High purity cyanides: A dying technology revived

164

Journal of Crystal Growth 75 (1986) 1 64- 172 North—Holland. Amsterdam

HIGH PURITY CYANIDES: A DYING TECHNOLOGY REVIVED MC. DeLONG and F. ROSENBERGER Department of Phesics, Unit’eriitr of Utah, Salt 1_ake Car, Utah 84/12, US~~1 Received 15 May 1985; manuscript received in final form 13 January 1986

Alkali cyanides are not generally available in adequate purity to permit single crystal growth. Commercially available NaCN and KCN may he purified by a combination of vacuum drying, gettering of oxygen-containing anions, melt filtration and/or 7.One refining. Alternatively, higher purity cyanides may be synthesized by a variety of techniques. The efficacies of various synthesis and purification schemers are ccmpared using infrared absorption for the detection of anions and atomic absorption spectrophotoinetrv br cations. A collection of physical properties of alkali cyanides is also presented.

I. Introduction Alkali cyanides are well suited model systems for many fundamental solid state research probems. They possess complete intersoluhility with halides. Hence, with increasing cyanide concentration in solid solution, one can observe a continuous evolution of molecular interaction and subsequent molecular ordering and phase transitions from isolated point defects through a spin-glass phase. In addition, vibrational emission from cyanide molecules isolated in an alkali halide matrix has been observed recently [11. Growth of high quality cyanide single crystals has been severely impeded by lack of adequate starting material. In order to obtain alkali cyanides with sufficient purity for single crystal growth, one must understand the methods used for commercial preparation and the chemistry whereby those products become contaminated, as well as alternalive techniques for their synthesis. further purification and characterization. These topics are the subject of this review. The most widely used commercial process for producing reagent grade NaCN and KCN is the so-called “wet” or neutralization process. wherein HCN is reacted with an aqueous solution of MOH (M = alkali metal, in this case Na or K) by a standard acid-base neutralization [2]: MOH(aq)

+

HCN

=

MCN

+

HOH.

(I)

By using purified HCN (liquid or vapor) and protecting the MOH solution from atmospheric carbonation, a product having a purity in the range of 98--99% is routinely attained by most manufacturers. Analyses of several lots of material from a variety of sources performed in our laboratory as well as a typical analysis of NaCN [2] are shown in tables I and 2. Although these purities significantly exceed the requirement of 95% specified for reagent grade NaCN or KCN [3], such salts are totally unacceptable for growing single crystals. RhCN and CsCN are not commercially available at all, LiCN is available only as a solvate in THF (Aldrich Chemicals, Milwaukee WI). An understanding of the chemical processes whereby cyanides degrade in purity is critical to

Tible I

Cyanide content of typical commercial reagents Manufacturer

Assay (%) ______________________

NaCN

KCN

Fisher

98.0, 98.4

98.7. 99.1

ROC/RIC Mallinckrodt

97.4 98.5, 98.0

Apache Sargent-Welch

94.1 983

98.7

Baker

98.9

97.7

°

0022-0248/86/$03.50 Elsevier Science Publishers By. (North-Holland Physics Publishing Division)

Multiple values refer to different lots

MC. DeLong, F. Rosenberger / High purity evanides

165

maintaining a high level of purity from synthesis through storage and handling, purification and single crystal growth. The attack of water on alkali cyanides is literally a textbook example for the hydrolysis of the salt of a weak acid and strong base [4]. In aqueous solution (including the surface

MOH: above 80°C the formate/ammonia reaction dominates. Hydroxide and formate produced by reactions (2)—~4)and (7), respectively, may themselves form carbonate via the reaction HCOOM + MOH = M

layer of water adsorbed on cyanides having varying degrees of hygroscopicity). the strongly ionic salt dissociates completely:

Reactions pertinent to vacuum heating of alkali cyanides contaminated with formate, with subse-

MCN

=

M~+ CN.

(2)

Simultaneous equilibria involve the dissociation of the weak acid HCN:

2CO,

2 HCOOM

=

H~+ OH-.

(4)

The equilibria of reactions (2), (3) and (4) are coupled. Effectively, equilibrium (4) increases [H~] and equilibrium (2) greatly increases [CN]. driving reaction (3) strongly to the left. If the HCN is allowed to escape (e.g. by storage of the cyanide in a plastic container which permits exchange of water and HCN with the environment), reaction (1) proceeds to the left, contaminating the cyanide with hydroxide. Such basic solutions are notorious for becoming contaminated with carbonate through an analogous absorption of atmospheric CO2:

CO2 + H20

=

H2C01,

H2CO1 + 2MOH

=

M.,C.,O4 + H2, M7C03 + CO, =

-

=

M2C05

+ H2O.

(8)

200-400°C,

-

(3)

as well as the dissociation of the water: HOH

H2.

quent formation of oxalate and carbonate were put forth by Hinks et al. [6]:

M2C704 HCN=H~+CN,

+

>

400°C.

(9) (10) They also claim that at the melting point of the cyanide, the carbonate decomposes according to MSCO,

=

M20

+

co2.

(11)

The CO liberated in reaction (10) and the CO., liberated in reaction (11) subsequently react with the alkali cyanide to produce cyanate and carbon: MCN+CO=MOCN+C,

(12a)

2 MCN + CO2

(12b)

=

2 MOCN + C.

As pointed out by Hinks et al. [6] and Neubert and Susman [7], and verified by our experience .vacuum melted MCN will be heavily con-

(5)

taminated with MOCN and C . The ubiquitous MOCN may also be generated directly. if the

(6)

cyanide is heated in an oxygen-contaminated atmosphere [3]:

A competing mechanism was reported by Ricca and D’Amore [5], who passed CO2-free air through aqueous cyanide solutions. They studied the hydrolysis by monitoring the composition of gaseous products entrained in the effluent gas as a function of the solution temperature. In addition to the mechanism described above, MCN hydrolyzes to formate and ammonia according to the reaction: NaCN + 2 H2O = HCOONa + NH1. (7) Ricca and D’Amore found that both reactions proceed at temperatures above 15°C.At 50°Cand below, the dominant products are HCN and

2MCN+O2=2MOCN.

(13)

Note that at each stage of Hinks’ decomposition, one volatile product is liberated. Hence heating under vacuum will indeed drive the reaction according to the stated path. If, on the other hand, the alkali cyanide is carefully vacuum dried below 50°C to remove water, then melted under an inert atmosphere in a closed vessel from which the volatile reaction products cannot escape, a different chemistry prevails. In this case, as reported by Lessing et al. [8], cyanide still hydrolyzes initially to formate, then to carbonate, but the only infrared-active contami-

166

SI.C. DeI_ong, 1. Rosenherger / Ihg/i purity crannies

nants found in their cyanide mixtures after melting are carbonate and OH -. This observation is partially corroborated by the unreferenced report of Jenkins [2] that NaCN is not extensively hydrolyzed by NaOH below 500°C.Above that temperature he claims the hydrolysis products to he carbonate, cyanamide, oxide and hydrogen. Independent of the details of the hydrolysis. it is clear that any attempt to grow a single crystal from reagent NaCN or KCN will result in a melt containing several percent carbonate, cyanate. and/or carbon. This level of these impurities would result in impurity concentrations in the

PRESSURE EQUALIZING SEPARATORY FUNNEL

HCN OH 6 6

ANTI-SIPHON VALVE

/

N

2GAS INLET

/

cyanide crystal in excess of the solubility limit. Such a two phase mixture would be useless for most single crystal applications. Hence, techniques for purification and/or synthesis of carbonate-free cyanides as well as analytical techniques for determin~in~~thyuc~ess of potential improvements

The most widely used modern technique for synthesizing high purity alkali cyanides on the laboratory scale dates to Meyer [9]. All are based on a modification of reaction (1) in which the alkali hydroxide is dissolved in anhydrous ethanol or a similar non-aqueous solvent. Unfortunately, Meyer’s recipe includes dissolution of the hydroxide in a small volume of water at the beginning of the synthesis. This step is inherently counter-productive and appears in neither later syntheses of NaCN by Guernsey and Sherman [10] nor of RhCN and CsCN by Sugisaki et al. [11] and in our laboratory. Our synthesis technique is similar to those of refs. [10,11]. The synthesis, illustrated and predominantly used for RbCN, consists of four steps: (1) purification of commercial RbOH, (2) purification and aliquoting of a defined quantity of HCN, (3) synthesis and (4) purification of RhCN. As mentioned earlier, all hydroxides display varying degrees of hygroscopicity. RbOH and CsOH being the most hygroscopic. The aqueous solution formed on the surface dissolves CO2 from the atmosphere, reacting with it to produce

I

\~_(\\,,~bOH/etOH

(//

ICE BATH

MAGNETC STRRER ..

.

1-ig. 1. Apparatus for synthesis of alkali cyanides from alkali hydroxides and liquid HCN -

carbonate. Commercial hydroxides (including reagent KOH) typically contain 10-30% water and several percent carbonate. To purify the hydroxide of water, we vacuum dry it, slowly increasing the temperature over a period of several days such that the system pressure remains less than iO~ Torr. The dried RbOH (Apache Chemteals, Seward IL) is then dissolved in the minimum quantity of anhydrous ethanol * (solubility approximately 0.64 RbOH/mI etOH) and filtered under a dry nitrogen atmosphere to remove the insoluble carbonates. A small quantity of ethanol is added to avoid saturation. The solution and stirring magnet are then transferred to a i-liter. 3-necked round bottom flask under N~ and . . Ethanolic hydroxides are much more caustic to skin and eyes then aqueous solutions; hence contact must he avoided.

Prolonged contact results in serious damage.

M. C. DeLong F. Rosenberger / High purity canides

167

PRESSURE ____

GAGE

LNTRAP

GRADUATED (VYCOR)

DFFU~ON

MECHAMCAt VACUUM ~MP

Fig. 2. Apparatus for aliquoting predetermined quantities of HCN.

-J

~

o

~

~

>-

)~

covered with a small volume of benzene (see fig 1) HCN is purified (see below) and measured in a defined quantity using the gas handling system depicted in fig. 2 [12]. A 200 ml cylinder of HCN (FUMICO Inc., Amarillo, TX) is attached to the system and all parts evacuated to the HCN cylinder valve. The Vycor HCN transfer vessel is then

cooled to 0°C, the pump sealed off and HCN transfer begun. A new HCN cylinder contains air, which eventually diffusion limits vapor transport.

At such times, the HCN cylinder is closed, the contents of the transfer vessel frozen to 77 K, and the system evacuated. Transfer is resumed until the volume of liquid HCN in the vessel reaches the predetermined level. The HCN is then frozen to 77 K, the vessel backfilled with N2, loosely stoppered and transported to a fume hood for synthesis. Prudent laboratory practice requires that an individual equipped with a gas mask be stationed at the entrance to the room in which liquid HCN is being handled. The HCN is gently warmed to melting and a comparable volume of benzene added to it. This solution is transferred to the separatory funnel on the reaction vessel (fig. 1) and slowly added to the ethanolic RbOH with vigorous stirring at 0°C. The resulting RbCN is then filtered from its ethanol/benzene matrix, rinsed on the filter several times with petroleum ether and crudely dried to

,. ,~:-

-

- -

1 68

.5!, ( ‘.

De/_ong, 1~Rosenherger / high puritr cra,iules

100°C in a vacuum oven. As a final purification step. it is dried under < 10~ Torr vacuuni to 1 50°C, then melted under Ar and pulled into a polycrystal by a Czochralski process. This removes carbon, reduces OCN , and greatly reduces the surface area available for hydrolysis during subsequent storage in a dessicator. CsCN has also been synthesized by the technique described above. However, because of its considerably greater hygroscopicity, a C’sC’N synthesis technique was developed which avoids water as a reaction product. The apparatus is depicted in fig. 3. The major innovation is the replacement of the ethanolic hydroxide by a solution of alkali metal in anhydrous liquid ammonia, All glass parts are predried in a drying oven, then evacuated on the system. The NH 5 is predried by storing overnight with freshly cut sodium metal. The HCN is dried by condensing onto P2O5 prior to use. The synthesis sequence consists of filtering the molten (7~= 28°C) cesium metal (Electronic Space Products Inc.. Los Angeles. CA) through a frit and into the 1-liter. 3-necked reaction flask equipped with a glass-covered stirring magnet. (Our first attempt demonstrated that teflon ignites spontaneously in contact with molten cesium at room temperature!) The cesium is then dissolved in approximately 300 ml NH1 at —78°C’ (dr’s ice/isopropanol slurry). Residual cesium above the frit is collected by dissolving in NH5 condensed above it. Synthesis proceeds as HCN entrained in Ar is passed into the stirred solution. The endpoint of the reaction is easily recognized by the (nearly complete) disappearance of the characteristic blue color of NFI1-solvated alkali metal. The solvent is removed by recondensing it into its original container. The CsCN is harvested in a N2-filled glovebag. carefully dried to 150°C, melted under Ar. and grown into a clear houle 3 dicontainingWesingle regions of mension. have crystal also synthesized NH1 cm 4CN hy a technique identical to that used for CsCN, except for the omission of the cesium metal. NH4CN is thermally unstable, completely decomposing to NH1 + HCN above 35°C[131. Small quantities of LiCN have been produced by a similar technique. except the higher melting point obviates melt filtration in our apparatus.

Numerous other techniques have been published for synthesizing LiC’N, which is a popular source of lithium in organic syntheses. All authors agree that because of its extreme hygroscopicitv. a rigorously dry environment is crucial to success. The most extensive studies were done by Perret and Perret [14]. who synthesized the salt by dropping lithium metal sheets into a 20% solution of HCN in henzene. (They also had no success itt reproducing Meyer’s method.) The same method was later used by Lely and Bijvoet [151 to prepare the crystals they used to determine the melting point and structure of LiCN. Johns and DiPietro [161 have synthesized LiCN with “quantitative yield” by adding a benzene solution of HCN to a commercially available hexane solution of n-hutyllithium. Great care must be taken in using Uhutyllithium, which is pyrophoric. Hence not onl~’ water hut also oxygen must he scrupulously cxeluded. Rossmanith [17] produced small quantities of the THF-solvate by displacing silver from AgCN in naphthalene and THF. After extensive refluxing, the silver is filtered from the hot solution. The THF-solvate crystallizes out as the solution cools. The THF may he removed by heating in vacuum: only 1% remains complexed at 60°C. Both of the preceding methods were utilized by Ismail et al. [181 to prepare materials for their matrix isolation speetroscopy studies. (The same group also reported that the molecular structure corresponds to the isomer LiNC [19].)

I Purification

Early attempts at cyanide synthesis and purification were reviewed by Thompson [20]. Purification techniques included dissolution in liquid ammonia and removal of less soluble carbonate and cyanate by filtration [21] as well as the same procedure using ethanol as a solvent. He coneluded that custom synthesis from hydroxides or alkali ethoxides and HCN gave a product superior to any achievable by recrystallization of commercial salt from any organic solution. His paper is also a valuable source of data on solubilities of cyanides and their contaminants in ethanol and

M. C. DeLeng, F. Rosenherger

/ High purity cyanides

169

methanol. Vacuum distillation was also found to be ineffective [21]. Modern purification of reagent NaCN and KCN began with Susman’s procedure for zone refining [6,7,22]. It is essentially a two stage procedure. Susman heats and melts the reagent alkali cyanide under vacuum, then filters the melt through a quartz frit. The resulting ingot is contaminated primarily with cyanate (at the percent

KCN and powdered silicon (Alfa Products, Danvers, MA) on a quartz frit is dehydrated at < i0~ Torr at room temperature for several days. In keeping with the conclusions of the previous analysis of the prevailing chemistry, we raise the temperature to only 150°Cunder vacuum over the course of several more days. The system is then backfilled with 0.8 atm gettered argon and the temperature held above the melting point for

level) as well as a small amount of silicon from

several hours, during which time the mixture is stirred by slowly passing Ar through it. Finally,

attack on the frit and some ferrithiocyanate as a result of iron contamination. Our experience is that iron is lot-specific, its presence having decreased markedly since 1979. The concentration of all contaminants is reduced by zone refining in a thin molybdenum boat under an inert atmosphere. One critical characteristic of Susman’s process appears to be his time-temperature-pressure program for bakeout. He pumps at room temperature for 2 h, purging periodically with helium, then heats under vacuum to 200°C in 4 h, holds 20 h at 200°C, then increases the temperature smoothly to 665°C”.An entirely different method has been used by Lessing et al. [8], incorporating an elegant modification patented by Bretschneider [23]. They add 3—30% metallic silicon powder to a eutectic mixture of NaCN: KCN, then dry them for two weeks at room temperature at a pressure of 1.3 Pa (10 mTorr). The mixture is heated only under 1 atm Ar. The silicon reacts with the molten cyanide, presumably gettering the carbonate cornponents as insoluble silicates and SiC, which are subsequently filtered from the melt. Their overall observations are that (a) if K5Na1_5CN is exten“

. .

.

the mixture is filtered through the quartz frit. The polycrystalline salt is then zone refined in a boat made of 0.005 inch thick molybdenum sheet (Schwarzkopf Development Co., Holliston, MA) in a configuration similar to that described by Neubert and Susman [22]. Several improvements have, however, been made. Our boat has a cylindrical cross-section obtained by rolling the sheet. This allows for the use of simple cylindrical endplugs machined from spectroscopically pure graphite (Ultra Carbon Corp., Bay City, MI). The sheet is supported on the outside by commercially available 1.5 inch OD X 0.020 inch wall stainless steel rings. In order to reduce boat deformation, the cyanide is kept soft by maintaining it at about 450°C.This is accomplished by directly resistively heating the boat (‘— 3 V, 200 A). The molten zone is then established with a narrow, travelling zone heater. The heating element is wound from Kanthal Al wire (Kanthal Corp., Bethel, CT) supported by a bed of Fiberfrax felt (Carborun-

dum Corp., Niagara Falls, NY) inside a LAVA (3-M Co., Chattanooga, TN) furnace housing. This

sively vacuum dried at room temperature, then

arrangement also allows vacuum bakeout of all

heated and melted under argon, the only infrared active impurities found are carbonate and hydroxide; (b) if subjected to the same procedure after admixture of silicon, no IR-active impurities are detected; and (c) cyanate and carbon are formed if and only if the carbonate-contaminated cyanide is heated to melting under vacuum. Our IR spectra of pellets pressed from “as received” reagent KCN show not only heavy carbonate contamination but also considerable amounts of cyanate. Our current procedure for purifying commercial reagent NaCN and KCN is an amalgamation of those described above. A mixture of NaCN or

contents at a temperature exceeding that to be used in subsequent zone refining. Neubert and Susman [71have estimated the effective segregation coefficient for OCN in KCN to be approximately 0.7. We observe a somewhat smaller value in NaCN, leading to greater zone refining efficacy. In RbCN, on the other hand, the segregation coefficient appears to be greater than 0.9. Hence zone refining is relatively ineffective for removing OCN from RbCN. A final comment on purification is warranted. It is widely known that many anionic impurities may be effectively removed from alkali halides by

-

-

170

81. C’. DeLong. F Rovenherger

/

high purity crannies

H2S04), HCN polymerizes to a brown-black mass

Cable 2 Typical analysis of NaCN

[21

Constituent

known as azulmic acid or polycyanogen. Value

Sodium cyanide. NaCN (%)

98—99

Sodium carbonate, Na,C’O., (Or) Sodium formate, HCOONa (%) Sodium hydroxide. NaOH (%) Sodium chloride, NaCI (%) Sodium sulfide, Na25 (ppm) Water, H20(g) Iran, as Fe (ppm)

4. Characterization

0.3-1.3 ((.4 (1.2

Three different techniques are useful for assessing the purity of cyanides at various stages of

~ <005

10—20

Table 3 Infrared absorption maxima for impurities in single crystal potassium cyanide 125]

processing. A wet chemical titration with standard silver nitrate, described in detail by Kolthoff et a!. [24], can he used to determine the total cyanide content in a given sample with a typical precision Results of application of this technique to NaCN and KCN from a variety of commercial of 0.1%.

Impurity

Occurrence

Frequency (cm )

suppliers were presented in tables 1 and 2. NaCN and KCN that we have purified by zone refining

OCN

Solid solution

12070 1296.2

routinely assay 100,0% ±0.1%. (Because of greater simplicity and sensitivity, we now use zone refined

NO.,

Solid solution Adsorbed on surface Surface layer Surface layer

1385

KCN to standardize silver nitrate solutions.) LiCN

H.,O K2CO., (anhydrous) K2CO., (hydrated)

we synthesized assayed 99.75 ±0.2% pure with respect to CN. RbCN synthesized as described earlier assayed 99,6 ±0.1% pure.

CO~

Solid solution

1632 1397 1360 1444 1414 1463

R—OCO, (covalenily bonded)

Surface layer

1485 shoulder

Determination of the presence of molecular anions is done qualitatively by IR spectrometry. Although difficult to quantify. this technique is

1563

extremely

useful

for comparing

CO~ and/or

OCN concentrations among samples from differtreatment near their melting point with the appropriate halogen gas and/or acid halide. Purifi-

ent sources or from different parts of a zone refining ingot. Neubert and Susman [25] have identified the spectra of the dominant anionic impurities in KCN. Their results are reproduced

cation of cyanides with the analogous cyanogen

in table 3. We use the optical density at 1296.2

[(CN)2] or HCN gas at elevated temperature, as attempted by several researchers , has failed. Both (CN)2 and HCN are unstable at elevated temperature. In the absence of a stabilizing acid (e.g.

cm

as a figure of merit for the quality of zone

refined alkali cyanides. Optical densities of less than 0.01/em are routinely achieved in the most pure parts of zone refined ingots.

Table 4 Analysis of alkali cyanides by atomic absorption spectrophotometry Sample

Source

RbCN LiCN (‘sCN NaCN

Single crystal Synth: Li+HCN Synth: CsOH + HCN Reagent, as-received Zone refined Reageni. as-received Zone refined

KCN

Impurity concentration (mole ppm) Li

Na

K

13

100 <18 200

3800 2.4 27

12 3.3 3.1 1.2 1.3

3413 847

Rh

9 510

Ca

Mg

3 5

0.5 4 0.5 (1.2 1.2 1

Fe

13 27 5.3 6.3

M. C. DeLong, F. Rosenberger / High purity cyanides

171

Table 5

Physical properties of alkali cyanides

LiCN NaCn KCN RbCN CsCN

Melting point (°C)

(Tm)

Structure

160 563.7 ±I 634.5 599 [261 528 ±2 [271

Ortho. 1131 fcc fec fee CsCl [281

Phase transition temperatures (K)

Density at 25°C 3) (g/cm

T~ 288 [29] 168 [29]

172 1291 83129]

132 [30]

193 ~~[28]

Relative hygroscopicity

1.0755 141 1.60 1.52 2.32 [301 3.41 1281

Extreme

Considerable Little Little High

‘° 1’~:fcc

transforms to ferroelastically ordered orthorhombic (NaCN, KCN) or monoclinic (RbCN) phase. to ferroelectric ordering. °~ CsCl: transforms to trigonal phase at 193 K. 5) ~I~2: transforms

Table 6 Physical properties of alkali cyanides

Solubilities

NaCN

KCN

1-120 (25°C) etOH (25°C) meOH (25°C)

41.7 g/100 g soln. [32] 1.9 g/l 30 g/I

NH., (— 33°C)

38.8 g/100 g soln. [31] 9.7 g/l 78 g/I 37.5 g/I [211

Vapor pressures NaCN 1331

0.1 960.5

KCN

1341

1.0 1089

10 1257

100 Torr at 1489 °C

In ‘>m (atm)= —25930/T+17.37 for monomers In P~(atm) = —29762/T+22.32 for dimers In P 1 (atm) = — 36796/T + 24.39 for trimers

The presence of metallic impurities in cyanides

cation schemes may be compared using infrared

may be determined with great accuracy and sensi-

absorption for the detection of anions and atomic

tively by atomic absorption spectrometry. Results

absorption spectrophotometry for cations.

of several of our analyses, showing the improve-

Some physical properties of the alkali cyanides

ments at each stage of purifying reagent NaCN and KCN, are given in table 4. As is the case with halides, the dominant impurities are other alkali

are listed in tables 5 and 6.

metals.

Acknowledgments

5. Summary Alkali cyanides are not generally available in

adequate purity to permit single crystal growth. Commercially available NaCN and KCN may be purified by a combination of vacuum drying, gettering of oxygen-containing anions, melt filtration and/or zone refining. Alternatively, higher purity cyanides may be synthesized from liquid HCN and the appropriate alkali metal or hydroxide. The efficacies of various synthesis and purifi-

We would like to acknowledge continuing support of this project by F. L’Uty under NSF Grants DMR 77-12675 and DMR 81-05532. We would also like to thank Dr. G. Kodama for technical assistance during development of the synthesis techniques.

References [11K.P. 5840.

Koch, Y. Yang and F. Lilly, Phys. Rev. B29 (1984)

1 72

81. C. DeLong, F Rosenherger

/

High purity crannies

[2] W.R. Jenkins. in: Kirk—Othmer Encyclopedia of (‘hemical Technology, 3rd ed. (Wiley, New York, 1979) pp. 320-331.

[20] MR. Thompson. NatI. Bur. Std. (US) J. Res. 6 (1931) 1051. [21] L. Hackspill and R. Grandadam. Ann. (‘him. 5 (1926)

[3] Reagent Chemicals. 1974 Committee on Analytical Reagents (American Chemical Society, Washington, DC, 1974) pp. 462. 565. [4] Fl. Remy. Treatise on Inorganic Chemistry Vol. I (Elsevier, Amsterdam, 1956) p. 823. [5] B. Ricca and G. D’Amore. Gazz. (‘him. ltal. 79 (1949)

218. [22] T.J. Neuhert and S. Susman, Rev. Sci. lnstr. 35 (1964) 724. [23] G. Bretschneider. Entfernen von (‘arhonaten and (‘yanaten aus Alkali oder Erdalkalicyaniden, German Patent I. 006. 839 (1957);

308.

see also Gmelins Handhuch der Anorganischen (‘bernie.

[6] D.G. Hinks. DL. Price. J.M. Rowe and S. Susnian, .J. Crystal Growth 15 (1972) 227. [7] Ti. Neubert and S. Susman, J. Chern. Phys. 41 (1964) 722. [8] J.G.V. Lessing, K.F. Fouché and T.T. Retief, Electrochim. Acta 22 (1977) 391. [9] J. Meyer. Z. Anorg. AlIg. Chem. 115 (1921) 203. [101 E.W. Guernsey and MS. Sherman, Am. (‘hens. Soc. 48 (1926) 695. [11] M. Sugisaki. T. Matsuo. H. Suga and S. Seki, Bull. Chcm. Soc. Japan 41(1968)1747. [12] F. Rosenberger. J.M. Olson and MC’. DeLong. J. Crystal Growth 47 (1979) 321. 1131 iA. Lely and J.M. Bijvoet, Rec. Tray. Chim. 63(1944)39. [14] A. Perret and R. Perrot, Helv. Chim. Acta 15 (1932) 1165.

Vol. 21. 1st Suppl.. Ed. E. Pietsch (Springer, Berlin. 1964) p. 324. [24] l.M. Kolthoff. E.B. Sandell. E.J. Meehan and S. Bruckenstein, Quantitative Chemical Analysis, 4th ed. (Macmillan. New York. 1969) pp. 811—812. [251Ti. Neuhert and S. Susman. J. (‘hem. Phys. 42 (1965) 1848. [26] V. Kondo, D. Schoemaker and F. Lilly. Phys. Rev. B19 (1979) 4210. 1271 F. Lilly. private communication (1979). 128] A. Loidl, K. Knorr, J.l. Kjems and S. Haussiihl, J. Phys. C’ (Solid State Phys.) 13 (1980) L349. [29] B. Koiller, MA. Davidovich, LC. Scavarda do (‘armo and F. L’ilty, Phys. Rev. B29 (1984) 3586. 13(1 J.M. Rowe. ii. Rush and F. Lilty, Phys. Rev. B29 (1984)

IlSI J.A. Lely and J.M. Bijvoet, Rev. Tray. (‘him. 61 (1952) 244.

2168. [311 GD. Oliver and S.E.J. iohnsen, i. Am. (‘hem. Soc. 76

1161 lB. Johns and HR. DiPietro, J. Org. (‘hem. 29 (1964) 1970. [17] K. Rossmanith, Monatsh. Chem. 96 (1965) 1690. [18] Z.K. Ismail. RH. Hauge and iL. Margrave. AppI. Spectrosc. 27 (1973) 93. [19] Z.A. Kafafi, RI-I. Hauge and J.L. Margrave. Polyhedron 2 (1983) 167.

(1954) 4721. [32] AS. Corbel, J. (Them. Soc. 129 (1926) 3190. [33] Landolt-Börnstein Zahlenwerte und Funktionen, 6th ed., Vol. 2. Part 2 (Springer. Berlin, 1960) p. 41. [341L.L. Simmons. L.F. Lowden and T.C’. Ehlert, J. Phys. Chem. 81(1977) 709.