Even though manganese(II1) seems to be the stable intermediate under the conditions used in this work, there undoubtedly exists an equilibrium betxeen manganese(II1) and manganese(1V) in solution : 2Mn(III)
= hhf2
+ Mn(1V)
(3) There is no accurate information about the extent to which this reaction might go to the right. In fact, if the commonly used Ea’s are made the basis of calculation, the equilibrium constant of Reaction 3 is found to be about lo9. This method of calculation is open to serious criticism on the grounds that the E O ’ S were measured using solid manganese dioxide and none was generally noticed in this work except in oxidation of solutions whose original manganous concentration was greater than 0.3 mm. The precipitation of solid manganese dioxide would presumably drive Reaction 3 to the right. No evidence for the existence of man-
ganese(1V) was found under the conditions of this work, so presumably it plays a minor role in both the chemical and electrical processes. HoiVerer, because precipitation of manganese dioxide was noticed at high manganese concentrations, it is recommended that sample sizes be limited so that 2.5 mg. or less of manganese is contained in 100 ml. of titration medium. ACKNOWLEDGMENT
The author thanks Merck & Co., Inc., for the argentic oxide used in this work and the Research Corp. for a grant which made possible the purchase of the Potentiostat. Thanks are also due to John Ganchoff, who performed a number of preliminary experiments. LITERATURE CITED
(1) Davis, I). G., Lingane, J. J., Anal. Chzm. ilcta 18, 245 (1958). (2) Flaschka, H., ilfikrochemie ver. Mikro-
chim. Acta’40,’42(1952).
(3) Iinuma, H., Yoshimori, T., Research Repts. Fac. Eng. Gzfu Univ., No. 6 , 88 (1956). (4) Kolthoff, I. N., Belcher, R., ‘Yolumetric Analysis,” 5-01, 111, p 629, Interscience, h’ew York, 1957. (5) Kolthoff, I. hf., Sandell, E. B , IND. ENG.CHELI., ANAL. E D . 1, 181 (19%). (6) Lang, R., Faude, E., 2.and. Chew 108, 181 (1937). ( 7 ) .Latimer, W. M., “Oxidation Potentials,” 2nd ed., p. 260, Prentice-Hall, New York, 1952. (8) Lingane, J. J., ANAL.CHEX.26, 1021
(1954).
(9) Lingane, J. J., “Electroanalytical Chemistry,” 2nd ed., p. 5G7, Interscienre. T \ ‘ P ~ -York. 1958. ...__ (10) Linginel- J. J., Davis, D. G., Anal Chim. ilcta 15, 201 (1956). (11) hleites, L., ASAL. CHEM.27, I l l 6
(1955). (12) SyrokomskiI, I-. S., Nayarooa, T. I,,J.Anal. Chem. V.S.S.R. 6 , L5(1951). (13) Tsubalii, I., Kiura, M., Japan Ana1vst 3, 137 (1954).
RECEIVEDfor review March 16, 1959. Accepted Axil 29, 1959. Division of Analytical bhemistry, 135th Meeting, ACS; Boston, hlass., April 1959.
Inte rna I EIec t roIysis - CouIomet ric Method f o r Determination of Small Quantities of Oxygen A p plication to I nd ividua I Sa m ples WILLIAM G. KNAPP Argonne National laborafory, lernonf, 111.
b Existing methods for the determination o f free oxygen, particularly in mixed gas and liquid samples, have many limitations peculiar to nuclear chemistry. A method that i s satisfactory for use with gases and/or liquids depends on the chemical reaction between oxygen and cuprous ions, and the electroreduction of the cupric ions so formed to the cuprous state. Cupric ions in the cuprous ammonium chloride reagent solution are electroreduced a t the platinum cathode o f a platinum-copper internal electrolysis cell. Conventional coulometric procedures are then used for the integration of the electrical current generated. This current is a linear function of the oxygen chemi. cally absorbed b y the oxygen-cuprous ion reaction, and, therefore, is c function of oxygen introduced to c reaction-analysis cell. Higt, sensitivity and good linearity over a wide range of oxygen concentrations are important features.
A
many conventional methods for the determination of small quantities of oxygen in individual samples are very sensitive and reliable for the applications for which they have been developed (1-4), they could not be readily modified to conform to limitations in the overlapping fields of radiation chemistry, nuclear reactor water, steam, and gas studies, and corrosion evaluations in both in-pile and out-ofpile test equipment. A new method was needed, that would be suitable for use with a variety of samples, gases, liquids, or a mixture of these, containing radionuclides above biological tolerance levels, or interfering compounds such as hydrogen peroxide. Because oxygen is occasionally encountered a t low levels, and sample quantity is frequently limited, the logical approach seemed to be to electroreducc the oxygen present directly or indirectly and to measure the coulombs required. According to Faraday’s law. the electroreduction of 1 y of oxygen would requirc LTHOUGH
201 pa. minutes, a quantity very easy to determine with conventional coulometric procedures. The limitations mentioned above, particularly as concerns mixed samples or the presence of radioactivity, suggested that the free oxygen be stripped from liquid samples or that gaseous samples should be moved (as in gas chromatography). by a n inert gashelium, nitrogen, or argon-into a reaction-analysis cell. The reagent used in this cell would absorb the oxygen and a n electrochemical determination could then be made on the reagent solution The liquid samples could then be recovered for further chemical tests, or ir the case of liquid samples of significant value (heavy n-ater) or IcontaininF precious materials (uranium-235) 07 radionuclides (fission products, etc.) i. simple recovery or disposal of the uncon taminat’ed material could be made. -4reagent \vhich haE: been used fo. colorimetric determinations of oxyce-’~: cuprous animonium cilioridt:-ia usw VOL 3’1 NC. 5 SEPTEMBEL I955
*
146:
to absorb free oxygen from the gas being passed through the reaction cell. The reaction forms cupric ions. The first attempt at electrochemical analysis of t t e cupric ions used a controlled cathode potential poa er supply to regenerate the cupric ions to the cuprous form. The current necessary to accomplish the electroreduction is a measure of the oxygen absorbed by the reagent. It was soon found that with a copper anode and a platinum cathode the electroreduction of cupric ions to cuprous ions proceeds by internal electrolysis, thus eliminating the need for an external poJTer supply. The oxygen-reagent reaction is: 02
+ 4Cuf+
+ CU'+
20'-
4 - 2 0 THD
?---
i
U5/50 J O I U T
-1
mm. C A P I L L I R I
mm O R
-&I
(1)
The cathode reaction is responsible for reduction of cupric ions: 4Cu2+
+4e
+
4Cu+
(2)
while at the anode, copper is taken into solution : 4Cu0
-4e
-t
4Cu-
(3)
S o t only are the cuprous ions regenerated by the cathode reaction, but additional reagent is generated by the anode reaction. Because the actual quantity of oxygen introduced to the cell is never significant, weightwise, this increase in concentration due to oxygen introduced is negligible. (At an oxygen absorption rate of 1 y per minute, 11.5 mg. of copper is dissolved per day.) The quantity of current generated by the cathode-anode reaction is a function of the oxygen absorbed by the reagent. The determination of the quantity of current requires that the current be integrated with time. The number of microampere minutes obtained can then be converted to the quantity of oxygen present in an individual gas, liquid, or mixed sample. Because the electroreduction proceeds by internal electrolysis exactly as it would if controlled cathode potential coulometric techniques were used, the plot of current us. time is typical of plots that are obtained under similar sample introduction conditions with other controlled potential coulometric techniques. The electroreduction current curve reaches a maximum value soon after the introduction of the sample, and the current then decreases exponentially n ith time to the original level. The same current integration methods are used here as with controlled potential coulometric methods. REAGENTS
Aqueous cuprous ammonium chloride used for the absorption of oxygen and as the electrolyte for the internal elcctrolysis-coulometric analysis. Cupric chloride is used in the preparation of the reagent berause I t is more readily soluble. Three grams of cupric chloride and 5 grams of ammonium chloride, IS
1464
ANALYTICAL CHEMISTRY
-
''L
Figure 1 . Spiral electrode dimensional details Electrolytic copper
FIUE FRIT LOCATE TO FEED INTO COPPER PPIR4L ELECTRODE
Figure 2. Reaction-analysis cell details
made to 250 ml. with 5% aqueous ammonium hydroxide, is used as a stock solution. This may be prereduced to the cuprous state by shaking the solution with metallic copper turnings, or the reagent may be reduced in the analysis cell. It requires about 1 hour for full reduction of the cupric stock solution in the cell. There is a distinct time advantage when the reduced (cuprous) stock solution is used in that the cell will come to equilibrium within 15 minutes after addition of new reagent. To realize this advantage, a glass connection between a stock solution reservoir and the reaction cell is necessary so that fresh reagent can be added without oxidation by air. APPARATUS
Various configurations of the reaction cell have been studied in an attempt to obtain the highest possible cell efficiency coupled with the fastest possible cell response. Although some cell design giving very good gas-liquid contact times of up to 90 seconds have given analysis efficiencies of loo%, the analysis of one sample in one of these cells required 2 hours or more. It m s necessary to sacrifice high efficiency t o decrease this time. The cell used for these studies contains a copper electrode as the anode of the electrolysis cell and provides a relatively long path to gas being bubbled through the solution, giving a gas-liquid contact time of about 15 seconds. Dimensional details of this electrode are given in Figure 1. A 45mesh platinum gauze about 4'/4 inches square is used as the cathode. This
gauze is fitted inside the cell n-hich is constructed from a 45/50 standardtaper joint. KO special pretreatment of the electrodes is required. The cell has a small stopcock for drainage and a 1-mm. capillary side-arm through which the gas is introduced into the cell. A fritted tip (adapted from a microfilter stick, Catalog Xo. 7310, manufactured by Microchemical Specialties Co.) is sealed on this capillary and projects into the cell so that the gas is introduced into the cell in fine bubbles which then follow the copper spiral up through the solution. The cell is diagrammed in Figure 2. A Teflon plug closes the cell and supports the anode. A platinum wire through the plug makes electrical contact with the platinum mesh cathode. The gas is vented through a hole in the plug. I n addition an exit gas trap cell, adapted from a conventional bubble counter, prevents air contamination through the vent. A sample cell (Figure 3) with provision for injecting water or gas samples is connected at the input of the reaction-analysis cell. I n these studies the sample is injected with a hypodermic syringe and fine needle through a vaccine cap at the top of the cell. Scrubbing gas is introduced through the fritted-glass bubbler at the bottom of the cell, Flow of inert gas is measured n-ith a Brooks rotameter, tube size 1A-15-1, with a borosilicate glass float. The rotameter is equipped with standard spherical (No. 12) ground joints for case in adapting into the laboratory analysis equipment. This tube size has a useful range of 10 to 200 ml. of air per minute. The current generated by the cell is measured as a voltage drop across a small resistor which short circuits the copper and platinum electrodes. A 10-
___
Figure 4. Circuit diagram
Amplifier not required for larger quantities of oxygen
Figure 3. Sample bubbler cell details
ohm precision resistor is generally used, though smaller values are sometimes desirable to extend the reactor range for higher currents, Figure 4 is a schematic of the circuit used. A Minneapolis-Honeywell Regulator Co. 0-1to (t51-mv. variable span recorder with *@50 mv. continuously variable suppression has been used for measuring and recording of the voltage drop, or, in other words, the cell current. Accessories include a stepwise variable chart speed selector (Insco Drive) giving up to 240 inches per hour, and a retransmitting slide-wire built into the recorder. An outboard integrator, also manufactured by Minneapolis-Honeywell, is used to integrate the current us. time. Various electronic computertype integrator instruments or recorder integrating devices recently introduced for use in electrochemical analyses or for gas chromatography can he adapted to this application. A Leeds & Northrup Co. microvolt amplifier with recorder output (0 to IO mv.) is used where currents are very small. Any stable amplifier with a gain of 10, a very high input impedance, and a suitable output for driving a recorder could he used if it has microvolt sensitivity. The amplifier input is connected in place of the ahove mcntioned recorder across the shorting resistor, with the amplifier output feeding the recorder. A signal gain of 5 or more is then possible. The complete assembly of the instrumentation is shown in Figure 5. METHOD
As in gas chromatography, an inert carrier gas is required. Helium, nitrogen, hydrogen, or argon is used. Bo&
tled gas of the highest purity commercially available is recommended so that background is kept to a minimum. A bottle of gas will last several months. Additional purification of the gas has not been necessary for studies to &te because background due to oxygen in the gas is electrically cancelled. A pressure-reducing valve and a needle control valve are used for gas flow control. The gas flow is set a t a rate a t which hackground levels and determinations will he made. Thirty milliliters per minute has been satisfactory for helium gas. The gas is passed through the sample bubbler cell shown in Figure 3. This cell is kept partially filled with water even between determinations so that the inert gas will he saturated with moisture and evaporation, and crystallization will not occur in the reaction cell. The inert gas carries gaseous samples, or strips free oxygen from liquid samples in this bubbler. Samples are introduced into this bubbler through the vaccine bottle-type stopper, using a hypodermic syringe. The prebubbler is calibrated so that after a sample determination is completed, a minimum quantity of liquid is withdrawn through the stopcock to allow room for the new sample to be injected. The liquid sample level in this bubbler is maintained as high as possible to eliminate holdup of liquid droplets containing free oxygen ahove the level of the liquid being stripped. Such a holdup of liquid sample would result in only partial stripping of oxygen content during the usual determination time, with slow attainment of the complete removal of oxygen from the sample. With gas samples this bubbler is kept filled with water so that void space will be minimized. The use of hypodermic syringes for sample collection and injection are less satisfactory for oxygen quantities of 0.5 y or less in a 10-ml. sample, because the introduction of this
Figure 5. omplifier
Integrator, recorder, and
Precision rerirtor at left, just below recorder
quantity of sample upsets the gas flow rate and thus upsets the reactionanalysis cell equilibrium temporarily. This departure from equilibrium becomes significant a t low oxygen levels mentioned above. It i,s necessary with these low coucentrations to bypass the bubbler while sample manipulation is being performed. Sample injection techniques and equipment similar to that used in gas chromatography are being designed and will he described in a future paper. For samples containing over 0.5 y of free oxygen in a IO-ml. sample, there is no difficulty in using the syringe method. The technique developed for handling the sample is important-for instance, it is necessary to have several milliliters of sample in the syringe in excess of that required for the determination. A rub ber eraser is used to stopper the syringe needle for transport of the sample. A slight pressure is maintained on the syringe plunger to maintain an ou& ward flow of sample during necessary manipulations, such as preparing for injection of a sample into the bubbler. Although various sampling methods are possible as determined by sample material and sampling conditions, the hypodermic syringe method described ahove has been suitable for the study and development of the method to its present form, and will be satisfactory for a great variety of applications. Prior to injection 01 the sample, the background current due to oxygen in the mert gas is allowed to come to VOL. 31, NO. 9, SEPTEMBER 1959
1465
quire 96,500 coulombs (ampere-seconds) for electroreduction, then:
SI*:-
nF
I
O2 = 32 grams X 60 sec./min. I
= 4 X 96,500 X 10-8
32 X 60 201 pa. min.
(5)
The oxygen content of the sample is linearly proportional to integrated current output, and can be calculated as follows:
i '
where I
K
figure 6. Typical current curves O a t a i n 'Table I
Table 1.
,Sample
Determination of W i i i e n c y Factor with Samples of Known Oxygen Content Sample O2Content pa. pa.
NO.
Size, M1.
1 2 3 4 5 6
10.0 10.0 7.0 3.0 1.0 0.2
a
Y
Counts
Minutes, Exptl.
Minutes, Theoretical
76.0 76.0 53.2 22.8 7.60 1.52
1733 1742 1206 526 172 35
3466 3484 2412 1052 344 70
15,280 15,280 10,690 4,580 1,530 306
7.6 P.P.M.,s Integrator
Efficiency Factor
0.227 0 228 0.226 0 230 0.225 0.229 Av. 0.228
Air-saturated water a t 28" C., 740 mm. Hg barometric pressure.
equilibrium. The zero suppression feature of the recorder is used t o adjust this current, indicated as the voltage drop across a 10-ohm resistor, to the recorder zero point. A 10-mv. full scale sensitivity is satisfactory for 10-ml. .samples containing from 2 to 100 y of oxygen. Smaller sample quantities should be used where neceesary to give an oxygen content not greatly in excess of 100 y . Greater recordtx sensitivity is necessary €or smaller quantities of oxygen. The linearity and full acale counh per minute value of the integrator will have been determined in advance of actual determinations. With the integrator on and zeroed, the a m p l e is injected. A small positive pip is usually encountered due to -displacement of inert gas by the sample, the inert gas being forced through the reaction cell. A significant reduction of current is then experienced momentarily. No explanation of this phenomenon (negative pip) can be given a t this time. Within a few seconds the oxygen will be detected and indicated as an increase in current flow through the cell. With a gas sample a current peak is reached almost instantaneously, while the peak current from oxygen in a water sample may not be reached until nearly 2 minutes after injection of the sample, depending on inert gas flow. The current then begins to decrease, partly as a function of oxygen still being stripped from the sample and introduced into the cell, and partly as a function of reattainment of the initial 1466
ANALYTICAL CHEMISTRY
equilibrium in the cell. When the oxygen is completely stripped from the sample and absorbed in the cell, the current continues to decrease due to the latter function until it levels off at zero on the recorder scale.
A family of curves is shown in Figure 6 representing from 1.52 to 76 y of oxygen in the sample. The same recorder sensitivity was used for illustration. I n practice the sensitivity would be increased for samples containing small quantities of free oxygen so that a significant integrated palue (counts) could be obtained. Integrator counts for the respective curves are given in Table I. Integrator counts are evaluated as follows: 1 count = ( B - A ) x 1000pa, min. (4) ( b - a) X R ,
where A = recorder scale zero millivolt value-zero suppression millivolts. B = recorder full scale millivolt value-span plus suppression millivolts. a = integrator counts per minute a t recorder zero (0 c.p.m. for equipment described). b = integrator counts per minute a t recorder full scale (500 c.p.m. for equipment described). R = series resistance in ohms across which voltage drop i s measured as a function of current. Because according to Faraday's law 1 gram equivalent of oxygen mill re-
=
integrator counts
= efficiency factor
For given analysis parameters (constant temperature, flow rate, etc.) all the terms except I are constant and can be grouped rn one value. Data givenin Table I was taken with {B - A ) equal to 10 mv., a equal to 0 c.p.m., b equal to 500 c.p.m., and R equal to 10 ohma, so 1 count is equal to 2 pa. minutes. The theoretical quantity of current that should have been obtained, according to Faraday's law (number of micrograms of oxygen present times 201 pa. minute) is given in column 6, Table I, while the quantity of current experimentally obtained as calculated above, Equation 4,is given in column 5, Table I. The efficiency theoretical is given for each deterexperimental minkion (column 7, Tabie I). The efficiency is a function of the chemical reaction between the free oxygen in the inert gas and the cuprous ions. This reaction can be promoted by decreasing bubble size, giving greater surface areas for reaction, lengthening time of the reaction, and increasing temperatures. A fritted-glass tip of fine porosity is used to give smallest possible bubble size, and the spiral electrode configuration gives the longest practical reaction time. A variation in gas flow rate will vary both bubble size and contact time so that the flow rate for all determinations must be the same as used for standardization. A decrease in the efficiency factor by 0.02 to 0.03 is experienced by doubling the flow rate. There must be a compromise on a rate fast enough to give reasonably short determination times, but slow enough that a maximum efficiency is obtained. With a given set af operating parameters, efficiency remains constant from batch t o batch of fresh reagent. Because temperature affects the cell efficiency, either temperature control or a correction factor is necessary. A small Dewar flask and glycerol give some degree of temperature stabilization and temperature correction factors can be experimentally determined (about 3y0 per degree), or a jacketed reaction cell can be used. Constant
)
temperature water circulated around the cell would eliminate the necessity for a correction factor. Both methods have been successful. The cell, in the presence or absence of free oxygen, is slightly photosensitive so that where the quantities of oxygen are very small, the cell should be coated with an opaque material OP otherwise shielded from light to prevent erratic results. A black cloth draped over a cell immersed in glycerol in a Dewar flask, and a painted jacketed cell have been used. Other than attempting to produce smaller bubbles and increase the reaction time as mentioned above, no effort has been made to increase the efficiency factor. This stability of this factor (Table I) is sufficient for quantitative application of the method. The useful life of the reagent is only 24 to 36 hours. It has been convenient to prepare the cell, using cupric stock solution, each evening. The gas flow rate is adjusted to about 10 ml. per minute or less and the electrodes are shorted through the series resistor. By the nest morning (or within an hour-if the cell should be needed) the cell current has reached equilibrium. The interferences encountered with carbon monoxide, hydrogen sulfide, and probably other reducing or oxidizing gases may be removed with suitable gas-scrubbing equipment. Using a single cell and one integrator, an experienced operator can run about four samples, each containing 50 to 75 y of free oxygen, per hour. This number per hour can be almost doubled by using two cells and one integrator because the remaining quantity of current necessary for attainment of equilibrium after about half of the required time for the analysis has elapsed, can be predicted with very good accuracy. By observing the cur-
fields for which it was developed-radiation chemistry, reactor water, steam and gas studies, and corrosion evaluation research. Samples will include water containing various materials (including radionuclides), and radioactive, and nonactive gases. The simplicity of the method suggests possible application to oxygen dissolved in petrochemicals, metals, etc., or to oxygen liberated or absorbed by chemical or photochemical reactions. This method can be used for continuous monitoring of gases but a majar drawback is the limited life of the reagent. The speed of response with this reagent makes it valuable for following changes of oxygen concentration 89 B function of some other physical or chemical parameter. An application where the method is used t o follow changes in water decomposition as a function of gamma radiation intensity and inert gas sweep rate will be given in a paper covering application of the general method to continuous monitoring of gases.
rent a t this time, the operator can refer to data obtained from previous determinations to estimate this quantity. He can then connect the second cell-resistor combination to the recorder-integrator, adjust the zero suppression as required, and inject a sample. I n the meantime, the first cell will attain equilibrium, although the current flowing through its shorting resistor is not being determined. After the second cell has reached a point where an estimation of quantity of current still necessary can be made, the recorder-integrator is again connected to the first cell-resistor assembly, and so on. Calculations can be made and the next samples prepared during the 6 to 10 minutes while one sample is being determined and the other cell is coming to equilibrium. APPLICATIONS
With specialized sampling equipment and techniques under study and construction, the method will be applicable to samples containing a wide.range of oxygen concentrations, and to samples containing a variety of materials which normally interfere with analytical methods r)r sample manipulations. Improved sampling methods are being devised for gases, liquids, or mixtures of these. Laboratory tests have shown that no interference is encountered with hydrogen or hydrogen peroxide which is frequently encountered in reactor studies. The sample volume required is small because of the high sensitivity of the method. Sampling manipulations can be automated Rhere required. Samples, whether liquid or gases, can be recovered. Liquids will not be contaminated. I n view of these features the method is being applied to studies in those
LITERATURE CITED
(1) Hersch, P., Dechema Monograph. 27, 299 (1956). (2) Silverman, L., Bradshaw, W., “Rapid and Precise Method for Determination of Oxygen in Certain Gases. Modified
Brady Method,” NAA-SR-1488 (May
15, 1956). (3) Wall, R. F., Ind. Eng. Chem. 49, No. 10, 77A (1957). (4) Wright, J. M., “Instructions for Installation, Operation, and Maintenance of Bettis Dissolved Oxygen Analyzer, Model 3,” WAPD-CDA(AD)-453 (Sept. 4,1958).
RECEIVED for review January 27, 1959. Accepted A ril 27, 1959. Division of Analytical 8hemistry, 134th Meeting, ACS, Chicago, Ill., September 1958. Work performed under auspices of the U. S. -4tornic Energy Commission.
Polarography of Mixtures Simultaneous Determination of Iron and Nickel SIDNEY 1. PHILLIPS and EVAN M O R G A N Research laboratory, International Business Machines Corp., Poughkeepsie, N. Y.
b Methods for rapidly analyzing mixtures of iron and nickel without the necessity of a separation are described. In ammoniacal or pyt idinical supporting electrolytes containing an excess of 5-sulfosalicylic acid, iron and nickel may b e determined with a precision to ~ 2 7 ~ .
T
HIS work was undertaken because of the need for a simple, rapid method of analyzing thin films of an ironnickel alloy. The usual methods for analyzing mixtures of iron and nickel are not attractive \?-hen only a small sample is -available. They employ either separate samples (aliquots) or a
separation. Spectrophotometric methods are tedious and time-consuming and may be subject to numerous interferences. In analyzing samples of low concentration, the methods of polarography are particularly appealing. No separation is required prior to the determination of a polarogram, and the VOL. 31, NO. 9, SEPTEMBER 1959
1467
