Noms
Jan., 1962
173
NOTES A REFERENCE ELECTRODE FOR CERTAIN MOLTEN SALT SOLUTIONS BY GEORGEW,I~ARRINGTON AND 13. T.TIEN Departman4 of Chemisfty, Temple University, Phaladelphia, Pa. Received M a y $9. 1061
Potentiometric investigations of molten salt systems have become increasingly popular because useful thermodynamic data can be obtained by these methods. The success of any potentiometric study depends on finding a suitable reference electrode. Several types of reference electrodes have bcen developed to meet this need. These electrodes may be divided into two groups. The first group consists of electrodes with liquid junctions such as the silver-silver nitrate electrode of Vlengas and Rideal, the silver-silver chloride clectrode of Senderoff and Brenner,2 and the Pt elcctrodc ol Lnitinen and Liu.a The second group involves a glass barrier or bulb. Notable among these are Delimarskii's glass elect rod^,^.^ filled with eithcr sodium amalgam or Na-Sn alloy, and the glass membrane type filled with molten electrolyte as devised by Bockris, et aL6 A very simple reference electrode composed of Pyrex glass has been developed in this Laboratory which is similar to the electrode of Delimarskii. It differs, howevcr, in that sodium is not necessary inside the bulb and the applicability is not limited to melts containing sodium ions. This is true provided the electrode has been equilibrated in the melt in which the measurements are to be carried out. A particular advantage of this electrode, apart from simplicity of construction, is that it can be used in different molten salt systems. Experimental Materials.-All chemicals were of rcagcnt grade and were used dirwtly without further purification except oven drying at 115". CoCl2 \vas obtained as the hexahydrate and was dchydratcd before use by oven drying at 115'. Cobalt wire W:LS 99.5% Co and 1%& S gage 18. The mercury in all elcrtrodes wits triply distilled. Preparation of Electrodes.-The cobalt indicator electrode was sirnplv cot)alt wire. The wire was polished electrolyticillly in 5O:nO HICl, in absolute ethanol, rinsed and dried. Cobalt amalgam also was used succcwfully as the indicator eltbctroclc but hccaufie of ease of handling the wire was preferred 111 most cases. Corisl ruction of the reference electrode was very simple. A bulb of approximately 15-mm. diameter wa8 blown on one end of a short length of 5-mm. 0.d. Pyrex tubing. The bulb then waa filled with either pure mercury, cobalt amalgam or Li-IC nilrate eutectic cont,aining 0.8 mole % CoC12. A platinum or tungsten wire then was iusertcd down the tube (1) S. N. Flongaa and E. Ridoal, Proc. Roy. Soc. (London), A2SS, 443 (1966). ( 2 ) S. Sendwoff and A. Brenner, J . Electrochem. Sac., 101, 31 (1954). (31 11. A. Laitinen and C. H. Liu, J . Am. Chem. Soc., 80, 1015 (1958).
(4) Yu. K. Delimamkii and R. 8. Khaimovkiok, Khim. Zhur., 15, 77 (1949). (6) Yu. K . Delimarskii and A. A. Kolott. ibid.. 16, 438 (1950). (6) J. O'M. Boakrie, G. J. Hilla, D. Inman and L. Young, J . Sei. Indcr., 33, 438 (1956).
into the liquid in the bulb in order to establish contact with the measuring circuit. Ap aratus -Using usual notation the cell may be written as: Li-K nitrate or KSCN, glass; Hg. The cell consisted of a 600-ml. tall form Pyrex' beaker, which was immersed in a salt-bath of stainless steel containing Li-K nitrat e eutectic. The cell was covered with an asbestos pad with holes provided for electrodes, nitrogen gas inlet, thermowell and a port through which increments of CoC12 could be added. The assembly was heated in a resistance furnace and the temperature controlled by means of a Powerstat. Temperature was measured by a Chromel-Alumel thermocouple and was kept constant during the course of a run to f1' of the initial value. Cell e.m.f. was measured by a Cary Vibrating-Reed Electrometer (Model 31) and wau recorded by a Varian G-10 recorder. The readings were accurate within 1%of scale setting. Procedure.-Five hundred grams of freshly prepared LiNOrKNOs eutectic was placed in the cell and the temperature was raised slowly to the desired value. Dry nitrogen was bubbled through the melt continuously to provide an inert atmosphere and to agitate the solution. Increments of CoCb, weighed to within 0.1 mg., were added and voltage readings taken after thermal equilibrium was reestablished. The procedure was the same when pure potassium thiocyanate was used as the solvent, in which case this salt was substituted for the eutectic.
80;
Results and Discussion The results presented below are based on ohservations made using Li-K nitrate eutectic with pure mercury inside the glass bulb. Exactly analogous results were obtained when KSCX was the solvent. The glass bulb filled with pure mercury gave the best precision compared to bulbs filled with either cobalt amalgam or Li-K nitrate. The temperatures of the runs discussed were within the range 168-180°, held to the limits mentioncd above. TABLE I -E.m.f.(fl Initial
290 249 22s 207 180 153
mv.), mv.Final
255 23 1 210 198 171 141
Log C o eoncn.
-3.10
-2.77 -2.44 -2.15 -1.91 -1.63
Table I is a presentation of data obt:iincd with a typical electrode. Identical results were found for each of eight other glass electrodes except, of course, for shifting of tghebasc line. The initial and h a 1 values rcfer to the e.m.f. just prior to and just after an addition of CoC12. When either the initial or final values are plotted us. log Co concentration straight lines are obtained. Neither of these lines, however, has a slope that agrees with the value predicted by the Nernst equation. It may be seen in Table I that when the initial value of any particular addition is compared to the final value for the preceding addition considerable drift has occurred. Approximately ten minutes elapsed between increments. However, when the difference between initial and final values is plotted cumulatively vs. log
174
NOTES
CO concentration a straight line is obtained having exactly the slope predicted by the Nernst equation for the cell involved. Thus it is seen that the cell containing the glass electrode is behaving in the proper fashion once the effect of drift is removed. I n order to be a useful reference, however, the electrode must be stable. It was observed that the drift decreased slowly with time, suggesting thqt an “aging” process was occurring. I n order to study this effect the following experiment was set up. Three bulbs, selected at random, were prepared as above and placed in the molten solvent. One of these was arbitrarily selected as reference and the other two measured against it over a period of several days. Voltage readings were taken periodically over each eighthour working day. For the first two days the voltages changed a t rates greater than 50 mv./hr. On the third day one bulb showed a change of 2.5 and the other 6.5 mv./hr. On the fourth and fifth days the rates for each had fallen to less than 0.5 mv./hr. Beyond this time period the e.m.f. showed no further change. Thus five days aging yields a reference electrode that is stable for direct potentiometric measurements. Removal from the melt, washing and drying had no effect on this stability. Several electrodes were selected a t random and aged for one week in Li-K nitrate eutectic. Each then was placed in a cell of the type described earlier and the e.m.f. measured as a function of CoClz concentration. Each cell yielded a linear Nernst plot having exactly the predicted slope. E o drift was observed in any of these measurements. On the basis of the information presented above it is not feasible io attempt a complete explantion for the mechanism occurring a t this electrode. Certain observations may be made, however. The fact that the material inside the bulb made little difference suggests that this material serves only to establish electrical contact. The glass does not act as a membrane separating electrode compartments. Equilibration seems to follow an exponential rate suggesting zi diffusion -controlled process. Once equilibration is established, however, it is apparently quite stabie since random agitation (Le,, K2 bubbling) or complete removal appears to have no effect. Additional investigations are being conducted to answer these questions more fully. It should be pointed out, however, that this investigation, while establishing a useful reference electrode, also has shown the ideal behavior of Co(I1) in the solvents studied. The maximum concentration used was 0.1 mole % and to this concentration the solutions behaved ideally. The presence of chloride ion added as CoClz apparently has no effect at these concentrations. H. T. Tien wishes to express his thanks to Dr. D. 0. Rudkin, Eastern Psychiatric Institute, for his kind permission to use some of the facilities of the Institute.
Vol. 66 NUCLEAR MAGNETIC RESONANCE STUDY OF BORON COoRDIh’ATION I N POTASSIUM BORATE GLASSES’ BY %-E. SVAITSON, E. FORSLIND AND
Research Group for N M R , Divzsion of Physzcal Chemistry,, The Raga1 Institute of Technology, Stocloholm, Sweden
J. KROGH-MOE Suedash Institute of Siltcute Research. Gothenburg, Sweden Received June 6, 1961
The coordination of boron in the alkali borate glasses has been studied earlier with n.ni.r. by Silver and bra^.^,^ The B1l-resonance indicates that the boron nuclei in these glasses have two different local environments. Due to an anisotropic distribution of electronic charge around the boron nucleus, the resonance line in vitreous boron oxide shows a second-order broadening effect of nuclear quadrupole interaction, and this broad line is assigned to boron in threefold oxygen coordination. By addition of alkali oxide a sharp resonance line is produced which is attributed to the appearance of boron in fourfold coordination belonging to BO4 tetrahedra with low quadrupole interaction. It would in principle be possible to calculate the amounts of threefold and fourfold coordinated boron in the sample after separation of the overlapping resonance lines and evaluation of the areas under each line. The results of Silver and Bray, however, were evaluated in a very simplified manner from the observed signal amplitudes. As pointed out by the authors themselves their method is likely to overestimate the fraction of fourfold coordinated boron at high alkali content, since the width of the broad line increases with increasing alkali content and, indeed, their measurements (filled circles in Fig. 1) lead to a fraction of fourfold coordinated boron definitely higher than that assumed by Warren4 and the one observed in some crystalline borates investigated by Krogh-R/Ioe.6t6 Warren assumed that boron changes from threefold to fourfold coordination as alkali oxide is added to the glass, supposing each oxygen bonded to two b o r a s at low alkali content, all oxygens being engaged in boron-oxygen bonds. Krogh-Moe6 has shown that the boron-oxygen networks of crystalline potassium pentaborate,6 cesium triborate6 and lithium diborate’ are built in agreement with the following rule regarding the coordination of boron and oxygen: each “molecule” of alkali oxide added to boron oxide converts two boron atoms from threefold to fourfoId coordination. This corresponds to a fraction of boron in fourfold coordination Nd = X / ( l - X) (shown by the solid line in Fig. l), where S is t8hemolar alkali concentration. The above rule and expression for N4 follows from (1) The Swedish Xatuial Science Research Council and the S t a t e Council of Technical Research have provided financial support and the n.m.r. apparatus cost has been defrayed by grants from the Knut and Alice Wallenberg Foundation. (2) A. H. Silver and P. J. Bray, J . Chem. Phys., 29, 984 (1968). (3) J. D. Mackennie, “Modern Aspects of the Vitreous S t a t e ” Vol. 1, Butterworths, London, 1960. (4) E. E. Warren, J Am. Ceram. Soc., 24, 256 (1841). ( 5 ) J . Krogh-Moe, Arkw Kemi, 14, 489 (1959). ( 6 ) J. Krogh-Moe, Acta Cwst.. 18, 889 (1960). (7) J. Kiogh-Moe, to be published.