C;%é\é{\\»cie~ = | l—' Technische Mitteilung TM-HL-235 Abte"ung: HL Bearbeiter: A . Dfindal" * /Skh Visum: j m Betrifft: Datum: 20.8.74 " Synthesis and Gas Chromatography of Selected Metal Chlorides. 45 Seiten Zeichnungen Contents page 1. Synthesis of Molybdenum Chlorides 2 1.1. Aim of Work 2 1.2. Introduction 2 1.3. Experimental 5 1.4 Results and Discussion 8 2. Gas Chromatography of Volatile Metal Chlorides at High Temperature 11 2.1, Aim of Work 11 2.2. Introduction 3.1 2.3. Experimental 20 2.4, Conclusions 27 5. References 29 * Privab: Arif Diindar, Anittepe, Ata Sokak 2-4 ANKARA - Turkey * Univ.: Middle East Technical University, ANKARA - Turkey Verteiier| Abteltung | Name Expl. | Abteilung Name Expl. GL Direktion 3 DO |Bibliothek 3 HL |Dr. J. Peter 1 Beserye 15 alle Gruppenleiter je 1 A. Dindar 2 W. Eberhard 1 Dr. M. Furrer 1 Frau Dr. Ianovici 1 Dr. K. Tunaboylu 1 CH R. Keil 1 ME G. Ullrich 3 -Dieses Dokument ist Eigentum des Eidg. Institutes fir Reaktorforschung- TM=HL=235 page 2 1. Synthesis of Molybdenum Chlorides 1.1. Aim of Work In the project of a Molten Salt Fast Breeder Reactor(l) the chloride of plutonium-239 (15%) diluted by sodium chloride is the molten fuel. Molten fertile material is UC1l_-238 (65%) plus 3 sodium chloride besides newly bred PuCl, and fission products which acts also as blanket material andsas coolant. Molybdenum is one of the fission products and forms low volatile chlorides with chlorides present. It is also one of the most important construction materials and will probably be used in this type of reactor. Therefore, our aim was to synthesize molybdenum (V,ITII,II) chloride-samples which could be used in the experiments for the project. 1.2. Introduction Experiments were done to synthesize molybdenum penta- and tri- chlorides. —— — —— — — — ———————————— — — ————— —————— ————————— — The method used was the reaction of molybdenum(VI)oxide with boiling hexachloropropene. MoO_, + 3 C 3 Cl6 + MoCl_. + 3 C 5 CluO + 1/2 C1 2 2 3 Porterfield and Tyree, Jr(z) showed that the method used by Herman and Suttle for the preparation of uranium(IV)chloride(B) from uranium(IV)oxide and hexachloropropene can be conveniently extended to other polyvalent metal chlorides. TM-HL-235 page 3 M2On + nCSCl6 +* 2 MCln + nCBCluO They added: "These chlorides are well characterized and have long been prepared by satisfactory means. However, the method described has the advantages of greater convenience and rapidity, indifference to small amounts of water on the oxides (the water is converted to hydrogenchloride and trichloro- acrylic acid) 03016 + 2 H20 - 0301302H + 3 HC1 relatively complete conversion of the oxide to the chloride, and products which are easy to separate and purify." Two methods were used for the synthesis of molybdenum trichloride. 1.2.2.1. Reduction of Molybdenum pentachloride with hydrogen: This reduction is carried out by passing a current of pure and dry hydrogen over molybdenum pentachloride. MoCl5 + H2 > MoCl3 + 2HC1 The purification of hydrogen is very important, because the formation of oxychloride must be avoided. It is also necessary to Keep the reaction temperature under 194°C in order to avoid the formation of a cake in the molybdenum pentachloride mass (see table 1). TM-HL-235 page U4 ASubstance mp , °c color MoCl5 194 black MoClu 317 brown MoCl3 1027 dark red MoCl2 decomp. yellow MoOCl3 green MoOClu green Table 1 Physical properties of molybdenum chlorides (4) and oxychlorides 1.2.2.2. Reaction of Molybdenum(VI)oxide with Hexachloropropene The analysis of molybdenum chloride which had been synthesized by the method (1.2.1.) showed that the stoichiometric ratio of chlorine to molybdenum was close to 3 (see 1.4.1.) that shows another possibility.to synthesize molybdenum trichloride by the same method but for a longer time as to complete the reduction. (5) Besides this Larson and Moore reported that in preparation of molybdenum tetrachloride by the reaction of molybdenum penta- chloride with refluxed chlorobenzene, substantial formation of molybdenum trichloride occured after extended reflux periods (greater than 10 hours). It is possible that hexachloropropene can be used instead of chlorobenzene in excess, due to its resem- blance also being unsaturated, and able to reduce molybdenum pentachloride to -trichloride. TM-HL-235 page 5 1.3. Experimental 200 ml of hexachloropropene (1.41 mole) and 20 g of molybdenum(VI) oxide (0.14 mole) were put in a 250 ml round-bottomed flask with a reflux condenser. The upper end of the condenser terminated in a drying tube. The flask was filled with argon. The reaction mixture was brought to the boiling temperature by means of a heating mantle and refluxed for 15 minutes, until reaction was complete shown by a color change from yellow to dark red. Argon was continuously flushed through the system. The more slowly the cooling process is performed, the larger is the size of the indi- vidual crystals obtained. Therefore, the flask was well insulated and allowed to cool to room temperature for 3 hours. The crystals were very dark-colored under a dark red solution. The flask was removed from the reflux condenser and closed immediately with a sintered glass filter tube. The upper end of the tube terminated in a round-bottomed flask in which the filtrate can be collected. Then the whole system was inverted and the metal chloride was separated from the solution by filtration. During this time, argon was still in charge into the reaction flask. The filtering opera- tion was completed slowly in about 3 hours. Before the filtration was completed, the crystals were washed three times with previously dried CClu which was added by opening the system when the crystals were still covered by solution. After the filtration process, the round-bottomed flasks at both ends of the filter tube were replaced by dry argon filled flasks. The molybdenum pentachloride crystals were then sucked and dried under vacuum. Finally, the system was transported into the glove- box and opened. TM-HL-235 page 6 ———————— —————— T ————— ———————————————————— - 1.3.2.1. Reduction of molybdenum pentachloride with hydrogen The reaction vessel was made of quartz glass. It was equipped with a porous glass try on which the molybdenum pentachloride crystals were put. After molybdenum pentachloride had been put inside the vessel, in the glove-box, the vessel was placed in a vertical electri- cal oven (Fig. 1). The vessel had a spiral tube in its lower part so that the entering gas (hydrogen/argon 1l:1) was preheated before coming into contact with molybdenum pentachloride. At its upper end the vessel terminated in a drying tube. The leaving gas was passed through water in order to absorb hydrochloric acid to be determined afterwards. The vessel was insulated with glass wool and the reaction was carried out at 167-175°C for 11 hours. The hydrogen and argon mixture was passed through the molybdenum pentachloride mass in order to prevent back sucking in a possible stop of the hydrogen current. During the reaction the vessel was vibrated continuously to prevent cake formation. 1.3.2.2. Reaction of Molybdenum(VI)oxide with Hexachloropropene 90 ml of hexachloropropene (0.63 mole) and 6 g of molybdenum(VI)- oxide (0.042 mole) were put in a 250 ml round-bottomed flask and the same procedure as for the synthesis of molybdenum pentachlo- ride was followed with the only exception that the mixture was refluxed for 20 hours. Pig. 1 vibrator Reduction of MoCl dryer = 5 to MoCl3 water TM-HL-235 page 7 thermoelement temperature controller TM-HL-2 35 page 8 1.4, Results and Discussion Molybdenum pentachloride produced by the method 1.3.1. was analy- zed colorimetrically for molybdenum and by a potentiometric method for chlorine ). It was found that the ratio in moles of chlorine to molybdenum was 3.64 and the mass balance of molyb- denum and chlorine gave 100.8 percent. An analysis was also carried out for the chlorine content of molyb- (6) denum pentachloride samples by the gravimetric method . The results are shown in table 2 (assuming the absence of impurities) chlorine molybdenum Cl/Mo no |sample weight |wt. g moles |wt. g | moles . .| in moles 1. ] 0.0766 g 10.0405 | 0.00114 | 0.0361 |0.00038 . 3,02 2 .. .4 0:X2TT & 0.0691 ] 0.00195 |0.0586 |0.00060 5.25 Table 2 Analysis of molybdenum pentachloride samples From the results of these analyses, the sequence of the following reactions seems reasonable and possible. C Cl6 C Cl6 5 MoCl, 2 0. Moc1 MoO., + C Cl6 + MoCl 3 3 B 3 On the other hand, because the water content in our glove-box was very high (39 ppm) the hydrolization and oxidation of molybdenum pentachloride could cause deviations in the results. *) Performed by R. Keil, EIR-CH TM-HL-23%5 page 9 1.4.2.1. Reduction of Molybdenum Pentachloride with Hydrogen This experiment was done two times with (a) the so called molyb- denum pentachloride synthesized in 1.3.1. and (b) the commercial molybdenum pentachloride. The first one was abondoned afterwards because the results of analysis of molybdenum pentachloride showed the stated composi- tion M0013—4‘ Therefore, it was tried to reduce the commercial molybdenum pentachloride to -trichloride. But during the prepa- ration of the reaction vessel, high water content in the glove- box (80 ppm) caused some amount of molybdenum pentachloride to hydrolize and as a result of this hydrolization, the green crystals of molybdenum oxychlorides were observed (see table 1). The eluted HCl was titrated with sodium hydroxide after 11 hours of reaction time. Table 3 shows the results. N HC1 «|N HC1/N HC1 No | N NoOH | VHC1, ml |V NaOH, ml | N HC1l |.(expected) .|(expected) . — e — - 1 0.20 25 10.86 0.0869] 0.4654 18.7 % 2 0.25 ho 16.19 0.1012] 0.4654 21..T ‘S Table 3 Determination of the degree of reduction of MoCl_ to 5 MoClB. * MoCl5 input 7 g (0.0256 mole) HC1l expected 2 x 0.0256 = 0.0512 mole water 110 ml HCl1l expected: 0.0512 (1000/110) = 0.4654 N However, it seems possible to us to perform this reaction under anhydrous conditions. TM-HL-235 page 10 1.4.2.2. Reaction of Molybdenum(VI)Oxide with Hexachloropropene Molybdenum trichloride produced by this method was analyzed again by the colorimetric method for molybdenum and by the potentio- metric method for chlorine* and it was found that the chlorine: molybdenum ratio, in moles, was 3.30 whereas the mass balance of molybdenum plus chlorine gives 100 %. A gravimetric analysis was as before, carried out for chlorine content of our molybdenum trichloride. Table 4 shows the results (assuming the absence of impurities). sample chlorine molybdenum Cl/Mo No |weight, g wt, g moles wt, g | moles in moles 1 0.2359 0.1284 0.00362 | 0.1075 | 0.00112 3.23 2 0.1887 0.1026 0.00289 } 0.0861 | 0.00090 dvel Table 4 Analysis of molybdenum trichloride samples From the mass balance of molybdenum and chlorine, which is 100 %, it is clear that our product contains no impurity, ie oxides, oxychlorides. In this case the molar ratio of chlorine to molyb- denum makes us to think about the incomplete reduction of molyb- denum pentachloride to -trichloride. In this case molybdenum trichloride can be separated from the other by means of an appre- ciable organic solvent in which only it is soluble (ie, alcohol, ether). * Performed by R. Keil, EIR-CH TM-HL-235 page 11 2. Gas Chromatography of Volatile Metal Chlorides at High Temperature 2.1. Aim of Work Our work is a part of the project of a Molten Salt Fast Breeder Reactor. How are the volatile metal chlorides formed? As already mentioned, the molten fuel, the molten fertile material, the coolant and the blanket material contain chlorides. From the fission products, germanium, arsenic, tin and antimony form vola- tile chlorides while zinc, gallium, rubidium, zirconium, niobium, molybdenum, cadmium, indium and cesium form low volatile chlo- (7) rides . Therefore it is possible to collect information about the formation of these volatile metal chlorides by gas chromato- graphy. 2.2. Introduction Gas chromatography is a technique for separating volatile sub- stances by percolating a gas stream over a stationary phase. It is in analytical application widely. Its greatest advantages are its speed, resolution, qualitative and quantitative analysis and its very great sensitivity. In general, gas chromatography has largely been concerned with the separation of volatile organic compounds and not too much has been done with metal chlorides. Three trends are apparent in the re- sults reported so far. They involve the use of fused salts, slightly volatile liquids and solid materials as the stationary phase. Fused salts, while useful at very high temperatures, are difficult to prepare on the support, especially in anhydrous form and are subject to reaction with the solutes. The slightly volatile liquids are lost from the column, redistribute on the support, may be thermally unstable, and are capable of reactions with the (8) solutes TM-HL~2 35 page 12 The main problem in the qualitative and quantitative application of gas chromatography in determination of inorganic volatile chlorides is the choice of an appropriate liquid phase which does not react with these compounds. In general, reactivity may (9) be a question of a) interaction between solute and the liquid phase (sometimes due to the metallic tubing of the columns) resulting in the crea- tion of interaction products b) sorption of the solute on the solid support, or ¢) the way by which the solutes are dissolved in the liquid phase. The reactivity of inorganic volatile chlorides with respect to packing materials and occasionally the difficulty in determining the magnitude of this effect has often led to different opinions on the stability of various substances that might be used as | packing materials for the separation of inorganic chlorides. Conclusions regarding the reactivity may be drawn from (a) obser- vation of the chromatograms obtained, (b) the possible changes occuring in the packing material after the experiments, (¢) the ratio of logarithms of specific retention volume to vapor pressure of liquid phase. The stability of the baseline and its deviation from zero, the presence or absence of artefact peaks or the non-appearance of expected peaks, the symmetry of the peaks looking for tailing and skewing are the points to be examined so as to draw conclusions regarding the reactivity. Possible retention of some of the solutes in the column can be ascertained by checking whether there is any difference . in the weight or color of the column packing material TM-HL-2 35 page 13 before and after any series of experiments. Finally one can obtain indications as to whether or not the solutes are ideally dissolved in the liquid phase, without any change of their mole- cular structure, by examining the plot of logarithm of specific retention volume vs logarithm of vapor pressure which should be a straight line with a slope equal to unity. Since in analytical gas chromatography, mg or sub mg amounts of sample are brought into contact with at least a thousand-fold amount of column material, it is evident that when reactive sub- stances are involved, the inertness of column materials has to meet severe demands. When quantitative results are aimed at, the use of metals and organic liquids except completely halogenated products, and diatomaceous supports is not allowed. The lack of inertness of metals such as copper and stainless steel is perhaps best illustrated by the ready attack of these by chlorinating agents. Further proof of the unsuitability of, for example, copper is the violent reaction observed between it and antimony (V) chloride, both in liquid and in gas phase. Diatomaceous supports are unsuitable for the present purpose because of their adsorptive properties. For example, large amounts of HC1l are almost irreversibly adsorbed by Sil-O0-Cel C 22 (chromosorb R or P). These supports also have the great ad- sorption affinity for metal halides. Even the white diatomaceous supports, which are usually regarded as being less adsorptive than the red ones, are unsuitable, because of the numerous metallic impurities which may react. It was proved that even rigorous acid treatment failed to remove these impurities completely(lo). Some of the chlorides, like SbCls, are active chlorinating agents, or moreover, free chlorine may be present in the sample. The stationary liquid should therefore be at least resistant to chlorination. This condition is not met by most organic stationary TM-HL-235 page 14 (10) phases. Sie et al. reported that a considerable amount of HC1 was formed when chlorine was contacted with silicone grease, even at temperatures as low as room temperature. (11) Freiser has reported the separation of tetrachlorides of tin and titanium at 102° by using n-hexadecane (31%) coated chromo- sorb. Both tin(IV) and titanium(IV) chlorides gave well defined, symmetrical bands. The ratio of vapor pressure of these two compounds was about the same as the inverse ratio of their reten- tion times, showing the absence of reactions with the stationary phase. ke ) by including This study was extended by Keller and Freiser niobium(V) and tantalum(V) chlorides and use of somewhat less volatile stationary phases (n-octadecane and squalane) at tempera- tures of 150 to 200°. Keller(s) apiezon T and silicon o0il as stationary liquids on red chromo- separately reported the use of sorb as a support in copper columns at temperatures of 100 to 2007 . The separation of the four chlorides by these authors is more of a qualitative nature. Studying the chromatograms shown by Keller, it is concluded from the presence of shoulders,tails and artefact peaks that decomposition of solutes have occurred to a large extent within the column and both apiezon grease and sili- con oil reacted with chlorides. Tin(IV)chloride on apiezon grease and titanium(IV)chloride on silicon oil at 100° did not appear. Keller concluded that the normal alkanes appear more satisfactory than branched chain compounds. Wachi(ls) at 125o using a stainless steel column packed with silicon grease was unable to elute tin(IV) and titanium(IV) chlorides or apiezon M on ground Sil-0-Cel C 22 insulating brick. Failure to obtain peaks was attributed to complete reaction of the chlorides TM-HL-235 page 15 with the greases and the walls of the column. Wachi then turned to fused inorganic eutectics in glass columns and reported the 5 and SnClu-Tlclu. Poor results with the latter pair were attributed to the insolu- separation of the pairs SnClu-SbCI bility of titanium(IV)chloride in eutectics. Wachi also reported the failure of nitrate eutectics to separate the tran- sition metal chlorides due to the oxidizing properties of the nitrates. He attempted the gas-solid chromatography of tin and titanium tetrachlorides by using C-22 firebrick as stationa- ry phase, but resolution was very poor. (14) used a eutectic mixture of BiCl3 and PbCl2 on Sil-0-Cel C-22 firebrick for a separation of titanium(IV) Juvet and Wachi and antimony(V) chlorides at 240°. A pronounced baseline in- stability was observed. Their work was not of quantitative nature. (15) reported interaction between the silicon products Tadmor and metal halides. He obtained well defined peaks attributable to reaction products when analyzing germanium(IV), tin(IV) and arsenic(III) chlorides with silicon grease as stationary phase. (16) In a later paper' ', Tadmor reported the separation of germa- nium(IV), tin(IV) and arsenic(III) chlorides and their inter- action on a column packed with Sil-0-Cel - C-22 insulating brick, by an isotopic exchange technique (hydrochloric acid labeled with Cl1-36). The chromatograms obtained with the uncoated support showed irregular peaks and imperfect separations. The incomplete recovery of the chlorides (65% for GeClu, 78% for AsClB) at the column's outlet was attributed to losses due to hydrolysis occurring within the column. Better separation was obtained by coating the support with 1 percent nitrobenzene. The chromato- grams shown exhibit peak shapes which are considerably distorted. TM~-HL~-235 page 16 (17) chloride and germanium(IV) chloride on 17 different stationary Tadmor also studied the separation of arsenic(III) liquids on Sil-0-Cel brick and reported the changes in weight of these immobile phases as a function of temperature, and the maximum temperatures at which no change in weight had been detected. The chromatogram of germanium(V) and arsenic(III) chlorides on uncoated Sil-0-Cel brick exhibits during 110 minu- tes only one peak for them. The ratio of the retention times of these chlorides is approximately the inverse of their vapor pressure ratio at 100°. This indicates the absence of reaction with support. Wilke et al.(l8) the tetrachlorides of silicon, tin and titanium by gas-liquid reported the determination of the purity of chromatography. They used nitrobenzene, silicon oil and apiezon N as stationary phases at about 100° on diatomaceous firebrick supports (sterchamol, diaphorit). They found that well resolved individual peaks for silicon (IV) and titanium(IV) chlorides could only be obtained by preinjecting a sufficient quantity of titatium(IV) chloride. Although a complete separation is reported with a column thus treated, the peaks still exhibit considerable tailing, as is evident from the chromatograms shown. No proof is given of complete recovery of solutes from the column. Dennison and Freund(lg) reported the separation of arsenic(III), tin(IV), germanium(IV) and titanium(IV) chlorides on halocarbon 6-00 coated (24.5% W/W) chromosorb W at 107° and 126°. They used new all-glass apparatus with a specially developed sampling chamber. Elutions were in the order of their boiling points. They also tried quantitative determination of arsenic(III) and tin(IV) chlorides. Maximum error was 6 percent which was believed to result primarily from the premature vaporization of the sam- ples as they were being injected. TM-HL-235 page 17 (20) Brazhnikov and Sakodynski reported certain regularities in the retention of chlorine-containing inorganic and organometallic compounds. They separated germanium(IV), silicon(IV), tin(IV) and titanium(IV) chlorides on the solid support polychrom, a type of teflon, which was coated with siloxane elastomer E-301, apiezon N, n-octadecane, polyorganosiloxane liquid VKZh-94, liquid polytri- fluoromonochloroethylene of the Kel F-10 type, Kel F-3 or AlBrS. From the chromatograms shown, the results are of only qualitative nature. (21) transition metal chlorides including arsenic(III), tin(IV), anti- Zado and Juvet reported the elution characteristics of 11 mony (III), molybdenum(V), niobium(V), zirconium(IV) chlorides on 12 inorganic fused salt mixtures, by using acid-washed chromosorb P, porous vycor, glass beads as solid support. They discussed that most of the solutes studied are capable of forming chloro-complexes with available chloride in the liquid phase melt, and the ratio of the dissociation pressures of two chlorocomplexes is, in general different from the relative volatility ratio of the uncomplexed solutes and finally concluded that the separation of solutes based on this difference in complex stability would be particularly desirable for situations in which solutes have similar boiling points. (10) studied 19 substances including germanium(IV), Sie et al. arsenic(III), tin(IV), antimony(III, V), gallium(III) chlorides on Kel F-40 as stationary phase which was coated (15% W/W) on haloport F as solid support. Except for the unstable tetra- chloride of vanadium and pentachloride of antimony, all inorganic chlorides examined have been found sufficiently stable to permit a.gas chromatographic separation. Besides small peaks attributable to HCl, no artefact peaks were observed at all. With proper choice TM~-HL~-235 page 18 of column materials and some adaption of apparatus and injection technique to the reactivity of these compounds, quantitative analyses are possible with an accuracy equal to that normally obtained in gas-liquid chromatography. (22) Stumpp reported that graphite was a suitable stationary phase for separating metal chlorides by the technique of gas-solid chromatography. A special construction permitted the injection of the solid metal chlorides. The chromatograms obtained exhibit tailing to a large extent and are far from being of quantitative nature. (9) Parissakis et al. investigated the interaction of some chlorides including tin(IV), germanium(IV), arsenic(III), antimony(V) chlori- des upon the stationary phases Kel F-10, Kel F wax, silicon oil DC-550, silicon rubber, apiezon L, phasepack P and graphite. They reported that with the exception of antimony (V) and vanadium(IV) chlorides, the chlorides tested had shown no interaction with the above mentioned phases. Kel F wax on celite gave the most symme- trical peaks with negligible tailing. Titanium(IV) chloride caused darkening of apiezon L. (23) In another paper , Parissakis et al. also demonstrated the appropriate operation conditions for good qualitative and quanti- tative separation of mixtures of five inorganic volatile chlorides on the same stationary phases mentioned in the previous paragraph. By gas chromatography, it is also possible to calculate some physico- chemical properties, ie heat of solution TM-HL-235 page 19 For any given solute, the retention values are related to the column temperature by the following relationship(eu). . J Fc(tR-tA) 273 & W, Te where Vg = specific retention volume J = pressure drop correction factor Fec = volumetric flow rate of the carrier gas measured at the outlet pressure and column temperature trR = retention time of substance tp = retention time of air tRl = t'R - tA Wr, = weight of liquid phase Tc = column temperature, % Through the experiments and temperatures, it can be assumed without any serious error that J and Wy, remain constant. On the other hand, correction for the gas flow rate is Te Fc=Fmi,a- where Fm and Td are the flow rate observed and the detector tem- perature, respectively. Substituting into the main equation gives Vg = IFB g0 273 | gop o wl "R T4 R where K is the new constant term defining (JFm 273/W1l Td). Taking logarithms InVg = LnK + Lntg' ™-HL,~235 page 20 (25) As known Vg & = === 4+ where AHs is the heat of solution of the substance in the liquid phase. Equating the last two equations Ln tg' = - === + C! where C' is equal to (C-Lnk) Rearranging AHs Log tR' = - + C" 2.303 RT The heat of solution may be determined directly from the slope of the straight line which will be obtained by plotting log tg' vs reciprocal of the column temperature. 2.3. Experimental The chromatograph used was a VARIAN 2100 equipped with a Ni-63 electron capture detector. Nitrogen was used as carrier gas at flow rates of 24-48 ml per minute. The flow rates were measured by the soap bubble method. The pressure regulator was used to assure a uniform pressure to the column inlet and thereby a con- stant rate of gas flow. TM-HL-235 page 21 Columns used were U-shaped glass tubes (to avoid reactions of the highly corrosive chlorides) 2 mm ID, 1/4 inch OD, 170-180 cm long. The recorder was A-25 VARIAN with different suitable chart speeds and mV ranges. All columns were packed, under anhydrous conditions when neces- sary, by adding the packing material to the column by tapping the side of the tube gently with a spatula until the column packing material did not settle any further. In the case of gas- liquid chromatography with n-hexadecane as the stationary phase and carbopack A as the solid support, 16 g of carbopack A was suspended in a solution of 2 g of n-hexadecane in 10 ml of toluene. The solvent was removed on the water bath under vigorous stirring (26) The columns were conditioned each time overnight (or 48 hours in the case of carbopack B) at the maximum recommended temperature. The effluent end of the column was not connected to the detector during the conditioning period. Column oven temperatures varied depending on the packing material and the elution characteristics of the substances. In the injec- tions, the temperature of the injection port was 30-50o higher than that of the column oven. The detector temperature was either 50o higher than that of the column oven or constant at 3000. The elution characteristics of the volatile metal chlorides, arsenic(III), titanium(IV), tin(IV) and antimony(V) chlorides were studied. They were obtained from commercial sources. As it is known they are highly corrosive, volatile and very easily hydrolized to insoluble hydroxides and basic salts even in the presence of traces of moisture. To avoid this, the samples were prepared under nitrogen atmosphere in the glove-box and stored in glass-stoppered flasks. TM-HL-235 page 22 Because of the very high sensitivity of the detector even for the smallest amounts of pure chlorides (for CCly, it is 19 - g our chlorides were introduced as their solutions in n-hexane. The metal chlorides used had satisfactory solubilities in n-hexane. The samples were introduced by means of Hamilton 701 or S.E.G. lo A-FN-GP syringes. Repetitive retention measurements were made for each solute at each temperature. Packing materials: . SE-30 (methyl silicon), 3% W/W on varaport 30, 100/200 mesh Carbopack A Carbopack A, hydrogen treated . n-hexadecane, 12.5% W/W on carbopack A Ui & W - L . . Carbopack B, hydrogen treated (commercially) Besides the metal chlorides, air, chlorine, hydrochloric acid and some hydrocarbon injections were carried out independently to compare them with the metal chloride chromatograms under the same conditions. For each of the metal chlorides, three solutions with different concentrations were prepared. AsCly: 1. 2.84 x 107° g/ml SnCl,: 1. 2.94 x 107° g/ml 2. 3.74 x 107" g/ml 3, 3.86 £ 1077 g/ml 3. 0.97 x 10 ° g/ml 3. 1.00 x 10°° g/ml TiCl,: 1. 2.28 x 1072 g/ml sbCl; 1. 3.10 x 1072 g/ml 3.00 x 107" g/ml 2. 4.08 x 107" g/ml 0.78 x 10°° g/ml 3. 1.06 x 10™° g/ml TM-HL-235 page 23 The packing material used first was SE-30 as the stationary phase on varaport as the solid support. This column had already been in use at the beginning of the present work. The retention time of air in this column was determined as 0.40 min. n-Hexane exhibited many unexpected peaks (Fig. 2). As shown having nearly the same area, two big peaks do not permit us to think about any impurity in the sample as the reason. The peaks show considerable tailing. The reaction of n-hexane with the packing material is probably the reason for these findings. Since no oxysorb was being used at that time, the impurities in the carrier gas might cause the small peaks to appear. Free chlorine and hydrochloric acid gases can play an important role in the molten salts reactions which take place in the reactor. This is the reason for examining them besides the metal chlorides as solutions in n-hexane. When chlorine was injected, three shoul- ders, one of which was Cl2, were obtained on the first peak of n-hexane (Fig. 3). The other shoulders may be attributed to possible reaction products. With HCl, no special peak resulted. It is quite possible that HC1l is hidden in one of the n-hexane peaks. When chlorine and HCl were injected as gases, they exhi- bited many peaks, showing interaction inside the column (Fig. 4). Arsenic(III) chloride solution was injected at temperature inter- vals from 90O to 200°. Two peaks besides those for n-hexane were obtained at 125, 130, 150, 175 and 200°. Figures 5-8 represents small peaks, shoulders and tailing to a large extent. However, the second peak disappeared at lower temperatures (90, 100, 1100) (Fig. 9-11). This could be explained in two ways, (a) since these temperatures are lower than the boiling point of arsenic(III) chloride (1300) it might be retained inside the column, (b) as a result of repetitive injections, the solution might oxidize and hydrolize completely and arsenic trichloride might disappear. as a result of both, the first peak is attributed to the reaction product. TM-HL-235 page 24 ¢, 9K 1/Te trs min tp-tp=tg', min Log tgr' 398 0.00251 11,30 10.90 1.04 403 0.00248 9«33 8.93 0.95 423 0.00236 4.45 .05 0.61 Lu8 0.00223 2.+20 1.80 0.26 h73 0.00211 Y2 0.85 -0.09 The logarithm of adjusted retention time was plotted vs. the reciprocal of absolute column temperature for arsenic trichlo- ride (Fig. 12). The slope of the straight line is 2815. Then AHs = -(2815)(2.303)(1.987) The heat of solution of arsenic(III) chloride in methyl silicon is -12.8 kecal per mol. When the most diluted solution of titanium tetrachloride was in- jected (0.78 x 10-8 g) at 130, 150, 175 and 2000, one big and two small peaks were obtained (Fig. 13). The retention time of the biggest peak is identical to the unexpected peak of the ar- senic(III) chloride chromatogram. When the more concentrated so- 78) at the same temperatures as above, too many peaks resulted, showing lution of titanium(IV) chloride was injected (3.00 x 10 the interactions inside the column with increasing concentration of chlorine (Fig. 14). TM-HL-2 35 page 25 When tin(IV) chloride was injected at 200 and 2250, a shoulder on the first peak of n-hexane was obtained with a retention time equal to those of the unexpected peaks of arsenic(III) and ti- tanium(IV) chloride chromatograms. Two more peaks resulted, the second one probably being the tin tetrachloride peak (Fig. 15). The case with antimony (V) chloride is similar to titanium(IV) chloride. When the least concentrated solution of antimony penta- chloride was injected (1.06 x 1078 g) at 130 and 150°, a big peak with a retention time again equal to those of the unexpected peaks were obtained (Fig. 16). When the more concentrated solution of antimony pentachloride was injected (4.08 x 1077 g) including the temperatures 100 and 115O too many peaks were obtained (Fig. 17). After the failure to determine and separate the metal chlorides with this column we turned to gas-solid chromatography and used carbopack A as the solid packing material. When n-hexane was injected into this column a lot of unexpected peaks, also including a negative peak, were obtained (Fig. 18). HC1l in gas form gave only one peak with considerable tailing. However, repetitive injections created new extra peaks. With chlo- rine injection in gas form four peaks resulted. Tin tetrachloride (1.0 x 1078 g) gave a small peak (Fig. 19). The column created two peaks itself when the baseline was drawn without any injection, by programmed temperature method (Fig. 20a). The number of the peaks was five when n-hexane was chromato- graphed by the same method (Fig. 20b). Then many organic solvents namely cyclohexane, heptane, iso-octane, toluol, benzol and decane were tested by both programmed temperature and constant temperature methods. But none of them exhibited chromatograms bet- ter than those of n-hexane. n-Hexanes from different commercial sources were also tried and with that from FLUKA three peaks TM-HL-2 35 page 26 including the negative one were eliminated (Fig. 21). With injec- tion of chlorine and HCl separately in n-hexane, many peaks were obtained. At 60°, HC1l and chlorine have the same retention time. (27) reported that the use of gas-solid Di Corcia and Bruner chromatography was extended to the elution of polar compounds by treating graphon, MT and FT with hydrogen at 1000°. "In com- paring the adsorption isotherms before and after treatment, it appears,that chemisorbed oxygen, always present in these adsor- bents, is removed. Peak tailing is eliminated or strongly redu- ced." Upon this, we treated carbopack A with hydrogen at 800°C for seven hours (we couldn't manage to reach 10000). After this treatment, we still couldn't eliminate the extra peaks for n-hexane, the number being 2 this time. Then we again turned to gas-liquid chromatography and used n- nexadecane coated on carbopack A. We injected different organic solvents into this column and obtained only one peak at 40° for all of them, that of n-hexane being the most ideal one. The nega- tive dip after this peak was attributed to detector contamination (Fleg: 22). When chlorine and HCl were injected separately in gas form, they gave only one peak with tailing to a large extent (Fig. 23, 24). Arsenic(III) chloride did not appear at 100° and the baseline exhibited pronounced instability. That was attributed to the column bleeding at this temperature which is quite higher than the maximum temperature allowed for n-hexadecane (500). Carbopack B treated with hydrogen at 1000O (commercial) was the last packing material we worked with. n-Hexane exhibited again two peaks with negative dips after them (Fig. 25). Arsenic(III) chloride did not appear at 140° in 80 minutes. TM-HL-2 35 page 27 2.4. Conclusions The major difficulty in the separation of metal halides is in finding a non-reactive packing material capable of use at the elevated temperatures required. At the end of our literature survey we have concluded that nobody had complete success about this subject quantitatively. Qualitative separations would not be sufficient for our work and quantitative results were aimed at without any further help of the literature. There is a close relationship between column packing material and the substances investigated by gas chromatography. Having a very sensitive detector for chlorides and the necessity to decrease the concentration of metal chlorides caused us to use an organic solvent and eventually created another class besides metal chlorides HC1l and chlorine. That was really our major difficulty and we had extra peaks for n-hexane when chromatographed on the packing ma- terials used except n-hexadecane. Since chlorine, HCl and some metal chloride peaks had retention times equal to those of extra peaks and quantitative results were aimed at, we insisted on eliminating these unexpected peaks by trying different packing materials and were able to do it with n-hexadecane only. With all our packing materials, four metal chlorides which were tried either did not appear or gave extra peaks. From the presence of these peaks, unstable baselines and tailings we conclude that there are interactions between the packing materials and the samples. Those peaks of the metal chlorides with the same reten- tion time on SE-30 make us to think that it is a common reaction product. The extra peaks were the results of not only these reac- tions but also of the reactions between packing materials and hydrolization and oxidation products of metal chlorides. With the introduction of chlorine and HCl, too many peaks were obtained. TM-HL~-235 page 28 When the concentration of chlorine was increased, the number of unexpected peaks also increased. The metal chlorides are highly corrosive, volatile and very easily hydrolized. To avoid this, we prepared the samples under nitrogen atmosphere. However, because the amounts of moisture and oxygen were quite high in the glove- box, hydrolization and oxidation of the metal chlorides could not be avoided. Opening the small bottles, to take metal chloride samples for injections naturally increased the degree of oxidation and hydrolization. This was observed easily from the precipita- tion of metal hydroxides in the bottles. In further working, the samples should be prepared under complete anhydrous conditions and bottles with penetrable septums should be used instead of those with stoppers used up to now to store metal chlorides. (10) In the results reported so far, Sie et al. -and Parissakis et al.(g) had success to a degree in separating metal chlorides quan- titatively. The packing materials which were used by them, namely Kel F-40, Kel F-iO, Kel F wax, silicon oil DC-550, silicon rubber, apiezon L, phasepack P should be tried. Although some of them had been ordered by us more than 2 months ago, to try, they still haven't arrived even after the completion of this report. One important point is that some reporters had completely failed with some of the above mentioned materials. Carbopack B (commercially hydrogen treated) should be continued trying. In the case of failure with all these packing materials, capillary colums may be given trial. 3. References (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) Taube M., Ligou J. Poterfield W.W., Tyree S.Y. Jdr. Hermann J.A., Suttle J.F. Fergusson J.E., Larson M.L., Moore F.W. Strouts C.R.N., Wilson H.N. Parry-Jones R.T. Taube M. Keller R.A. Parissakis G., Vranti-Piscou D. Kontoyannakos J. Sie S.T., Blumer J.P.A., Rijnders G.W.A. Freiser H. Keller R.A., Freiser H.F. Keller R.A. TM-HL-2 35 page 29 Molten Chlorides Fast Breeder Reactor, Problems and Possibilities. EIR-Bericht Nr. 215, Eidg. Institut fir Reaktorforschung, Wirenlingen, 1972 Inorganic Synthesis, Vol. 9, P 135 Inorganic Synthesis, Yol: 5, p. 143 in V. Gutman (ed), Halogen Chemistry, Academic Press, London and New York, 1967, p. 249 Inorganic Chemistry, 3(1964) 285 Chemical Analysis Clarendon.Press, Oxford, 1962, p. 117 The Possibility. of Continuous in-core Gaseous Extraction of Volatile Fission Products in a Molten Fuel Reactor. EIR-Bericht Nr. 257, 1974, p. 5 J. Chromatog., 5 (1961) 225 J. Chromatog., 52 (1970) 461 Separation Sci., 1 (1966) 41 Anal. Chem., 31 (1959) 1440 in R.P.W. Scott (ed) (1960), Butterworth, Gas Chromatography London, 1960, J. Chromatog., 5 (1961) 226 p. 301 TM-HL-235 page 30 (14) Juvet R.S., Anal. Chem., 32 (1960) 290 Wachi F.M, (15) Tadmor J J. Inorg. Nucl. Chem., 23 (1962) 158 (16) Tadmor J. Anal. Chem., 36 (1964) 1565 (17) Tadmor J. J. of Gas Chrom., 2 (1964) 385 (18) Wilke J., J. Chromatog., 18 (1965) U482 Losse A., Sackmann H. (19) Dennison J.E., Anal. Chem., 37 (1965) 1766 Freund H. (20) Brazhnikov V., J. Chromatog., 38 (1968) 244 Sakodynski K. (21) Zado F.M, in A.B. Littlewood (ed), Gas Chromatography Juvet R.S. 1966, Elsevier, London, 1967, p.283 (22) Stumpp E. z. Anal. Chem., 242 (1968) 225 (23) Parissakis G., Chromatographia, 3 (1970) 541 Vranti-Piscou D., Kontoyannakos J. (24) Tranchant J. Practical Manual of Gas Chromatography Elsevier, London, 1969, p. 5 (25) Coates V.J., Gas Chromatography, Noebels H.J., Academic Press, New York, 1958, p. 20 Fagerson I.S. (26) Ambrose D., Gas Chromatography Ambrose B.A. George Newnes, London, 1961, p. 16 (27) Di Corcia A., Anal. Chem., 43 (1972) 1634 Bruner F. \ - ’ min mia Fig. 2 n-Hexane Packing mat. Te = 150°C I. _ 3 SE-30 Fe: 24 ml/min. TM-HL-235 page 31 J 4%\:4 S ma Fig. 3 012 in n-Hexane Packing mat. SE-30 Te = 150°C Fe: 24 ml/min Fig. 4 b— TM-HL-235 page 32 . _— e — T 012 in gas form Packing material = SE-30 Te = 150°¢C Fe: 24 ml/min. TM-HL-235 Cl2 in gas form Packing material = SE-30 Te = 150°¢C Fe: 24 ml/min. page 32 o) = ==t | Gg) o s hD ~ ¢ { e i S e T { 1 i - ':! imf\ ‘Lw—\ m}; F_ 5% ‘“ 1 TM-HL-235 page 33 — ——] A sCl3 Packing material = SE-30 Te = 125°C Fe: 24 ml/min. TM-HL-235 page 34 o - =y \ : \ * = \ F\ l - — I - ‘101-)‘ a.mh pbin by Bmor Ban Fax *.«-4 ! Fig. 6 AsCl3 Packing material = SE-30 Te = 150°¢C Fe: 24 ml/min. - - I = J ~ ! - NF\ q | & ; [ 2 min S G- £ Fig. T ASCl3 Packing material SE-30 Te = 175°C Fe: 24 ml/min. TM-HL-235 page 35 Fig. 8 AsCl3 Packing material: Te = 200°¢ Fe: 24 ml/min SE-30 TM-HL-235 page 36 | B X | | o | N | - W | | f | o t K | ‘n % % ] =} { ! | ) : - o | ~J -nh # Fig. 9 AsCl, ‘Fig. 10 AsCly Packing material: SE-30 Packing material: Te = 90°C ot Fc: 24 ml/min i 18 Fe: 24 ml/min L°‘3 (“ R LA\ Q.10 Fig. 12 200 249 o 30 Cho S0 6o VTG_ x \_Q" . — -1 Log (tg-t,) vs T = for AsClB on SE-30 TM-HL-235 page 37 o A sCl3 Packing material SE-30 Te-110°C Fe: 24 ml/min. H- P 4'0\.}; 3:&5\.- Fig. 13 T:'LCl,4 Packing material: SE-30 Tec = 150°C Fe = 24 ml/min. b : .n‘\h\ an z j mia ~Min ?m.‘«, Fig. 14 AsCl3 (more concentrated) Packing material: SE-30 Te = 150°C Fe: 24 ml/min e Q¢ o3ed G¢e-TH-NWL £ ? o ’ ~ o — e : = o — . - ~ W e - N 2 min n men $ ”";’1 " ) “ \\ ) . im}; 2’»»‘«. 3 —— e, Packing material = SE-3%0 Te = 200°C Fig. 16 SbCl5 Packing material = SE-30 Te = 150°C Fe: 24 ml/min. Fe: 24 ml/min 6¢ o3ed G¢2-TH-IL TM-HL-235 page 40 v W W i T TS ) ,m,4 m S w | — \ - Qo ?‘ o Rig. 17 SbCl5 Packing material = SE-30 Te = 150°C Fe = 24 ml/min. Fig. 18 n-Hexane Packing mat.: Carbopack A Te 200°¢C 24 ml/min. o L e Pig. 19 n-Hexane Packing mat.: Te = 200°C Fe 24 ml/min. Carbopack A Th @3ed G¢2-TH-NWL TM-HL-235 page 43 S & i | ! : | o o ; N - = & | * | | § ~ w | | | o b o N . .--—n—yu-u-T—J &—-stLi!'F l . = Fig. 21 n-Hexane Packing mat.: Carbopack A Te = 60°C Fc = 24 ml/min. Fig. 22 # i LR n-Hexane Packing mat.: Te = 40%C n-Hexadecane U 36 tty o3ed G¢2-TH-IWL TM-HL-235 page 45 L EE——— \\“h-.. min 2 . 4 : | C Fig. 23 012 gas Fig. 24 HC1 gas Packing mat.: Packing mat.: n-Hexadecane n-Hexadecane Te = 40°C Te = 40°C n-Hexane Packing mat.: Carbopack B-HT Te = 100°C Fc = 48 ml/min. f | . I = ‘ | o ' | o 54 (=2 Ea L w -] N [40) - | y & Mmn ’100‘\:-4 20 man 0 ki 9t °3ed G¢2-TH-NWL - 45— — —— [l | | n oG N o B — e » - : S i e———ty i - ‘*"x—” ‘ L w ~ | - = ‘ | g OO |- — 1_ : —_—-[—-’ === = Fie) 20 Progra ) TEMPERATURE R - &) WITHUT ECTION - - B) N-HEMNE, | +— —--——:~fc == : ‘mt*lM[Nl* e S E—1— o o | = s — R — * » 1 = i e \ 3 | ‘ =~ lh\ w ' _L — - ~ w - fl =+ ¥ : : | [ — = (-] N