8,2 RESISTANCE THERMOMETERS 401

TABLE 8.3 Temperature Range, Sensors, and Interpolation Equations for the
International Practical Temperature Scale

Temperature
Range (°C) Sensor Fixed Point Equation

-190 to 0 Platinum thermome- Oxygen, ice, steam, Reference equa-

ter sulfur tion
0 to 660 Platinum thermome¬ Ice, steam, sulfur Parabola
ter
660 to 1063 10% rhodium plati¬ Antimony, silver, Parabola
num thermocouple gold
Above 1063 Optical pyrometer — Planck’s Law

is the selection of the temperature sensor, its installation, the instrumentation

system for recording and displaying the output signal, and the use of the tem¬
perature data in product design or process control. This chapter deals with the
issues of importance to the measurement of temperature for engineering appli¬
cations.

8.2 RESISTANCE THERMOMETERS

Resistance thermometers consist of a sensor element that exhibits a change

in resistance with any change in temperature, a signal conditioning circuit that
converts the resistance change to an output voltage, and appropriate instru¬
mentation to record and display the output voltage. Two different types of
sensors are normally employed: resistance temperature detectors (RTDs) and
thermistors.
Resistance temperature detectors are simple resistive elements formed of
such materials as platinum, nickel, or a nickel-copper alloy known commercially
as Balco.^ These materials exhibit a positive coefficient of resistivity and are
used in RTDs because they are stable and provide a reproducible response to
temperature over long periods of time.
Thermistors are fabricated from semiconducting materials such as oxides
of manganese, nickel, or cobalt. These semiconducting materials, which are
formed into the shape of a small bead by sintering, exhibit a high negative
coefficient of resistivity. In some special applications, where very high accuracy
is required, doped silicon or germanium is used as the thermistor material.
The equations governing the response of RTDs and thermistors to a tem¬
perature change and the circuits used to condition their outputs are different,
therefore, they will be treated separately in the following subsections.

’ Balco is a trade name for a product of the W. B. Driver Co.

402 TEMPERATURE MEASUREMENTS

Resistance Temperature Detectors (RTDs)

A typical RTD consists of a wire coil for a sensor with a framework for
support and a sheath for protection, a linearizing circuit, a Wheatstone bridge,
and a voltage display instrument. The sensor is a resistive element that exhibits
a resistance-temperature relationship given by the expression

R = R„i\ + 7,r + + • • • + (8.1)

where 7i, 72. • • 7«are temperature coefficients of resistivity.

/?„ is the resistance of the sensor at a reference temperature T„.

The reference temperature is usually specified as = 0°C.

The number of terms retained in Eq. (8.1) for any application depends
upon the material used in the sensor, the range of temperature, and the accuracy
required in the measurement. Resistance-temperature curves for platinum, nickel,
and copper, which illustrate the nonlinearity in resistance R with temperature
T for each of these materials, are shown in Fig. 8.1. For a limited range of
temperature, the linear form of Eq. (8.1) is often used to relate resistance change
to temperature change. Equation (8.1) is then expressed as

(8.2)

When error due to the neglect of nonlinear terms becomes excessive, either
linearizing circuits can be used to compensate for the nonlinearities, or additional
terms can be retained from Eq. (8.1) to relate the measured A.R to the unknown
temperature T. Retaining the temperature coefficients 71 and 72 from Eq. (8.1)

Nickel

Copper
o

0I2-1-1-1_1_I_

-200 0 200 400 600 800 1000

Temperature (°C)

Figure 8.1 Resistance-temperature curves for nickel, copper, and platinum.

408 TEMPERATURE MEASUREMENTS

Figure 8.9 Deviation from linearity for a 50-0 RTD sensor with a 150-0 shunt resistor
when the sensor is bonded to 1018 steel. (Courtesy of Micro-Measurements.)

wire circuit (7?^ = 0.5 O) with a 50-0 nickel sensor will produce an apparent
temperature offset of only 0.2°C (0.4°F). All error due to lead wires can be
eliminated by using either the three- or four-wire circuits illustrated in Figs. 8.5
or 8.6.
Stability of the sensors is usually assured by aging of the elements during
the manufacturing process. Stability may become a source of error when the
upper temperature limit of the sensor is exceeded either by design or accident.
Anytime the upper temperature limit of a sensor is exceeded, any new temper¬
ature measurements should be repeated until stable and reproducible readings
are obtained. Stability can also be affected by the polymeric carrier used with
bondable RTDs. These carriers have a finite life and lose their strength at
temperatures in excess of 120°C (250°F).
Self-heating errors are produced when excitation voltages or currents are
used in the signal conditioning circuits. Usually there is no reason for large
excitation signals, since an RTD is a high-output sensor (a typical output is 0.9
mV/V • °C or 0.5 mV/V • °F). Self-heating errors can be minimized by limiting
the power dissipation in the RTD to less than 2 mW. In those applications where
small temperature changes are to be measured and very high sensitivity is re¬
quired, sensors with large surface areas should be employed. These sensors with
large surface areas can dissipate larger amounts of heat; therefore, higher ex¬
citation voltages can be used without introducing self-heating errors.
Bonded RTD sensors resemble strain gages and, in fact, they respond to
strain. Fortunately, the strain sensitivity of the sensor is small in comparison to
the temperature sensitivity. A bonded RTD with a nickel sensor exhibits an
apparent temperature change of 1.7°C (3°F) when subjected to an axial tensile
strain of 1000 |xm/m along the filaments of the gage grid. The magnitude of the
strain effect is such that it can be neglected in most applications.

Thermistors
Thermistors are temperature-sensitive resistors fabricated from semicon¬
ducting materials, such as oxides of nickel, cobalt, or manganese and sulfides
of iron, aluminum, or copper. Thermistors with improved stability are obtained
 8.2 RESISTANCE THERMOMETERS 409

when oxide systems of manganese-nickel, manganese-nickel-cobalt, or man¬

ganese-nickel-iron are used. Conduction is controlled by the concentration of
oxygen in the oxide semiconductors. An excess or deficiency of oxygen from
exact stoichiometric requirements results in lattice imperfections known as Schottky
defects and Frankel defects. N-type oxide semiconductors are produced when
the metal oxides are compounded with a deficiency of oxygen that results in
excess ionized metal atoms in the lattice (Frankel defects). P-type oxide semi¬
conductors are produced when there is an excess of oxygen that results in a
deficiency of ionized metal atoms in the lattice (Schottky defects).
Semiconducting materials, unlike metals, exhibit a decrease in resistance
with an increase in temperature. The resistance-temperature relationship for a
thermistor can be expressed as

In {RIR,) = p(l/r - l/TJ

or

R = R,e^^^''^ - (8.4)

where R is the resistance of the thermistor at temperature T.

Rg is the resistance of the thermistor at reference temperature T^.
P is a material constant that ranges from 3000 K to 5000 K.
T and T„ are absolute temperatures, K.

The sensitivity 5 of a thermistor is obtained from Eq. (8.4) as

^R/R _ p (8.5)
AT '

For p = 4000 K and T - 298 K, the sensitivity S equals -0.045/K, which is

more than an order of magnitude higher than the sensitivity of a platinum
resistance thermometer (S = +0.0036/K). The very high sensitivity of ther¬
mistors results in a large output signal and good accuracy and resolution in
temperature measurements. For example, a typical thermistor with R^ = 2000
n and a sensitivity S = -0.04/K exhibits a response Ai^/AT = 80 fi/K. This
very large resistance change can be converted to a voltage with a simple bridge
circuit. The voltage change associated with a temperture change as small as
0.0005 K can be easily and accurately monitored.
Equation (8.4) indicates that the resistance 7? of a thermistor decreases
exponentially with an increase in temperature. Typical response curves for a
family of thermistors is shown in Fig. 8.10. Since the output from the thermistor
is nonlinear, precise determinations of temperature must be made by measuring
the resistance R and using a calibration table similar to the one presented in
Appendix A (Table A.l). Linearity of the output can be improved by using
modifying potentiometer and/or bridge circuits; however, these circuits reduce
the sensitivity and the output of the thermistor.
416 TEMPERATURE MEASUREMENTS

lead wires. The sensitivity of a thermistor is high; therefore, the change in

resistance A/? - resulting from a temperature change is much greater than the
7

small change in resistance of the lead wires due to the temperature variation.
Also, the resistance of the thermistor is very large relative to the resistance of
the lead wires {RjIRi^ ~ 1000); consequently, any reduction in sensitivity of the
sensor due to lead-wire resistance is negligible.
Errors may occur as a result of self-heating since the power {P — PRt)
dissipated in the thermistor will heat it above its ambient temperature. Rec¬
ommended practice limits the current flow through the thermistor to a value
such that the temperature rise due to the PRj power dissipation is smaller than
the precision to which the temperature is to be measured. A typical thermistor
with Rj- = 5000 ft is capable of dissipating 1 mW/°C above the ambient tem¬
perature. Thus, if the temperature is to be determined with an accuracy of 0.5°C,
the power to be dissipated should be limited to less than 0.5 mW. This limitation
establishes a maximum value for the current / at

I = ^/PTRr = VO.0005/5000 - 316(10-^)A = 316 p.A

In this example, it would be prudent to limit the current 1 to approximately 100

|jlA. Adequate response can be obtained, even at these low currents, because
the sensitivity of a thermistor is so very high. Precise measurements of can
be made easily with a digital millivoltmeter.

8.3 EXPANSION THERMOMETERS

The expansion (or contraction) per unit length A///of a material experiencing
an increase (or decrease) in temperature AT is given by the expression

A/// = a^T (8.20)

where a is the thermal coefficient of expansion of the material. Since a is very

small for metals (it ranges from 1 to 26 |xm/m • °C), direct measurement of A/
to infer AT is difficult. In order to circumvent this sensitivity problem, devices
that utilize the differential expansion between two different materials have been
developed to measure temperature change AT. Devices of this type include the
familiar liquid-in-glass thermometer, bonded bimetallic strips of two metals, and
pressure thermometers. Each of these devices is described in the following sub¬
sections.

Liquid-in-Glass Thermometers
The well-known and widely used glass thermometer provides a simple,
convenient, and inexpensive means for measuring temperature in many appli¬
cations. The thermometer consists of an indexed glass capillary tube with a bulb
at one end to hold a supply of fluid. The fluids commonly used are mercury and
 8.3 EXPANSION THERMOMETERS 417

alcohol. Mercury can be used for temperatures between — 39°C ( —38°F) and
538°C (1000°F). When a lower temperature limit is needed, alcohol permits
measurements at temperatures as low as — 62°C ( —80°F); pentane can be used
for measurements as low as — 218°C ( —360°F).
Glass thermometers are designed for either partial or full immersion. As
the name implies, full-immersion thermometers are calibrated to read correctly
when the thermometer is completely immersed in the fluid whose temperature
is being measured. Partial-immersion types are marked and should be immersed
only to the depth indicated by the immersion mark.
The accuracy that can be achieved with a glass thermometer depends upon
the quality and range of the particular thermometer being used. Also, strict
attention must be paid to immersion requirements, since corrections must be
made when these requirements are not satisfied. With a good-quality, full-im¬
mersion thermometer having a range from 0°C to 100°C, the temperature can
be determined to within ±0.1°C.
Glass thermometers provide a low-cost means for measuring temperatures
with reasonable accuracy over the range from about -200°C to 500°C. Since
the readout is visual, they are not used in automatic data systems or in auto¬
matically controlled processes in industry.

Bimetallic Thermometers
The sensing element in a bimetallic thermometer consists of a bonded com¬
posite of two materials, as illustrated in Fig. 8.14. Materials is usually a copper-
based alloy with a large coefficient of thermal expansion, while material B is
usually Invar (a nickel steel), which has a very small coefficient of thermal
expansion. When the bonded bimetallic strip is subjected to a temperature
change, the differential expansion causes it to bend into a circular arc. The radius
of curvature of the arc is given by the expression

[3(1 + 6)^ + (1 + ee)(e^ + l/9c)]f . .

6(a^ - a5)(l + ej^AT

where 0 = G/G is the thickness ratio.

e = Eg/E^ is the modulus ratio.

Bimetallic elements in the form of cantilever beams, spirals, washers, and helixes
are inexpensive and deform significantly with relatively small changes in tem¬
perature; therefore, they are used in a wide variety of temperature sensing and
temperature control devices. In thermostats, they are used to control temper-

~r
t
; Material A

' to Material B
1 ‘B

Figure 8.14 Beam-type bimetalic element.

418 TEMPERATURE MEASUREMENTS

ature by switching the heat source on and off. As overload switches in electrical
equipment, they are activated by excessive current flows and turn off the equip¬
ment. Finally, they are often used in conjunction with a linear-displacement
sensor such as a potentiometer or linear variable-differential transformer (LVDT)
to provide a temperature indicating instrument.
The accuracy of bimetallic thermometers varies; therefore, they are usually
used in control applications where low cost is more important than accuracy.
For those applications where accuracy is important, high-quality bimetallic ther¬
mometers are available with guaranteed accuracies of about 1 percent.

Pressure Thermometers
A typical pressure thermometer, illustrated schematically in Fig. 8.15, con¬
sists of a bulb filled with a liquid such as mercury or xylene, a capillary tube,
and a pressure sensor. When the bulb is subjected to a temperature change,
both the bulb and the fluid experience a volume change. The differential volume
change AF^ is proportional to the temperature change AT. In a closed system
completely filled with liquid under an initial pressure, the pressure changes in
response to the differential volume change. The pressure is transmitted through
the capillary tube to a pressure measuring transducer, such as a bourdon tube,
bellows, or diaphragm. Movement of the bourdon tube or bellows can be trans¬
mitted through a suitable linkage system to a pointer whose position relative to
a calibrated scale gives an indication of the temperature. The bourdon tube or
bellows can also be used with a potentiometer or linear variable-differential
transformer (LVDT) to construct a temperature measuring and recording in¬
strument. Similarly, an electrical resistance strain gage on a diaphragm provides
the sensor for a temperature measuring and recording instrument.
The dynamic response of a pressure thermometer is poor because of the
thermal lag associated with the mass of fluid in the bulb; therefore, such in¬
struments cannot be used to measure temperatures in fluids undergoing rapid
changes in temperature. The pressure thermometer can, however, provide a
 8.4 THERMOCOUPLES 419

simple, low-cost, reliable, and trouble-free method of measuring temperature

in systems undergoing relatively slow changes in temperature.
Pressure thermometers filled with mercury cover the range from - 39°C to
538°C (-38°F to 1000°F), while those filled with xylene are used for the range
from - 100°C to 400°C (- 150°F to 750°F). The response is linear over a large
portion of the range. Capillary tubes as long as 60 m (200 ft) have been used
sucessfully for remote measurements. Temperature variations along the capillary
tube and at the pressure sensing device require compensation. A common com¬
pensation scheme utilizes an auxiliary pressure sensor and capillary tube. Bimetal
elements can also be used to effect partial compensation. The accuracy of pres¬
sure thermometers under the best of conditions is approximately ±0.5 percent
of the scale range.

8.4 THERMOCOUPLES
A thermocouple is a very simple temperature sensor, consisting essentially
of two dissimilar wires in thermal contact, as indicated in Fig. 8.16(3. The op¬
eration of a thermocouple is based on the Seebeck effect, which results in the
generation of a thermoelectric potential when two dissimilar metals are joined
together to form a junction. The thermoelectric effect is produced by diffusion
of electrons across the interface between the two materials. The electric potential
of the material accepting electrons becomes negative at the interface, while the
potential of the material providing the electrons becomes positive. Thus, an
electric field is established by the flow of electrons across the interface. When
this electric field becomes sufficient to balance the diffusion forces, a state of
equilibrium with respect to electron migration is established. Since the magnitude
of the diffusion force is controlled by the temperature of the thermocouple
junction, the electric potential developed at the junction provides a measure of
the temperature.
The electric potential is usually measured by introducing a second junction
in an electric circuit, as shown in Fig. 8.16^, and measuring the voltage across
one leg with a suitable voltmeter. The voltage across terminals M-N can be

Material A Material A

Material B M N

Figure 8.16 Thermocouple sensor and circuit for measuring the temperature difference
Tj - Tj. (a) Single junction, (b) Dual junction.
 420 TEMPERATURE MEASUREMENTS

represented approximately by an empirical equation having the form

= C,iT, - T,) + C^{T\ - Tl) (8.22)

where Cj and C are thermoelectric constants that depend on the materials

2

used to form the junctions.

Tl and T2 are junction temperatures.

In practice, junction 1 is used to sense an unknown temperature Tj, while

junction 2 is maintained at a known reference temperature T2. Since the ref¬
erence temperature T2 is known, it is possible to determine the unknown tem¬
perature Tl by measuring the voltage E„. It is clear from Eq. (8.22) that the
response of a thermocouple is a nonlinear function of the temperature. Also,
experience has shown that Eq. (8.22) is not a sufficiently accurate representation
of the voltage-temperature relationship to be used with confidence when precise
measurements of temperature are required. For this reason, thermocouples are
calibrated over the complete range of temperature for which they are useful and
tables are obtained which can be used to relate temperature Tj to the thermo¬
electric voltage E^. Thermocouple tables are presented in Appendix A for iron-
constantan (Table A.2), Chromel-Alumel (Table A.3), Chromel-constantan (Table
A.4), and copper-constantan (Table A.5) thermocouples. It is important to note
that the reference temperature is T2 = 0°C (32°F) in these tables.

Reference Junction Temperature

Since a thermocouple circuit responds to a temperature difference {T^ -
T2), it is essential that the reference junction be maintained at a constant and
accurately known temperature Tj. Four common methods are used to maintain
the reference temperature.
The simplest and most popular technique utilizes an ice and water bath, as
illustrated in Fig. 8.17. The reference junction is immersed in a mixture of ice
and water in a thermos bottle that is capped to prevent heat loss and temperature

Thermos bottle and cap

Material A Copper

Measuring
junction
Material B
Ip Copper

Reference Readout
junction

Figure 8.17 The ice bath method for maintaining a reference temperature at 0°C (32°F).
434 TEMPERATURE MEASUREMENTS

Temperature
sensor

-O E„0-

Figure 8.26 Two-terminal temperature sensor circuit with leadwire resistance and a
series output resistance with trim potentiometer for standardizing sensitivity.

mV/K or 10 mV/K. This trim adjustment also permits the sensor’s calibration
error at a given temperature to be adjusted so as to improve accuracy over a
given range of temperatures as shown in Fig. 8.27.
Unfortunately, the two-terminal integrated-circuit temperature transducer
is limited to use in the range of temperatures from -55°C to 150°C. In this
temperature range, it is an excellent temperature measuring device.

8.6 RADIATION METHODS (PYROMETRY)

As the temperature of a body increases it becomes increasingly difficult to

measure the temperature with resistance temperature detectors (RTDs), ther¬
mistors, or thermocouples. The problems associated with measurement of high
temperatures by means of these conventional methods (lack of stability, break¬
down of insulation, etc.) provided the motivation for initial developments in
pyrometry (inferring temperature from a measurement of the radiation emitted
by the body). As the art of pyrometry developed, two other applications emerged.
In certain measurements of temperature, the presence of the sensor affects the
temperature; therefore, a noncontact method of measurement, such as pyrom-

Temperature (°C)
Figure 8.27 Typical nonlinearity and calibration error for a two-terminal integrated-
circuit temperature transducer.
 8.6 RADIATION METHODS (PYROMETRY) 435

etry, is desirable. Another application that has developed is the measurement

of complete temperature fields. Thermocouples, thermistors, and RTDs provide
data only at a point. Temperature distributions over the entire body are often
needed, and the radiation emitted from a body provides this information if it is
properly recorded and interpreted. Thus, pyrometry is a useful method for the
measurement of very high temperatures, for providing a noncontacting method
for measurement, and for obtaining full-field temperature distributions.

Principles of Radiation Measurements of Temperature

When a body is heated it radiates energy that can be detected and related
to the temperature of the body. The relationship between intensity of radiation,
wavelength of the radiation, and temperature is known as Planck’s Law and can
be expressed as

2'nc^h (8.26)
1) - 1)

where is the spectral radiation intensity for a black body (W/m^).

X is the wavelength of the radiation (m).
T is the absolute temperature (K).
h is Planck’s constant = 6.626 (10(J • s).
c is the velocty of light = 299.8 (10^) (m/s).
k is Boltzmann’s constant = 1.381 (10“^^) (J/K).

Cl and C are constants.

2

From Eq. (8.26) it is evident that

C, = 2ttc2/z = 3.75(10-16) yy ■ m-

C 2 = hdk = 1.44 (10-^) m • K

The spectral radiation intensity is the amount of energy emitted by

radiation of wavelength X from a flat surface at temperature Tinto a hemisphere.
It is evident from Eq. (8.26) that the spectral radiation intensity W, depends
upon both wavelength X and temperature T. A plot of versus X for several
different temperatures is shown in Fig. 8.28. Note that peaks at a specfic
wavelength, which depends on temperature, and that the wavelength associated
with the peak IT;, increases as the temperature decreases. The wavelength X^
associated with the peak in VF;, can be expressed as

X^ = 2891(10-6)/7’ (8.27)
436 TEMPERATURE MEASUREMENTS

Figure 8.28 Blackbody radiation at different temperatures.

The area under each of the curves in Fig. 8.28 is the total power W emitted at
the particular temperature T. Thus

IT = W^d\ - 5.67(10-«)r^ W/m2 (8.28)

JK

Equation (8.28) is the Stefan-Boltzmann Law with the emissivity e equal to

unity (e = 1). From the previous discussion it is evident that

1. The total power W increases as a function of the fourth power of the tem¬
perature.
2. The peak value of spectral radiation intensity VF;, occurs at shorter wave¬
lengths as the temperature increases.

Both of these physical principles are used as the basis for a measurement of
temperature.

The Optical Pyrometer

The optical pyrometer, illustrated schematically in Fig. 8.29fl, is used to
measure temperature over the range from 700°C to 4000°C (1300°F to 7200°F).
The radiant energy emitted by the body is collected with an objective lens and
focused onto a calibrated pyrometer lamp. An absorption filter is inserted into
the optical system between the objective lens and the pyrometer lamp when the
temperature of the body exceeds 1300°C (2370°F). The radiant energy from both
 8.6 RADIATION METHODS (PYROMETRY) 437

the hot body and the filament of the pyrometer lamp is then passed through a
red filter with a sharp cutoff below X = 0.63 ixm. The light transmitted through
the filter is then collected by an objective lens and focussed for viewing with an
ocular lens. The image observed through the eyepiece of the pyrometer is that
of the lamp filament superimposed on a background intensity due to the hot
body. The current to the filament of the pyrometer lamp is adjusted until the
brightness of the filament matches that of the background. Under a matched
condition, the filament disappears (hence the commonly used name—^disap-
pearing-filament optical pyrometer), as illustrated in Fig. 8.296. The current
required to produce the brightness match is measured and used to establish the
temperature of the hot body. Pyrometers are calibrated by visually comparing
the brightness of the tungsten filament with a blackbody source of known tem¬
perature (e =1).
When the brightness of the background and the filament are matched, it is
evident from Eq. (8.26) that

8 1
(8.29)
^C2l\rT _ Y ~ QC2l\rT[ _ ^

Absorption filter when

temperature
low
b
Figure 8 29 Schematic illustration of the optical system and filament brightness ad¬
justment in an optical pyrometer, (a) Schematic illustration of an optical pyrometer. (6)
Filament brightness adjustment in an optical pyrometer.
438 TEMPERATURE MEASUREMENTS

where k, is the wavelength of the red filter k^ — 0.63 fJ-m.

e is the emissivity of the surface of the hot body at X = 0.63 ixm.
Tfis the temperature of the filament.
T is the unknown surface temperature.

When T < 4000°C (7200°F), the term » 1 and Eq. (8.29) reduces to

X^ (In e)/C2 + lITf

It is obvious from Eq. (8.30) that T = 7} only when e = 1. If e ¥= 1, then T

7= Tf and Eq. (8.30) must be used to determine the temperature T from the
temperature Tf indicated by the pyrometer. The emissivity of a number of ma¬
terials (oxidation-free surface) are listed in Table 8.8.
If the emissivity of a surface is not known precisely, then an error will occur
when Eq. (8.30) is used to determine the temperature T. The change in tem¬
perature as a function of change in emissivity is obtained from Eq. (8.30) as

dT _ kT de
(8.31)
y“ ~ Q T
Since kTIC2 <0.1 for T< 2000°C (3630°F), errors in temperature determinations
are mitigated considerably with respect to errors in emissivity. For example, at

TABLE 8.8<^ Emissivity e of Engineering Materials at X = 0,65 pm

Material Solid Liquid Material Solid Liquid

Beryllium 0.61 0.61 Thorium 0.36 0.40

Carbon 0.8(M).93 — Titanium 0.63 0.65
Chromium 0.34 0.39 Tungsten 0.43 —

Cobalt 0.36 0.37 Uranium 0.54 0.34

Columbium 0.37 0.40 Vanadium 0.35 0.32
Copper 0.10 0.15 Zirconium 0.32 0.30
Iron 0.35 0.37 Steel 0.35 0.37
Manganese 0.59 0.59 Cast Iron 0.37 0.40
Molybdenum 0.37 0.40 Constantan 0.35 —

Nickel 0.36 0.37 Monel 0.37 —

Platinum 0.30 0.38 90 Ni-10 Cr 0.35 —

Rhodium 0.24 0.30 80 Ni-20 Cr 0.35 —

Silver 0.07 0.07 60 Ni-24 Fe- 0.36 _

Tantalum 0.49 — 16 Cr

From ASME Performance Test Codes PTC 19.3, 1974.