National Oceanic
and Atmospheric
Administration
United States
Environmental Protection
Agency
Research and Development
National Ocean Survey
Test and Evaluation Laboratory
Rockville MD 20852
Office of Energy, Minerals, and
Industry
Washington DC 20460
EPA-600 7-79-057
March 1979
Test and Evaluation of
Potassium Sensors in
Fresh- and Saltwater
Interagency
Energy/Environment
R&D Program
Report
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from the
effort funded under the 17-agency Federal Energy/Environment Research and
Development Program. These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects; assessments of, and development of. control technologies for energy
systems; and integrated assessments of a wide range of energy-related environ-
mental issues.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/7-79-057
March 1979
TEST AND EVALUATION OF POTASSIUM SENSORS IN FRESH- AND SALTWATER
by
Gary K. Ward
National Ocean Survey
National Oceanic and Atmospheric Administration
Rockville, Maryland 20852
Interagency Agreement No. D5-E693
Project No. EAP-78-BEA
Program Element No. 1 NE 625C
Project Officer
Gregory D'Allessio
Office of Energy, Minerals, and Industry
U.S. Environmental Protection Agency
Washington, D.C. 20460
This study was conducted
as part of the Federal
Interagency Energy/Environment
Research and Development Program
Prepared for
OFFICE OF ENERGY, MINERALS, AND INDUSTRY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
-------
DISCLAIMER
This report was prepared by the National Ocean Survey's Test and
Evaluation Laboratory, National Oceanic and Atmospheric Administration,
reviewed by the U.S. Environmental Protection Agency, and approved for
publication. Approval does not signify that the contents necessarily
reflect the views and policies of the National Oceanic and Atmospheric
Administration or the U.S. Environomental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement or
recommendation for use.
-------
FOREWORD
The Test and Evaluation Laboratory of the National Ocean Survey's Office
of Marine Technology, National Oceanic and Atmospheric Administration,
conducts work to:
o test and evaluate new or state-of-the art sensors for use in the
marine environment.
o determine the error bounds on chemical sensors' performance to
obtain data quality assurance information
o determine suitability of new sensors for in situ or field use
o evaluate new methods for the chemical analysis of seawater
o develop standards and calibration equipment and procedures to
maintain a quality assurance program for measurements in the
marine environment.
New chemical sensors are under continual test and evaluation to assure
that the most accurate results possible are obtained. If specific ion
electrodes are to be used to measure chemical parameters in seawater,
they must be evaluated directly in the medium of interest, i.e.,
seawater. This report investigates the basic characteristics of specific
ion electrodes in a variety of water types to determine their suitability
for in situ or monitoring applications.
Eugene M. Russin
Chief, Sensor Test Branch
Test and Evaluation Laboratory
111
-------
ABSTRACT
Three different types of potassium ion-selective electrodes,
manufactured by three different companies, were evaluated for suitability
for application in monitoring or in situ chemical analysis systems. Each
sensor was tested for the following parameters: accuracy, precision,
temperature dependence, short- and long-term stability, durability,
sensitivity to variations in light intensity and flow conditions,
response time as a function of temperature and potassium concentration,
and variations between different manufacturers. The three sensors
(glass-membrane single electrode, glass-membrane combination electrode,
and liquid ion-exchange electrode) were evaluated at 10°C and 25°C in
freshwater, synthetic seawater (35-, 20-,and 5-ppt salinity), Atlantic
Ocean water (35- and 20-ppt salinity), and Chesapeake Bay water (5-ppt
salinity). A description of the devices, the theory of their operation,
and a summary of the tests and results are included. Although all three
electrodes performed well in freshwater, the results with the liquid ion-
exchange electrode were significantly better in seawater than those with
the two glass-membrane electrodes. An accuracy of 5% in concentration
could be achieved with some of the sensors when properly and frequently
calibrated. The response times (95%) were unexpectedly long for all the
sensors and were generally greater than 10 minutes. While none of the
electrodes were affected by changes in light intensity, the two glass-
membrane sensors were sensitive to external motion and flow variations.
IV
-------
CONTENTS
Foreword 1 ][1
Abstract iv
Figures vl
Tables VI1
Abbreviations and Symbols lx
Acknowl edgment s X
1. Introduction I
2. Cone lus ions 3
Calibrations 3
Freshwater 3
Seawater *
Response t imes 4
General conclusions 5
3. Recommendat ions 6
Reliable data acquisition 6
Future development 6
4. Experimental Procedures 7
Instrument description 7
Test procedures and rationale 11
5. Results 13
Cal ibr at ions 13
Freshwater - 25 °C 16
Freshwater - 10°C 17
Seawater - 25°C IS
Response times 20
Drift 25
Environmental effects. . , 25
Accuracy 29
References 31
Bibliography 32
Appendix - Instrument Theory 33
-------
FIGURES
Number Page
1 Orion, Markson, and Thomas potassium electrodes .... 37
2 Thomas electrode response in freshwater at 25°C as
a function of the logarithm of the potassium activity . 38
3 Thomas electrode response in freshwater at 25°C as
a function of concentration 38
4 Markson electrode response in freshwater at 25°C as
a function of the logarithm of the activity 39
5 Markson potassium electrode response as a function of
concentration at 25°C 39
6 Orion potassium electrode response as a function of the
logarithm of the activity at 25°C 40
7 Orion electrode response in freshwater as a function of
concentration 40
8 Temperature dependence of the Thomas potassium electrode
in freshwater 41
9 Effect of temperature on the response of the Markson
electrode in freshwater 41
10 Temperature effect on the Orion potassium electrode
in freshwater . 42
11 Effect of salinity on the Thomas potassium electrode
at 25°C in seawater 42
12 Salinity effects on the Markson potassium electrode at
25°C in seawater 43
13 Effect of salinity on the Orion potassium electrode
response at 25°C in seawater 43
14 Time response of the Thomas potassium electrode in
Run #1 at 25°c in freshwater 44
15 Time response of the Thomas potassium electrode for
various concentrations of potassium in Run #3 in
freshwater 44
VI
-------
16 Time response of the Thomas electrode at 10°C in
freshwater solutions 45
17 Time response of the Thomas electrode in 20-ppt
salinity seawater 45
18 Time response of the Thomas electrode in 5-ppt
salinity seawater at 25°C 46
19 Time response of the Markson potassium electrode at
25°C in freshwater (Run #1) 46
20 Time response of the Markson electrode at 25°C in
various freshwater solutions (Run #3) 47
21 Time response of the Markson electrode at 10°C in
freshwater solutions 47
22 Markson electrode time response in 5-ppt seawater
at various potassium concentrations 48
23 Time response of the Markson electrode in 20-ppt
salinity seawater 48
24 Time response of the Orion potassium electrode in
freshwater at 25°C (Run #1) 49
25 Time response of the Orion potassium electrode in
freshwater solutions at 25°C (Run #3) 49
26 Orion electrode time response at 10°C in freshwater
solutions 50
27 Time response of the Orion electrode in 20- and
35-ppt salinity seawater solutions 50
28 Orion electrode time response in 5-ppt salinity
seawater for various potassium concentrations 51
29 Time response of the Orion potassium electrode in
35-ppt standard seawater at 25°C 51
Vll
-------
TABLES
Number Page
1 Manufacturers' Specifications for Potassium Sensors • • • 8
2 Coefficients for Nernst Equation for Potassium sensors . .14
3 Empirical Calibration Coefficients for the Potassium
Electrodes 15
4 Response Times of Potassium Electrodes 21
5 Drift of Electrode Potentials in Freshwater- ....••• 26
6 Electrode Drift in Seawater 27
7 Environmental Effects 28
8 Accuracy of Potassium Electrodes in Seawater 30
vm
-------
LIST OF ABBREVIATIONS AND SYMBOLS
ABBREVIATIONS
ACS —
avg —
cm —
elect -
h
IAPSO -
IUPAC -
log —
min. —
min —
ml
mV
ppt —
SYMBOLS
A
a.
AgCl —
Cl
E
E°
F
I
°K
K+
KC1
M
Na+
R
S
T
Tl
T95
z
Y
American Chemical Society
average
centimeter
electrode
hour
International Association for the Physical Sciences of the Oceans
International Union of Pure and Applied Chemistry
logarithm
minimum
minute
milliliter
millivolts
parts per thousand
activity
ionic activity
silver
silver chloride
— chloride ion
— electrode potential
— electrode constant
— Faraday's constant
— ionic strength
- degrees Kelvin
- potassium ion
— potassium chloride
- molarity
— sodium ion
— universal gas constant
- electrode slope
— temperature
— response time within 1 mV
- 95% response time
- ionic charge
- activity coefficient
— approximately
ix
-------
ACKNOWLEDGEMENT
The author wishes to acknowledge the professional and dedicated
efforts of Charles White and Paul Eichelberger, project technicians, and
Wesley Jue and Lynn Moses for their assistance in data collection and
analysis, without which this study could not have been completed. The
author gratefully acknowledges the support of this study by the
Environmental Protection Agency and the National Ocean Survey of the
National Oceanic and Atmospheric Administration.
-------
SECTION 1
INTRODUCTION
The ideal analytical instrument in potentiometric chemical
measurements would be an electrode which is sensitive only to a single
substance among all the other constituents in the system. This
requirement is met, to a certain extent, when membrane systems form the
basis of ion-selective electrodes. These devices develop an electrical
potential proportional to the logarithm of the "activity" of that ion in
solution. The "activity" of an ion is proportional to the concentration
of the ion. Generally, an electrode of this type is "selective" for one
chemical parameter, i.e. although the membrane potential can be affected
by more than one type of ionic species, the electrode selectively favors
one specific ion on a response basis. Due to this selectivity, as well
as for durability and compactness, potentiometric sensors are attractive
devices for continuous water quality monitoring, either as in situ probes
or as components in chemical analysis systems. They would eliminate the
use of discrete samples and thereby provide an ideal method for obtaining
truly representative measures of water parameters. The environment
itself would become the sample, eliminating the sampling contamination
and time lag which occurs between sampling and analysis and during which
chemical and biological alterations take place.
Presently, ion selective electrodes are available for many chemical
parameters and are used in marine chemistry research, effluent and
pollution monitoring, mixing studies, and baseline surveys. Several
commercial water quality systems offer capabilities for specific ion
electrode measurements in freshwater, estuarine, and seawater
environments.
In addition to the recent availability of specific ion electrodes
for many chemical parameters, there are several other advantages to their
use as analytical tools:
1. simple and relatively fast measurements
2. inexpensive and portable
3. capable of non-destructive analyses
4. adaptable to automatic measurands
5. capable of real-time measurements
6. continuous response for monitoring applications
7. small sample requirements
8. potential in situ appplication
Unfortunately, there are also some problems in electrode
measurements. The major difficulty is that the electrodes respond to the
activity of the ion, not the total concentration. The "activity" of an
ion can be thought of as the "effective concentration" of the free ion in
-------
solution, and is affected by temperature, pressure, ionic strength (a
chemical parameter related to the number of other ions in the solution),
and ionic interactions such as ion-pair formation or complexation. Even
if the electrodes were truly specific, they would respond only to the
free ion activity and, since many chemical species in natural waters are
complexed with other ions or adsorbed on particulates, a large fraction
may not be measured by these sensors. It is held by some scientists,
however, that it is the concentration, or activity, of the free ion that
is actually needed to evaluate the true effect of the species in the
water column. This controversy is purely academic since at this time
there is no method to accurately calculate individual ionic activities in
solutions of mixed composition. Calibration problems occur, therefore,
when the electrodes are used in complex solutions such as seawater.
A second disadvantage with specific ion electrodes is that they
were not developed as highly accurate devices and are subject to a number
of possible interferences (particularly in seawater) in addition to a
certain amount of instability or drift. Most of these problems can be
overcome with future development of selective membranes and frequent
calibration with the appropriate standard solutions. There is a need for
the development and evaluation of simple, rugged sensors for in situ
measurements as opposed to jury-rigging the already complicated
colorimetric or titrimetric analyses.
Recently, there has been a great deal of interest in ion-selective
electrodes but very little work has been done on the performance and
behavior of these devices in seawater or on their suitability as in situ
sensors. The accuracy, precision, reliability, and durability of these
electrodes must be known before they can be used for meaningful
measurement s.
The potassium ion-selective electrode can be used directly in
seawater or freshwater without sample pretreatment and therefore lends
itself to direct in situ applications. In this study, three potassium
electrodes were evaluated:
1. Potassium, Specific Ion Electrode #93-19 manufactured by Orion
Research, Inc.
2. Potassium Electrode #4202-Q10 from Arthur H. Thomas Company
3. Combination Potassium/Ammonium Selectro-Mark Electrode #1018
from Markson Science, Inc.
-------
SECTION 2
CONCLUSIONS
CALIBRATIONS
Freshwater
(1) None of the electrodes exhibited a pure Nernstian behavior
with a Nernst theoretical slope, although the Thomas Potassium Electrode
did exhibit the theoretical slope in one run in freshwater. Because of
this erratic behavior, the calibration slope must be determined
experimentally, and meters which assume the theoretical slope cannot be
used for reliable data collection.
(2) Since none of the potassium electrode responses were linear
with concentration, an experimental working curve of millivolts versus
known concentration must be constructed. At least two calibration points
should be used to closely bracket the potassium concentration in the
unknown sample since the calibration non-linearity can become quite
severe under certain conditions.
(3) All three electrodes had significant drift problems (Orion the
least) and therefore require frequent calibration (at least twice daily).
Because of the drift, the sensors cannot be used for continuous
monitoring unless a method is provided for frequent, periodic
recalibration.
(4) Whenever possible, the sample and standard should be measured
at the same temperatures. If this is not possible, a temperature
correction to the calibration curve can be generated. The Markson
Potassium Electrode was relatively unaffected by temperature variations,
particularly at high concentrations, and therefore its potential readout
could easily be corrected for temperature changes.
(5) Potassium electrodes cannot be expected to give results with
greater than 5% accuracy in freshwater or seawater.
(6) Electrodes from the same manufacturer have the same basic
characteristics, but the electrode potentials for the potassium activity
can vary significantly between sensors. This requires that each sensor
be evaluated for its individual characteristics before being employed as
a sensing device.
(7) Electrode stability increased after approximately 2 weeks of
use for the Markson and Thomas electrodes. The Orion liquid ion-exchange
electrode became remarkably stable after dropping to a low temperature
(10°C) and returning to room temperature.
-------
Seawater
(1) The glass membrane sensors (Thomas and Markson) did not
perform as well in seawater as the liquid ion-exchange electrode (Orion).
The latter was nearly linear in seawater at various salinities and in
freshwater, while the glass sensors had very nonlinear responses in
seawater. For all three sensors, however, frequent calibration in
standard seawater solutions at the salinity of the sample is necessary.
The Markson electrode was very sensitive to salinity variations and must
be calibrated carefully. The Orion and Thomas potassium electrodes were
relatively insensitive to salinity variations in seawater samples higher
than 20-ppt salinity, and therefore could be calibrated in standard
seawater solutions within the general salinity range of the sample.
(2) Background and matrix effects can be eliminated or reduced by
either: (a) the standard addition method, or (b) duplication of the
sample matrix in the standard calibration solutions. For in situ
applications, the standard addition method is difficult and impractical.
Reduction of the matrix problems was accomplished fairly well in this
study with carefully prepared synthetic seawater standards for salinities
20 ppt or higher.
Response Times
(1) Electrode response was much slower than expected and was
sometimes difficult to determine due to high drift rates. In general,
the sensors do not reach 95% of the final value for a minimum of 10
minutes. A rapid response does occur in some cases but usually involves
an initial over-response followed by a slow recovery to some final value.
(2) All the sensors responded faster to an increase in
concentration than a decrease in concentration, indicating the
possibility of a slight memory effect.
(3) For some electrodes (notably Orion), the time response curves
were independent of potassium concentration. In these situations, a
generalized response curve, over a range of potassium values, could be
developed and the final value obtained by extrapolation of the first few
minutes of sensor readings along that curve.
(4) Neither the response time nor the stability of the sensor is a
function of whether an internal or external reference electrode is used.
(5) Response was much slower at low temperatures.
4
-------
GENERAL CONCLUSIONS
(1) The electrode potentials were not affected by the height of
the filling solution as long as they were kept above the level of the
sample solutions to maintain a positive outward flow.
(2) The Orion Potassium module did not last as long as expected
and failed after 3 months. (Module is under guarantee for 6 mo.)
(3) The glass electrodes were not as durable as the plastic Orion
electrode. The body of the Markson sensor cracked in shipment, but was
quickly replaced by the manufacturer. The Thomas potassium electrode
developed an extreme sensitivity to motion of the connecting wire and was
also replaced promptly by the manufacturer.
(4) The glass membrane electrodes were more sensitive to
variations in flow around the sensor than the liquid ion-exchange sensor.
-------
SECTION 3
RECOMMENDATIONS
RELIABLE DATA ACQUISITION
(1) Soak electrodes for at least 1 week in media similar to the
samples, and subject them to low temperatures (10°C) during this time to
"break" them in.
(2) Perform a detailed calibration in standard solutions with the
same background matrix as the sample and at the same temperature.
(3) Recalibrate the electrodes at two points near the sample
concentration at least twice a day.
(4) Wait at least 15 minutes for the electrode response to
stabilize.
(5) Before using the sensors, evaluate each one for general
response characteristics, calibration linearity, daily drift, and
temperature dependence. This would not take long and could be easily
done in the lab before laboratory or field use.
FUTURE DEVELOPMENT
(1) Reinforce wire connections to the electrode.
(2) Reduce time response or make the response more reproducible.
(3) Stabilize electrode potential over a short time (1 d)
(4) Develop a combination electrode with refillable and removable
reference electrodes, which could eliminate some interference problems by
allowing variations in the filling solution and would also permit
continued use of the sensor if the reference should fail.
-------
SECTION 4
EXPERIMENTAL PROCEDURES
INSTRUMENT DESCRIPTION
Three different types of potassium sensors were evaluated in this
study. (See Figure 1.) The Thomas Potassium Electrode #4202-Q10,
manufactured by Arthur H. Thomas Company, employed a glass bulb membrane
and required an external reference electrode. The Combination
Potassium/Ammonium Selectro-Mark Electrode #1018, manufactured by Markson
Science, Inc., was a combination electrode (requiring no external
reference electrode) with a glass bulb membrane. The internal reference
electrode was a sealed silver/silver chloride reference electrode and
required no reiuvenation of the internal filling solution. The Orion
Model 93-19 Potassium Specific Ion Electrode, manufactured by Orion
Research, Inc., was a liquid ion-exchange electrode with a porous
•organophilic membrane which also required an external reference
electrode. The Orion sensor consisted of an electrode body and a
replaceable "pretested" sensing module which contained a gelled internal
filling solution, a membrane saturated with liquid ion-exchanger, and a
reservoir of liquid ion-exchanger. To activate the sensor, the module
was simply screwed into the electrode body.
The external reference electrode was an Orion 90-01 Single Junction
Reference Electrode constructed as a sleeve-type Ag/AgCl reference
electrode designed for precision measurements in conjunction with
specific ion electrodes. (See Appendix-Instrument Theory.) With this type
of sensor, we were able to compare the glass-bulb membrane against the
liquid ion-exchange electrode, and a combination electrode against a two-
electrode system.
The electrode potential was read on an Orion Model 801A Digital
pH/mV meter and a Corning Digital 112 Research pH meter. Both meters had
a range of +1999.9 mV in 0.1 mV increments, with +0.1 mV repeatability.
The high impedance input signals from the electrodes were amplified and
fed into an analog-to-digital converter which produced a number of
digital pulses proportional to the analog voltage input. These pulses
were counted, and the digital information was decoded and used to
determine the numbers displayed on the meter. The meters feature high
input impedance, extremely low drift, and an output in absolute
millivolts. The meter drift was determined by placing a shorting strap
across the terminals. The drift of both meters was less than 0.2 mV in 6
months. Response time lag was found to be less than 5 seconds by rapidly
switching from standby to millivolt measurement.
The sensing electrodes were secured in a container top which sealed
onto a polypropylene sample container, and all sensors were submerged
-------
TABLE 1. MANUFACTURERS* SPECIFICATIONS FOR POTASSIUM SENSORS
Orion
Markson
Thomas
Concentration range
pH range
Temperature range (°C)
1 to 10 5M
1-12
0-50
1 to 10 4M
6-9
—
1 to 10 4M
5-9
0-70
Elect, resistance (megohms) 0.1 - 0.3
Sample
Min. sample size
Storage
Life
a
Electrode length
Cable length
Electrode diameter
Soaking time (h)
Temperature effect
Reproducibility
Response time
aqueous only aqueous
3-ml
dry (air) dry
6 months —
+2%
< 1 min (99%)
aqueous
dry
13.5 (13.9)
75 (84)
1.2 (1.5)
0
2%
(12.7)
(74)
(1.6)
1
__
13.4 (13.4)
76 (80)
1.6 (1.6)
24
—
Values in parenthesis are actual dimensions.
Temperature effect for a 1° change.
°With calibration every hour.
For concentrations greater than 10 M.
-------
approximately 4 cm into the sample solution. The submersion depth was
kept constant throughout the tests, and the sample container was sealed
to prevent evaporation. The sample container was submerged in a non-
metallic constant temperature bath and rested on a Troemner submersible
magnetic stirrer. A star-head Teflon magnetic stirrer agitated the
solution inside the sample container. The non-metallic temperature bath
was contructed to eliminate any possible stray electrical interferences
which often occur in metal baths. The temperature was controlled to
+0.01°C with a Fisher Proportional Temperature Controller and a Neslab
FBC-4 Bath Cooling Coil. A Hewlett-Packard quartz crystal thermometer
monitored the temperature and was periodically calibrated against a
platinum resistance thermometer and Mueller bridge. Sample solutions
were suspended in the bath prior to each run to eliminate temperature
effects on the electrode response.
The seawater samples were measured for salinity on a Guildline
Model 8400 Laboratory Salinometer (Autosal) which has been evaluated and
described in detail by the National Oceanographic Instrumentation Center
(NOAA Technical Memorandum NOS 18, July 1976). The seawater samples were
analyzed for potassium on a Perkin-Elmer Model 503 Atomic Absorption
Spectrophotometer. The calibration curves and fits were determined by a
least-squares regression program on a Hewlett-Packard "9825" Calculator.
The freshwater standard solutions were prepared with Fisher
Certified ACS Reagent grade potassium chloride which was oven-dryed for
several hours and cooled in a desiccator. The standard solutions were
prepared by weight with Millipore ion-exchanged (18 megohm) water and
dried potassium chloride without further purification. Following the
formula of Kester (1967), the synthetic seawater was prepared from Fisher
Certified ACS reagents: sodium chloride, sodium sulfate, potassium
chloride, sodium bicarbonate, potassium bromide, boric acid, sodium
fluoride, magnesium chloride (hydrated)- calcium chloride, and strontium
chloride (hydrated). The synthetic seawater at 20- and 5-ppt salinity
were prepared by weight-diluting the artificial 35-ppt seawater with pure
water.
The "unknown" natural water samples were IAPSO standard seawater
(P66 27/7, 1974-C1 ppt. = 19.3675), Atlantic Ocean water (32.176-ppt
salinity from a station at 38° 40.3'N and 74° 20.0'W), and Chesapeake Bay
water (4.159-ppt salinity). The IAPSO standard seawater was weight-
diluted with pure water to obtain 20- and 5-ppt salinity seawater. The
Atlantic Ocean water was prepared in a similar manner to 20-ppt salinity.
The diluted IAPSO seawater samples provided a test for salinity effects
with no compositional changes. The Chesapeake Bay water provided samples
with polluted estuarine waters.
Since the purpose of this study was to evaluate the potential use
of potassium sensors directly in natural water environments, no reagents
were added to the samples to remove ionic strength effects or ionic
interferences. In pure water, theoretical calibration curves were
-------
determined using activities of potassium ions calculated from the
activity coefficient data of Robinson and Stokes (1965) to determine if
the sensors truly exhibited Nernstian behavior. In seawater samples,
however, a working curve was prepared as a function of salinity, since
the theoretical Nernst slope is valid only in pure water.
10
-------
TEST PROCEDURES AND RATIONALE
Each electrode was evaluated in freshwater at 10°C and 25°C, and
in five different water types at 25°C: pure water, synthetic seawater,
IAPSO standard seawater, Atlantic Ocean Water, and Chesapeake Bay water.
The pure water runs were performed to observe basic electrode
characteristics with no chemical interferences and only slight ionic
strength effects. The synthetic seawater runs were used to study the
electrodes in a known-composition solution which simulated natural
seawater and its possible matrix effects. The IAPSO standard seawater
runs were made at 35-, 20-,and 5-ppt salinity to determine the salinity
effect without possible compositional changes. Clean Atlantic Ocean
water was used to study differences between IAPSO seawater (treated
Atlantic water) and natural ocean water (for possible organic effects);
the Chesapeake Bay water provided samples with an estuarine matrix and
possible pollutant interferences.
Calibration curves in pure water were obtained for each sensor by
measuring the potassium activity in seven standard solutions ranging from
1 xlO~3M to 1.0 M KC1. Two calibration runs were made at 25°C, the first
in decreasing concentration steps, the second in increasing concentration
steps. With this procedure, we could determine the effect of increasing
and decreasing concentrations on calibration curves and response time in
addition to the short-term drift, or reproducibility, between the two
runs. The electrodes were then calibrated at 10°C in the same standard
solutions and again at 25°C. From these runs, we could determine the
effect of increasing and decreasing temperature on calibration curves and
response times, long-term drift at 25°C (from the first two runs), and
durability from temperature changes. During each calibration the sensors
were tested for: (1) short-term drift (3 h ) ; (2) sensitivity to flow
variations; (3) sensitivity to changes in light intensity; (4)
sensitivity to motion of the electrode and the connecting wire; (5)
variations in electrode potential due to changes in filling solution
heights; and (6) response time.
Calibration curves in synthetic seawater solutions were obtained
for each electrode by measuring the potassium activity in four standard
seawater solutions, which were prepared by adding known amounts of KC1
salt to the synthetic seawater mixture. Immediately after calibration,
potassium activity readings were obtained in the two natural water
samples. This procedure was completed at three salinities: 35, 20, and 5
ppt. The natural water samples were: (1) IAPSO standard seawater and
Atlantic Ocean water at 35-ppt salinity, (2) diluted IAPSO standard
seawater and diluted Atlantic Ocean water at 20-ppt salinity, and (3)
Chesapeake Bay water and diluted IAPSO standard seawater at 5-ppt
salinity. The concentrations of K+ in the natural water samples were
also measured by atomic absorption spectrophotometry. From the electrode
seawater runs, we determined the effect of salinity on the electrode
potentials and response times, stability in various water types,
11
-------
sensitivity to changes in light, organic and/or other interferences, and
suitability for possible in situ monitoring in freshwater, estuarine, and
seawater environments.
12
-------
SECTION 5
RESULTS
CALIBRATIONS
Three calibrations were run at 25°C, one was run at 10°C, and one
was run in 35-, 20-, and 5-ppt salinity seawater. Each electrode was
soaked in a 0.01 M KC1 solution for 1 week prior to the tests. All
concentrations are given in molarity, defined as moles of potassium per
liter of solution.
The electrode potentials in each solution varied a great deal
between the different sensors. In 0.01 M KC1 at 25°C, for example, the
sensors had potentials of -99.5, -0.3,and -131.5 mV for the Thomas,
Markson, and Orion electrodes, respectively. Electrode potential also
varied between electrodes from the same manufacturer. For example, in
0.01 M K (the potassium concentration in seawater), two sensors from the
Thomas Company had values of -99.5 and -19.8 mV, and two Orion sensors
had values of -131.5 and -74.0 mV in the same standard solutions.
Obviously, the absolute value of the potential in millivolts is very
dependent upon the characteristics of each individual sensing electrode
and the reference electrode used for measurements, and can only be
determined empirically by calibration in standard solutions.
In pure water, the electrode potential was first plotted as a
function of concentration and activity. The activities of the potassium
ion in freshwater were determined from the mean activity coefficient (Y+)
data for KC1 from Robinson and Stokes at 25°C. The Mclnnes convention
was employed to separate Y+_ into ionic components
[i.e.,Y-|£K ) = Y- (Cl~)]. The logarithm of the potassium activity was
fitted to the electrode potential with the equation
E = E° + S log A (1)
where S should be the theoretical Nernst slope (59.16 at 25°C), E° is the
intercept, and A is the activity of potassium ions. The coefficients, E°
and S, are given for all the sensors at 25°C in Table 2. The precision
is included as the standard deviation.
The logarithm of the potassium concentration (M) was also fitted to
the electrode response. Since the curves were not always linear, the
data were fitted to the equation
E(mV) = A + B log M + C (log M)2 (2)
where A is the intercept, B is the empirical slope, and C is the non-
linearity, or deviation, from the linear slope. Since the activity of
the potassium ions was not equal to the concentration, all the
calibration curves were non-linear; the coefficients for Equation (2) are
given in Table 3 for the freshwater and seawater calibration runs.
13
-------
TABLE 2. COEFFICIENTS FOR THEORETICAL NERNST EQUATION
(E = E° + S log A) at 25°C FOR POTASSIUM SENSORS
Calibration run
Standard deviation
FW
#1
#2
#3
34.13
31.40
110.88
Thomas
65.32
59.95
65.15
3.4
3.5
0.3
FW
#1
#2
#3
93.48
84.26
84.38
Markson
49.84
46.56
43.33
3.8
2.3
1.3
FW
#1
#2
#3
1.95
-68.97
60.34
6.1
6.5
2.0
14
-------
TABLE 3. EMPIRICAL CALIBRATION CURVE COEFFICIENTS FOR THE
POTASSIUM ELECTRODES
Calibration
Freshwater
(25°C) #1
#2
#3a
(10°C) #1
Seawater
35 ppt
20 ppt
5 ppt
Freshwater
(25"C) #1
#2
#3
(10°C) #1
Seawater
35 ppt
20 ppt
5 ppt
Freshwater
(25UC) #1
#2
#3
(10°C) #1 .
Seawater
35 ppt
20 ppt
5 ppt
A
24.45
16.51
96.04
48.30
115.32
122.84
123.86
79.28
72.47
73.84
74.11
66.27
57.49
58.82
-16.56
-19.17
42.07
42.38
45.99
44.72
53.68
B
Thomas
60.81
48.22
56.98
34.74
71.50
70.21
85.38
Markson
35.41
36.88
36.44
36.78
-7.89
7.87
35.76
Orion
54.47
42.17
50.77
40.46
39.91
41.80
59.53
C Standard deviation
-0.44
-2.40
-0.80
-5.12
15.94
12.58
12.88
-3.60
-2.06
-1.09
-0.43
-7.66
0.74
6.56
1.32
-6.70
-3.46
-4.69
-4.99
-3.06
1.31
4.5
3.1
0.3
1.2
0.1
0.2
— —
2.9
2.9
1.0
1.0
0.1
0.4
0.4
5.6
2.7
1.1
1.2
0.4
0.5
0.1
sensor.
15
-------
Freshwater - 25°C
Thomas Potassium Electrode—
The Thomas sensor was very unstable at high concentrations in Run
#1, although it had been soaking in a KC1 solution for 7 days prior to
the evaluation (6 days of soaking was recommended by the manufacturer as
the time required for optimal stability). The electrode response is
plotted as a function of potassium activity in Figure 2. The slopes of
the calibration curves (electrode response as a function of the log of
the activity) for the three freshwater runs were 65.3, 60.0, and 65.2 mV,
respectively, compared to the theoretical value of 59.2 mV calculated
from the Nernst equation (Equation 3). The third calibration run was
completed with a new Thomas electrode (the original sensor had ceased to
function) and behaved very much like the first electrode. Both sensors
had a linear response to the logarithm of the activity, indicating that
the electrodes do respond in a Nernstian manner.
The calibration curves in freshwater are shown in Figure 3 as a
function of the log of concentration. Calibrations 1 and 2 were done
with the first Thomas electrode and were 6 days apart. The coefficients
for the calibration curves changed considerably from Run #1 and Run #2:
The intercept (term A in Equation 4) decreased 33% (8 mV); the slope
(B term) decreased 21% (12.6 mV); and the non-linearity (term C in
Equation 4) increased from -0.4 to -2.4. The large decrease in the slope
reflects a reduction in the sensitivity with usage. The significant
drift that occurred in 6 days emphasized the point that the sensor must
be calibrated at least once a day to obtain results within 10%. It
should also be noted that although the slope of the new electrode in Run
#3 was only 6% different from that of the original in Run #1, there was a
large difference in electrode potential (70 mV), indicating that large
variations are possible in potentials of sensors from the same company.
Markson Potassium Electrode—
The Markson sensor was also unstable during the initial tests, but
settled down after 1 week in operation. The manufacturer's instructions
had recommended only 1 hour of soaking before use. Although the
electrode responses were a linear function of the log of the potassium
activity (see Run #1 in Figure 4), the experimental slopes of 49.8,
46.6, and 43.3 mV for Runs #1, #2, and #3, respectively, did not compare
favorably with the theoretical Nernst value of 59.2 mV; this indicates
that the sensor was less sensitive to changes in potassium activity than
expected from theory.
The electrode response as a function of concentration gave
encouraging results, however, as all three calibration curves were very
similar (Figure 5). The slope term changed only 4% over a period of
several weeks (only 1% between Runs #2 and #3), and the intercept
16
-------
increased only 2%. The non-linearity decreased slightly with each run
until the electrode response was nearly linear by Run #3. In freshwater
the Markson electrode showed very little drift after 1 week of operation,
resulting in less frequent calibration requirements, but appeared to be
less sensitive than the others to changes in potassium ion
concentrations.
Orion Potassium Electrodes—
The Orion sensor also had stability problems in Run #1,
particularly at low concentrations. However, the electrode became
remarkably stable after the 10°C calibration, which could be due to
either some stabilizing effect of the temperature drop or merely
increased stability from usage. At 25°C, the electrode response was a
linear function of the potassium activity (Figure 6) and behaved as a
Nernstian device, although the standard deviation was somewhat larger for
the Orion than for the other sensors. The experimental slopes for Runs #1, #2,
and #3 were 66.9, 55.0, and 66.0 mV, respectively, and compared favorably
to the theoretical slope of 59.2 mV.
From the calibration curves as a function of concentration, shown
in Figure 7, it is apparent that the electrode potential did not drift
once it had stabilized at concentrations X).01 M K (the concentration of
potassium in seawater) for the first two runs. At lower concentrations,
however, the Orion sensor failed to stabilize to less than +1 mV in Runs
#1 and #2, and therefore was very difficult to calibrate below 0.01 M K+.
After the 10°C run, the subsequent 25°C calibration had a slope within 7%
of Run #1, although the entire curve was offset by 58 mV at the
intercept.
In general, the slope of the Orion electrodes was linear and
changed relatively little over a period of weeks in freshwater solutions;
therefore, a daily recalibration at only one concentration is sufficient
(for an accuracy of 10%) to determine the offset from the complete
calibration curve derived weekly. For greater accuracy (5%) at least two
recalibrations per day, with at least three standard solutions are
required.
Freshwater - 10°C
Theoretically, the slope of a device exhibiting Nernstian behavior
should decrease by 3 mV/decade for the temperature drop from 25°C to
10°C. Two of the electrodes under evaluation were more than twice as
sensitive to the temperature change as predicted by theory.
Thomas Potassium Electrode—
The calibration slope decreased by 6.5 mV/decade for the 15-degree
temperature drop. (See Figure 8.) In addition to this, however, there was
a large offset (46.2 mV at 0.05M) in electrode potential at 10°C and a
17
-------
marked increase in non-linearity, indicating a sensitivity of Thomas
potassium sensors to temperature fluctuations.
Markson Potassium Electrode—
The slope of the Markson potassium electrode changed only 2.6
mV/decade at 10°C (Figure 9), and the temperature effect on the intercept
was only 0.3 mV. Other than the predictable slope change, the electrode
appears to be relatively insensitive to temperature variations,
particularly at high concentrations.
Orion Potassium Electrode—
The slope of the Orion potassium electrode decreased slightly more
at 10°C than the Thomas sensor (7.5 mV decade, see Figure 10), but was
similar to the Markson electrode in that the intercept changed only 0.3
mV for the temperature increase from 10°C to 25°C. Although the non-
linearity was greater at 10°C, the electrode stability was significantly
improved at 10°C. While the intercept (E° in Equation 1) changed
considerably (58mV) from 25°C down to 10°C, it did not change with the
increase in temperature back up to 25°C.
In general, the slopes of the electrodes response decreased by 2 to
7 mV/decade for the 15-degree drop in temperature from 25°C to 10°C.
While the intercept (in Equation 2) for the Thomas sensor changed
significantly, the temperature variations had little effect on the
intercept for the Orion and Markson sensors (0.3 mV). The results
indicate that the most reliable data can be obtained by measuring samples
at the same temperature (+0.5°C) as the calibration solutions. Since the
intercept (E°) for the Markson and Orion electrodes was not temperature
dependent, sensor readings may be corrected for temperature fluctuations
with only an adjustment of the slope term.
Seawater - 25°C
Thomas Potassium Electrode—
The seawater calibration curves for the Thomas sensor at 25°C are
shown in Figure 11 as a function of salinity and potassium concentration.
The non-linearity decreases with salinity until, at low salinity, a
nearly linear relationship is observed. The large salinity effects
emphasize that the electrodes must be calibrated in standard solutions
with a background matrix similar to that of the sample. If the
freshwater calibration were used for the seawater samples, the errors in
potassium concentration would have been 210%, 203%, and 37% in 35-, 20-,
and 5-ppt salinity IAPSO seawater, respectively. It is important to
note, however, that the differences in electrode potential between the
20- and 35-ppt standards were not large (8%); therefore, a calibration of
the electrode with standard solutions near 35-ppt salinity could be used
for most ocean samples (30-37 ppt) without introducing significantly
18
-------
large errors, since the high salinity appears to swamp out the ionic
strength effects. Below 20 ppt, however, small variations in salinity
can cause large errors unless the standard solutions are prepared close
to the salinity of the unknown sample. Due to the increased non-
linearity of the calibrations in seawater, the sensor should be
calibrated with standard solutions close to and bracketing the
concentration of potassium in the samples.
Markson Potassium Electrode—
The Markson sensor did not perform well in seawater solutions. It
is apparent from Figure 12 that the sensor is severely affected by
fluctuations of the background salinity. The non-linearity changed from
-7.7 at 35-ppt salinity to +6.6 at 5-ppt salinity and resulted in some
unusual calibration curves. The errors resulting from using the
freshwater calibration curves at 0.01 M (K+) were 210%, 183%, and 66% at
35-, 20-, and 5-ppt salinity, respectively. The data clearly show that
the Markson sensor must be calibrated not only in standard solutions with
the same background salinity, but also very near the potassium
concentration in the sample to eliminate errors from the non-linearity of
the curves.
Orion Potassium Electrode—
Figure 13 shows that not only was the Orion sensor relatively
insensitive to large changes in salinity, but also the electrode response
was very nearly linear at all salinities. The effect of salinity (5 to
35 ppt) was very small (only 1.5% error at 0.05 M K+) and, therefore,
standard solutions near 35-ppt salinity could be used for a wide range of
concentrations and salinities in oceanographic or estuarine applications.
It is still necessary, however, to calibrate the sensor in a seawater
matrix, since errors resulting from using only the freshwater calibration
would have been 144%, 91%, and 56% in 35-, 20-, and 5-ppt salinity,
respectively.
In general, the glass bulb membrane sensors were severely affected
by salinity changes and must be carefully calibrated in seawater
solutions near the salinity of the samples. Based on the calibration
data, the liquid ion-exchange sensor appears to be suitable for
oceanographic use after calibration with seawater solutions at salinities
within +_ 5 ppt of the samples.
19
-------
RESPONSE TIMES
The response times of the potassium electrodes are summarized in
Table 4 in the following forms: internal 95% response time, Tgc (l);
external 95% response time, Tgc (E); and the ITJPAC-recommended response
time, Tj . The first two forms are the 95% response times, which are
defined as the time required for the system output to attain 95% of the
asymptotic value when subjected to a step input, where Tgc, (I) differs
from TQC (E) by the type of step change. The 95% response is equal to
three time constants" (1-e ) of a pure exponential response. The third
response time, T-, , is the time required for the sensor to reach a value
within +1.0 mV ( 3% in concentration) of the final electrode potential.
The electrode output was monitored continously for 3 hours in each test
solution, initiating from the moment the sensor was immersed. The TQC
(internal), or TQC (I), is the time required for the electrode response
to reach 95% of the change between the initial and final electrode
potentials in the same standard solution. The TQC (external), or TQC
(E), is the time required for the sensor output to reach 95% of the total
change between the final potential in two consecutive test solutions at
two different concentrations. The Ti response time is considered to be
the most useful parameter for evaluating the electrode equilibration
time, since it is independent of the magnitude of the concentration step-
change. All three methods of expressing the response time, however, are
presently in use by various investigators.
The responses of each potassium sensor in freshwater and seawater
have been plotted as "delta mV" versus time (minutes) at 10°C and 25°C.
The "delta mV" term is the difference between the initial electrode
potential obtained immediately after immersion in the solution and the
potential output at time, T. In this manner, the electrode response at
various concentrations (the Bottle numbers in the figures correspond to
various levels of potassium concentration) could be easily depicted on
one graph.
The calibration #1 at 25°C was performed by starting with the
highest concentration standard, and proceeding in decreasing
concentration steps to the lowest concentration standard. Freshwater
Runs #2 and #3 at 25°C and the 10°C calibration were completed in reverse
order, with progressively increasing concentration steps. The molarity
of the standards correspond to the bottle numbers in the response curve
plots as follows: 0.5 M in #2, 0.1 M in #3, 0.05 M in #4, 0.01 M in #5,
and 0.005 M in #6.
Thomas Potassium Electrode—
The electrode response is shown as a function of time and potassium
concentration in Figure 14 for freshwater solutions at 25°C# The
response in Run #1 at high concentrations (>0.01 M) is characterized by a
rapid response, a negative overshoot, and then a recovery to the final
stable value. The recovery from the overshoot was usually very slow, and
20
-------
TABLE 4. RESPONSE TIME (MINUTES) OF P.OTASSIUM ELECTRODES
Freshwater
25°C. #1
#2
#3
10°C #1
Seawater
5 ppt
20 ppt
35 ppt
Thomas
51a(79)b,56c
13(45) ,24
28(52), 28
32(66), 24
25(30)
41(29)
20(19)
Markson
30(43), 32
11(21), 12
9(29), 19
21(32), 22
55(49), 22
56(41), 21
42(55), 33
Orion
57(82), 56
38(76), 54
1(29), 6
57(75), 50
14(54), 6
1(11), 1
3(20), 7
a
Tqc. with initial potential equal to the electrode potential in the
previous standard solution, Tq (E).
T_5 (in parentheses) with initial potential at 0.3 minutes in the
same solution, T (I).
Q
Time required for electrode response to reach +lmV of the final value T
(recommended by IUPAC).
21
-------
the response did not level off for more than 90 minutes. In the low
concentration samples, the overshoot disappeared, and a "normal" response
curve was observed. The electrode response in Run #2 was very similar to
those at low concentrations in Run #1. Although the response curves in
freshwater Run #3 (Figure 15) were obtained with a new Thomas electrode,
it also exhibited the overshoot characteristic of the original electrode.
This second sensor had a long positive overshoot at high concentrations,
which decreased with decreasing concentration and eventually became
negative, similar to that observed at low concentrations in Run #1. It
should also be noted that, after the initial response period (60 min),
the sensor drifted constantly~0.2 mV every 15 minutes.
At low temperatures (10°C), the overshoot observed in the 25°C runs
became much larger and longer (see Figure 16), resulting in longer
response times. At both temperatures, the potential in the low
concentration samples was less stable than in high concentrations.
The sensor response in 35- and 20-ppt salinity synthetic seawater
was characterized by a .rapid response, followed by a general positive
drift of~0.4 mv/15 min (see Figure 17) without the overshoot found in
the freshwater samples. In 5-ppt salinity seawater, however, a long (30
min) overshoot was observed prior to the resumption of the positive drift
found at the higher salinities (Figure 18). In IAPSO standard seawater,
Atlantic Ocean water, and Chesapeake Bay water, the response was the same
as in synthetic seawater at equal salinities, except that the drift was
slightly greater. The similar responses in synthetic seawater and
natural seawater indicate a general insensitivity of the sensor to
dissolved organic components in natural seawater and possible pollutant
interferences in Chesapeake Bay water; therefore, these electrodes could
be used in natural waters after proper calibration with synthetic
seawater (with respect to the response time). The similarity of the
curves in seawater at high salinities would allow the characterization of
a general response curve, then the calculation of a final value by
extrapolation along that curve after a few minutes of initial readings.
This would not be possible in freshwater, however, since the shape of the
curves changed drastically with variations in concentration (Figures 14
and 15).
Markson Potassium Electrodes—
The response curves from the Markson sensor, given in Figure 19 for
Run #1 in freshwater, were characterized by a rapid response, a positive
overshoot, and finally a return to a stable reading «0.1 mV/30 min ).
The time required for the sensor to recover from the overshoot, however,
increased as the concentration decreased. The response in Run #2 was the
same as in Run #1 except that the overshoot disappeared at high
concentrations. The sensor stability, in contrast to the Thomas
electrode, was not affected by concentration. The electrode response in
freshwater for Run #3 (Figure 20) was similar to Run #1 except that the
overshoot was not as pronounced and became negative at high
22
-------
concentrat ions.
The low temperature calibration (10°C) was characterized by the
same positive overshoot observed at 25°C that increased with decreasing
concentration (Figure 21). The recovery from the overshoot, however, was
much slower than that in the 25°C calibration.
The electrode time responses in synthetic seawater solutions
(Figures 22 and 23) were remarkably independent of salinity, but varied
considerably with changes in potassium activity. The electrode
potentials were much less stable in the seawater runs and drifted in the
negative direction by 0.2-0.9 mV every 15 minutes. The response in the
high-and medium-salinity Atlantic Ocean water and standard seawater were
similar to those in the synthetic seawaters at equal salinities except
that the sensors were more stable, exhibiting drifts of only 0.2 mV/15
minutes or less. In the polluted low-salinity Chesapeake Bay water,
however, the electrode rarely stabilized and, after 1 hour, continued to
drift at a rate of 0.4 mV every 15 minutes.
In general, the response times were much slower in both synthetic
and natural seawater, and the drift problems were much more severe. In
unclean waters, the glass sensing membrane appears to be significantly
affected, resulting in longer response times and an increase in drift.
Since the response curves changed with potassium concentrations, it is
not possible to prepare a general response curve in freshwater or
seawater.
Orion Potassium Electrode—
The response curves of the Orion electrode in Run #1 of the
freshwater calibrations at 25°C turned out to be rather unusual (see
Figure 4). At the high and low concentrations, the overshoot
characteristic for the other potassium electrodes was absent. In the
middle concentration range, however, a negative overshoot was observed,
and the electrode was generally unstable at high and low concentrations.
In Run #2, similar behavior was found. After more than 2 hours, the
electrode potential was still unstable (^.4 mV/15 minutes). Fortunately,
Run #3 was completed after the 10°C runs and was found to be remarkably
stable compared to the first two runs, with final values that varied less
than 0.1 mV/15 minutes. (See Figure 24.) Although an overresponse was
still found at concentrations >0.01M, the response curves were almost
identical regardless of potassium concentrations. The temperature drop
to 10°C between Runs #2 and #3 seems to have stabilized the sensor. (See
Figure 26.)
The Orion electrode also behaved well in the synthetic seawater
samples. TheAmV curves in 35 and 20 ppt for all K+ concentrations were
nearly identical «0.3 and 0.9 difference inAmV at 35 and 20 ppt
respectively) and are shown in Figure 27 as a function of potassium
concentration at 35-ppt and 20-ppt salinity. After approximately 20
23
-------
minutes, there was a small negative drift of ~ 0.1 mV/15 minutes.
Although the characteristic overshoot was not observed at 35 ppt, a
slight overshoot was found at low concentrations in 20-ppt salinity
seawater, which increased as the potassium concentration decreased. At
5-ppt salinity, the response curves were quite different from those at
the higher salinities, but did not change significantly with variations
in potassium concentration. (See Figure 28.) At this low salinity, all
time response curves exhibited a negative over-response, characteristic
of many electrodes in freshwater, which also increased with decreasing
concentrat ion.
The electrode time response was almost identical in Atlantic Ocean
water and IAPSO standard seawater at equal salinities. However, unlike
the synthetic seawater samples, the sensors responded very rapidly «20
sec) in 35- and 20-ppt salinity samples then remained very stable for a
short period (~10 min before the electrode potential began to drift
0.3 mV every 15 minutes. (See Figure 29.) The response in 5-ppt natural
seawater (IAPSO standard seawater and Chesapeake Bay water) was very
similar to that in the 5-ppt salinity synthetic seawater samples and
therefore had the characteristic negative overshoot and recovery,
followed by a positive drift. The large drift (0.6 mV/15 min), found in
Chesapeake Bay water, was significantly greater than that in IAPSO
seawater (0.4 mV/15 min) and indicates a susceptibility of the sensor to
interferences in polluted waters.
In general, once the Orion sensor was stabilized at 10°C, the
electrode response in freshwater and saltwater was almost independent of
variations in potassium concentration and would permit the development of
a general response curve from which final values could be extrapolated.
In seawater, the sensor is also unaffected by salinity changes in
solutions over 20 ppt and could be calibrated in solutions approximately
equal in salinity to the unknown samples.
24
-------
DRIFT
The drift of the electrode potential was monitored over three
different time spans. The short-term stability was determined from the
change in response from TIQQ for a period of 2 hours. The drift over 1
day was determined from the reading taken 3 hours after electrode
immersion until the reading of the following day. The drift over 1 week
was determined from the difference between the three calibration runs
which were completed 1 week apart in the same standard solutions. Due to
the large variations in absolute values for electrode potentials, longer
drift tests were deemed unnecessary.
The hourly, daily, and weekly drifts are given in Table 5 for each
sensor in freshwater. Since the electrode potentials rarely stabilized
completely in the seawater solutions, the drift after 1 hour of readings
is given in Table 6 for results in natural and synthetic seawater as a
function of salinity. Also included in Table 6 is the average value for
drift in freshwater at 10°C and 25°C. In general, the greatest drift in
potential occurred at low concentrations during the first 5 hours, after
which the sensors stabilized to 0.1 - 0.2 mV per hour.
The Thomas electrode had the worst short-term drift over the period
of 1 day (16% error) in freshwater and was particularly bad in high-
salinity (20 and 35 ppt) natural waters. The sensor had less drift in
low salinity (5 ppt) seawater and freshwater. The Markson potassium
electrode settled down with use (See Table 5), but exhibited excessive
drift in IAPSO standard seawater at 35- and 20-ppt salinity. The drift
in the potential of the Orion potassium electrode dropped significantly
after the 10°C calibration but gradually increased with continued use in
seawater solutions. ( See Table 6.) There was also slightly less drift for
the Orion electrode in synthetic seawater, relative to the natural
seawater. The Orion electrode drift, however, was remarkably constant
and predictable at all salinities (See Figure 27) and therefore could be
treated as a drifting baseline.
ENVIRONMENTAL EFFECTS
The sensitivity of each sensor to light and flow variations and to
external motion of the electrode body and wire is given in Table 7.
Although none of the sensors were sensitive to changes in light
intensity, the Markson electrode and the first Thomas electrode were very
dependent on flow around the sensor. The Orion was fairly stable under
all environmental conditions.
25
-------
TABLE 5. DRIFT (mV)a OF ELECTRODE POTENTIAL IN FRESHWATER
Time range
T100 to 2 hours
Run #1
Run #2
Run #3
1 dayc
1 week (average)
c
1 week
Thomas
0.6
0.7
0.4
4.3
5.7
6.0
Markson
0.8
0.3
0.2
3.4
7.3
11.7
Orion
0.8
0.7
0.2
2.6
4.4
18.6
aDrift of 2 mV corresponds to an error of 7.5% in concentration.
^T,„„ is the time at .which sensor stabilized after initial response.
cWorst case values.
26
-------
TABLE 6. ELECTRODE DRIFT IN SEAWATERC
Water type
Thomas
Markson
Orion
35 ppt
20 ppt
5 ppt
0 ppt
(1) synthetic
(2) IAPSO
(3) Atlantic Ocean
(1) synthetic
(2) IAPSO
(3) Atlantic Ocean
(1) synthetic
(2) IAPSO
(3) Chesapeake Bay
(1) 25°Cb
(2) 10°C
0.3
0.4
0.6
0.4
0.0
0.6
0.2
0.3
0.1
0.4
0.3
0.5
1.2
0.1
0.3
0.8
0.1
0.4
0.1
0.4
0.1
0.1
0.2
0.3
0.2
0.1
0.3
1.3
0.3
0.4
0.5
0.4
0.5
aDrift in terms of change in mV every 15 minutes after 45 minutes of
readings (0.2 mV = 0.8%).
'Average value for all three runs.
27
-------
TABLE 7. ENVIRONMENTAL EFFECTS3
Parameter
Light variation
Flow variation
External motion
Thomas
0
9.9,0.1
+14, +0.4
Markson
0
6.0
+4.5
Orion
0
0.2
+0.2
aChange in electrode potential (mV).
Two values are given for Thomas. The first value corresponds to the
first Thomas sensor in Runs #1 and 2. The second value is for the second
Thomas electrode after the first sensor failed at 10°C.
28
-------
ACCURACY
The accuracy of each sensor in natural waters was determined by
comparing the value obtained from the electrode, after calibration in
seawater, to the value obtained by atomic absorption analysis. The
errors are given in Table 8 as percent deviation of the electrode values
from the true concentration of potassium in each of the natural water
samples. The electrode concentration values were obtained from the
electrode potential after 1 hour of response. The electrodes were
calibrated with four standard seawater solutions, followed immediately by
the determination of the unknown concentrations the same day. The large
errors from the glass-membrane sensors may be due to the drift of the
electrode potential in seawater. (See Table 6.)
29
-------
TABLE 8. ACCURACY3 OF POTASSIUM ELECTRODES IN SEAWATER
Salinity (ppt)
35
20
5
IAPSO
Station A
IAPSO
Station A
IAPSO
Chesapeake
aAccuracy is
Thomas
25.6
36.8
43.6
6.7
1.3
Bay 57.7
expressed as percent
Markson
27.0
35.1
44.9
5.7
99. 7b
47.9
error, % error =
Orion
1.0
2.5
3.0
0.2
28.0
11.0
100 (observed-true) .
Extreme drift problems.
(true)
30
-------
REFERENCES
Boyd, J.E. 1976. Performance Evaluation of Guildline Model 8400 Laboratory
Salinometer. NOAA Technical Memorandum NOS 18 (PB 259696), National
Ocean Survey, NOAA, Rockville, Md. 20 pp.
Kester, D.R. 1967. Preparation of Artificial Seawater. Limnol. Oceanog. 12,
176.
Robinson, R.A. and R.H. Stokes. 1965. Electrolyte Solutions. Butterworths,
London. 571 pp.
31
-------
BIBLIOGRAPHY
Bates, R.G. 1973. Determination of pH - Theory and Practice. John Wiley
and Sons, New York. 479 pp.
Durst, R. 1969. Ion-selective electrodes. National Bureau of Standards
Publication 314, Washington, D.C. 452 pp.
Koryta, J. 1975. Ion-selective electrodes. Cambridge University Press,
Cambridge. 207 pp.
Riley, J.P.,and G. Skirrow. 1975. Chemical Oceanography (Vol. 4), Academic
Press, New York. 363 pp.
32
-------
APPENDIX - INSTRUMENT THEORY
Ion-selective electrodes generally have a sensing membrane which allows
only the ion of interest (potassium in this study) to pass from the sample
solution at the outer membrane surface to an internal solution in contact
with the inner membrane surface. The internal solution contains a fixed
activity of the ion to which the membrane is permeable. When the electrode
is placed in a sample solution, there is a momentary flux of ions across the
membrane in the direction of the solution containing the lower activity of
the mobile ion. Since the ions carry a charge, an electrical potential is
set up which opposes further ion migration. Eventually, an equilibrium is
established in which the potential across the membrane is exactly that re-
quired to prevent further net movement of ions.
Changes in the membrane potential can be measured by making electrical
contact to the inner solution with a suitable reference electrode. At the
same time, the sample solution is in electrical contact with a second
reference electrode via a salt bridge. In some electrodes, this second
reference electrode is contained inside the body of the ion-selective sensor
which is therefore referred to as a "combination electrode." A high input
impedance voltmeter connected across the two electrode leads will indicate
a potential given by the Nernst equation
E = E° + 2.3 RT log A (3)
zF
where E is the potential in millivolts developed by the system. The term
E° is a constant which depends on the particular choice of reference
electrode used, the choice of ion-activity in the inner solution, and the
small potential due to the liquid-liquid junction at the salt bridge
connection. The term RT/zF is the Nernst factor where R is the universal
gas constant, T is the temperature (°K), and F is Faraday's constant.
The value of the Nernst factor at 25°C is 59.16 mv per decade. The term
A is the activity of the ion (id") to which the membrane is permeable, and
z is the charge of that ion (+1). The activity is equal to the product
of the ionic activity coefficient ( Y +) and the concentration of the ion
(M),
A =Y+ M (4)
and Y + is a function of ionic strength and ionic interactions (such as
complexation). Ideally, the electrode pair should be calibrated with
standard solutions of known activity and fit to Equation (3). In a plot
of E (in millivolts) versus the logarithm of the activity, the slope
should be equal to the Nernst slope (2.3 RT/zF) and the intercept equal
to E° if the sensor has a true Nernstian response. For practical
applications, the electrode potential (E) is plotted as a function of the
log of the concentration (M) to provide a working curve for concentration
measurements in unknown solutions. Since the ionic activity sensed by
the electrode is not equal to the true concentration due to background
interferences, the electrode must be calibrated in standard solutions
33
-------
which duplicate the background matrix effects of the sample. For
example, an electrode measurement in seawater requires calibration of the
sensor in standard seawater solutions.
Two types of potassium electrodes were evaluated in'this study: two
glass electrodes and a liquid ion-exchange electrode, all of which were
quite different in construction and selective processes. The glass
electrodes have a sensing element made of a thin-layered bulb prepared
from special glass. The glass electrodes are usually made from mixtures
of oxides of elements with oxidation state of 3 or greater (i.e.,
silicon, aluminum) and oxides of elements of oxidation state 1 or 2
(i.e., K , Na ). When they are melted and subsequently cooled, these
oxides form a 3-dimensional solid in which the most mobile charged-
species are the monovalent cations. A membrane made of this glass is,
therefore, permeable almost solely to the cation and functions as a
cation exchanger. The result is that a Nernst potential is observed when
such a membrane separates two solutions of a single salt at two different
concentrations. The electrode is prepared by filling a thin-walled glass
bulb with a solution of the salt (K ) of constant composition. The
electric potential measured using such a half cell depends only on the
activity of the cation (a^) in the external solution and, therefore,
E = constant +2.3 RT (log a^)
zF
This provides the means to measure the activity of the cation in
different solutions as long as an appropriate reference electrode is used
to complete the circuit.
The liquid ion-exchange electrodes have a sensing element
consisting of an organic liquid ion-exchanger dissolved in an organic
solvent. The ion-exchange liquid is held in contact with the sample by
means of an organophilic porous membrane surrounded by a circular porous
plastic reservoir saturated with the ion-exchanger. The internal aqueous
reference salt solution is on the inside of the membrane. The transport
of an ion through the membrane is dependent upon the process by which the
ion enters the membrane and the movement of the ion within the membrane
phase. Selectivity and rejection of other ions is achieved by blocking
the ability of the ion to pass through the membrane solution interface or
to move across the membrane. The selected ion can move freely in the
membrane by diffusion. Since the liquid phase is in contact with the
aqueous sample solution, it must be water insoluble and have a low vapor
pressure to prevent significant evaporation. In most cases the liquid
phase is a relatively low dielectric-constant, high molecular-weight
organic liquid allowing the site and the ion to move together through the
membrane phase. At the membrane interface, a process of ion exchange
takes place between the ions of the ion-site in the organic phase and the
free ions in the aqueous phase.
The reference electrode is that half of the electrode pair which
provides a constant reference potential regardless of solution
composition. The potential developed by the sensing electrode is
34
-------
measured against the reference potential to yield an overall system
potential that is proportional to the activity of free ions.
Two types of reference electrodes were evaluated in this study:
the first type was contained within the sensing electrode referred to as
a "combination electrode;" the other type was an external single-junction
reference electrode with the internal filling solution chosen to form a
low junction-potential liquid junction with the sample solution. A
sleeve-type reference electrode was used for those sensors requiring an
external reference electrode. The filling solution of the sleeve-type
reference electrodes contacts the sample solution by means of a narrow
ring-shaped opening between an outer sleeve and the inner body of the
electrode. The space between the body and sleeve widens above the tip to
form a conical reservoir of filling solution. This type of electrode
provides exceptionally stable junction potentials, making them especially
suitable for specific ion and precision pH measurements. The junction
area, due to its size and the high leak rate of the internal filling
solutions, does not become easily clogged as in the case of the frit-
type, or ceramic, junction reference electrodes.
Many of the problems with ion-selective electrodes can be traced to
the liquid junction in the external reference electrode. The problem can
be clarified if one compares the ion-selective electrode measurement with
a typical pH measurement. Most laboratory pH measurements are made to an
accuracy of 0.1 pH units, which corresponds to approximately 6 mV in
electrode potential. However, with specific ion sensors, the analyst
usually tries to obtain 1-2% in the activity of the ion being measured.
This requires an accuracy of 0.1 to 0.2 mV. Any irreproducibility or
drift in the liquid-junction potential becomes much more apparent in
specific ion measurements. In this evaluation, we used the same
reference electrode for all measurements to eliminate possible
differences in electrode response due to the reference electrode.
Duplicate runs were also made with another reference electrode. Both
reference electrodes gave the same results for calibration curves and
response times.
Significant liquid-junction potentials arise anytime two solutions
of different compositions are brought into contact (i.e., internal
filling solution and the sample solution). The potential results from
the interdiffusion of the ions in the two solutions. Since different
ions diffuse at different rates, the electrode charge will be carried
unequally across the solution boundary, resulting in a potential
difference between the two solutions. For reliable electrode
measurements, liquid-junction potentials in the standardizing solutions
must be equal to that in the sample solutions. Sensors for seawater
analyses, therefore, must be calibrated in standard solutions with a
background matrix similar to seawater. If this procedure is not
followed, the change in liquid-junction potential between the standard
solutions and the sample will appear as an error in the measured
electrode potential. The structure of the reference electrode can also
35
-------
cause liquid-junction problems. A small continous flow of the reference
filling solution into the sample solution must be maintained. The
velocity of the flow must be just enough to overcome back-diffusion of
sample ions into the junction itself. Otherwise, there would be a steady
buildup of sample ions inside the junction which would, in time, give
rise to variations in the junction potentials. The reference electrodes
used in this evaluation were constructed to eliminate the problem.
36
-------
Piqure 1. The Orion, Markson, and Thomas potassium electrodes,
37
-------
20 -y-
-10 --
-40 --
-70 --
-100 --
-130 --
-160
-3.0
Figure 2.
90 -j-
40 --
-10 --
-60 --
x*
-110 --
-160
-3.0
Figure 3.
-2.5
-2.0
-1.0
-0.5
0.0
-1.5
Log A
Thomas electrode response in freshwater at 25°C as a function
of the logarithm of the potassium activity.
-2.5
-2.0
-1.0
-0.5
0.0
-1.5
Log M
Thomas electrode response in freshwater at 25°C as a function
of concentration.
38
-------
90 -r
60 --
30 --
0 --
-30 --
-60
-3.0
-2.5
-2.0
-1.0
-0.5
0.0
-1.5
Log A
Figure 4. Markson electrode response in freshwater at 25°C as a function
of the logarithm of the activity.
90 -r
60 --
30 --
0 --
-30 --
-60
« RUN 1
+ RUN 2
o RUN 3
-3.
0
+
-1.0
-0.5
Figure 5.
O ^ —O 171 — 1 ^
Log M
Markson potassium electrode response as a function of
concentration at 25°C.
39
0.0
-------
-25 -r
-55 --
-85 --
-115 --
-145 --
-175 --
-205
-3.0
Figure 6.
35 -r
-2.5 -2.0 -1.5 -1.0 -0.5
Log A
Orion potassium electrode response as a function of the
logarithm of the activity at 25°C.
0.0
-205
-3.
-2.5
-2.0
-1.0
-0.5
-1.5
Log M
Figure 7. Orion electrode response in freshwater as a function of
concentration. 49
0.0
-------
40 -r
' -10 --
-60 --
-110 --
•160
25 C #2
10 C
-3.0
-2.5
-2.0
-0.5
-1.5 -1.0
Log M
Figure 8. Temperature dependence of the Thomas potassium electrode in
freshwater.
90 -r
60 --
30 --
0 --
-30 --
0.0
-60
-3.
* 25 C #3
+ 10 C
0
-2.5
-2.0
-1.0
-0.5
0.0
-1.5
Log M
Figure 9. Effect of temperature on the response of the Markson electrode
in freshwater.
41
-------
50 -r
10 --
0.0
-1.5
Log M
Temperature effect on the Orion potassium electrode in
freshwater.
Figure 10.
35 PPT
20 PPT
5 PPT
Figure 11.
-2.0 -1.5 -1.0
Log M
Effect of salinity on the Thomas potassium electrode
at 25°C in seawater.
42
-0.5
-------
70 -r
55 --
40 - =
25 --
10
-3.0
.
-2.5
-0.5
-2. 0 -1. 5
Log M
Figure 12. Salinity effects on the Markson potassium electrode at 25°C.
-115
-35 --
-55 --
-75 --
-95 --
-3.0
-2.5
-1.0
-0.5
-2.0 -1.5
Log M
Figure 13. Effect of salinity on the Orion potassium electrode response
at 25°C in seawater.
43
-------
8 -r
-4
-8
1 1 1 1 1 1
1 1 I 1
3 30 60 90
#5
1 #6|
120
o
-P
r—I
-------
25 -r
16 --
o
-P
t—4
O
Q
-2
Figure 16
80 -r
30
60
Time (min)
90
120
Time response of the Thomas electrode at 10°C in freshwater
solutions.
#1
60 --
40 --
20
0
Figure 17.
Time (min)
Time response of the Thomas electrode in 20-ppt salinity
seawater.
45
-------
o
-P
i—i
-------
D
+>
i—i
0>
90
J5
#4
#3
#2.
30 60
Time (min).
Figure 20. Time response of the Markson electrode at 25°C in various
freshwater solutions (Run #3).
25 -r
120
-10
Figure 21.
30 60
Time (min)
Time response of the Markson electrode at 10°C in freshwater
solutions.
47
-------
o
-p
15 --
10 --
5 --
0
-5 --
-10 --
-15
0
Figure 22.
5 -r-
20 40 60
Time (fflin)
Markson electrode time response in 5-ppt seawater at various
potassium concentrations.
-11
0
Figure 23.
Time (rain)
Time response of the Markson electrode in 20-ppt salinity sea-
water,
48
-------
-4
30
90
120
60
Time (min)
Time response of the Orion-potassium electrode in freshwater
at 25°C (Run #1).
#2
=»
-20 --
-25
Figure 25.
1 1
1 t
3 30
1 -—
1
e'0
1 1
' 90
#6
1 1
120
Time (min)
Time response of the Orion potassium electrode in freshwater
solutions at 25°C (Run #3).
49
-------
10 -I-
0
(D
CD
-10 -h
-15 -h
-20
30
60
Time (min)
90
Figure 26. Orion electrode time response at 10°C in freshwater solutions.
4 -r
20 PPT
. . 35 PPT
-2
Figure 27.
Time (min)
Time response of the Orion electrode in 20-
seawater solutions.
50
and 35-ppt salinity
-------
-6
Figure 28.
5 --
3 --
ime
(min)
Orion electrode time response in 5-ppt salinity seawater for
various potassium concentrations.
o
1 --
Q -1 -h
-3 --
-5
Figure 29.
—» 1 1 1 1 1
20 40 60
Tim© (min)
Time response of the Orion potassium electrode in 35-ppt standard
seawater at 25°C.
51
------- |