2. The Thermal Conductivity Detector (TCD)
2.1 DETECTION MECHANISM
The thermal conductivity detector is among the most commonly used measuring
devices in gas chromatography for monitoring substances separated in the column.
This detector measures changes in the thermal conductivity of the carrier gas, caused
by the presence of the eluted substances.
The thermal conductivity, If, of component S is determined by the gas density, e,
the mean molecular velocity, ij, the mean free path, 2, and the specific heat at constant
It follows from equation (2.1) that the thermal conductivity is a function of the size
of the molecules, their mass and the temperature, as
where D is the diffusion coefficient. The dependence of the thermal conductivity on
the molecular masses of substances S and C is given by the relationship
Thermal conductivity values for some compounds are given in Table 2.1.
The thermal conductivity of binary mixtures is given by the thermal conductivities
of the components and their molar fractions, x :
where A and B are constants and .yS = 1 - sc.
By solving equation (2.4) for a change in the conductivity of a binary mixture ,
an expression that quantitatively describes the non-linear variation of the conductivity
with the concentration is obtained (see Fig. 2.1)
THE THERMAL CONDUCTIVITY OF SOME COMPOUNDS
Compound KS x lo3 [W/cm. K] Compound KS x lo3 [W/cm. K]
FIG. 2.1. Thermal conductivities of some binary mixtures.
For very low concentrations of the eluted substance, all the terms in the expanded
Taylor series (equation (2.5)) can be neglected except for the first, thus giving a linear
relationship between the concentration of the eluted substance and the change in the
conductivity of the binary system.
Measurement with the TCD is based on monitoring changes in the electric con-
ductivity of the filament, caused by variations in its temperature during passage of
the gaseous mixture through the measuring cell. A temperature gradient is established
due to transfer of thermal energy by the medium. Under stationary conditions, the
amount of heat transferred, Q, is proportional to the thermal conductivity of the
flowing gaseous mixture and to the difference in the temperatures of the filament, T,,
and the cell walls, T,:
During the design of practical measuring devices, it was found that the overall
amount of heat transferred is not given by the thermal conductivity of the medium
alone, but that the molar heat capacity and other factors also play a role. The follow-
ing processes contribute to the overall heat change, i.e., to the measured signal:
- the thermal conductivity of the medium;
- forced convection of the gaseous mixture;
- free convection and diffusion of the gas;
- the thermal conductivity of leads and connections;
- thermal radiation.
The participation of the thermal conductivity and of forced convection of the gas
in the overall heat transfer can be distinguished only with great difficulty. Forced
convection represents heat transfer coupled with mass transport under the dynamic
conditions present in the gas chromatograph and amounts to about 25% of the
overall change. This part of the transferred heat is proportional to the volume velocity
of the gas, u, and to its heat capacity, C,:
Hence the TCD signal depends on the flow-rate of the carrier gas.
The contribution of free convection corresponds to energy transfer in the concen-
tration gradient and is negligible compared with the other factors.
Heat transfer by the leads and electrical connections is determined by their cross-
section, length and thermal conductivity. Therefore, it is desirable that the leads to
the sensor should be as short as possible and have as small a cross-section as possible.
It has been found that this parameter does not have a large influence on the overall
change in the TCD signal in devices with heated filements, but that its importance
increases when thermistors and transistors are used as sensors:
Heat radiation depends on the surface area of the measuring element, on its
temperature and on its material quality. The literature gives values of less than 4%
for the contribution of thermal radiation to the overall heat transfer.
2.2 TCD SIGNAL
Measurement with the thermal conductivity detector is based on monitoring changes
in the resistance of the sensor, R,. A current I passes through the sensor and the
thermal equilibrium in the measuring cell, through which a gaseous mixture with
a thermal conductivity of K passes, can be expressed by the equation
12Rf = J d(T, - T,)
where a is the geometric constant of the measuring cell and J is Joule’s constant.
It follows from the experimental arrangement that part of the thermal energy is lost
during the passage of the gaseous mixture through the measuring cell and is manifested
by a decrease in the sensor temperature, T,, as the temperature of the cell walls, T,,
can be considered to be constant because of their high heat capacity. The sensor
resistance is a linear function of the temperature:
R, = RP(1 + XT,)
and therefore the measured changes in the resistance are given by the relationships
AR, = aRP. AT,
AT, = - (T, - T,)
The signal of the thermal conductivity detector, SrcD, is proportional to the change
in the voltage of the measuring bridge, AE:
srcD N AE -f --
- AR a(T, - T,)
R, K (1 + aT,)
From equation (2.13) it follows that the signal depends on a number of parameters,
the most important being
- the voltage of the measuring bridge, E;
- the geometric constant of the measuring cell, a;
- the properties of the construction material, expressed by a;
- the temperatures of the sensor and the cell walls, T, and T‘, respectively;
- the thermal conductivity of the carrier gas, gc, and its flow-rate;
- the thermal conductivity of the eluted substance, AK.
As has already been mentioned, the TCD signal is not given by changes in the
thermal conductivity of the gaseous mixture alone; forced convection of the material,
which has a certain heat capacity, is also important. It is obvious that this contribu-
tion will become smaller with increasing participation of diffusion in the mass trans-
port in the measuring cell. Therefore, the term containing heat capacities is most
significant with flow-through cells. Bearing this in mind, the following expression has
been derived for SJCD [28, 291:
where constants A and 6 are determined by the effective collision cross-sections and
by the molecular weights of the eluted substances and of the carrier gas, where
0 < A < 1 < 6, and u, is the molar flow-rate.
The signal of the thermal conductivity detector is therefore strongly dependent on
the molecular weights of the eluted substance and the carrier gas. It generally holds
that the thermal conductivity decreases with increasing molecular weight, whereas
the heat capacity increases. If hydrogen is the carrier gas, then all eluted substances
will have lower thermal conductivities and higher heat capacities. The difference in
the thermal conductivity term in the proportionality (2.14) increases with increasing
molecular weight of the eluted substance and is always negative; the difference in the
thermal capacity term increases and is positive. Hence the two processes have opposite
effects and under certain circumstances their sum is equal to zero.
2.2.1 TCD background current
The background current of the thermal conductivity detector is given by the compo-
sition of the carrier gas, by its flow rate and by the detector temperature. As all
practical designs employ a compensation bridge circuit, it is pointless to specify
background current values. The TCD noise is caused by fluctuations in the detector
temperature and in the carrier gas flow-rate . For this reason, the measuring and
the reference branches of the thermal conductivity detector are usually connected in
a bridge circuit. This circuit is indispensable with flow-through and semi-diffusion
cells, while diffusion cells yield results that are independent of the flow-rate and thus
they have the lowest noise level.
2.2.2 TCD response
The TCD response is the sum of the signals, SJCD, over the substance elution time.
It follows from equations (2.13) and (2.14) that the relative magnitude of the signal
depends on the character of the eluted substance; it decreases with increasing mo-
lecular weight. A criterion frequently applied during evaluation of the effect of the
structure of the eluted substance on the signal of the thermal conductivity detector
is the relative molar response (RMR) [2, 3, 471. The RMR values are proportional
to the various measuring sensitivities for various substances caused by their different
thermal conductivites. The literature gives the dependences of STCD on the molar and
weight percent content of a substance. When light carrier gases are employed, then
the proportionality of the signal to the weight per cent of the substance is usually
THE RELATIVE MOLAR RESPONSE TO ORGANIC OXYGEN-CONTAINING
COMPOUNDS AND TO ALKYL-AROMATIC HYDROCARBONS 
M - molecular weight
Number of carbon
Equation for RMR
n-Alkyl-aromatic c6-c10 0.741.M + 17.0
Primary alcohols c1-c4 0.672.M + 25.4
R = Co-C,;
R’ = Ci-C4
0.630.M + 25.9
0.688.M + 24.6
0.473.M + 49.5
used. In any event, the use of relative responses is to be recommended for quantitative
measurements, as they depend very little on experimental conditions such as the
bridge voltage, cell temperature and flow-rate. The RMR values for homologous
FIG. 2.2. The concentration dependence of the TCD response to some hydrocarbons,
N, - carrier gas, 1 - pentane, 2 - heptane, 3 - octane .
series of hydrocarbons, alcohols, etc., are given in Table 2.2 . These values are
obtained using hydrogen as the carrier gas and only a narrow range of molecular
masses is covered. When the values are compared with those given in the literature
[25, 341, substantial differences are found. The validity of the published relative re-
sponse rates must be judged carefully, as data concerning the measuring cell geometry
and the experimental conditions are sometimes not specified. RMR values differing
by as much as 18% between flow-through and semi-diffusion cells have been found
. It should be emphasized that, when using nitrogen as the carrier gas, even the
significance of empirical RMR values is doubtful. Under these conditions, direct
calculation of the concentration from the temperature dependences of il in the
carrier gas employed [SO] has only limited validity.
The molecular weight of the eluted substance significantly affects the character and
magnitude of the JCD response. As the thermal conductivity decreases with increasing
molecular weight, while the heat capacity increases, these two parameters have oppo-
site effects and lead to conversion of negative peaks into positive peaks. This conver-
sion is important with high-molecular-weight carrier gases, e.g., nitrogen. Fig. 2.2
shows that the change in the signal polarity is attained earlier with increasing mole-
PARAMETERS CHARACTERIZING THE THERMAL CONDUCTIVITY
DETECTOR WITH VARIOUS SENSORS
heated filament thermistor transistor
Sensitivity [mV/mole] 4 x 104 4 x 105 2.5 x 106
Linear dynamic range 5 4.4
The lower detectable amount
f 3 [I1 & 10 56
5 0.2 0.012
ImVIKI 6 2900
The basic parameters of the thermal conductivity detector depend on the exper-
imental conditions used and therefore they are discussed in greater detail in the
following paragraphs. Generally, it can be stated that the thermal conductivity
detector is a universal measuring device with a wide linear dynamic range. However,
the minimum detectable amounts are sometimes large and then other detectors are
preferable. Some parameters characterizing the thermal conductivity detector are
listed in Table 2.3. The linearity and linear dynamic range must be given in log-log
coordinates because of the exponential form of equation (1.23).
2.3 EFFECT OF EXPERIMENTAL PARAMETERS
ON THE MAGNITUDE AND SHAPE OF THE TCD SIGNAL
2.3.1 Carrier gas
In measurements with the thermal conductivity detector, hydrogen, helium, nitrogen,
argon , carbon dioxide and mixtures of various gases, e.g., air or nitrogen with
5-10% of hydrogen or helium , are employed as carrier gases. The suitability
of various carrier gases can be evaluated with reference to the discussion related to
If hydrogen or helium is used as the carrier gas, the difference in the thermal
conductivities is always large and always negative. When the eluted substance is an
inert gas with a low molecular weight, then the term containing the heat capacities
plays almost no role and the TCD value is given chiefly by the change in the thermal
conductivity of the gaseous mixture.
When nitrogen is used as the carrier gas, the changes in the conductivity caused by
the presence of an eluted substance are small, but the difference in the heat capacities
increases significantly. It is evident that the magnitude of the TCD signal is inversely
proportional to the molecular weight of the eluted substance and generally SL, is
greater than SLY, as the thermal conductivity and heat capacity terms in equation
(2.14) are comparable for nitrogen.
The purity of the carrier gas effects the magnitude of the TCD signal . If
a sample is injected into a carrier gas containing impurities, then, owing to changes
in the partial pressures, the impurities are replaced by the eluted substance and the
resulting change in the signal is small. For this reason, pure carrier gases should be
employed, although identical sensitivities for measurement with pure and impure
carrier gases have sometimes been reported .
The pronounced dependence of the signal of the thermal conductivity detector on
the gas flow-rate follows from the detection mechanism, involving removal of heat
by forced convection of gases in the measuring cell. As already mentioned, this
effect decreases with increasing participation of diffusion processes in the measuring
cell; the dependence of the TCD signal on the flow-rate is negligible with the diffusion
type of cell, expecially when a light carrier gas is employed . It follows from
equation (2.14) that the thermal conductivity term is independent of the flow-rate.
The change in the signal with increasing flow-rate will be greater the greater is the
(CP, - CP~) value, i.e., the greater is the molecular weight of the eluted substance.
This change, leading to peak inversion, will naturally be greater and more frequent
when nitrogen is used as a carrier gas because of the similarity of the conductivity
and heat capacity terms in equation (2.14).
A decrease in the signal is not always observed during a change in the carrier gas
flow-rate; in fact, over a certain range of flow-rates, there is virtually no signal change.
This is due to the fact that the elution peak becomes narrower with increasing flow-
ate and hence the fdet/d t ratio increases with increasing detector volume. Consequently,
the mean concentration of the eluted substance in the effective volume of
the detector increases and a maximum concentration is attained.
The dependence of the response on the flow-rate, which is regulated either before
or after the column, is depicted in Fig. 2.3, from which it follows that the two
dependences have the same shapes, i.e., the products of Cs and of the integration time
t ' I-
FIG. 1.3. The dependence of the JCD response on the flow-rate of the gaseous mixture;
A - flow-rate varied before the column, B - flow-rate varied after the column, before
the detector inlet .
From these dependences, it follows that the type of carrier gas used is of basicimpor-
tance. It is advisable to avoid the use of nitrogen and to work with hydrogen or helium.
When using heavier carrier gases, the relationship between the response and the concen-
tration may be markedly non-linear in the region of chromatographically significant
concentrations, resulting in distortion and inversion of chromatographic peaks.
2.3.2 Construction of the JCD
The measurement of thermal conductivity is carried out with a sensor, the resistance
of which is strongly dependent on the temperature of the medium (Fig. 2.4). The
sensor itself is at a temperature 7;, which is higher than the temperature of the walls
of the measuring cell T,. The sensor temperature is constant with stationary heating
conditions and a constant flow-rate of an unchanging gaseous mixture. A change in
the composition of the flowing gases is reflected in a change in the sensor temperature,
causing a change in the sensor resistance, R,, thus providing an electrically treatable
signal. At present, heated filaments, thermistors and transistors are employed as
FIG. 2.4. The dependence of the sensor resistance on temperature; a - heated filament,
b - thermistor, c - transistor.
The heated filament was the first sensor to be used in the thermal conductivity
detector. The dependence of its resistance on temperature is linear over wide temperature
range (up to 400" C). This advantage is partly offset by its relatively low
sensitivity to temperature changes (see Table 2.3.).
In addition to heated filaments, thermistors with negative thermal coefficients are
used as sensors. The thermistor resistance is an exponential function of temperature
and the maximum temperature used is about 100 "C; above this value, the sensitivity
of the thermistor towards ,temperature changes is very low. If the thermistor resistance
equals RP at a particular standard temperature, To, (usually 25 'C), then the resistance
at temperature T, is given by the relationship.
R, = RP exp [ - (A - 31 (2.15)
A transistor was first used as a sensor in the thermal conductivity detector in 1972
, its use being based on the direct conversion of changes in the sensor temperature
into an electric signal. As transistors cannot be heated directly, they are maintained
at a temperature T, by an external source. The steady-state thermal equilibrium can
be expressed in terms of the transistor collector current, I,:
I, E = 0 . R (T, - T,) (2.16)
The sensor temperature, T,, changes during the elution and the change is manifested
in a change in the collector current, AI,, which is measured directly. A bridge circuit is
not used and this method leads to a considerable increase in sensitivity.
18.104.22.168 Seiisor heating voltage
In contrast to the heated filament detector, the dependence of the SrcD obtained
with a thermistor on the voltage exhibits a maximum. The optimum value of the
heating voltage is given by the relationship
E,,, - J (RO . 6. Rc . T,) (2.17)
Under constant thermistor parameters, i.e., standard-state resistance, Ro, and ma-
terial constant, 6, and with a given carrier gas characterized by a thermal conductivity
Kc, the optimum temperature of the supply voltage depends solely on the temperature
of the detector walls (see Fig. 2.5).
FIG. 2.5. The dependence of the relative
magnitude of SrCD on the heating voltage
and the detection block temperature;
a, b - thermistor, c - heated filament.
Fig. 2.6. The dependence of STCD on the
carrier gas quality; 1 - nitrogen,
2 - hydrogen.
A change in the carrier gas results in a change in the optimum heating voltage.
As follows from the introductory section, carrier gases with lower molecular weights
absorb more heat and therefore a larger heat supply is necessary in order to maintain
the same sensor temperature and a similar sensitivity. This dependence is depicted
schematically in Fig. 2.6.
22.214.171.124 Sensor pararneters
The signal of the thermal conductivity detector with a heated filament depends on the
properties of the filament, especially on its specific resistance, e, and its thermal
coefficient, a. The signal is related to these values by the relationship
STCD - alJe
The values of a, e and STCD for some materials are summarized in Table 2.4.
THE SPECIFIC RESISTANCE, THERMAL COEFFICIENT
AND THE RELATIVE RESPONSE OF SOME MATERIALS
USED AS TCD SENSORS
Material a. 103[K-'] p [Q. mm2/m] sTCD
Platinum 4.0 0.106 13
Pt-Ir 90-10 1.2 0.24 2.5
80- 20 0.8 0.31 1.5
Tungsten 4.54 0.058 19
Nickel 4.91 0.072 18
One of the most commonly used materials is platinum, which, however, has poor
mechanical properies; the filament diameter is therefore generally rather large (about
0.02 mm). Alloys of platinum with iridium or rhodium have better mechanical proper-
ties but yield smaller signals then platinum sensors. After platinum, tungsten is the
most commonly used filament material. Its relative response is comparable with that
of platinum and it also has very satisfactory mechanical properties, so that very thin
filaments (down to 0.006 mm) can be used. However, it is oxidizeh by atmospheric
oxygen at temperatures above 500 "C. Nickel filaments are occasionally used in
corrosive media, but their use is limited by their poor mechanical properties [8, 9,
13, 15, 36, 441.
A number of workers have tried to prolong the life-time of sensors, decrease the
noise level and prevent drift of the background current. In addition to passivation of
the W-Re filament at 330 "C by formation of oxides , the surface of the filament
was gilded  and treated with HF  or CH,Cl,  vapour. These modifi-
cations led to an increase in the detector stability. The measuring filament has also
been covered with a palladium catalyst , increasing the sensitivity for both
a- and p-olefins.
Thermistors are characterized by the values of Ro and the material constant. The
thermistor resistance varies within a wide range, from 1 to lo6 0. In general, the
detector can be used at higher temperatures if a high-resistance thermistor is em-
ployed, but a recorder with a high input impedance must be used. As the sensitivity
of the thermistor to temperature changes is highest at low temperatures, thermistors
with low resistances (of the order of lo3 to lo4 62) can be employed.
The thermistor material constant corresponds to the energy required to increase
the temperature by AT,. Hence it is evident that, with increasing material constant,
the detector time constant will increase, thus rendering the results less reliable.
126.96.36.199 Cell geometric constant
The amount of heat removed from the surface of a heated filament is proportional to
its surface area (see equation (2.6)). For a cylindrical body, this geometric constant
is given by the relationship
where 1 is the length of the heated filament, rc is the internal radius of the cell and rf
is the external radius of the filament. It follows from equations (2.6) and (2.19) that
STCD increases with increasing length and radius of the heated filament and is inversely
proportional to the cell radius. These dependences must be evaluated correctly, as it
is impossible simultaneously to lengthen the filament and make the cell volume smaller.
If the filament is lengthened, the cell must be larger and the detector volume increases;
this leads to an increase in the detector time constant. Existing commercial detectors
represent a compromise among the above requirements. The effect of the geometric
constant on the magnitude of the TCD signal is represented schematically in Fig. 2.7.
0 M 15 -a
FIG. 2.7. The dependence of the FIG. 2.8. The dependence of STCD on the
TCD signal on the geometric constant
at various temperatures of the heated
temperature of the filament and of the cell
walls; the difference (Tf - T,) is plotted
filament. on the x-axis.
When a thermistor is employed as the sensor, it is assumed to be spherical and the
geometric constant is expressed by the relationship
a = - - 4nrt
rc - rf
Therefore, the signal of a detector employing a thermistor is directly proportional to
Temperatures of the sensor and the cell walls
From the principle of the thermal conductivity detector, it follows that T, must
generally be larger than T, (when a heated filament is used), so that (T, - T,) 2
I - 200 "C. The detector signal is directly proportional to the difference between the
temperatures of the heated filament and the cell walls, from which follow important
conclusions concerning the adjustment of the experimental conditions.
At a selected detection block temperature, SrCD is proportional to the temperature
of the heated filament, i.e., to the heating intensity (Fig. 2.8); on the other hand, at
a given temperature of the heated filament, i.e., with constant heating, STCD is
inversely proportional to the detection block temperature. In practice, it is most
advantageous to maintain the temperature of the heated filament as high as possible
(taking care not to burn the filament) and the detection block temperature as low as
possible (avoiding condensation of eluted substances in the cell).
When working with columns with programmed temperatures, the possibility of
a decrease in the TCD signal must be borne in mind if the detection block is connected
to the column thermostat. During an increase in the temperature of the column
thermostat, the temperature difference (T, - T,) decreases and consequently STCD
The above rules also hold for thermistor sensors. It has been found experimentally
that the highest measuring sensitivity is attained for a small difference between the
sensor and cell temperatures, (T, - T,) = 35 - 50 "C, while the sensor temperature
should not exceed 100 "C. STCD is strongly dependent on the detection block temperature:
STcD 81.5R1l2T-2 f c (2.21)
Therefore, it is preferable to maintain the detection block temperature as low as
Time constant of the TCD
The magnitude of the time constant depends on the effective volume of the detector.
The requirement that the time constant of the measuring device should be as low as
possible, leading to the smallest possible distortion of the elution curves, is partially
in opposition to the requirements regarding the magnitude of the detector signal
concerning, for example, the length of the heated filament or a large thermistor radius.
The cell time constant is expressed by the relationships
vdet = ' * 'dst
7 = 0.632td,,
These relationships are valid, however, only for cells in which transport occurs
exclusively through convection. A number of thermal conductivity designs have been
proposed in which both diffusion processes and mass convection towards the sensor
surface participate. According to this criterion, cells are classified as flow-through,
semi-diffusion and diffusion. The shapes of these cell types are depicted in Fig. 2.9.
FIG. 2.9. Various shapes of thermal conductivity cells; a - flow-through,
b - diffusion, c - semi-diffusion.
The expression for the time constant of flow-through cells reflects the significant
dependence of their signal on the gas flow-rate. Therefore, manostats are placed
before the cells in order to stabilize the flow-rate and the pressure . The smallest
distortion of the shape of the elution curve is achieved with a low time constant;
designers of thermal conductivity cells thus attempt to make the flow-through cell
volume as small as possible. Cells with volumes of 20 pI , 2.6 ,d  and even
1 pL1 have been described.
TIME CONSTANTS OF VARIOUS CELL TYPES
IN THERMAL CONDUCTIVITY DETECTORS
Cell 7 [sec] Notes
Flow-through 0. I- 1 most frequently used, most
sensitive to all changes
Diffusion 10-20 unsuitable for modern
Semi-diffusion o1 /r2 > 1 up to 10 properties of the flow-through cell
CI/U2 J. 1 up to 20 properties of the diffusion cell
It is obvious that the time constant will increase with increasing participation of
diffusion in the transport process. The time constant of a semi-diffusion cell can be
expressed by the relationship
z = 0.632t,,, -
where trl is the gas flow-rate through the measuring branch (see Fig. 2.9). The time
constants of the cell types discussed are listed in Table 2.5. In addition to the time
constant of the measuring cell, the sensor time constant, determined by its mass and
heat capacity, must also be considered. Thus the sensor requires a certain time to
record a change. The time constant of a heated filament is given by the relationship
As the heat capacity of the filament is proportional to its volume, i.e., mC, - $1, the
requirement of high TCD sensitivity (see equations (2.9) and (2.19)) leads to an
increase in the sensor time constant. The heat capacity of the thermistor approxima-
tely equals the material constant, 6, and varies around 1 sec for most sensors of this
type. In a provisional arrangement, the transistor sensor had a large time constant
of 10 sec , which the authors felt could be decreased to 0.6 sec.
In some designs, the heated filament is sealed in glass in order to decrease the
catalytic action of the heated filament on the thermal decomposition of substances
and to prevent corrosion of the sensor. Similarly, materials other than glass have also
been employed, e.g., fluorinated plastic . These modifications lead to an increase
in the lifetime of the sensor and make its use at higher temperatures possible, but the
THE TIME CONSTANT OF THE THERMAL CONDUCTIVITY DETECTOR
filament in thermistor transistor
Length of cell
- sensor [cm]
2 10 2 0.2 0.2
Diameter of cell
- sensor [cm]
0.5 0.002 0.1 0.2 0.2
Gas flow-rate [ml/min] 30
Time constant [sec] 0.45 0.01 4 1 9.6
time constant of the sensor and consequently that of the whole detector are increased
considerably due to an increase in the heat capacity of the measuring element. Cell
time constants and those of individual sensor types are given in Table 2.6, from
which it follows that a heated filament has the lowest time constant. It should be
used in combination with a flow-through cell whenever theoretical or quantitative
empirical conclusions are to be made on the basis of the measured results.
188.8.131.52 Measuring circuits
If a heated filament or thermistor is used as a sensor in the thermal conductivity
detector, it is connected in a Wheatstone bridge for compensation measurements.
When a voltage E is brought to a bridge consisting of the sensor resistance, R,,
and a reference sensor resistance, R,, then, for the current passing through the circuit
E = I(R, + R,)
A signal corresponding to the voltage change due to a change in the sensor resistance
can be measured at the bridge output.
FIG. 2.10. Scheme of the
FIG. 2.1 1. Scheme of the TCD circuit with a constant
TCD constant-current circuit. sensor temperature.
The above principle is used in both constant-current and constant-voltage circuits,
with either two or four sensors. A single thermistor in a voltage-powered bridge has
also been employed ; however, the only advantage of this circuit was its simplicity,
the JCD parameters not being improved. In addition to the measurement of d.c.
voltage and current, square-wave JCD operation has also been proposed. Constant-
current bridges yield a broader linear dynamic range than constant-voltage bridges.
Recently, a thermal conductivity detector with a constant sensor temperature was
introduced . In this circuit, a single variable resistor is connected in the bridge
(Fig. 2.11). The uncompensated output voltage is amplified and fed back to the bridge
through a transistor controlling the current passing through the bridge. In this
way, the current is instantaneously adjusted so that the sensor resistance, R,, and
consequently its temperature, T,, do not change.
It has been found experimentally that a device with a constant-temperature
sensor has a number of advantages over constant-voltage or constant-current circuits.
The lowest detectable amount is decreased ten-fold, the linear dynamic range is
broader by one order of magnitude and the detector time constant is decreased.
Concentrations of 10% by volume were measured in this way without a change in the
The pronounced improvement in the JCD parameters when a constant temperature
is used can be explained by the fact that, owing to the constant sensor temperature
during the elution, the entire elution curve is measured under constant conditions
with constant sensitivity. With constant-voltage or constant current circuits, the
sensor temperature gradient follows the concentration gradient of the eluted substance.
With increasing amounts of eluted substances, the measuring sensitivity decreases
owing to a decrease in the sensor temperature.
2.4 APPLICATIONS OF THE TCD
The thermal conductivity detector is one of the most commonly used detectors in gas
and liquid-solid chromatography . In addition to a number of reviews [17, 19,
EXAMPLES OF THE APPLICATION OF THE THERMAL CONDUCTlVITY
Substance determined Material analyzed The lowest detect- Ref,
Permanent gases in air
Permanent gases in chlorine
in high purity ethylene
N2, 02, c02
in exhaled air
NO, in air
0 in organic compounds
N in organic compounds
Alkanes, alcohols, aliphatic acids
Olefins, branched olefins
0.002 1.11 10, 23,
1 1.1g 39
1.9 x mole 5
8.7 X mole 5
1.4; 2.7; 3.5ppm 32
24, 33, 34, 42, 481, the excellent treatment of the TCD in the book by Jentzsch and
Otte  should be mentioned. Some examples of TCD applications are given in
Table 2.7. When the TCD is compared with other detectors, for example the discharge
detector [l], E CD  or FP D , better parameters are usually found for the TCD.
Comparison of the TCD with the FID [4, 35, 371 depends on the structures of the
eluted substances. The TCD is, of course, more suitable for inorganic gases such as
CS,, COS, H,S and SO, . During measurement of the signal ratio, S"/SrCD,
FIG. 2.12. Analysis of air using a TCD with a transmodulator .
values greater than unity were obtained for alkanes, and values smaller than unity for
oxygen-containing compounds . It can be presumed that the thermal conductivity
detector will find further use in analyses of inorganic gases, of silylated samples, etc.
. A thermal conductivity detector with a palladium transmodulator  gave
excellent results in analyses for rare gases in air (Fig. 2.12).
I. Arnikar H. J., Rao T. S., Karmarkar K. 13.: J. Chrornatogr. 38, 126 (1968)
2. Barry E. F., Rosie D. M.: J. Chrorwtogr. 59. 269 (1971)
3. Barry E. F., Jr.: Chenz. Abslr. 76, 50 486c (1972)
3a. Beskova G. S., Filipov V. S.: Zaaod. Lab. 38, 154 (1972)
4. Bogoslovskii Yu. N., Razin V. L.: Zlz. Fiz. Khim. 45, 2490 (1971)
5. Brazhnikov V. V., Sakodynskii K. I.: Clzern. Absrr. 74, 94 013j (1971)
6. Castello G., DAmato G.: J. Chromatogr. 32, 625 (1968)
7. Chowdhury B., Karasek F. W.: J. Chroniatogr. Sci. 8, 199 (1970)
8. Cieplinski E. W., Spencer S. F., Illingsworth W. L.: US Pat. 3,537,914 (Nov. 3, 1970)
9. Delew R. B.: J. Chromatogr. Sci. 10, 600 (1972)
10. Ediz S. H., van Swaay M., McBride H. D.: Chem. Instruni. 3, 299 (1972)
11. Gerdes W. F.: US Pat. 3,474,661 (Oct. 28, 1969)
12. Goryunov Yu. A,, Zalkin V. S., Okhotnikov B. P., Rotin V. A., Rozanova L. I., Maksimov
B. G.: USSR Pat. 371 511 (Feb. 22, 1973)
13. Guillot J., Bottazzi H., Gyuot A., Trambouze Y.: J. Gas Chromatogr. 6, 605 (1965)
14. Gutsche B., Herrmann R.: 2. Anal. Chem. 249, 168 (1970)
15. Hachenberg H., Gutberlet J.: Brensf. Chem. 49, 242 (1968)
16. Hara N., Katsuda M., Kato H., Hasegawa K., Ikebe K.: Chem. Abstr. 68, 65 5D4r (1968)
17. Hartmann C. H.: Anal. Chenz. 43, 113A (1971)
17a. Jadrijevic V., Deur-Siftar D.: Chromafogruphia 7, 19 (1974)
18. Jentzsch D., Otto E.: Detektoren in der Gas Chromatographie, Akademische Verlaggeselschaft,
Frankfurt a. M. 1970
19. Johns T., Stapp A. C.: J. Chromatogr. Sci. 11, 234 (1973)
20. Karp S., Lowell S.: Anal. Chern. 43, 1910 (1971)
21. Kiefer M. E.: Ger. Offen. 2.222 617 (Nov. 30, 1972)
22. Lindsay A. L., Bromley L. A,: Ind. Eng. Chem. 42, 1508 (1950)
23. Lovelock J. E., Simmonds P. C., Shoemake G. R.: Anal. Chem. 43, 1958 (1971)
24. McNair H. M., Chandler G. D.: J. Chromatogr. Sci. 11, 454 (1972)
25. Monfort J. P.: Chim. Anal. (Paris) 53, 646 (1971)
26. Morrow R. W., Dean J. A., Shults W. D., Guerin M. R.: J. Chromatogr. Sci. 7, 572 (1969)
27. Nikolaeva N. A., Dolgopolskaya P. I., Rezler R. Yu.: Chem. Abstr. 70, 111 448s (1969)
28. Novik J., WiEar S., Janak J.: Coll. Czech. Chem. Commun. 33, 3642 (1968)
29. Novak J., Janiik J.: Coll. Czech. Chem. Commun. 35, 212 (1970)
30. Noviik J., JaniEek M.: Chem. Listy 65, 739 (1971)
31. Otte E., Gut J.: Chromatographia 5, 246 (1972)
32. Pecsar R. E., De Lew R. B., Iwao K. R.: Ana. Chem. 45, 2191 (1973)
33. Richmond A. D.: J. Chrornatogr. Sci. 9, 92 (1971)
34. Rosie D. M., Barry E. F.: J. Chromatogr. Sci. 11, 237 (1973)
35. Schaeffer B. A.: Anal. Chenz. 42, 448 (1970)
36. Seibel A. C., Johns T.: US Pat. 3,533,858 (Oct. 13, 1970)
37. Sokolov D. N., Golubeva L. K.: Zauod Lab. 35, 143 (1969)
38. Tamura H., Hozumi K.: Chem. Abstr. 72, 128 311f (1970)
39. Thuerauf W.: 2. Anal. Chem. 250, 11 1 (1970)
40. Vermont J., Guillemin C. L.: Anal. Chem. 45, 775 (1973)
41. Versino C., Fogliano L., Giaretti F.: Riu. Combust. 21, 389 (1967)
42. Villalobos R.: Chem. Eng. Progr. 64, 55 (1968)
43. Volkov E. F., Rabinovich S. I., Breshchenko V. Ya.: Ref. Zh. Khim. 1969, 5 G 14
44. Wade R. L., Cram S. P.: J. Chromatogr. Sci. 10, 622 (1972)
45. Walsa J. T., McCarthy K. J., Merritt C., Jr.: J. Gas Chromatogr. 6, 416 (1968)
46. Wassiljewa A.: Physik. 2. 5, 737 (1904)
47. Watanabe S., Nakasato S., Kuwayama H., Sasamoto Y., Shihaishi S., Seino H., Nagai T.,
Negishi M., Hayano S.: Yukagaku 22, 102 (1973)
48. Winefordner J. D., Glenn T. H.: Aduan. Chromatogr. 5, 263 (1968)
49. Wittebrood R. T.: Chromatographia 5, 454 (1972)
50. Zalkin V. S.: Zauod. Lab. 36, 129 (1970)