ANALYTICAL CHEMISTRY
264 probably with little change in the sol portion.
molecular weight than the untreated sol, probably because milling breaks down both sol and gel. Two separate samples taken from the same piece of government specification GR-S (Table I, standard GR-S, A and B ) were carried through the complete osmotic pressure measurement procedure to establish the reproducibility of results. The last two columns of Table I show the precision obtained. One determination gave a molecular weight of 115,000 and the second yielded a value of 116,000. While these results are undoubtedly fortuitous, in view of the many possible sources of error, they indicate that good precision can be expected. SUMMARY
ACKNOWLEDGMENT
In contrast to
this effect, natural rubber milled to the same extent has a lower
AND CONCLUSIONS
Osmotic pressure apparatas and technique have been developed which can be used to determine the number average molecular weights of polymers in the range 50,000 to 500,000. Simplicity of construction and over-all convenience in use, combined with reliability of results, are the outstanding advantages. Reproducibility of the results obtained (of the order of 5 % ) is best judged from the osmotic rise values of the four different measurements for each of the four concentrations given in Table I. The molecular weights of German Buna S-3 and GR-S treated in various ways are reported: Buna S-3 sol, 166,000; heatsoftened Buna 5-3, 98,200; milled whole Buna S-3, 224,000; GR-S, A 115,000; and B 116,000.
The authors wish to thank F. R. Stavely for his continued interest and suggestions throughout this work. Grateful acknowledgment is made to F. S. Grover and L. 0. Stauffer of the Laboratory Machine Shop for fabricating the osmometers Appreciation is expressed to the Firestone Tire and Rubber C‘o for permission to publish this work. LITERATURE CITED
(1) Baker, Walker, and Pape, GR-16; PB9687,Office of Technlral Services, U. S. Department of Commerce (June 29, 1944). (2) Boissonnas and Meyer, Helv. C h i m . Acta, 20, 783 (1937). (3) Carter and Record, J . Chem. Soc., 1939,660. (4) Cragg, J . Colloid Sci., 1, 261 (1946). (5) French and Ewart, IND. ENG.CHEM., ANAL.ED.,19, 165 (1947. (6) Fuller, Bell S y s t e m Tech. J., 25,374 (1946). (7) Fuoss and Mead, J . Phys. Chem., 47,59 (1943).
(8) Herzog and Spurlin, 2. p h y s i k . Chem., A, Bodenstein-Festhann 239 (1931). (9) Montonna and Jilk, J . Phys. Chem., 45,1374 (1941). (10) Schulz, 2. physik Chem., A176, 317 (1936). (11) Sturtevant, “Weissberger’s Physical Method8 of Organic Chemistry”, Vol. I, p. 327, New York, Interscience Puh. lishers, 1945. (12) Wagner, IND. E m . C H E M . , ANAL.ED., 16, 520 (1944). (13) Wall, Banes, and Sands, J . Am. Chem. Soc., 68, 1429 (1946) (14) Weidlein, Chem. Eng. N e w s , 24,771 (1946). PRESENTED before the Division of Rubber Chemistry a t the 110th Meeting of the AMERICAXCHEWCAL SOCIETY,Chicago, Ill. Investigation carried out under the sponsorship of the Office of Rubber Reserve, Reconstructlor Finance Corporation, in connection with the Government Synthetic Rubbe* Program
Apparatus for Rapid Conductometric Titrations Determination of Sulfate d
d
LLOYD J. ANDERSON1 AND ROGER R. REVELLE*, Scripps Institution of Oceanography, La Jolla, Calif. This paper describes the development and construction of electrical apparatus suitable for conductometric titrations, and presents a method of using the apparatus for macro- and microdeterminations of sulfate. When the precipitation is controlled by seeding, amounts of sulfate as small as 1 mg. can be determined to *l%even in the presence of more than fifty times as much chloride.
N
UMEROUS attempts have been made to develop simple and rapid procedures for determining dissolved sulfate. Simple and direct titration procedures with barium solutions have been unfeasible because of the lack of reliable color-changing end-point indicators. As a result, a great many indirect methods have been devised in which the unknown sulfate is precipitated by a suitable reagent and a n equivalent amount of some other ion, easier to determine, is liberated. The methods of Muller ( 7 ) , Hinman (S),and Webb ( I d ) , though usually less time-consuming than the conventional gravimetric procedure, leave something to be desired in the way of reliability and accuracy. One color indicator for direct barium titration has been used by Robertson and Webb (9), who state, however, that “the simplicity of the method is to some extent outweighed by the capricious behavior of the indicator”. Several electrometric methods have been developed in which the end point is recognized by the aid of potentiometric or conductometric indicators. Reversible sulfate electrodes are not suitable for routine work because they require extreme care, in both construction and use, in order to prevent “oxygen poison1 Present 9
address, Present address,
U. s. Navy Electronics Laboratory, San Diego, Calif. U. S. Navy Bureau of Ships, Washington, D. C.
ing”. Bimetallic electrodes (IS) are often suitable for pure solutions but become insensitive in mixtures. Dutoit ( 2 ) was the first to apply the conductometric titration technique to sulfate determination and numerous others have reported using the method. Kolthoff and Kameda ( 5 ) investigated the reliability of the conductometric titration for sulfate and concluded that the procedure is too inaccurate for most analytical work, but that coprecipitation of sulfate, the usual nemesis of sulfate determinations, does not occur and hence cannot account for their low results. They also found that the mere presence of the barium sulfate precipitate greatly affected the conductance changes meaeured. during the titration. I n the present method, the precipitation is seeded with a small amount of pure barium sulfate. The seed crystals were precipitated from a solution of C.P. barium hydroxide by adding a slight excess of C.P. sulfuric acid and allowing the washed precipitate to stand for several weeks. These factors tended to produce comparatively large seed crystals. With such crystals as precipitation nuclei the effects of adsorption should be minimized. One would also expect seeding to produce precipitates that were much more uniform from one titration to the next. Perhaps the more uniform results reported herein are due in large part t o
V O L U M E 19, NO. 4, A P R I L 1 9 4 7
265
used for such purposes and in this case have proved very effective in keeping the power delivered to the bridge constant within narrow limits. I n order to prevent direct contact between oscillator and bridge, the alternating current is supplied through a n impedancematching output transformer. It was felt desirable to mount the oscillator and power packonachassisseparate from the rest of the circuit in order t o reduce further the possibility of inducing random voltages into the bridge, and to decrease the amount of equipment needed on the titration bench itself. The power ie fed from the oscillator to the bridge through a flexible cable fitted with jacks for convenience in disconnecting the two unite when desired. The bridge is a continuous deflection type rather than the conventional slide-wire and null indicator. This type of bridge requires no manipulation during a titration, since conductance changes are read directly from the meter. Such operation eliminates the necessity of a calibrated precision slide-wire and aids greatly in reducing the sine and ex4PPARATU S pense of the apparatus. I n this case a simple carbon-surfaced volume control was found to be perfectly satisfactory as the The electrical circuit can be divided into three more or less disvariable arm of the bridge. This unit, Rlj, is compact enough Tinct units: source of alternating current, Wheatstone bridge,. to allow it and the other components of the bridge circuit, except and vacuum tube voltmeter. the conductance cells themselves, to be mounted inside the chassis containing the voltmeter. Thus all the electrical controls requiring adjustment during a series of titrations are located on one The alternating current wurce IO a conventional type of repanel on the front of the compact bridge voltmeter unit. 3istance-tuned oscillator driving a 6F6 power tube, and supplies In the circuit diagram, two conductance cells are indicated. 20 volts of 1000-cycle alternating current to a 500-ohm bridge. Titrations are carried out in only one of them. The function of i n inductance-type oscillator is avoided to reduce induction of the other is merely to balance out incidental conductance changes jpurious currents in the bridge. I n the circuit diagram (Figure due to temperature drift and any other changes not directly con0 the oscillator and its power pack are shown a t the top. The nected with the titration itself. Theoretically, complete comjutput of the power pack is stabilized against line voltage changes pensation will result only if both cells have the same volume and ov gas-discharge voltage-regulator tubes. These are commonly conductance, since only under these conditions would the temperature rise be identical in both cells. Fortunately this condition can be easily OSCS L L A T O R AND POWER PACK attained: First, the variable bridge resistor, R16, is set a t its electrical center and thedial is marked for future settings. Then one cell is filled with the solution to be titrated and the second cell is q l e d with a slightly smaller volume of a similar solution. Kow, concentrated sodium chloride or distilled water is added until the voltmeter indicates that the bridge has reached balance,
aeeding the titration mixtures. From the standpoint of convenience, the greatest advantage in seeding lies in accelerating the precipitation rate and thus reducing the titration time from 30 to 5 minutes per sample. The chief limitation to conductometric titration of sulfate lies In the concentrations of anions other than sulfate in the solution. When these are large, it is necessary to use a concentrated barium Jolution and microburet for titration in order to maintain a welliefined “break” in the titration curve of meter reading versus volume of reagent added. In the present method, the breaks in -he curves were sufficiently ne11 defined J o obtain an accuracy of -1% even in solutions where the ratio of chloride to sulfate was high as 66 to 1.
VR30-105
. Figure 1. RI Ra
.
B R I D G E
I.
VOLTMETER
Schematic Diagram of Electrical Circuit
15,000 o h m s 2,000 o h m s Ri. 10,000 ohms Two 6-watt 120-volt lamps in &. aerie. (ca. 1000 ohma) Rr. 100,000 o h m s &a. 500,000 o h m s RT. 500,000 o h m s Ra. 1,000 o h m s R,. 5,000 o h m s Ria. 20,000 ohm. &I. 500,000 o h m s Rll. 660 o h m s Ria. 500 o h m s Rid, RN. 100 ohms Rn. 1,000 ohma &r. 3,000,000 o h m s Rit. 25 ohms Ra. 500,000 o h m s Rm. 250,000 ohms RZI. 10 ohma Rw. 10,000 ohms
.
AND
VR30-150
0.01 mfd. 600 volt c1,cz. 1.0 mfd. 680 volt cz. 0.5 m i d . 600 volt C4,C;. C6,CS. 10.0 mfd. 25 volt electrolytic
c6.
0.01 mfd. 600volt electrolytic 8.0 mfd. 600 volt electrolytic 0.20 mfd. 608 volt electrolytic Cll. 0.10 mfd. 600 volt electrolytic ClZ. 0.02 mfd. 600 volt electrolytic CII. Two flashlight cells i n series vi T y o heavy duty 1.5-volt cells V? in series T w o heavy duty 45-volt cells Vi. i n series 120 volt primary, 5.3 volt, 6.3 TI. volt, and 660 volt c.t. accondary Tz7000-ohm primary, 500-ohm secondary Microammeter, 0-500 microamperes. 130-ohm coil; Triplett Model 521
G,G.
..
If a series of titrations is to be made on solutions of nearly the same conductivity, it will not be necessary to readjust the resistance of the control cell each time. Practice indicates that differences even as high as 25% do not cause very serious drifting of the bridge balance. It is well, however, to start a series with the conductance? nearly equal. The vacuum tube voltmeter, used here to indicate conductance changes, is shown schematically a t the lower right of Figure 1. It is a common type of capacity-coupled voltmeter, but has numerous features adapting it to the present purpose. Most important of these, from the standpoint of convenience and increased sensitivity, is the so-called I‘D-battery” circuit composed of Vz, Sz,and R3Y. The function of this circuit is to by-pass most of the plate current of the final tube but to allow the fluctuations to go virtually undiminished through the meter. This arrangement permits the use of a more sensitive meter, thereby increasing the effective sensitivity of the instrument. Actually, this device is not a nen nor even a recent development. Kinney and Garman ( 4 ) ,Morton (6), and others (8,13)have reported using it in electrochemical instruments and its use is also commonplace in purely electrical apparatus. In the present circuit, the use of the D-battery increased the effective sensitivit! of the voltmeter eightfold.
A further advantage of the D-battery in a conductance voltmeter is that it affords an independent control for placing the meter needle a t any desired point on the scale. This is particularly useful in spacing a series of titration curves on a plotting sheet.
266
ANALYTICAL CHEMISTRY and amounts to only a small percentage of the total change to be measured. Furthermore, although it is not a novel one, the two-cell method of compensating for temperature and other incidental effects has proved effective with present apparatus. When the cell conductances are properly equalized and the apparatus is set in operation, the meter readings remain constant, except for the small voltmeter drift. When changes of conductance are set up in one of the cells, the meter responds a t the rate of 9-mm. deflection for each 1% conductance change. This makes possible the detection of changes of about 0.0270. The conductance ceds constitute the remainder of the electrical equipment. Two sizes were used: a pair of 500-ml. cells for large samples and a 100-ml. pair for micro work. They were made from ordinary reagent bottles and each cell had a pair of pure silver electrodes sealed on opposite sides of the cell near the bottom. In D the 100-ml. pair the electrodes were 6 mm. in diameter and in the 500-mi. pair were 18 mm. in diameter. These dimensions were found optimum to obtain maximum oscillator output and without undue polarization difficulties. The electrodes were sealed in the cell with De Khotinsky cement and connecting wires were brought out through small holes drilled through the cell wall.
:Figure 2. Over-all Sensitivity Curve of Electrical Circuit Presumably the voltmeter sensitivity could be further increased by using more stages of amplification. The practical limit of such a procedure is the point where random fluctuations begin to register on the meter. The sensitivity of the present voltmeter is almost a t this limit. Meter fluctuations occur, for example, if the stirring of the cells is not sufficiently smooth or if large bubbles or bits of precipitate approach the electrode surfaces too closely. With reasonable care, however, the meter action is smooth and steady. The actual sensitivity is about 2500 microamperes per volt or, with the meter selected, about 600 mm. per volt. Figure 2 shows the over-all response of the instrument to changes in conductivity in one of the cells. It is seen that the response is linear until the readings are yithin 150 microamperes of the null point. Inside this limit the sensitivity falls rapidly, becoming zero a t balance, then rising again as the balance point is passed. The reason for this behavior lies in the voltmeter itself. The microammeter responds to differences in the slopes of adjacent segments of the S-shaped characteristic grid voltageplate current curve. As the input voltage from the bridge is reduced near balance, these adjacent segments encompass a smaller and smaller total portion of the curve and hence the difference in their slopes approaches zero and the voltmeter becomes more and more insensitive. The response of the voltmeter could undoubtedly be made virtually linear if it were necessary, but in this case the restriction imposed offers no serious disadvantage in conductometric titration work. In actual practice, the electrical circuit as described above is very satisfactory. As in most other vacuum tube instruments, a certain amount of drift in meter readings is noted just after the apparatus is turned on. K i t h fresh batteries, this amounts to about 4 microamperes per minute a t first and after 15 minutes is only a tenth of this. Such a drift amounts to only 2 or 3 microamperes for an entire titration, while the total change to be measured is of the order of 100 to 500 microamperes. As the plate batteries age, the drift increases. For example, after 6 months of normal battery use it is about 80 microamperes per minute initially and drops to 2 microamperes per minute after half an hour. Even this latter drift is not troublesome, since it is uniform for a given titration
One pair of cells was made using platimum electrodes instead of silver. It was found that, on passing 60-cycle current through these cells, gas bubbles appeared on the surface of the platinum when only a few volts were applied, whereas with silver electrodes 20 volts could be applied with no gas liberation or conductance instability of any kind. Although the frequency was subsequently raised t o 1000 cycles, silver electrodes were still favored because of their reduced tendency to cause electrolysis of the solutions. The only other part of the apparatus needing special mention is the microburet, which was made according to a description by Scholander (10) and has proved very precise and reliable. In principle, the buret operates by the displacement of mercury from a reservoir by the spindle of a micrometer caliper. This, in turn displaces the reagent from the buret into the titration cell. When calibrated for irregularities in the diameter of the spindle, the buret was found to deliver to *0.0003 ml. It has a total volume of about 0.69 ml. TITRATION PROCEDURE
With S1 open and SBclosed, the oscillator and voltmeter are set in operation by plugging the oscillator cord into a 60-cycle 115-volt outlet and closing S2 on the voltmeter chassis. Pr’ext, the sample to be titrated is introduced into the cell and diluted to a convenient level. The conductance of the control cell is adjusted roughly t o that of the titration cell by fillin it with nearly the same volume of a similar solution. Switch if is closed allowing current to flow through the cells, and then the knob & Rzzis rotated slowly to the right until the meter reading reaches a maximum and decreases to about 30 microamperes. Now the meter is thrown on high sensitivity by opening 8,. The null point is located on the meter by adjusting the bridge knob, R1,until the point of maximum reading is found. Next, with the $ridge set a t its electrical center, a small amount of water or concentrated salt solution is added until the null point is just reached by the meter. This will bring the two conductances very nearly to equality and requires but little time to perform. If the cells were made nearly identical, the volumes of solution in the two should also be the same, ensuring effective compensation of all incidental drifts occurring in the cells. By now the electrical circuits will have settled down to a steady state and the titration can be started. h small amount of reagent is added to the sample in order to determine the direction in which the meter will deflect during the titration. The bridge knob should be set on the side of balance giving a needle deflection away from the null point. I n order to keep the meter readings on the range of linear response, the bridge is adjusted so that the needle is about 150 microamperes away from the null point a t the start of the titration.
It is usually sufficient to plot only about five points before the end point and an equal number after. Figure 3 shows a series of check runs made on sea water, representative of the straight lines
V O L U M E 19, N O . 4, A P R I L 1 9 4 7 Table I. Bottl e No.
1st
MZ. 1
2 3 4 5
3.270 3.279 3.377 3.282 3.282
267
and weighed into the volumetric flask containing sodium, magnesium, and calcium chlorides. S x - Probnble Before addition of the sulfate and Error CI dilution to volume, the chlorides were tested for sulfate by adding concentrated 0,1397 -0.0005 barium nitrate to a sample of the con0,1391 -0,0007 centrated chloride solution. No trace 0,1400 -0. 0008 0.1398 -0,0005 of cloudiness was observed. When 0.1 0.1397 *0,0005 mg. of sulfate was added to the mixture 0.1396 * 0.0006 a decided cloudiness appeared, indicating that previously the salts were sufficiently sulfate-free for the purpose. The barium nitrate solution was made up with Baker's C.P. barium nitrate which had been dried for 72 hours a t 140" C. Both solutions were standardized gravimetrically by adding a slight excess of sulfate or barium, respectively, to known volumes of the solutions t o be standardized. The precipitation was carried out by adding one reagent dropwise to a hot solution of the other and digesting the precipitate for 2 hours on a steam bath. The initial concentration of the unknown ion was approximately 0.015 M . Enough hydrochloric acid was added to the mixture to make it 0.3 M in hydrochloric acid. These conditions were found by Crowell (1) topbe optimum for avoiding absorption of nitrate by the barium sulfate precipitate. Two runs on the standard barium solution agreed to 0.15% and two runs on the artificial sea water agreed to 0. 12m0.
Sulfate Analyses on 25-iIfl. Samules of Sea Water
0 208 41 Ba(XOa)z 2nd 3rd ,1fZ. .w1.
3.251. 3.287 3.370 3.311 3.270
3.281 3.304 3.352 3.299 3.287
4th ,111. 31320 3,328 3.288 3.304
Average
Sulfate
MI.
Mg.
3,267 3.297 3.357 3.295 3.286
Chloride
65.50 66.10 67.30 66.06 65.88 Average
Mg.
469.0 475.5 480.5 472.5 471.5 ratio
formed by the experimental points. The dotted lines represent theoretical slopes and the theoretical end-point level. EXPERIMENTAL RESULTS
Macromethod. Table I shows the results of sulfate analyses carried out on a series of samples of sea water obtained off the coast of Southern California. The average weight ratio of sulfate to chloride found for all the samples was 0.1396 h0.0006. This agrees well with 0.1394, the accepted ratio for the major oceans of the world (11). The work on sea water showed that calcium interferes seriously with conductometric determinations of sulfate. Analytical results come out about 5y0low in sea water, for example, because of ita presence. Fortunately, the effect of calcium can be neutralized by adding sodium oxalate before titrating, provided the ratio of sulfate to calcium is a t least 6 to 1. The effect of oxalate appears to be due to mass action equilibria. One may consider that a competition exist.s between barium and calcium for the sulfate, which results in partial substit,ut,ion of calcium for barium in the barium sulfate crystals. Such behavior becomes the more likely because of the chemical similarity of barium and calcium as well as the isomorphism of their sulfates. R h e n oxalate is added, another competition is set, up between calcium and barium for the oxalate, resulting in coprecipitation of barium with the calcium oxalate. Khen the ratio of oxalate to calcium is about right, the amount of sulfate used up by the calcium is equal to the barium used up by the oxalate, and the results come out equal to theoretical. For sea water the optimum ratio of calcium to oxalate is empirically found to be about 3-6 to 1. If much more oxalate than this is used, the results are too high, while if much less is added, they are too low. When the calcium-sulfate ratio is greater than 1 to 6, titration end points become erratic and oxalate seems to be ineffective. Owing to the low initial concentration of sulfate in the titration mixtures (0.0015 M), it was found advantageous to seed the precipitation with a lit,tle pure barium sulfate. The time required to make a tit,ration without seeding is about 30 minutes but is reduced to about 5 minutes in a seeded mixture. Seeding also seems to be responsible for minimizing coprecipitation effects and for producing precipitates of more uniform character. The seed is prepared by adding C.P. sulfuric acid to C.P. barium hydroxide until the mixture is acidic. The precipitate is then washed on a sintered-glass filter with 370 acetic acid until no cloudiness is observed on mixing a little of the clear filtrate with concentrated barium solution. The product is stored in 3c"c acetic acid in order to prevent the absorption of carbon dioxide and the subsequent formation of barium carbonate. The addition of a small amount of acetic acid is necessary in sea water titrations, in order to prevent precipit,ation of barium by the carbonate and bicarbonate present in the sea water. The presence of a small amount of acet,ic acid should have no deleterious effect upon most ot,her types of sulfate titrations, while its presence in the seed suspension ensures its remaining pure on standing. When the suspension is made just thin enough t o be easily pipetted, about 5 ml. are sufficient for 500 ml. of titration mixture. The accuracy and precision of the method were checked using artificial sea water, both with and without calcium and with a known concentrat,ion of sulfate. I n the artificial sea water mixture, the required weight of Baker's C.P. sodium sulfate was dried for 72 hours a t 110' C.
A secondary method for standardizing the barium nitrate solution also gave good agreement. This method involved evaporating a known volume of the solution to dryness on a steam bath and heating a t 140" C. for 72 hours. Two runs by this method checked to O . l % , but the value arrived a t was 0.3% higher than that given by the precipitation method. Owing to the possibility of occluded water in the barium nitrate crystals, the precipitation results were favored. Numerous conductometric runs were made on the artificial sea water t o determine the per cent recovery and the probable error of individual titrations. Sext, the sulfate content was increased and further runs were made (Table 11). I n the water containing calcium, sodium oxalate equivalent to l / 5 of the total
Id0
Figure 3.
zdo
M ICROAMPERES
3d 0
Plot of Typical Titration Data
Total aulfate approximately 65 mg.
ANALYTICAL CHEMISTRY
268 calcium present was added before each run. The data indicate a n average of 99.74y0 recovery of the total sulfate added and an average probable error of
Table 111. Microdeterminations of Sulfate
’’S F
10.32%. Micromethod. All the previous ti-
1.25
804 Added
1st
Mo.
Mo .
Mo.
MQ.
Mg.
16.07 15.72 15.42 11.81 11.95 4.02 1.064 1.078
16.03 15.92 15.70 11.83 11.85 4.09 1.067 1.075
15,60 15,90 15.70 11.97 11.95 4.02 1,040 1,056
15.67 15.90 15.58 12.07 11.95 4.03 1.070
16.00
3.75 16.00 trations were made with 500-ml. cells 3.75 16.00 and a 5-ml. buret for the barium nitrate. 5.00 12.00 5.00 12.00 Cn order to adapt the method to mi15.00 4.00 56.25 1.067 crodeterminations, it was necessary to 56.25 1.067 reduce both sizes. A pair of 100-ml. cells was made similar to the larger ones, and the microburet mentioned above was also constructed. I n the investigation of the suitability of the method for micro work, all sulfate solutions were merely mixtures of sodium sulfate and sodium chloride in varying proportions. Since sea water titrations were as successful as those of simpler mixtures in the macro work, it was tacitly assumed that.the sulfate-chloride mixtures would afford as good a test of the method as more complex solutions. Table I11 shows the results obtained from the investigation and indicates the suitability of conductometric titrations for sulfate
Sulfate Found in Titrations 2nd 3rd 4th
,...
5th Mg.
-
15:78
ii:ii 11.92 4.01 .
..
I
Average Mg.
15.84 15.84 15.60 11.92 11.92 4.03 1,060 1,070
Recovers and Probable
Error % 99.0*l.Z
99.0 *o. &
97.5 *O.: 99 4t0.6 99.4+O. 2 100.8+0 99.3*1 1 99 7 + 0 9 I
micro analyses. It is seen, for example, that 1 mg. of sulfate cac be determined to *l% even in the presence of 56 times as murk chloride. I n Figure 4 are plotted the titrations tabulated on line 7 of Table 111. The experimental points in this case form just as good straight lines as those in Figure 3, where about sixty times as much sulfate was present. With still smaller cells and more refined technique, there is no reason why accurate determinations on considerably smaller amounts of sulfate could not be carried out. The apparatus described in this paper has also been used in chloride titrations with considerable success. Results will be presented in a subsequent paper. Comparison indicates that the conductometric method of recognizing the chloride-silver nitrate end point is capable of precision and accuracy fully comparabl~ to those of precise chemical methods. SUMMARY
I00
M
Figure 4.
3
2 00 ICROAMPERES
Plot of Sulfate Microtitrations
Total sulfate approximately 1 mg.
The method, with suitable apparatus for determining amount$ of dissolved sulfate as small as 1 mg. with an accuracy of *1.OL7( even in the presence of considerable chloride, is a modification of the familiar conductometric titration. It deviates from the classical method in that the apparatus registers conductance changes automatically. Temperature drift in the conductance cell is compensated without the necessity of a thermostatic bath. Calcium interferes with the determination, but its effect can be eliminated by use of oxalate, provided the ratio of sulfate to calcium is a t least 6 to l. The time required for a titration can be shortened from 30 to 5 minutes by the addition of a small amouni of pure barium sulfate. In macrodeterminations, when the precipitation was controlled by seeding, recovery averaged 99.74’% of the total sulfate present, with a probable error of *0.32%, I n micro amounts (1 to 16 mg.) recovery averaged 99.3y0 and the probable error wae *0.8%. These results shorn less probable error than is usual in conductometric sulfate determinations. This uniformity is probably due to the presence of more uniform precipitates, which in turn were produced by providing seed crystals on which t8hr precipitation could take place. LITERATURE CITED
Table 11.
Sulfate Determination on Artificial Sea Water
Sea Water, CalciumFree
67.05
Sea Wat,er with Calcium and Oxalate Mg. of Sulfate Added 68.00
Sea K a t e r , Calcium-Free with Added Sulfate
134.67
Mg. of Sulfate Recovered in Individual Titrations
67.08 66.40 66.76 67.26 66.64 67.10 67.00 66.82
67.32 67.97 67.86 68.20 68.07
...
134,28 134,34 133.92 134.30 134.12 134.12
.. ..
% Recovery and Probable Error of Individual Titrations 89.75*0.39 99.82+0.45 99.64+0.11
(1) Crowell, W. R.,“Notes on Quantitative Analysis”, 1st ed., DP 46-7, Ann Arbor, Mich., Edwards Bros., 1935. (2) Dutoit, P.,J . chim. phys., 5 , 27 (1910). (3) Hinman, Am. J. SCi. Arts, 114,478 (1877). (4) Kinney, G. F., and Garman, R. L., J . Chem. Education., 13,190 (1936). (5) Kolthoff, I. M., and Kameda, T., IND.ENG.CHEY.,ANAL.ED. 3, 129 (1931). (6) Morton, C., J . Chem. SOC.,1931,2983. (7) Muller, W., Ber., 35, 1587 (1902). (8) Pope, C.G., and Gowlett, F. W., J . Sci.lnstrumentr,4,380(19.271, (9) Robertson, J. D., and Webb, D. A,, J. Expt. Bwl., 16, 173 (1939). (10) Scholander, P. F.,Science, 95,177 (1942). (11) Thompson, T. G., Johnston, W, R., and Wirth, H. E., J. c o w d t intern. sxploration mer, 6, No.2, 246-51 (1931). (12) Webb, D.A.,J. Ezptl. Biol., 16,438 (1939). (13) West, L.J., and Robinson, R. J., IND.ENQ.CHEM.,ANAL.E n 12,476 (1940).
