence, each step of a cumulative
intrusion or extrusion volume versus pressure plot can yield a unique D value that
characterizes the particular range and type of pores being filled or emptied at given
pressure ranges. Once pores of a distinct size become filled or emptied, the
mercury volume ceases to change significantly and dVrd P decreases, thereby
complicating fractal calculations in the transition regions between pore ranges of
materials with multimodal pore size distributions.
More recently, Neimark proposed a thermodynamic method for the evaluation
of surface fractal dimensions Dfs . from mercury porosimetry or gas sorption data
w61x. The approach essentially involves expressing the interface area of the adsorbed or the intruding fluid S . as a fractal function of the radius of curvature of the
meniscus of the pores affected at equilibrium a ., i.e.
S s K a 2yD f s.

29.

where K is a proportionality constant. In gas sorption experiments, a is given by


the Kelvin equation and S is calculated from the Kiselev equation w1,2x. The

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

369

successful application of the thermodynamic approach in the capillary condensation region of gas sorption isotherms led to its incorporation into commercial
software under the name Neimark-Kiselev NK. Method w62x. It is anticipated
that the analogous approach, when using the Rootare]Prenzlow equation w63x to
calculate the interface area S, will result in its inclusion in commercial mercury
porosimetry software as the NRP method w64x.
2.4.7. Sample compressibility
Mercury intrusion involves subjecting samples to hydrostatic pressures which are
applied equally in all directions. This means that upon intrusion the walls of all
pores penetrated by mercury at any given pressure are uniformly affected by
similar stresses. Hence, a collapse of the pore walls as they are filled with mercury
is in general unlikely. On the other hand, solid samples could in principle compress
and thus generate additional volume for mercury intrusion to take place. The
influence of sample compressibility on mercury porosimetry data can be assessed
by means of a solid compressibility factor b , defined as the fractional change in
solid volume Vs per unit of pressure,

bsy

1 dVs
Vs d P

s r He

dV
dP

30.

Most solids exhibit very low compressibilities typically of the order of 10y1 1
m2rN. which vary fairly linearly with pressure w46,65x. This implies that a typical
sample could compress by about 0.5% of its original volume when subjected to an
applied pressure of 60 000 psi. In agreement with these observations, mercury
porosimetry curves often exhibit small yet finite dVrd P slopes at pressures higher
than those required to fill all accessible pores in many solids. Hence, the compressibility factor b can be estimated from the slope of the linear portions of high-pressure mercury intrusion or extrusion curves. Similarly, a bulk modulus of elasticity
s 1rb . can be derived as a correlator of the elastic reversible deformation. or
plastic irreversible deformation. behavior of solids in general.
As illustrated in Table 6, ongoing work on oxygen-plasma treated activated
carbons Garca
et al., in preparation.. has uncovered a not unexpected. relationship between crystalline XRD parameters such as the average crystallite height Lc
and mercury compressibility. Interestingly, Table 6 also shows that there appears to
be an inverse correlation between compressibility and the adjusted extrusion
contact angle itself a function of surface chemistry w1x. that yields overlapping
mercury intrusion and extrusion curves. The implications of the latter correlation
are to be discussed elsewhere Garca
et al., in preparation..

3. Concluding remarks
The classical interpretation of mercury porosimetry data has been largely restricted to the generation of pore volumes and pore size distributions. This has

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

370

Table 6
Characterization of oxygen-plasma treated activated carbon via mercury porosimetry
Burnoff
%.

SA BET.
wm2 rgx

SA Hg.
wm2 rgx

Lc XRD.
wnmx

Vpore Hg.
wccrgx

Compressibility
=109 wm2 rNx

Extrusion contact
angle 8.

0.00
3.12
7.40
9.40
19.40

1050
1022
1021
1010
955

138.3
139.2
138.1
129.6
125.0

1.22
1.18
1.23
1.21
1.40

1.95
1.49
1.71
1.77
1.83

1.31
1.38
1.31
1.23
1.19

101.5
102.9
100.3
99.1
98.9

proven to be an unnecessary handicap imposed on a technique that is ideally suited


to handle the generation of high quality, high throughput characterization data.
The usefulness of applying alternative literature models to extract information
from mercury porosimetry curves has been demonstrated. Among the most promising models tested thus far, those generating particle size distributions, permeabilities and compressibilities have already led to useful correlations with performance
and other properties of the porous materials investigated. Undoubtedly the incorporation of such literature models into commercial software will provide answers to
many questions that could not be resolved through the application of the Washburn equation alone. It is expected that the development of reliable standard
reference materials and the analysis of solids of ideal structures will contribute
tremendously towards the long overdue maturation of a technique as versatile and
powerful as mercury porosimetry has already proven to be.

Acknowledgements
The author is privileged to have enjoyed very insightful discussions with Professor S. Lowell founder of Quantachrome Corporation. on the nature of mercury
porosimetry and other solid characterization techniques. Thanks are also due to Dr
M.A. Thomas for useful comments and to Dr J.M.D. Tascon
and his research
group at CSIC-Oviedo Spain. for supplying the oxygen-plasma treated carbon
samples.

References
w1x S. Lowell, J.E. Shields, Powder Surface Area and Porosity, 3rd ed., Chapman and Hall, New York,
1991.
w2x S. Gregg, K.S.W. Sing, Adsorption, Surface Area and Porosity, 2nd ed., Academic Press, London,
1982.
w3x T. Allen, Particle Size Measurement, Vol. 2: Surface Area and Pore Size Determination, 5th ed.,
Chapman and Hall, New York, 1997.
w4x J. Rouquerol, D. Avnir, C.W. Fairbridge, D.H. Everett, J.H. Haynes, N. Pernicone, J.D.F. Ramsay,
K.S.W. Sing, K.K. Unger, Pure Appl. Chem. 66 1994. 1739.
w5x ASTM D4284, Standard Test Method for Determining Pore Volume Distribution of Catalysts by

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372

w6x
w7x
w8x
w9x
w10x
w11x
w12x
w13x
w14x
w15x
w16x
w17x
w18x
w19x
w20x
w21x
w22x
w23x
w24x
w25x
w26x
w27x
w28x
w29x
w30x
w31x
w32x
w33x
w34x
w35x
w36x
w37x
w38x
w39x

371

Mercury Intrusion Porosimetry, ASTM Committee D-32 on Catalysts, Vol. 05.03, Conshohocken,
PA, 1994.
ASTM D2873, Standard Test Method for Interior Porosity of PolyVinyl Chloride. PVC. Resins by
Mercury Intrusion Porosimetry, ASTM Committee D-20 on Plastics, Vol. 08.01, Conshohocken,
PA, 1994.
ASTM D4404, Standard Test Method for Determination of Pore Volume and Pore Volume
Distribution of Soil and Rock by Mercury Intrusion Porosimetry, ASTM Committee D-18 on Soil
and Rock, Vol. 04.08, Conshohocken, PA, 1994.
DIN 66133, Determination of Pore Volume Distribution and Specific Surface Area of Solids by
Mercury Intrusion in German., Deutsches Institut fur
Normung e.V., 1993.
J. Rouquerol, F. Rodriguez-Reinoso, K.S.W. Sing, K.K. Unger Eds.., Characterization of Porous
Solids III COPS III., Studies in Surface Science and Catalysis, Vol. 87, Elsevier, Amsterdam, 1994.
B. McEnaney, T.J. Mays, in: J.W. Patrick Ed.., Porosity in Carbons: Characterization and
Applications, Halsted Press, New York, 1995, p. 93.
C.N. Satterfield, Heterogeneous Catalysis in Industrial Practice, 2nd ed., McGraw-Hill, New York,
1991, p. 131.
R.M. German, Particle Packing Characteristics, Metal Powder Industries Federation, Princeton,
1989, p. 275.
F.A.L. Dullien, Porous Media: Fluid Transport and Pore Structure, 2nd ed., Academic Press, New
York, 1992.
H.M. Rootare, Advanced Experimental Techniques in Powder Metallurgy, Plenum Press, 1970, p.
225.
A.W. Adamson, Physical Chemistry of Surfaces, 4th ed., Wiley, New York, 1982.
P.S. Laplace, Mechanique Celeste, Supplement to Book 10, 1806.
T. Young, in: G. Peacock Ed.., Miscellaneous Works, Vol. 1, J. Murray, London, 1855.
E.W. Washburn, Phys. Rev. 17 1921. 273.
E.O. Kraemer, in: H.S. Taylor Ed.., Treatise on Physical Chemistry, Van Nostrand-Reinhold, New
York, 1931, p. 1661.
J.W. McBain, J. Am. Chem. Soc. 57 1935. 699.
H. Giesche, in: S. Komarneni, D.M. Smith, J.S. Beck Eds.., Advances in Porous Materials, Mater.
Res. Soc. Symp. Proc., Vol. 371, Materials Research Society, Pittsburgh, 1995.
H. Giesche, K.K. Unger, U. Muller, U. Esser, Colloid Surf. 37 1989. 93.
G.P. Matthews, C.J. Ridgway, M.C. Spearing, J. Colloid Interf. Sci. 171 1995. 8.
C.D. Tsakiroglou, A.C. Payatakes, J. Colloid Interf. Sci. 137 1990. 315.
W.C. Conner, Jr., A.M. Lane, A.J. Hoffman, J. Colloid Interf. Sci. 100 1984. 185.
W.C. Conner, Jr., A.M. Lane, J. Catal. 89 1984. 217.
S. Lowell, J.E. Shields, Powder Technol. 38 1984. 121.
S. Lowell, J.E. Shields, J. Colloid Interf. Sci. 83 1981. 273.
S. Lowell, personal communication, 1994.
Available in different disk sizes and pore diameters from Structure Probe, Inc., SPI Supplies, West
Chester, PA, USA.
H. Iwata, I. Hirata, Y. Ikada, Langmuir 13 1997. 3063.
A.A. Liabastre, C. Orr, J. Colloid Interf. Sci. 64 1978. 1.
A. Neimark, personal communication, 1996.
H.L. Ritter, L.C. Drake, IEC Anal. Ed. 17 1945. 782.
L.C. Drake, H.L. Ritter, IEC Anal. Ed. 17 1945. 787.
R.C. Reid, J.M. Prausnitz, B.E. Poling, The Properties of Gases and Liquids, 4th ed., McGraw-Hill,
New York, 1987.
A.W. Neumann, R.J. Good, in Surface and Colloid Science, vol. 11, Plenum Press, New York, 1979,
Chapter 2.
J.R. Anderson, K.C. Pratt, Introduction to Characterization and Testing of Catalysts, Academic
Press, New York, 1985.
D.M. Grove, in: R.L. Bond Ed.., Porous Carbon Solids, Academic Press, New York, 1967, Chapter
4.

372

w40x
w41x
w42x
w43x
w44x
w45x
w46x
w47x
w48x
w49x
w50x
w51x
w52x
w53x
w54x
w55x
w56x
w57x
w58x
w59x
w60x
w61x
w62x
w63x
w64x
w65x

C.A. Leon
y Leon
r Ad . Colloid Interface Sci. 76]77 (1998) 341]372
D.N. Winslow, S. Diamond, J. Mater. 5 1970. 564.
D.N. Winslow, J.J. Shapiro, ASTM Bull. TP49 1959. 39.
E.A. Guggenheim, Trans. Faraday Soc. 36 1940. 397.
O. Kadlec, Adv. Sci. Technol. 1 1984. 177.
J.J.F. Scholten, in: R.L. Bond Ed.., Porous Carbon Solids, Academic Press, London, 1967, Chapter
IV.
R.H. Perry, D.W. Green Eds.., Chemical Engineers Handbook, 6th ed., McGraw- Hill, New York,
1984.
R.C. Weast Ed.., CRC Handbook of Chemistry and Physics, 64th ed., CRC Press, Boca Raton, FL,
1984.
T. Lyman Ed.., Metals Handbook, 8th ed., American Society for Metals, Metals Park, OH, 1961,
vol. 1.
R.P. Mayer, R.A. Stowe, J. Colloid Interface Sci. 20 1965. 893.
D.M. Smith, D.L. Stermer, Powder Technol. 53 1987. 23.
R. Pospech, P. Schneider, Powder Technol. 59 1989. 163.
R. Ben Aim, P. Le Goff, Powder Technol. 2 1968. 1.
D.M. Smith, D.L. Stermer, J. Colloid Interface Sci. 111 1986. 160.
C.L. Lawson, R.J. Hanson, Solving Least-Squares Problems, Prentice-Hall, NJ, 1974.
J.P. Butler, J.A. Reeds, S.W. Dawson, SIAM J. Numer. Anal. 18 1981. 381.
S.C. Carniglia, J. Catal. 102 1986. 401.
A.E. Scheidegger, The Physics of Flow Through Porous Media, University of Toronto Press,
Canada, 1974.
W.C. Conner, C. Blanco, K. Coyne, J. Neil, S. Mendioroz, J. Pajares, in: K.K. Unger, J. Rouquerol,
K.S.W. Sing, H. Kral Eds.., Characterization of Porous Solids COPS I., Studies in Surface Science
and Catalysis, vol. 39, Elsevier, Amsterdam, 1988, p. 273.
F.A.L. Dullien, Porous Media, Academic Press, New York, 1979.
D. Avnir Ed.., The Fractal Approach to Heterogeneous Chemistry, Wiley, New York, 1989.
P. Pfeiffer, D. Avnir, J. Phys. Chem. 79 1983. 3558.
A.V. Neimark, Adv. Sci. Technol. 7 1990. 210.
Quantachrome Corporation, AUTOSORB Gas Sorption Analyzer Software, Boynton Beach, 1996.
H.M. Rootare, C.F. Prenzlow, J. Phys. Chem. 71 1967. 2733.
Quantachrome Corporation, POREMASTER Mercury Porosimeter Software, Boynton Beach,
1997.
E.M. Suuberg, S.C. Deevi, Y. Yun, Fuel 74 1995. 1522.

Você também pode gostar