8.

57

Thermal Conductivity Detectors

J. E. BROWN

(1969, 1982)

B. G. LIPTK

(1995, 2003)

AT
TH.
CONDUCTIVITY

TO
RECEIVER
Flow Sheet Symbol

Applications:

Gases or vapors; best suited for binary gas applications such as detector on chromatographs or leak detection or for hydrogen and helium analyses, because these
gases have high thermal conductivity

Design Pressure:

Near atmospheric

Sample Temperature:

35 to 110F (2 to 43C)

Cell Materials of
Construction:

Brass, stainless steel, Monel

Range:

The full span of the analyzer should correspond to a minimum of a 2% change in

the thermal conductivity of the gas mixture.

Inaccuracy:

Inaccuracy is 1 to 2% of full scale for binary samples when the thermal conductivity
of each constituent is accurately known. Published thermal conductivity data can be
in error by as much as 5%. Interpretation of readings on multicomponent mixtures
require additional measurements and analysis.

Cost:

A portable leak detector with 10 cc/sec sensitivity costs $1600; industrial analyzers
for binary mixtures range from $2500 to $6000; analyzers with higher sensitivity
and/or in corrosion-resistant materials cost from $6000 to $12,000.

Partial List of Suppliers:

ABB Process Analytics (www.abb.com/analytical)

Cole-Parmer Instrument Co. (www.coleparmer.com)
Gow-Mac Instrument Co. (www.gow-mac.com)
Ion Track Instruments (www.iontrack.com)
MSA Instrument Div. (www.msa.com)
Siemens Applied Automation (www.sea.siemens.com)
Teledyne Analytical Instruments (www.teledyne-ai.com)
Thermco Instrument Corp. (www.thermco.com/)

INTRODUCTION
Composition measurement by detecting the thermal conductivity of gases is one of the simplest and oldest methods of analyzing process streams. Early developments by the British
resulted in an instrument of this type, which was called a katharometer or catharometer. The name still persists in Europe.
This technique takes advantage of the facts that different
substances have a varying capacity to conduct heat energy
from a heat source. This ability differs for each gas. It is
called thermal conductivity and can be expressed in various
2
2
unit systems such as BTU/hr/ft /F/in.; W/sec/cm /C/cm;
2
kiloergs/sec/cm /C/cm, and so on.

This is a simple, rugged, inexpensive, reliable, and easily

maintained, but nonspecific, analyzer that can determine the
composition of only binary mixtures. It is not very sensitive,
nor is it very fast, but it is well suited for many chromatographic
and leak detection applications.
THERMAL CONDUCTIVITY
Thermal conductivity is often expressed as a factor relating
the ability of a particular gas to conduct heat to that of air at
various temperatures (Table 8.57a). In practice, continuous
thermal conductivity analyzers measure a change in heat dissipation by comparing the change with a reference condition.
1653

2003 by Bla Liptk

1654

Analytical Instrumentation

TABLE 8.57a
Thermal Conductivity Factors
R 0*

R100*

TABLE 8.57b
Ranges of Gas Mixture Compositions Suitable for Measurement
by Thermal Conductivity

Acetone

0.406

0.546

Acetylene

0.776

0.900

Air

1.000

1.000

Gases in the Mixture

Ammonia

0.897

1.086

Air in carbon dioxide

Argon

0.709

0.725

Air in helium

Range of
Concentrations of
the First Gas in
the Second

An Error
of 1% of
Full Scale
Corresponds To

05%

0.05%

02.5%

0.025%

Benzene

0.370

0.573

Air in oxygen

00%

0.4%

Carbon dioxide

0.614

0.690

Air in sulfur dioxide

01%

0.01%

Carbon monoxide

0.964

0.962

Argon in nitrogen

07%

0.07%

Chlorine

0.322

0.381

Carbon dioxide in air

07%

0.07%

Ethylene

0.735

0.919

Carbon dioxide in nitrogen

07%

0.07%

Ethane

0.807

0.970

Carbon dioxide in oxygen

06.5%

0.065%

Helium in air

00.5%

0.005%
(50 ppm)

Helium

6.230

5.840

Hydrogen

7.130

6.990

Methane

1.318

1.450

Helium in hydrogen

012%

0.12%

010%

0.1%

00.3%

0.003%
(30 ppm)

Nitrogen

0.996

0.996

Hydrogen in helium

Oxygen

1.043

1.052

Hydrogen in nitrogen

Pentane(n)

0.520

0.702

Refrigerant 12

0.354

0.356

Sulfur dioxide

0.344

0.377

*R0, R100 = Thermal conductivity of gas/thermal conductivity of air

at 0 and 100C, respectively.

Nitrogen in argon
Nitrogen in carbon dioxide

05%

0.05%

05%

0.05%

Nitrogen in hydrogen

02.5%

0.025%

Nitrogen in oxygen

055%

0.55%

Oxygen in air

038%

0.38%

Oxygen in carbon dioxide

04.5%

0.045%

Oxygen in nitrogen

052%

0.52%

Sulfur dioxide in air

03%

0.03%

Measurement Ranges
It has been known for more than a century that the heatconducting ability of various gases differs considerably.
Therefore, by measuring the thermal conductivity of a
binary mixture, one can determine the composition of the
mixture. The accuracy of the measurement is a function of
the reliability of the thermal conductivity data used for the
gases making up the mixture, and those data are not always
accurate.
Some of the common thermal conductivity analyzer
applications include the measurement of hydrogen in air,
nitrogen, carbon dioxide, carbon monoxide, argon, blast furnace gases, and reformer gases; helium in air or nitrogen;

methane in air; propane in air; Freon in air; and carbon

dioxide in air, nitrogen, or flue gases.
A general rule is that the full span of the analyzer should
correspond to a minimum 2% change in the thermal conductivity of the gas mixture. Table 8.57b gives the full-scale
ranges of some binary gas mixtures that will result in at least
a 2% change in the thermal conductivity of the mixture. It
can be noted that, as the thermal conductivity of a gas
(helium, hydrogen) increases, so does the sensitivity of measurement; therefore, the range can be narrower.

2003 by Bla Liptk

THE TCD ANALYZER

The thermal conductivity detector (TCD) can measure the
composition of binary mixtures at moderate sensitivity.
Main Components
The major components of a thermal conductivity analyzer
are the measuring cell, regulated power supply, Wheatstone
bridge, and case temperature control.
The Detectors The measuring cell consists of a relatively
large mass of metal to provide a stable heat sink. Through
the metal block, flow passages are drilled or formed, and a
recessed cavity is machined for inserting a heat-source and
sensing element, such as a hot-wire filament. The cell material must be compatible with the process gas sample and must
also have a high thermal conductivity coefficient. Stainless
steel is generally employed.
The detector transducers can be resistance wires (filaments)
or thermistors (beads of metallic oxide). The operation of these
heated elements is similar except that filaments have positive,
and thermistors have negative, resistance coefficients. This

8.57 Thermal Conductivity Detectors

means that, as the temperature rises, the resistance of filaments

also rises, but the resistance of thermistors drops.
The choice between them is based on sensitivity and
temperature considerations. In general, thermistors are used
for ambient and subambient applications, whereas resistance
wire filaments are used at higher temperatures. In terms of
sensitivity, the thermistors are superior.
Hot-wire filaments are in prevalent use now as a result
of improved filament designs. Up to 1965, thermal conductivity analyzers frequently used thermistors to achieve desired
sensitivity. Thermistors are beads of metallic oxides with a
thin coating (typically glass) over the surface. This coating
tends to crack with excess heat and, when this element is
used in gases with high hydrogen content, the oxides are
reduced by the hydrogen, and drift is experienced. Glassbead thermistors develop frequent failures, particularly when
used in high-hydrogen-containing gases. The hot-wire filaments develop surface temperatures between 400 and 750F
(204 and 400C) and are sometimes used with a catalyst
coating to promote catalytic cracking of hydrocarbons to
further increase system sensitivity.
Once the selection is made, the next step is to pick the
right material to withstand the corrosiveness of the sample
gas. The filaments are usually made from tungsten or platinum alloy materials, but nickel, rhenium-tungsten, and goldplated tungsten filaments are also available. The wire diameter is about 0.001 in (0.03 mm).
The TCD Cells Figure 8.57c illustrates a TCD block assembly
with the measuring filaments inserted into both the measuring
and reference chambers. This is a flow-through design using
a single (replaceable) measuring and a single reference filament configured into a classical Wheatstone bridge circuit.

1655

ELECTRICAL CONNECTIONS
TO WHEATSTONE BRIDGE

FILAMENTS

REFERENCE
GAS
SAMPLE

FIG. 8.57d
Four-element thermal conductivity cell.

REGULATED POWER SUPPLY

AMMETER
(BRIDGE CURRENT)
SAMPLE
GAS FLOW

MEASURING
FILAMENT

REFERENCE
GAS FLOW

REFERENCE
FILAMENT

ZERO BALANCE
CORRECTION

REFERENCE
FILAMENT

MEASURING
FILAMENT

AMMETER
(BRIDGE UNBALANCE)

SIGNAL
OUT

DC
POWER

FROM
COLUMN

REFERENCE
FILAMENT

TO VENT

FROM
REFERENCE

FIG. 8.57c
The design of a two-element, flow-through TCD cell.

2003 by Bla Liptk

CALIBRATION
ATTENUATOR

FIG. 8.57e
Typical Wheatstone bridge measuring circuit for thermal conductivity analyzer.

SENSING
FILAMENT

TO VENT

RECORDER

In addition to the two-element design, four-element thermal

conductivity cell designs with recessed hot-wire filaments are
also available, as shown in Figure 8.57d. The elements are used
in pairs, one in the sample stream and one in the reference gas.
One or two pairs are normally used, but some cell designs
include eight pairs to improve sensitivity. Vertical mounting
is preferred to prevent sagging of the wire elements. The
recessed elements generally provide an improved noise level
but poorer response speed.
The response speed of the TCD is a function of the internal
volume of the detector cell. The flow through the cell must
be constant (usually in the range of 15 to 100 ml/min for
chromatographic and 500 ml/min for process analyzers).

1656

Analytical Instrumentation

SENSOR

50K

1
REFERENCE
REF

10K
3

OUTPUT
SIGNAL

R
2
1, 2, 4 = OP. AMPLIFIER
3 = INSTRUMENTATION AMPLIFIER
The current is determined by the value of Eref and R; i = Eref/R

FIG. 8.57f
The electric circuit for providing constant current to a TCD utilizing thermistor bead elements. (From Annino, R. and Villabolos, R., Process
Gas Chromatography: Fundamentals and Applications, Research Triangle Park, NC: ISA, copyright 1992 ISA. Used with permission. All
rights reserved.)

AIRSUPPLY

V0 + V

R
THERMISTOR
SENSOR (TO OVEN
TEMPERATURE REGULATOR)

Rs

EXPLOSION PROOF
HEATER ASSEMBLY

nR
ASPIRATOR
(PROVIDES MIXING
AND CIRCULATION)

FIG. 8.57g
The electric circuit for providing a constant resistance/temperature
circuit for a TCD. (From Annino, R. and Villabolos, R., Process Gas
Chromatography: Fundamentals and Applications, Research Triangle
Park, NC: ISA, copyright 1992 ISA. Used with permission. All rights
reserved.)

Bridge Circuits The sensing system can be a basic

Wheatstone bridge that uses a high-quality regulated power
supply (Figure 8.57e) rated between 100 and 300 mA.
Analyzer stability is primarily a function of power supply
voltage regulation. In addition to the means of improving
sensitivity, it is more practical to use low-noise operational
amplifiers on the bridge output. However, a low signal-tonoise ratio is required of the basic bridge output.
To provide faster response and to protect the filaments if
the flow of carrier gas fails, the latest TCD designs favor
constant current or constant resistance bridges (see Figures
8.57f and 8.57g).
Temperature Control Case temperature control provides a
constant temperature environment for the measuring cell to
enhance stability. Various temperature control systems are
used, ranging from off/on thermal switches with bare strip
heaters to the more refined ones for chromatographic ovens

2003 by Bla Liptk

120 VAC
REGULATED BY
CONTROLLER

ANALYZER
OVEN

FIG. 8.57h
Oven temperature regulator using thermistor-type sensor provides
sensitive temperature control.

(see Figure 8.57h). Analyzer stability can be adversely affected

by inadequate temperature control.
Operation
The thermal conductivity analyzer, when used as a chromatographic detector, operates with a metered sample of 50 to
200 cc/min (flow controlled). The sample flow passes through
the measuring cell and across the filaments or glass-bead
thermistors, which are hot from being heated by the
Wheatstone bridge. This resistance heating provides an elevated temperature on the surface of the filament.
Heat energy is conducted from the filament, through the
flowing gas, and to the walls of the cell. The quantity of heat
thus conducted is a function of the thermal conductivity of
the flowing gas. When a sample stream of lower thermal
conductivity than that of the zero standard gas is introduced,
less heat is conducted away, and the filament surface temperature (and its resistance, if it is a hot-wire filament) increases.

8.57 Thermal Conductivity Detectors

This causes an unbalance in the Wheatstone bridge. The

degree of unbalance can be calibrated in terms of composition.
If a thermistor-type detector is used, it has a negative
temperature coefficient, so it will unbalance the bridge in the
opposite direction, thus requiring a polarity reversal for the
ammeter connections. Otherwise, the operation is identical.
The Reference Filament Reference filaments are used to
provide compensation for temperature and barometric pressure variations. Because the reference filaments are in opposite legs of the bridge relative to the measuring filaments,
small temperature variations in the cell should affect both
filaments equally and therefore cancel out. The reference gas
can be sealed for this purpose, but a flowing reference provides additional compensation, because the venting pressure
reflects any variations in barometric pressure.
The reference gas is usually a single gas that is the same
as the major component in the sample gas. Reference flow
is generally less than the sample flow, 40 to 100 cc/min.
Packaging and Calibration
The packaging of thermal conductivity analyzers varies with
suppliers, but most manufactorers separate the electronics
(power supply and bridge) from the temperature-controlled
case. In some cases, the two parts can be physically separated
by a distance of up to 100 ft (30 m). In most designs, the
cabinets require little more panel or wall space than do conventional transmitters.
Calibration is accomplished using known samples to
establish an empirical calibration reference.
LIMITATIONS
Although simple in design, this analyzer has a major limitation:
only binary mixtures can be accurately measured by it. The
analyzer is nonspecific; because it measures the total thermal
conductivity of the process sample, it cannot distinguish or
identify the component that causes the conductivity change in
multicomponent mixtures. Therefore, its applications are limited to binary or binary-like mixtures. Some industrial gas
streams are binary mixtures and do require analysis.
Hydrogen in Hydrocarbons
One notable exception to the above-mentioned general rule
is the analysis of hydrogen in hydrocarbons. From Table 8.57a,

2003 by Bla Liptk

1657

it can be seen that hydrogen has a very high R0 value relative

to most hydrocarbon gases. In such cases, the sample often
can be considered to be a hydrocarbon-hydrogen binary
mixture and calibrated using an average background of
nonhydrogen components. The error in measurement will
depend on the true concentrations of the other components
but, in many applications, 5% full-scale inaccuracy is easily
attainable.

CONCLUSIONS
The advantages of the thermal conductivity analyzer include
its low cost, simplicity, reliability, and reasonable speed of
response. Its main limitation is that it can measure only binary
mixtures. In addition to its nonspecific nature, the need for
empirical calibration further restricts its use.
It is also recommended that all water vapors be removed
from the measurement sample by drying. Because of the
above-listed limitations, its applications are few and usually
involve binary mixtures for applications such as leakage
measurement and chromatography or the detection of
hydrogen or helium in applications where the thermal conductivity of the background gases is low and relatively
constant.

Bibliography
Annino, R., Process gas chromatographic instrumentation, Am. Lab., 21(10),
6071, 1989.
Guild, L., Design and performance of thermal conductivity detectors, Victoreen, 3.
Hoffman, Computerized dry-air leak testing for process control, Med. Dev.
Diagn. Ind., January 1996.
Krigman, A., Process chromatography: difficult becomes routine, InTech,
30(10), 3455, 1983.
McNair, H. M., Process gas chromatography, Am. Lab., 19(1), 1720, 1987.
Tye, R. P., The art of measuring thermal conductivity, InTech, March 1969.
Villalobos, R., Process gas chromatography, Anal. Chem., 47(11),
983A1004A, 1975.
Weiss, M., Three keys to keeping process chromatographs on line, Control,
May 1991.
Yarborough, D. W., Thermal Conductivity, Kluwer Academic/Plenum, New
York, 1988.
Yaws, C. L., Handbook of Transport Property Data, Viscosity, Thermal
Conductivity, Gulf Professional Publishing, Houston, TX, 1995.