MEASUREMENT OF THE RESISTIVITY OF

ULTRAPURE WATER AT ELEVATED

TEMPERATURES
Authors: K. R. Morash, Vice President; R. D. Thornton, Chairman; C. H. Saunders,
Engineer; Anthony C. Bevilacqua, Chemist, Thornton Associates, Inc., 1432 Main Street,
Waltham, MA 02154, (617) 890-3399;
T.S. Light, Consultant, 4 Webster Road, Lexington, MA 02173,(617) 862-3048.

This paper was presented at the ULTRApure Water EXPO May 1994 and was
published in the Ultrapure Water Journal in December of 1994. It is preprinted
with permission of Tall Oaks Publishing.

Abstract: The use of high temperature ultrapure water (70 to 95°C) has become more
prevalent and important over the past few years. This paper examines the special
requirements of instruments and sensors for measuring the resistivity of ultrapure water at
elevated temperatures.
The resistivity of theoretically pure water over a wide temperature range has been a
challenging study for many years. The high sensitivity of resistivity to ionic impurities has
been a dominating factor in the widespread use of the resistivity measurement for
determining water quality. This sensitivity, however, declines with increasing temperature.
At 25.0°C, pure water has a theoretical resistivity of 18.18 MΩ-cm. If 1.0 µg/L of
impurity, calculated as sodium chloride, is added to pure water, the resistivity decreases
3.9% to 17.5 MΩ-cm. However at 100°C, where the resistivity of pure water is 1.29 MΩ-cm,
1.0 µg/L of impurity decreases the resistance only 0.8% to 1.28 MΩ-cm.
To simplify the comparison of solution purity at different temperatures, the industry
refers resistivity measurements to a standard temperature of 25.0°C. Modern
instrumentation contains sensors for both conductivity and temperature and an algorithm for
calculating what the resistivity would be at the standard temperature. This paper will
discuss the influence of sensor, instrumental and algorithm variables on the accuracy of
ultrapure water measurements at elevated temperature.
The resistivity and conductivity of ultra pure water have been redetermined both
theoretically and experimentally and the new data summarized.

Note - Since this paper was presented at the Ultrapure Water Expo on May 10, 1994 the
authors have continued to investigate the resistivity of ultrapure water at high
temperatures. The additional data collected is included in the analysis in this paper.

Introduction
Henley and others have noted the increasingly important part played by hot deionized
water in the manufacture of semiconductors (1-3). The trend, appearing in the last 7 to 8
years, has raised the temperature of the wafer washing water from 40 to 80°C or higher.
Some benefits that accrue from the hot water include the inhibition of bacteria growth that is
observed at temperatures under 50°C, as well as the usual benefits that come from the use of
hot water -- increased solubility of impurities and decreased viscosity of water which permits
penetration of the smaller spaces created by increased circuit densities.
The modern use of high temperature ultrapure water demands measuring instruments
that can detect impurities of only a few tenths of a part per billion (ppb), roughly the same
levels that have been measured in the past at lower temperatures (4-7). Unfortunately it is
more difficult to measure the purity of hot water than it is to make the same accuracy
measurement on cold water. There are a surprisingly large number of sources of error, any
one of which can render the measurement inadequate. This paper examines these sources of
Measurement of the Resistivity of Ultrapure Water at Elevated Temperatures page 2

error and presents results of new calculations and measurements for the temperature
dependence of the resistivity of water. The focus is on measurement in the range of 60 to
100°C but data is also presented over the range of 0 to 100°C.

Measured Results

In 1989 Thornton and Light (8) conducted a study to calculate and measure the intrinsic
resistivity of ultrapure water from 0 to 100°C. The results showed agreement to values
calculated from other measurements within 0.25% for temperatures below 30°C but
differences rose to almost 3% at temperatures near 100°C. This discrepancy implied that at
elevated temperatures the resistivity change for impurity levels below about 1 microgram per
liter (µg/L or ppb) could not be calculated. Tests performed on several commercially available
resistivity instruments have verified this observation. This is an unacceptable limitation for
modern resistivity instrumentation, so we conducted a new study with detailed attention to
all of the known issues that affect measurement and calculation accuracy. This section
summarizes the measurement phase of the study.

Sources of Measurement Error

The objective was to obtain improved measurements with errors of less than 0.25% over
the entire temperature range. To achieve this level of accuracy we found there were a
number of important sources of error. This section discusses these sources.
Temperature sensitivity of conductivity
Figure 1 shows the importance of accurate temperature measurement. At 0°C the
sensitivity of conductivity (or resistivity) of ultrapure water to temperature is 7.4%/°C.
Therefore an error of 0.1°C in temperature measurement is equivalent to a conductivity error
of 0.74%. At 100°C this sensitivity drops to 2.3%/°C, a factor of three reduction. It might be
inferred from Figure 1 that accurate temperature measurement is less important at high
temperatures, but the issue is more complex than this because the sensitivity of the
measurement to impurity concentration has not been considered.

8
7

6
5
% / °C

3
2

1
0
0 10 20 30 40 50 60 70 80 90 100
Temperature (°C)

Figure 1. Sensitivity of pure water conductivity to changes in temperature.

Measurement of the Resistivity of Ultrapure Water at Elevated Temperatures page 3

The temperature dependence of the conductivity of the impurity and the sensitivity of
the temperature compensation algorithm to temperature errors both have a major effect on
temperature compensated conductivity measurements. For the first effect assume that we
have NaCl impurities measured in µg/L of NaCl. Figure 2 shows how this sensitivity changes
with temperature, and shows a factor of 12 decrease in sensitivity when the temperature
increases from 0 to 100°C. Most important, at 25°C the sensitivity is 4%/ppb while at 85°C it
is 1%/ppb, so the instrument must be four times as accurate to be as useful at 85°C as it is at
25°C. This means cell constant calibration, instrument and sensor accuracy, and calculation
procedures must all be four times as accurate.

10
9
8
7
6
%/ppb

5
4
3
2
1
0
0 10 20 30 40 50 60 70 80 90 100

Temperature (°C)

Figure 2. Sensitivity of conductivity change to NaCl impurities as a function of

temperature.

The sensitivity of the compensated conductivity to the temperature measurement and to

the conductance measurement is shown in Figure 3. A 1°C temperature error at 25°C causes
a 4.9% error in compensated conductivity while at 100°C the same temperature error causes
a 11.6% error in compensated conductivity. Figure 3 also shows that a 1% resistance
measurement error causes about half the percentage error in compensated conductivity as a
1°C temperature error does. Figures 1 and 3 show that as temperature increases a
temperature measurement error has less of an effect on the measured resistance but has a
greater effect on the compensated conductivity.
Measurement of the Resistivity of Ultrapure Water at Elevated Temperatures page 4

12
% Compensated Conductivity Error
10 % per °C

% per %K
8

0
0 10 20 30 40 50 60 70 80 90 100
Temperature (°C)

Figure 3. Sensitivity of the compensated conductivity to errors in the conductance

and temperature measurements.

Figures 2 and 3 show that at 85°C, temperature must be measured to within 0.1°C and
conductivity to better than 1% to get even a rough measurement of impurity with a
measurement sensitivity of 1 ppb. The goal of measuring to a few tenths of a ppb is almost
beyond the capability of today's instrumentation. Note that at 25°C we only needed to
measure to about 3% accuracy to achieve the same instrument accuracy. Thus the
temperature measurement only had to be accurate to about 0.5°C. We thus reach the
somewhat surprising conclusion that both conductivity and temperature measurement must
be more accurate at high temperatures.
Temperature measurement
Typical resistivity cells available today use either a thermistor or Resistance
Temperature Detector (RTD) to measure temperature. A thermistor is a semiconductor
device whose resistance decreases exponentially with increasing temperature. An RTD is a
metal wire whose resistance increases linearly with increasing temperature. Thermistors are
generally less precise and more subject to drift than RTDs, but they require less accurate
resistance measurement. For the most accurate measurement of temperature it is normally
preferable to use a platinum RTD because of its chemical, mechanical and electrical stability.
The platinum RTD has been selected by the National Institute of Standards and
Technology (NIST) as the working temperature standard. The accuracy of RTDs has been
quantified by the American Society for Testing and Materials (ASTM). The ASTM has
defined interchangeability curves for two classes of RTDs (IEC 751 standard). For Class A
and Class B 100 ohm platinum RTDs the allowed deviation between measured and actual
temperature is:
Class A: ±(0.15 + 0.002 [T]) °C
Class B: ±(0.30 + 0.005 [T]) °C

Considering only RTD inaccuracy, a Class A 100 ohm Pt RTDs can produce a
temperature error of 0.35°C at 100°C even if the resistance is measured perfectly. Individual
temperature sensors can be calibrated to provide better accuracy than required by the ASTM
interchangeability curves, and the stability of platinum makes such calibration possible.
Clearly a Class A platinum RTD or the equivalent is required for the accuracy of temperature
measurement required for sensitive impurity determination. In fact, even a Class A RTD
must be individually calibrated to achieve the goal of 1 ppb impurity measurement
sensitivity.
The measurement of temperature depends not only on the inherent errors in the sensor,
but also on errors in the resistance measurement. A platinum RTD has a resistance that
Measurement of the Resistivity of Ultrapure Water at Elevated Temperatures page 5

changes about 0.3%/°C change in temperature. An RTD resistance measurement with an

accuracy of 0.06% is required to measure temperature to an accuracy of 0.2°C. Since all error
sources are cumulative, the actual demands on the instrument are for an accuracy of better
than 0.02% in resistance measurement. This accuracy can be achieved by calibration, but
only if the measurement has sufficient repeatability in the presence of changes in
measurement environment.
For highly accurate temperature measurements a correction must be made for the
resistance of the lead wires to the RTD. The use of a 1000 ohm Pt RTD over a 100 ohm RTD
will reduce the effect of the lead wire resistance. However, for the most accurate
measurements, three or four wire resistance measurements of the RTD should be made.
An additional error factor is the self-heating of the temperature sensor due to
measurement currents in the sensor. Very precise resistance measuring instruments can
easily cause self-heating of more than 0.1°C, so it may be necessary to sacrifice accuracy for
reduced self-heating.
Conductivity cell calibration
Most modern conductivity or resistivity instruments either read the cell constant from a
smart sensor or allow the user to enter either a cell constant or cell constant correction
factor. Unfortunately it is difficult to calibrate a conductivity cell for ultrapure water
applications because there is no accepted standard solution that is suitable. A standard must
have a resistivity in the same range as the solution being measured or there can be errors
from a number of sources; examples include surface effects and measurement frequency and
waveform. Well known and accepted standards (9,10) have too low a resistivity. Higher
resistivity solutions are generally not stable in the presence of air. Two high resistivity
standards close to the desired range have been proposed (11,12) but are not accepted yet.
In our experience the best method of calibrating a resistivity cell is to use ultrapure
water. The cell should be initially calibrated in water at 40 to 50°C. It is easy to obtain
ultrapure water at this temperature and the sensitivity to impurity is reduced so the
calibration can be more accurate. However, calibration depends on knowing the intrinsic
resistivity in this temperature range, and there is still some uncertainty that limits the
accuracy of high temperature calibration above about 50°C.
The temperature of the water should then be reduced to 25°C where the resistivity of
the water is accurately known to be 18.18 Megohm-centimeters (MΩ-cm) on the basis of well-
known physical parameters. The cell constant should be verified at 25°C.
The quality of the water can be verified by reducing the temperature and showing that
the resistivity increases according to the known temperature relation (see Tables A and B).
At the lower temperatures any impurity will have a greater effect on resistivity (see Figure 2)
and hence signal the fact that the water was not sufficiently pure to be used for calibration.
Measurement method
A prior paper (8) has discussed the importance of using the right electrical frequency
and voltage waveform in the measurement. Additionally, since there is a 66:1 change in
ultrapure water resistivity between 0 and 100°C it may be necessary to change the
measurement frequency and range resistor in the instrument to achieve adequate accuracy
over this whole range. The need for accurate resistance measurement for temperature
determination has been discussed, but the ac measurement required for resistivity
determination tends to be more difficult than the dc measurement of RTD resistance.
Cleanliness and aging
Sensor cleanliness is very important. A sensor newly installed in an ultrapure water
system must typically operate for several hours before it can be used for reliable
measurement. We have observed that there is an initial aging effect that causes the cell
constant to change slightly the first time it is used at high temperatures. Once this change
has taken place there does not appear to be significant long term drift and the calibration
will stay constant for many months.
Temperature gradient in the cell
When the cell is measuring high temperature water it is common for the end of the cell
furthest removed from the water to be substantially cooler than the water. This temperature
Measurement of the Resistivity of Ultrapure Water at Elevated Temperatures page 6

gradient can cause the temperature sensor to be as much as a few tenths of a degree cooler
than the water with a resulting temperature error. For successful high temperature
measurements it is necessary to use special RTD mounting in the cell and to wrap insulation
around the outside of the cell. Cell construction must be such that the temperature sensor is
thermally isolated from the cell fitting and cable.
Thermal time lag
Sensors for commercial use should be designed to respond rapidly to temperature
changes in order to measure compensated resistivity accurately at high temperatures.
Fortunately it is possible to design sensors that have a dominant thermal time constant of a
small fraction of a minute, so reasonably fast measurements are possible.
Arithmetic calculations
With modern microprocessor based measurement it is not hard to perform measurement
and compensation calculations with sufficient accuracy, but there are pitfalls to avoid. For
example, a simple way to perform temperature compensation is to use table lookup
procedures with table values interpolated from data at uniform temperature intervals. The
calculation of intrinsic resistivity by this method can lead to unacceptable error at points
between table values; hence other methods of interpolation or calculation must be used.
Most conductivity and resistivity instruments now use some method of filtering or
averaging. Averaging over long time periods is useful for alarm actuation. However, real time
data is often desirable for display purposes. We believe that the best results are obtained
with a median filter that excludes occasional extreme measurements but otherwise does not
delay the display of changes in measurement more than is necessary.

Measurement approach
Apparatus
The experimental apparatus has been described previously (8) and some modifications
to the measuring equipment have been made to permit the required improvements in
accuracy. We used a DI loop with an array of conductivity sensors and a platinum RTD at
each end of the array, as shown in Figure 4. At temperatures above 50°C a titanium heat
exchanger with a high-powered electric heater was used to heat the water rapidly to the
measurement temperature. The water was cooled to a temperature suitable for the resins in
the ion exchange purification equipment.

WATER
HEAT
PUMP EXCHANGER
ION
(HEATING)
EXCHANGE
RESIN

WATER
STORAGE WATER
TANK FLOW

WATER
HEAT CONDUCTIVITY CELLS
EXCHANGER AND
(COOLING) TEMPERATURE SENSORS

Figure 4. Schematic of measuring system.

In order to measure the temperature accurately we used RTDs at each end of the array
of conductivity sensors calibrated to 0.0025°C and were able to interpolate to find the precise
temperature of each conductivity cell. The entire loop was wrapped with insulation, but there
was still a measurable temperature drop that varied from a few tenths of a degree to about a
degree from the first sensor to the last.
Measurement of the Resistivity of Ultrapure Water at Elevated Temperatures page 7

High precision resistance measuring equipment was used to measure the RTD
resistance and the conductivity sensors. All instruments were accurately calibrated.
Repeated measurements at 25°C were used to verify that no changes occurred between
successive measurement runs. A resistivity sensor was installed after the cooling heat
exchanger to verify the water purity at high temperatures. The water resistivity at this point
remained at 18.18 MΩ-cm when the water at the cell bank was at high temperatures thus
verifying that the heating was not introducing any impurities into the water.
Measurement procedure
After a stable measurement environment was established, we logged all measurements
for a substantial time. Computers were used to average the data at each data point for each
of the resistivity sensors. Each cell was calibrated to read 18.18 MΩ-cm at 25°C and
resistivity was measured to an accuracy of ±0.1%.

Measurement results
The data for all sensors was converted into a single determination of resistivity versus
temperature. This was done using cubic spline approximating functions with break points at
25°C intervals from 0 to 100°C. The values at the breakpoints were determined by
minimizing the mean square deviation of all measurements from the approximating function,
and the second derivative at the end points, i.e., 0 and 100°C, were determined so as to
minimize the discontinuity in the third derivative at the break points. This method of
approximation has been demonstrated to give excellent results when applied to known
functions with characteristics similar to the measured data.
The numerical results are given in Table A and the differences between various
measured and calculated data are shown in Figure 6. The deviation from the fitted curve for
the high temperature data from four cells is shown in Figure 5. The data in Figure 6 is the
calculated and observed data from a prior paper by Thornton and Light (8) and new
calculations that will be described in the next section.
Measurement of the Resistivity of Ultrapure Water at Elevated Temperatures page 8

Table A. Comparison of measured and calculated resistivity.

this work this work TL calculated

T ρ measured ρ calculated % diff ρ calculated % diff
0 86.19 86.07 -0.137 85.90 -0.334
5 60.48 60.27 -0.353 60.16 -0.521
10 43.43 43.26 -0.399 43.23 -0.465
15 31.87 31.75 -0.373 31.76 -0.356
20 23.85 23.78 -0.284 23.82 -0.131
25 18.18 18.14 -0.202 18.18 0.026
30 14.09 14.07 -0.115 14.12 0.182
35 11.09 11.08 -0.054 11.12 0.294
40 8.849 8.851 0.025 8.888 0.441
45 7.154 7.159 0.064 7.185 0.426
50 5.853 5.858 0.084 5.879 0.436
55 4.840 4.846 0.120 4.867 0.554
60 4.042 4.049 0.176 4.066 0.594
65 3.407 3.415 0.245 3.428 0.611
70 2.896 2.906 0.351 2.918 0.763
75 2.482 2.493 0.455 2.502 0.822
80 2.142 2.156 0.634 2.164 1.018
85 1.862 1.877 0.808 1.886 1.289
90 1.630 1.645 0.946 1.655 1.534
95 1.436 1.451 1.060 1.461 1.706
100 1.274 1.287 1.024 1.298 1.915

0.40

0.30

0.20

0.10

0.00

-0.10

-0.20

-0.30

-0.40
60 65 70 75 80 85 90 95
T e mpe ra ture

Figure 5. Deviation between measured results and the fitted curve.

Measurement of the Resistivity of Ultrapure Water at Elevated Temperatures page 9

2.00%
1989 TL Calculated
1.50%
1989 TL Observed
1.00% 1994 Calculated

0.50%

0.00%

-0.50%

-1.00%

-1.50%

-2.00%
0 10 20 30 40 50 60 70 80 90 100
T e mpe ra ture (°C)

Figure 6. Comparison to 1994 curve fit.

Theoretical calculations
Basic equations

Methods of calculating water resistivity from basic principles have been covered in a
number of prior papers (8) and only summary results are presented here.

w
κ = ρ −1 = κ H2O + 10 −3 ( λ Na + + λ Cl − ) Eq. 1
mNaCl

where: κ is the conductivity of pure water with NaCl impurities in S/cm

ρ is the resistivity of pure water with NaCl impurities in ohm-cm
λx is the specific conductivity of ion x in S-cm2/mole
w is the impurity concentration in g/liter
mNaCl is the molecular weight of NaCl or 58.45 grams/mole

The conductivity of pure water can be written in a similar way:

κ H2O = ρ −H12O = 10 −3 ( λ H + + λ OH − ) d K w Eq. 2

where: d is the density of pure water in g/cm3

Kw is the dissociation constant of water
Measurement of the Resistivity of Ultrapure Water at Elevated Temperatures page 10

Published data
In order to calculate the resistivity of pure water as accurately as possible, we have done
an extensive survey and critical evaluation of published measurements of the four physical
parameters on the right side of Equation 2. In reviewing the literature we were surprised at
the relative scarcity of original data for some of the parameters. In recent years there has
been a tendency for researchers to reinterpret past data and fit approximating functions to
describe the data. In many cases the approximating functions are fitted for a wide range of
conditions and the accuracy is not adequate for our purposes. Thus we chose to take the
original data and perform our own interpolation. The conclusions are summarized here for
each parameter, and the final values are given in Table B.
Density
The density of water is known to relatively high accuracy. We selected the data prepared
by Schmidt (13), which is given for 1°C intervals over the 0 to 374°C range. This data is
sufficiently accurate as to not cause even a 0.1% error in the resistivity calculation.
Dissociation constant
There are several recent references with minor differences. Most recent researchers
have used data from a 1974 paper by Sweeton, Mesmer and Baes (14). The results agree with
other researchers to within about ±0.004 pK units, so errors from this source are estimated to
be less than 0.5%.
Specific conductivity of the hydrogen ion
A 1980 publication by Strong (15) is used as the source for specific conductance of the
hydrogen ion. His results are relatively recent and the method used appears to be capable of
good results. We estimate that the accuracy of this data is about 0.25% at 25°C but the
accuracy is less at higher and lower temperatures.
Specific conductivity of the hydroxide ion
The best data for λOH- was found in a 1964 paper by Marsh and Stokes (16). This data
was reported for temperatures of 15, 25, 50 and 75°C. Although low temperature data is
scarce it is less important because the contribution of the hydroxide ion to conductivity is
only half as large as the contribution of the hydrogen ion and the intrinsic conductivity of
water is very low to start with. However, at 100°C the OH- ion contributes over 40% of the
conductivity of pure water and the intrinsic conductivity is larger, so errors are more
important. The only data for 100°C appears to be a recomputation of data reported by Noyes
in 1907. Marsh and Stokes give a value of 455 Siemens-centimeter squared per mole (S-
cm2/mole), but others have recomputed Noyes (17) data to arrive at values as low as 427.
This is a difficult parameter to measure at high temperatures and we now believe that values
inferred from our high temperature water measurements may be the most accurate
determination of λOH- to date. We have used the value of λOH- (T=100°C) = 455 S-cm2/mole
in our calculation and then computed what it should be if our other data is correct.

Calculation results
Tables
The final result of our calculation is shown in Table B and a comparison of our measured
and calculated resistivity is shown in Table A.
Measurement of the Resistivity of Ultrapure Water at Elevated Temperatures page 11

Table B. Physical parameters and calculated conductivity and resistivity.

T λH+ λOH- -log(Kw) d κH2O ρH2O

0 225.04 118.30 14.9412 0.99980 0.01162 86.09
5 250.27 133.81 14.7287 0.99990 0.01660 60.25
10 275.38 149.71 14.5286 0.99970 0.02312 43.25
15 300.29 165.94 14.3400 0.99920 0.03150 31.75
20 324.92 182.45 14.1618 0.99830 0.04204 23.79
25 349.19 199.18 13.9933 0.99711 0.05510 18.15
30 373.03 216.09 13.8337 0.99572 0.07104 14.08
35 396.38 233.11 13.6823 0.99404 0.09021 11.09
40 419.18 250.20 13.5385 0.99226 0.1130 8.850
45 441.39 267.30 13.4019 0.99020 0.1397 7.157
50 462.98 284.35 13.2718 0.98804 0.1708 5.856
55 483.90 301.31 13.1480 0.98571 0.2064 4.845
60 504.13 318.21 13.0301 0.98319 0.2470 4.049
65 523.67 335.07 12.9176 0.98049 0.2928 3.416
70 542.49 351.93 12.8103 0.97771 0.3440 2.907
75 560.60 368.82 12.7079 0.97475 0.4010 2.494
80 578.00 385.78 12.6102 0.97163 0.4639 2.156
85 594.71 402.85 12.5169 0.96843 0.5328 1.877
90 610.75 420.05 12.4279 0.96516 0.6079 1.645
95 626.14 437.42 12.3429 0.96163 0.6892 1.451
100 640.92 455.00 12.2618 0.95813 0.7768 1.287

Calculated Resistivity
The calculated results given in Table A are plotted in Figure 6 and are compared with
previously published data and the results of our measurement. For temperatures from about
5 to 60°C the calculated values of ρH2O are estimated to be accurate to about ±0.25% with
the probable error rising to more than ±0.5% in the 80 to 100°C range.
The measured and calculated resistivity of pure water as given in Table B can be
expressed with an accuracy of ±0.01% by means of the exponential polynomial approximating
function in Equation 3.

a + a t + a t2 + a t3 + a t 4 + a t5
ρ =e 0 1 2 3 4 5 Eq. 3
H O
2

where: t = T/100.

For calculated results use:

a0 = 4.45518, a1 = -7.39323, a2 = 5.48557
a3 = -3.74602, a4 = 1.94396, a5 = -0.49311

For measured results use:

a0 = 4.45656, a1 = -7.33064, a2 = 5.02097
a3 = -2.56203, a4 = 0.0643445, a5 = 0.0140405
Measurement of the Resistivity of Ultrapure Water at Elevated Temperatures page 12

New values for λOH-

The values for λOH- at high temperatures used in this paper were calculated by
extrapolating the data of Marsh and Stokes (16) and are therefore not as accurate as desired.
New values for λOH- at high temperatures can be calculated using the measured ρH2O from
Table A and the values of λH+, Kw and d from Table B. We report the value of λOH- at 75°C
to be 373 S-cm2/mole and at 100°C to be 466 S-cm2/mole. These values for λOH- are
improved based on the higher accuracy of the measured conductivity. The accuracy for
determining λOH- by this method is limited by the accuracy of Kw and λH+.

Conclusions

The resistivity and conductivity of ultrapure water have been determined both
theoretically and experimentally with particular emphasis on temperatures in the range 50
to 100°C. The theoretical calculations are based on an extensive literature survey for the best
original data for the temperature dependence of the density and ion product of water and the
specific conductivity of the hydrogen and hydroxide ions. This search showed a paucity of
good data on the hydroxide ion at high temperatures and our measurements may allow a
more accurate computation than values published in the literature.
Measurement results are given and are believed to be some of the most accurate
reported to date. The results are presented in both tabular and equation form suitable for
calculation of compensated resistivity. These measurements agree with calculations to about
±0.4% accuracy at temperatures below 70°C but differ by up to 1.0% in the 75 to 100°C range.
It is believed that the difference at high temperature is due to both inaccuracies in the
available data for the basic physical parameters and the inherent difficulty of making
accurate measurements at elevated temperatures.
Sources of error are discussed in some detail and include: accurate calculation of the
temperature dependence of resistivity; conductivity cell calibration; measurement method;
cleanliness, aging and temperature gradient in the cell; and thermal time lag. Calculations
show the sensitivity of the measurement to the error sources, and it is shown that both
temperature and resistivity must be measured with much higher accuracy at 75 to 100°C
than is necessary at 25°C to achieve the same accuracy in impurity determination. For best
results the sensor and instrument must have capability beyond that present in many
instruments designed for lower temperature applications. With careful attention to detail it
is possible to detect impurities at levels less than 1 ppb at elevated temperatures, but it is
significantly more difficult than to make the same determination at low temperatures.

Acknowledgments

The authors wish to express their sincere thanks to John DeLorey and Mark
Cruickshank for their assistance in collecting the data.
Measurement of the Resistivity of Ultrapure Water at Elevated Temperatures page 13

References

1. Harfst, W.F.; "Stringent Water Purity Requirements for the Pharmaceutical

Industry", ULTRAPURE WATER 10(7), pp. 21- (October 1993)
2. Henley, M.; "Semiconductors Makers Increasing Use of Heated High-Purity Water
for Rinsing Applications", ULTRAPURE WATER, 10(4), (May/June 1993).
3. Homnick, M.; Fulford, K.; Browne, S.; "Assessing the Design and Performance of a
Central Hot Ultrapure Water System", MICROCONTAMINATION, (February
1993).
4. Gray, D.M.; "Advances in High Grade Conductivity Measurement for Water
Treatment", ULTRAPURE WATER, 8(2), (March 1991).
5. Gray, D. M.; "Comparative Conductivity Measurement in Water Treatment",
ULTRAPURE WATER, 7(1), (January/February 1990).
6. Pate, K. T.; "DI Water Resistivity Versus Trace Ion Levels", ULTRAPURE WATER,
8(1), (January/February 1991).
7. Zabarsky, O. "Temperature-Compensated Conductivity--A Necessity for Measuring
High-Purity Water", ULTRAPURE WATER, 9(1), (February 1992).
8. Thornton, R.D; Light, T.S.; "A New Approach to Accurate Resistivity Measurement
of High Purity Water", ULTRAPURE WATER, 6(5), (July/August, 1989).
9. ASTM D1125-91, "Standard Test Methods for Electrical Conductivity and Resistivity
of Water", AMERICAN SOCIETY FOR TESTING AND MATERIALS, Philadelphia
(1992).
10. ASTM D5391-93 "Electrical Conductivity and Resistivity of a Flowing High Purity
Water Sample", AMERICAN SOCIETY FOR TESTING AND MATERIALS,
Philadelphia (1993)
11. Light, T., Thornton, R. "Conductivity Cell Standard for Low Ionic Strength
Solutions", ULTRAPURE WATER, 8(3), pp. 59-61, (April 1991).
12. Light, T.; Atwood, E.; Driscoll, J.; Licht, S.; "A New Low Conductivity Standard
Solution", ANALYTICAL CHEMISTRY, 65, pp. 181-182, (January 1993).
13. Schmidt, E.; "Properties of Water and Steam in SI-Units", Springer-Verlag, New
York, (1969).
14. Sweeton, F.H.; Mesmer, R.E.; Baes, C.F.; "Acidity Measurements at Elevated
Temperatures. VII. Dissociation of Water", JOURNAL OF SOLUTION
CHEMISTRY 3(3), pp. 191, (1974).
15. Strong,L.E.; "Aqueous Hydrochloric Acid Conductance from 0 to 100°C", JOURNAL
OF CHEMICAL ENGINEERING DATA, 25(2), pp. 104-106, (1980).
16. Marsh, K.N.; Stokes, R.H.; "The Conductance of Dilute Aqueous Sodium Hydroxide
Solutions from 15° to 75°", AUSTRALIAN JOURNAL OF CHEMISTRY, 17, pp. 740-
9, (1964).
17. Noyes, A.A.; The Electrical Conductivity of Aqueous Solutions, Carnegie Institution
of Washington, Washington, D.C., (1907).