due to the nonstoichiometric reaction of manganese dioxide with manganous sulfate to form manganic ion. Under the experimental conditions used manganic ion or manganese dioxide was always in excess, so that the decomposition of either oxidant may be responsible. The data on titration efficiencies show that results may be quantitative if a reductant is present. I n these cases results are more favorable because the oxidant is consumed as rapidly as it is formed. I n addition, the partial interaction of current with the reductant is undoubtedly responsible to some extent for the improved results. Of all of the conditions studied in the investigation, the direct oxidation of manganous to manganic ion in mixed phosphoric-sulfuric acids had the highest current efficiency. Similar conclusions were reached by Horn (1.4) in a previous study. The process, however, is efficient only at low current densities. The addition of phosphoric acid to the solution apparently lowers the diffusion current of manganous ion (Figure 5, curves A and B ) by complexation. Theoretical current efficiencies of nearly 100% are obtained only at current densities less than 1.5 ma. per sq. em. This upper liiit is dependent, of course, on the concentrations of phosphoric and sulfuric acids and particularly on the concentration of the manganous salt. Either manganic ion or ferrous ion may
be generated first with little difference in precision or accuracy of the results under the optimum conditions. A dual intermediate system utilizing manganic and ferrous ions was found practical. Such a system would be useful for oxidations wherein a n excess of reagent is necessary. After this work was completed, Selim and Lingane published a paper on the generation of manganese(II1) in sulfuric acid medium (28).
REFERENCES
(1) Adams, R. N., Reilley, C. X., Furman, N. H., ANAL.CHEM.25, 1160 (1953). (2) Belcher, R., West, T. S., Anal. Chim. Acta 6,322 (1952). (3) Campbell, A. N., Trans. Faraday SOC. 22, 46 (1926). (4) Cooke, W. D., Reilley, C. N., Furman, N. H., ANAL.CHEM.23, 1662 (1951). (5) Delahay, P., “New Instrumental Methods of Electrochemistry,” p. 36, Interscience, New York, 1954. (6) Domange, L., Compt. rend. 208, 284 (1939). (7) Drummond, A. Y . , Waters, W. A,, J . Chem. SOC.1953, 435.
( 8 ) Fenton, A. J., Jr., Furman, N. H., ANAL.CHEM.29, 221 (1957). (9) Furman, N. H., Adams, R. N., Ibid.,
25, 1564 (1953). (10) .Furman, N. H., Fenton, A. J., Jr., Ibad., 29, 1213 (1957). (11) Grube, G., Huberich, K., 2. Elektrochem. 29, 8 (1923). (12) Hobart, E. W., private communication, 1956.
(13) Holluta, J., J . physik. Chem. 115, 145 (1925). (14) Horn, H., Ph.D. thesis, Korthwestern University, Evanston, Ill., November 1954.
(lij-Ikegami, Hiroshi, J . Chem. SOC. Japan 52,173 (1949). (16) Kolthoff, I. M., Lingane, J. J., “Polarography,” Vol. I, p. 217, Interscience, New York, 1952. (17) Kolthoff, I. M., Watters, J., IND. ENG.CHEM.,ANAL.ED. 15, 8 (1943). (18) Lingane, J. J., Anson, F. C., Anal. Chim.Acta 16, 165 (1957). (19) Lingane, J. J., Karplus, R., IND. ENQ.CHEN.,ANAL.ED.18, 191 (1946). (20) Lingane, J. J., Kennedy, J. H., Anal. Chzm. Acta 15, 465 (1956). (21) Morse, H. N., Hopkins, A. J., Walker, M. S., Am. Chem. J . 18. 401 (1896): (22) Parsons, J. S., Seaman, W., Amick, R. M., ANAL.CHEM.27, 1784 (1955). (23) Reilley, C. S., Adams, R. A., Furman, N. H., Ibid., 24,1044 (1952). (24) Reilley, C. N., Cooke, W. D., Furman, N. H., Ibid., 23,1030 (1951). (25) Saito, K., Saito, N., J . Chem. SOC. J a v a n 55. 59 (1952). (26) ’Selim, ‘R. &.) Lingane, J. J., And. Chim. Acta 21,536 (1959). (27) Sem, M., 2. Elcktrochem. 21, 426 (1915). (28) Tutundzic, P. S., Mladenovic, S., Anal. Chim. Acta 12, 382, 390 (1955). (291 Ubbelohde. A. R. J. P.. J . Chem. ‘ SOC.1935., -1605. - (30) Watters, S., Kolthoff, I. M., J . Am. Chem. Soc. 70,2455 (1948). (31) Willard, H. H., Smith, G. M., IND. ENG.CHEM.,ANAL. ED. 11, 305 (1939).
RECEIVED for review December 21, 1959. Accepted March 18, 1960. Taken from the P h D . dissertation of A. J. Fenton, Jr., Princeton University, 1958.
EIectroIytic Determination of Microgram Quantities of Water in Paper R. G. ARMSTRONG, K. W. GARDINER,’ and F. W. A D A M Central Research and Engineering Division, Confinental Can An instrumental analytical procedure has been developed for the rapid and accurate determination of water in the 1 1 - to 200-7 range. The method is designed for use with moisture-bearing samples as small as 0.6-mg. total weight without elaborate classical microanalytical procedures. It has been applied to determination of the moisture content of paperboard disks ranging in diameter from ‘/la to 6/32 inch, which contained from less than 1 to over 6% moisture. The procedure involves use of a micro oven, in which the water contained in the small paper sample is vaporized and carried into a commercially available electrolytic hygrometer by a dry nitrogen stream. The method of calibration is described
752
ANALYTICAL CHEMISTRY
Co.,Inc.,
Chicago 20, 111.
and data show the precision and accuracy attained during a typical practical application.
A
ISSTRUMEKTAL analytical technique for the determination of water in the range of 11 to 200 y in small paper samples-e.g., 0.6 mg.-has been developed to provide a rapid and easily applied means for analyzing selected specific areas of sheets and formed paper products. The electrolytic measurement of water, on which this procedure is based, has been reported by Keidel ( 2 ) , Taylor (S), and Cole et al. ( 1 ) . This principle has been adapted for the treatment of paper samples. The water in the sample is
N
vaporized in a micro oven by a controlled heating program. The released water vapor is carried hy a stream of dry nitrogen into the electrolysis cell, where the water is absorbed by a thin continuous film of suitably anhydrous phosphoric acid located between two platinum helical electrodes. The absorbed mater is electrolyzed by impressing a potential on the electrodes. The signal from the electrolysis instrument is proportional to the current passed during the electrolysis and is recorded on a strip chart, while the area under the signal-time curve is simultaneously integrated by a digital 1 Present address, Consolidated Electrodynamics Corp., Pasadena, Calif.
read-out integrator. Thus, both the area under the plotted curve and the total integrated count are a measure of the total amount of water driven out of the paper sample. A complete determination can be achieved in as little as 10 minutes by this method, in contrast to the 1 t o 2 hours required by standard oven drying and weight-loss techniques. The electrolysis process is essentially specific for water; only certain basic materials, such as ammonia and some organic compounds, are possible interferences. Also, materials which can be absorbed by the phosphoric anhydride and undergo an oxidation or reduction under the test conditions will give erroneous results. Fortunately, such interference possibilities rarely are encountered in paper testing. The electrolytic determination of water has particular advantage over a weight-loss method for the treatment of samples which, upon heating, may lose other volatile components in addition to water. DESCRIPTION O F APPARATUS
The apparatus, shown schematically in Figure 1, consists of a tank of dry nitrogen, a magnesium perchlorate drying tube for the incoming nitrogen stream, a brass micro oven, a Consolidated Electrodynamics Corp. Model 26-301 Moisture Monitor, a flowmeter, a Minneapolis-Honeywell Model 51R10 Integrator, and a Model G l l A Varian recorder. The details of the micro oven employed to vaporize the moisture in the paper samples are shown in Figure 2. The micro oven heater is regulated by an automatic timer and an adjustable voltage control. The output signal from the Moisture Monitor is recorded by the 10-mv. Varian strip chart recorder and is simultaneously integrated by the Minneapolis-Honeywell Integrator, which has a range of 0 to 1000
counts per minute corresponding to a signal output range of 0 to 10 mv. il typical moisture determination is plotted in Figure 3. CALIBRATION TECHNIQUE
The system first is calibrated by introducing a measured volume of nitrogen which is equilibrated with water vapor a t a known temperature. The Jyater vapor equilibrium chamber used to obtain the calibration sample consists of a 10-ml. Mohr pipet immersed in a constant temperature bath controlled by a Bronwill Scientific constant temperature bath assembly. The bath temperature is accurately measured with an Anschutz thermometer and a Beckman differential thermometer. The bottom tip of the pipet is cut off and the top is attached to a three-way stopcock through which dry nitrogen is introduced to displace a measured volume of mater in the pipet. After equilibration, the gas volunie is forced into the gas introductory system of the electrolysis instrument by allowing water to rise in the pipet. The calibration sample is then carried into the electrolysis cell by a dry nitrogen stream a t 10-p.s.i. pressure and a f l o rate ~ of 25 ml. per minute. The water bath, which serves as both the constant temperature medium and the source of water vapor for the calibration sample, was maintained a t 24.00" =t 0.06" C., with a resulting water vapor pressure of 22.38 & 0.08 mm. of mercury. From the ideal gas law, the water concentration was calculated to be 21.8 y per 1 ml. of gas volume. ,--YICRO
y = 155.70
4- 11.1952
(1)
where 2 = micrograms and y = integrator count. The possible effect of variations in gas sample temperature on individual calibration readings was determined by analyzing known volumes of waterequilibrated nitrogen after they had passed through the micro oven heated to 72", 102', 1 3 5 O , and 176' C. For example, the average number of integrator counts obtained with replicate 5-ml. gas volumes a t these temperatures had a standard deviation of only 0.6%, indicating that variations in sample temperature, within reason, do not affect the accuracy of the method.
IO 9F
OVEN C A P
/
-THERMOCOUPLE
WELL
0
2
4
6
8
1
0
MINUTES
/\
LCARTRIDGE
Y
nitrogen samples ranging in volume from 0.5 to 8.5 ml., and shows the piecision of the data. The equation for the best straight-line fit of thcse data was calculated by the method of least squares to yield:
HE A T E R
Figure 3. Typical electrolysis curve for 149 y of water in 2-mg. paper sample
Figure 2. Micro oven for heating paper samples EXPERIMENTAL
I
Figure 1. Schematic diagram of instrumentation for determining water in paper by electrolysis A.
Oil-pumped nitrogen tank Magnesium perchlorate drying trap C. Micro oven D. Moisture Monitor E. Flowmeter F. Adjustable voltage control for micro oven heater G. Strip chart recorder H. Integrator
E.
The time required for 5-ml. volumes of oil-pumped tank nitrogen to attain water vapor pressure equilibrium was determined with the electrolysis system by plotting integrator counts us. exposure times ranging from 4.5 to 90 minutes (Figure 4). The plot shows that equilibration was attained after approximately 10 minutes of exposure. All subsequent calibration samples were conditioned for 40 minutes to ensure complete equilibration. The calibration curve shown in Figure 5 for quantities of water ranging from 11 to 184 y was made from a series of 49 points of integrator count us. micrograms of water contained in water-equilibrated
I n general, the handling procedure for an unknown sample involves the replicate sampling of the paper product using a circular hand punch. A sample is weighed and placed in the micro oven which has previously been flushed with dry nitrogen to eliminate contamination by moisture from the ambient air. The heating cycle is started and the determination is ended when the integrator automatically stops counting. The oven is cooled by a compressed air stream and the system is then ready for the next run. The total amount of water in the initial sample may not fall within the instrument range. I n such cases the paper product is again sampled, using a different size paper punch. VOL 32, NO. 7, JUNE 1960
753
Subsequent data will show that generally samples punched from commercial paper board show less per cent weight variation than the precision of the water determination. It is, therefore, possible to use a predetermined average paper sample weight for a series of per cent moisture determinations. To establish such experimental parameters as the maximum and minimum amounts of water that can be handled reliably by the process, the time and temperature limits of the heating cycle to be used, and the accuracy and precision of the method as compared to conventional weight loss procedures, replicate disks 0.25 inch in diameter of 15-point white tag board were conditioned a t approximately 43% relative humidity (R.H.) by storing them over a saturated solution of potassium carbonate a t 25" C. The moisture contents of three sets of samples were then determined as described in Table I, which also summarizes the results.
Table 1.
a
1
d U
l I
z soof zNO$ I I I
200
I
00
10
20
I
I
I
I
I
30
40
50
60
70
EXPOSURE
Figure 4.
99O c., 1 hr. 5.51 5.42 5.39 Av. 5.44 u 0.05
Time required to attain water vapor pressure equilibrium in
Electrolytic Method, Electrolysis Value Corres onding to 136" Heating Program
8.
5.41 5.66 5.44 5.50 5.37 5.31 5.41 5.49 0.03 0.14 The heating program involved the application of 85 volts to the micro oven heater for 3 l / 2 minutes, resulting in a cycle wherein the micro oven temperature reached 100' C. in 2112 minutes, rose to a maximum of 136' C., and immediately began to fall, but remained over 100' C. for at least 3 8 / 4 minutes before finally cooling to room temperature.
754 *
5 81 5 73 5.54 5.69 0.11
ANALYllCAL CHEMISTRY
5 ml.
of nitrogen for calibration of electrolytic moisture method
Weight-Loss Methods Vacuum oven Micro oven drying, heating program 93' C., 1 hr.. to give 136" C. 15 mm. Hg max. temp."
These data clearly show that the moisture value as determined by electrolysis is in good agreement with the value obtained by conventional weightloss techniques. Other micro oven heating programs resulting in temperatures higher than 136" C. gave similar results but were not in as good agreement as that achieved by employing the 136' C. cycle, and all subsequent electrolysis determinations were made with the 136" C. heating program. To determine Ihe electrolysis method's effective operating range and the precision achieved a t various moisture levels, 100 punched disk samples of 15point white tag board ranging from to 6/32 inch in diameter were conditioned in desiccators containing anhydrous calcium sulfate (approximately 0% R.H.), saturated potassium carbonate solution (approximately 43% R.H.), and saturated ammonium chloride solution (approximately 79% R.H.). The samples were obtained with four
90
TIME, MINUTES
Per Cent Water in '/c-lnch Disks of 15-Point White Tag Board Conditioned a t 43% R.H.
Air oven drying,
I
80
paper punches I / I ~ , a/32, '/a, and '/32 inch in diameter, providing a simple means of producing uniform samples. To report per cent moisture in paper by the electrolysis system, i t is necessary to know the initial sample weight. As a microbalance weighing of each sample for every determination would defeat the simplicity desired, a series of uniformally punched paper samples was weighed to determine whether or not the per cent weight variation of such samples would be within limits compatible with the established precision of the electrolytic method. A series of eight l/&xh-diameter disks punched out of a commercial 15-point white tag board, conditioned a t 43% R.H., was found to have an average weight of 2.652 mg. The standard deviation of these weighings was 0.083 mg., or 3.1%. As this variation is less than the variation in the actual moisture determinations as shown in column 3 in Table 11, it is feasible to make a series
of moisture determinations on the basis of a predetermined average weight for any given paper sample. The moisture contents of the 100 punched disk samples were determined simultaneously by the electrolytic method and by their weight loss after treatment in the micro oven. Table I1 includes the results obtained for the entire range of samples studied. DISCUSSION
The data of Table I1 show that a t the lower limit of the operating range there is a decrease in the precision of the determinations, as normally would be expected. Also, in each group of sample sizes the standard deviation of the results of determination of water by electrolysis is less than that by weight loss. Finally, a comparison of the previously mentioned weight variations of the punched disk samples (3.1%) to the data of column 3 in Table I1 shows that the sample weight variation is less than the precision of the water determined by electrolysis. I t is believed that this conclusion justifies the use of an average sample weight to report per cent moisture content without weighing every sample. The question arose in the course of this work as to whether or not a direct coulometric measurement could be made of the current passed through the electrolysis cell during a water determination. A successful measurement of this type would eliminate the indirect calibration procedure previously described. To test this possibility, the measuring circuit of the Consolidated Moisture Monitor was disconnected from the electrolysis cell and a %ohm precision re-
sistor was placed between the cell and the instrument ground. The voltage drop across the resistor was measured by a high impedance electrometer during a number of water determinations of paper samples conditioned to a known moisture content. The electrolysis current was plotted as a function of time and the area under the resulting curve was integrated n-ith a planimeter to give coulombs. The experimentally determined number of coulombs were compared with the theoretical number of coulombs corresponding to the known quantities of water in the samples.
Table ll. Comparison of Moisture Content for 15-Point White Tag Board as Determined by Electrolysis and Weight Loss to Establish Operating Range and Precision of Method
Mg; electrolysis loss electrolysis loss Not enough water, counts fall below lower range of calibration curve A '/I0 =/32 1.479 0.004 f O . O O 1 0.030 f 0 . 0 0 9 0.24 1 0 . 0 6 2.03 f 0 . 6 3 '/a 2.506 0.015 = t O . O O l 0.031f0.010 0.60 f 0 . 0 3 1.24 f 0 . 4 0 5/32 4.321 0.032 izO.002 0 . 0 3 1 f 0 . 0 1 2 0.74 f 0 . 0 3 0.71520.27 R 0.681 0.031 f0.003 0.033 fO.019 4.54 3ZO.36 4.85 1 2 . 5 4 . _ ~ 32; 1.528 o.oSi f o . 0 0 4 0.094520.0ii 5 27 h 0 . 0 9 6 . i 3 ~ 0 . 6 0 '/a 2.662 0.159 1 0 . 0 0 7 0.152 f 0 . 0 1 8 5.96 3ZO.08 5 . i 2 520.71 5/82 4.564 0.275 f 0 . 0 1 4 0.250 f 0 . 0 2 0 6.02 520.06 5 47 f 0 . 2 3 c '/le 0.738 0.048 f 0 . 0 0 4 0.063 f 0 . 0 0 6 6.44 3 ~ 0 . 2 5 8.55 1 0 . 7 1 1/52 1.560 0.130 f O . O 1 O 0.132 f 0 . 0 1 3 8.31 520.20 8.44 f 0 . 6 8 1/S 2.803 0.254f0.014 0.245 520,018 9.06 f 0 . 2 7 8.74 f 0 . 3 8 Too much water, counts fall beyond upper range of calibration curve 6/32 A. Samples conditioned over CaSO,, approximately C% R.H. B. Samples conditioned over saturated solution of K2COs,approximately 43% R.H. C. Samples conditioned over saturated solution of NH,Cl, approximately 79y0 R.H. ' / , A
This comparison shoived that for samples containing less than 70 y of water the experimental values were high. For samples containing from 70 to 160 y of water the experimental determinations were less than theoretical. To determine vhether or not the observed deviation from theoretical is a true characteristic of the electrolysis process, a Beckman electrolytic hygrometer mas substituted for the CEC Moisture Monitor. The measuring circuit of this instrument was disconnected and the current passed during a water determination was measured and plotted as described above. The experimentally determined number of coulombs were again compared with the theoretical number which would correspond to the known quantities of water in the samples. This comparison also showed that the
experimental values were high for samples having low water contentse.g., 80 y or less-but low for samples having higher water contents-e.g., 80 to 275 y. A plot of the known amount of introduced water us. the amount of water determined coulometrically shows a line whose slopeis not the same as the theoretical 100% coulometric line and which extrap-
::i 2000
a 0 I-
olation would not pass through the origin-i.e., the experimental line crosses the theoretical line. This is an apparently impossible situation for which we have no explanation a t the present time. The difference between the amount of introduced water and coulometrically determined water is mentioned primarily to point out that the 100% coulometric efficiency attained in a steady state was not attained in our batchwise operation. To eliminate the possibility that the low reading for samples of high moisture content was due to an incomplete absorption of water in the electrolysis cell, the effluent gas stream of the Beckman unit during an actual determination was passed directly into the Moisture Monitor; no water was detected by the second unit. Neither of the two instruments mas used under the specifications recommended by the manufacturers to yield the factory calibration. The reason for operating a t flow rates slower than those recommended is that a t the higher moisture contents of interest, the water concentration in the gas stream may initially exceed the range of the electrolysis instrument if the manufacturer's recommended flow rate is used. LITERATURE CITED
(1) Cole, L. G., Czuha, RI., Mosley, R. W., Sawyer, D. T., ANAL.CHEM.31,
td 2
w I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
2) Keidel, F. A., Ibid., 31, 2043 (1959). Taylor, E. S., Refrig. Eng. 64, 41 3)2048 (1956). RECEIVED for review September 14, 1959. Accepted March 31, 1960. Division of Analytical Chemistry, 135th Meeting, ACS, Boston, Mass., April 1959. VOL. 32, NO. 7, JUNE 1960
755