Ah' ACCURATE FORM OF COPPER COULOMETER BY HAROLD PFLAUM MATTHER'S A S D IAN WILLIAM WARK
There has not hitherto been described any accurate and easily manipulated form of coulometer, suitable for the measurement of currents greater than those that can be passed through the standard silver coulometer-that is, greater than one ampere. Even where i t is applicable, the silver coulometer is not easily manipulated. Consequently a more rapid and yet reliable form of coulometer would be welcome, especially should it be capable of dealing with currents greater than one ampere. Marshall' has described a copper coulometer for which he claims a high order of accuracy for currents over ten amperes. His claims are not, however, supported by adequate experimental work, and his method (the best available a t the time) of checking the coulometer against a standard Weston ammeter is, perhaps, not the most satisfactory. While engaged on some electrochemical work extending over a period of three years, we have had the opportunity of studying the practical performance of the Marshall type copper coulometer. The work performed had to be kept subservient to the main problem under investigation and therefore no ordered study along theoretical lines was possible. At the outset it was hoped that the instrument, as described by Marshall, would prove sufficiently accurate for our purposes. It soon became apparent, however, that though the instrument was satisfactory for the high currents for which it was designed, it failed for lower ones. It was retained in circuit in the hope that when the causes of failure had been ascertained and eliminated it would be possible to rely exclusively on it; it is satisfactory to be able to report that this position has now been reached. In all, some three hundred tests have been made on its accuracy under various conditions; not least important are a number of tests over a period of twenty-four hours. Under the conditions recommended later, current can be measured with a maximum error of one or two parts per thousand. The best results have been obtained with the simplest running conditions. Thus it is unnecessary generally to rotate the cathode; no depolariser is necessary in the electrolyte nor is any accurate temperature control necessary. It is strange that Oettel's artifice of adding alcohol t o the coulometer electrolyte has been followed so closely by later experimenters, seeing that it now appears unnecessary except in certain well-defined cases. It will be advisable to outline the method used in checking the coulometer. In the same circuit was placed a standard I 'IO ohm resistance,* the volt drop Trans. Faraday SOC.,21, 297 (1925). * This was certified by the makers, Leeds and Korthrup, to be correct to within 1/40%,. I t was checked in Melbourne against the University Sub-Standard of the Natural Philosophy Dept. and found to differ by 0.06Yc7,.The makers' value was accepted.
2346
HAROLD PFLAUM MATTHEWS A N D IAN WILLIAM WARK
across which was measured at frequent intervals by a calibrated potentiometer. The duration of the passage of current was estimated by means of a calibrated stopwatch. In tests of one hour's duration, at least fifteen current measurements were recorded. In the longer tests the period between readings was increased to a maximum of fifteen minutes. If the current varied, especially a t the beginning of the test, the frequency of measurement was increased. By drawing current from the battery through an auxiliary circuit for a few minutes prior to the test, the familiar irregularities on closing circuit were largely eliminated. In no case was the variation in current between readings sufficient to lead to any significant error. The current determinations by these means are certainly correct to within 0.17~ and probably to within o.oj7c. Description of Coulometer Construction: The anode was a sheet of 1,'16 inch electrolytic copper bent to form a cylinder 12 cm. diameter by 1 2 cm. high, with suitable lugs. It was thickened by electrodeposition of copper to the extent of over zoo ampere hours, and re-plated whenever necessary. It was hung by the lugs mentioned above in a 3-litre beaker, which formed the cell; from z to 2 . j litres of electrolyte were required. The cathode was a cylinder of platinum gauze of 2 . j cm. diameter by j cms. high. I t could be rotated about a vertical axis, also of platinum. Electrolyte: At first Oettel's solution was used. This consists of 150 grms copper sulphate crystals 50 grms sulphuric acid
50 grms alcohol I litre water
Subsequently variations were introduced. Dissolved alcohol was replaced, where specified, by the passag@ r hydrogen during the test and later both mere dispensed with. The acidity was reduced by one half in some tests and even to zero during two tests. The effects of t'hese variations are discussed later. Operation: The cathode was freed from copper by immersion in concentrated nitric acid. It was washed in distilled water, dried by heating in a blowpipe flame (which treatment detects any undissolved copper) and cooled in a desiccator before weighing. It was placed in position with the cell on open circuit and, where necessary, was set in rotation before the circuit was closed. Approximately one minute elapsed after the circuit was broken before the cathode could be removed from the cell. It was then washed in distilled water and finally in alcohol, the adhering film of which was ignited to dry the deposit, which was cooled in a desiccator before weighing. Experimental Results The first series of tests was undertaken with an electrolyte of the composition of Oettel's solution; no record of the temperature is available as the
AN ACCURATE FORM OF COPPER COULOMETER
2347
usual procedure of running a t room temperature was followed. The cathode was rotated a t 200-300 r.p.m. (The higher speed, 1000r.p.m., recommended by Marshall caused a vortex to penetrate the hollow cylinder of the electrode, offering facilities for the aeration of the electrolyte). Currents were lower than those for which Marshall designed the instrument, and it is to be emphasized that the errors reported under these conditions do not in any way condemn the instrument for higher currents.
TABLE I Tests in Oettel’s Solution, Rotating Cathode Test No.
Time Hours
Current
amp
Coulometer Error
+o.3%
I
I
I
2
I
I
Xi1
3
I
I
+o. I
4 6
I
I
+0.6
I
I
7 8
I
I
+0.2 Xi1
I
I
+0.2
9
0.8
1.5
+O.I
I2
I
I
f0.o
I3
I
I
I4
I.0j
I
$0.6 +0.8
Nil
16
I
I
I7
I
I
-1.0
5
I
0.5
-1.5
I1
I
0.2j
-1.5
The low values indicated by the coulometer in Nos. j and 1 1 are to be attributed to the low current. With the exception of S o . 17, all other results are high. Some recent results of Fuseya and Muratal indicate that the alcohol may have been responsible for these high results. These authors have shown that citric and other hydroxy-acids may be deposited with copper up to the extent of 6 5 of the weight of the copper. It was suspected that the hydroxyl group of alcohol might be responsible for its co-deposition with copper also. Accordingly, in another series of tests, no alcohol was added but as the literature is insistent on the use of some depolariser, hydrogen was bubbled through the solution instead. Between 4 and 7 litres of hydrogen per hour were used. I
Trans. Am. Electrochem. Soc., 50, 235 (1926).
HAROLD PFLAUM MATTHEWS AND I A S WILLIAM WARK
2348
TABLE I1 Preliminary Tests in Alcohol-free Solution Hydrogen bubbled. Cathode rotated Test S o .
Time
Current
Coulometer Error
20
I
hour
I
-0.2
21
I
'(
I
-0.2
29 30 3'
I
Ii
I
-0. I
I
'(
I
-0. I
I
'I
22
0 . 2 j
23
0.2;
25
0.25
26
0.25
27
0.25
28
0.2;
I
-0.
4 4 4 4 4 4
-0.1
I
Si1 +0.2
s0.35
Nil +0.2
Although high errors had been eliminated a t currents of I ampere, with higher currents they were still sometimes encountered. Further tests were therefore undertaken with currents of 4 amperes but the errors now disappeared.
TABLEI11
i
Alcohol absent
summary of Tests a t 4 Amperes Hvdrogen buhhled Cathode rotated
Test No. 22
23 25
26 27
28 33 38 39 40
93 94
95 96 97 98 I10 I11
112
Time 0.2j
Current
hrs 4 amps.
11
I1
11
IL
11
1L
11
11
Coulometer Error -0.
Remarks
1yc
Si1 +0.2
+O.
03
Si1
1L
/I
11
Ll
LL
I1
Ll
iL
+o.
LI
(1
fo. I
+0.2
Si1 Nil I
LL
(1
IL
II
LL
LL
I1
LL
(1
I1
IL
LL
-0.16
Ll
Ll
+ o . 06
I1
LL
Si1
iL
li
-0.04
-0.15 -0. I3 -0.13
-0.11
-0. I O
Fresh solution tests were conducted a t about z j"C A. fresh solution of one half normal acidity now used
A S ACCURATE FORM O F COPPER COULOMETER
2349
The acidity of the solution was reduced a t Test No. I I O because it was thought that, if re-solution of copper was responsible for low results, they might thereby be partly eliminated. Though the three different solut'ions seemed to give slightly different results,' it was concluded that at this current density the Marshall coulometer is a trustworthy instrument. The instrument was then tested further with lower currents. After giving fairly consistent results for twenty test,s, the performance became erratic and errors up to 2 C ; were not uncommon. It was some considerable time before the cause of the variations was discovered; meantime, fifty tests were made in which the error was sometimes high, sometimes low. The period of good results coincided wit'h the winter season, and ultimately the variat'ions were shown to be due to fluctuations in room temperature. I t was ascertained later that the actual temperature was the irhportant factor and not fluctuations during or immediately prior to the test. The results during this period are set out in Table IT.
TABLE IV /Cathode rotated
Summary of Early Results a t
hlonth
Sept. 1 9 2 7 OCt. I 9 2 7 Xov. 1 9 2 7 Dee. (First half)
Sumher of Tests
I
Mean Error
Ampere Mean Max. Daily Temp. of Weather Bureau
hlaximum Error
17
+ o . I:%
18' C.
14
-0.4jq
21°
15
-I
6y'
2
So
-2.1yc
-0
7%
2 -to
-0.9%
8
j (twice) -1.8yc
+o
During September almost all errors were on the high side; after September all were on the low side. As it now appeared probable that temperature had an influence on the performance of the coulometer, direct tests were undertaken to test this theory, the cell being arranged to function as a thermostat. It became desirable to extend the testing to include other variables and work has now to be reported covering the effects of (a) Rotation of cathode or stirring of solution (b) Temperature (c) Bubbling of hydrogen and, over more restricted ranges, of (d) Acidity of solution (e) Filtering the solution Later work has failed to reveal any differences between different solutions provided that they be given a short preliminary electrolysis to remove impurities.
2350
HAROLD PFLAZTM MATTHEWS A S D IAN WILLIAM WARK
(f) Duration of deposition up to 24 hours (9) Current density variations. (a), (b) and (c) were considered together, as illustrated in Table T'.
TABLE F' Current I Ampere Standard Marshall Coulometer (Tests from f to 1 2 hours duration) Hydrogen buhhled Cathode Cathode rotated stationary
Hydrogen not bubhled Cathode Cathode rotated stationary
Room Temp. Low Results (below 30') (IS tests)
Good Results ( 5 tests)
High Results (7 tests)
Good results ( 4 0 tests)
Higher Temp. High results (above 30') (6 tests)
Slightly high results ( 5 tests)
Very high results (7 tests)
High results (8 tests)
This compilation represents average results only, full experimental details are given later. The following deductions may be drawn:-
E$ect of Rotation: With this current, rotation leads to incorrect results; rotation in conjunction with high temperatures leads a t times to very high errors; these can be reduced by passing hydrogen through the cell. With a stationary cathode, variations, though in the same direction, become much smaller. At higher current densities, rotation might become essential, but it has already been shown that rotation is then no longer sufficiently detrimental to vitiate results. The range of the coulometer would therefore be increased if facilities were provided to rotate the cathode. E$ect of Temperature: Below 20' results are generally good, but if the cathode be rotated, errors mount with the temperature until, above 30°, they reach 5yG. TT ith stationary cathode, results are reliable at room temperature, even up to 30°C. Room temperature seldom exceeds t'his value in Melbourne. Efect of Hydrogen: JF-ith stationary cathode, hydrogen is unnecessary at room temperature. If the cathode be rotated or the temperature be raised hydrogen exerts a corrective influence. In extending the use of the instrument to higher current densities a t intermediate values-where rotation has just become necessary-it would be advisable to bubble hydrogen past the cathode. Table VI-XI11 show in detail the effects of these factors:
AN ACCURATE FORM OF COPPER COULOMETER
Y
>
h
$ * a -
N
0
I
HAROLD PFLAUM MATTHEWS AND IAX WILLIAM WARK
2352
TABLE VI1 Hydrogen bubbled, Cathode not rotated, Room Temperature (Current I Ampere) Test So.
Temp. Hours Error %
212
221
232
233
234
16
I4
14
I
I
14 5
5
I4 5
Xi1
Xi1
Iiil
-0.2
-0.
I
TABLE VI11 Hydrogen not bubbled, Cathode rotated, Room Temperature (Current I Ampere) Test No.
173'
175
174
Temp. 24 24 23 1 1 I Hours 2 s s ErroryG (-2.0) -0.6 -0 2 * Fresh solution, not previously electrolysed
176
207
209
211
I9
I7
17
I7
-1
si1
1
1 +I
I
+o
2
I
8
+I
4
TABLE IX Hydrogen bubbled, Cathode rotated, Temperature above 30' (Current I Ampere) Test No.
1.53
160
161
170
171
I72
Temp. Hours Error 2
36
d3
33
33
33
33
I
6
4
I
I
-0.8
3
.+
$0
+0.3
I
+0.4
+1.3
7
+I
TABLE X Hydrogen bubbled, Cathode not rotated, Temperature above 30' (Current I Ampere) Test No.
Temp. Hours Error Yc
235
236
40-26
28-36
I
238
239
53-36
34--30 5
5
5
-0. I
+O.IS
5 -0.
237
34-37
5
+o. I
+o.z
TABLE XI Hydrogen not bubbled, Cathode rotated, Temperature above 30' (Current I Ampere) Test No.
189
190
205
244
258*
32
35
30
241 31
243
Temp. Hours Error %
42
32
I
j
I
5
I
I
34 6
+2.6
* Solution stirred
$2.1
by hand every
+4.9 $0. j +2.7 +3.4 minutes instead of cathode being rotated.
f4.4 Ij
AN ACCURATE FORM O F COPPER COCLOMETER
2353
+
N o - w o i;"
-9
vi
z
a -
N
i
0
g m - - 0 N
-
--
9
+ 4
0
I
-+ vi
I
vi
I
g;oc N
-
Y
0
HAROLD PFLAUM MATTHEWS AND I A N WILLIAM WARK
2354
TABLE. XI1 Hydrogen not bubbled, Cathode not rotated, Temperature above 30' (Current I Ampere) Test No
186 39
Temp. Hours Error (5:
187
188
191
192
193
203
204
35
3.5
36
37
35
33
I
I
35 3
I
I
I
> -0
15
+O
15
+O
1.5
KII
+O
65
+O
4
I
+O
I
+O
I
Acidity of Solution: Even when no free acid was present-in Tests Nos. 261 and 262 of Table XIII-results were quite satisfactory. Previously no
difference was observed when the acidity was reduced by one half. Evidently the acidity has little if any influence on results. As almost a minute elapses before the cathode can be removed from the electrolyte after the circuit is broken, it was thought that there might be introduced an error due to slow re-solution of copper in the acid electrolyte. Therefore the rate of re-solution was measured for several deposits; the cathode was rotated during each estimation of re-solution rate.
TABLE XIV Rate of Re-solution on Open Circuit Test N o
Rate of re-solution
Coulomc t?r Error
o 6 mg min 0.8
-3.0
93 '37
0 7
-0.
0.8
138
0 9 0 7
-1.7 -1.6
87 91
I39
-4.55 I
-1.2
There is no connection between error and rate of re-solution. With the cathode stationary the re-solution loss would be much less than in these tests. Even with rotation the error due to re-solution need not amount to more than 0.057~ for a deposit corresponding to I ampere hour. Filtration: Filtration of the electrolyte made no difference to results. Owing to the formation of an anode slime, filtration is necessary from time to time, especially if the cathode be rotated, (Filtration does influence the form of the deposit', however.) Dicratz'on: From Table SI11 it follows that duration, up to 2 4 hours, has no effect on the reliability of the coulometer. Current Density: Current density is difficult to estimate for a gauze cathode. We have therefore reported our results in terms of the current passing and the size of the cathode. Results for four and one amperes respectively have been reported above. Results for lower currents are set out in the following graph-
A N ACCURATE FORM OF COPPPER COULOMETER
02
04
08
06
l0
2355
/'a
FIG.I
Procedure now recommended: The conditions of Table XI11 are recommended for currents of from one to three amperes and a cathode of the specified size. The cathode should not be rotated, there is no need to pass hydrogen, though it does no harm. (Tables VI1 and X) and results are equally satisfactory for all room temperatures between 13 and 30'. The mean error of 41 determinations under these conditions was zero. The maximum error of 0 . 2 7 ~ occurred only on three occasions and on two of these, the tests having extended over 2 4 hours, the comparison method of current measurement may have been slightly in error, as the variation between readings was greater than usual. For higher currents it is advisable to rotate the cathode, but in this case hydrogen should be bubbled past the cathode. High temperatures are to be avoided and in hot weather cooling would perhaps be necessary. Marshall's results indicate that the coulometer would be applicable for currents of r j amperes, but further testing is desirable above 4 amperes. For lower currents the addition of alcohol would increase the range of the instrument (See Fig. I ) . X more satisfactory procedure would be to decrease proportionally the size of the cathode and cell. I t is to be emphasised that the results reported above apply only for a cell of the given size; variations are to be expected with variation in the cell dimensions. Explanation of Results: Low results might be occasioned either (i) by liberation of hydrogen at the cathode, or (ii) by an increase in the cuprous ion concentration a t the cathode, caused by the action, Cu'+ e = Cu'
+
2356
HAROLD PFLAUM MATTHEWS A N D IAN WILLIAM WARK
High results might be occasioned by precipitation of cuprous ions. This implies either (i) a decrease in the Cu+/Cu++ ratio in solution (which appears unlikely), or (ii) Cu+ ions forming a t the anode in compensating amounts. The proper working of the coulometer demands firstly that the Cu+,’Cu++ equilibrium should not be upset by electrolysis, and secondly that only cupric ions should be dissolved at the anode and precipitated on the cathode. A complete study of the working of the coulometer would necessitate weighing of both anode and cathode and carefully following the changes in composition of the electrolyte. The anode sludge should also be weighed and analysed. For the reasons cited earlier we were unable to do this. summary
The Marshall copper coulometer has been subjected to more than three hundred tests over a wide range of current density. For high (see text) current density it is satisfactory as described by Marshall. For lower current densities certain specified modifications in procedure are necessary to keep the error below 0.2%. Temperatures higher than 30’ cannot be tolerated and it is unwise to rotate the cathode with low current densities; but the errors resulting from either of these causes may be considerably reduced by bubbling hydrogen through the electrolyte. Alcohol additions are not generally necessary, though they are helpful in extending the range of the instrument on the low current density side. Comparatively large variations in the acidity and lengthy periods of electrolysis-up to 2 4 hours in fact-are without effect on the performance of the coulometer. We are much indebted to Sir David Masson, under whose direction this work was carried out, for his stimulating interest and helpful advice. To Dr. A. L. Marshall our thanks are due for his help in obtaining for us a cathode similar to that used by him in his earlier work. We have also to thank Professor E. J. Hartung for having made available the University laboratory for this research, which was carried out while we were in the service of the Electrolytic Zinc Company of Australasia, Ltd. Department of C h e m i s t ~ y , The Cniversity, ;~dbOUTnE.
