Crumpler and Claiborne outlined the method used in the preparation of the amidinothiourea derivative. They reported that the precipitate formed is colloidal silver (1). I n the present study, when a 1% aqueous solution of silver nitrate was used with amidinothiourea, the silver was immediately reduced to the metal. Under these conditions amidinourea formed a stable compound. Palladium Derivatives. The palladium derivatives of amidinourea and amidinothiourea were prepared; however, the results of the analyses were inconclusive.
ACKNOWLEDGMENT
The author thanks D. C. Sievers and K. J. Fraley for their helpful suggestions and H. S. Pridgen, Jr., for preparing the 1-amidino-2-thiourea. The author also thanks Tennessee Eastman Co. for the cooperation of the Research Laboratories in completing this project. LITERATURE CITED
(1) Crumpler, T. B., Claiborne, E. B., VwginiaJ. Sci. 3, (No. l), 29-30 (1942).
(2) Feigl, Fritz, “Chemistry of Specific, Selective, and Sensitive Reactions,” pp. 205-8, Academic Press, Inc., New York, 1949. (3) Hillebrand, W. F., Lundel!, G. E. F., “Applied Inorganic Analysis," Wiley, New York, 1953. (4) Hoste, J., Anal. Chim. Acta 4, 23-7 (1950). (5) Steyermark, A,, “Quantitative Or-
ganic Microanalysis,” BlakLton, New
York, 1951.
RECEIVED for review January 12, 1962. Accepted August 1, 1962. Work done aa part of the Kingsport Science Seminar affiliated with the Joe Berg Foundation.
Determination of Oxygen in Mixed Fluorides by Inert Gas Fusion J. LEON POTTER, JAMES E. MURPHY, and HOWARD H. HEADY Reno Metallurgy Research Cenfer, Bureau o f Mines, U . S. Department of fhe Inferior, Reno, Nev. An inert gas fusion procedure is described for determining oxygen in mixed fluoride electrolytes used in electrowinning uranium and the rare earth metals. Main factors discussed: the temperature required to remove oxygen quantitatively from various fluorides, a water and acid trap to entrain the volatilized fluorides, a sample preparation technique for heterogeneous fluorides, and use of a dry crucible instead of a platinum flux. Relative error of the method in the 0.05 to 0.5% range is about 5%.
R
in molten salt electrolysis behavior of uranium and the rare earth metals and investigations to determine optimum conditions for electrowinning the metals from oxides in various electrolytes are in progress. Initial studies have resulted in an improved electrowinning process for producing higher purity cerium (8). A challenging analytical problem inherent in this research is the determination of oxygen in the fused fluoride baths, before and after the electrowinning operations. The importance and the need for an oxygen analysis method are emphasized by the research efforts of other laboratories, which have resulted in the publication during the past year of reports describing three different analytical techniques-vacuum distillation ( 6 ) , high temperature fluorination with potassium bromotetrafluoride ( d ) , and inert gas fusion ( I ) . Apparently, any one of these techniques could be applied to the analysis of oxygen in mixed fluoride salts. However, the inert gas fusion method has ESEARCH
obvious advantages in speed, simplicity, and convenience of analysis. Presented is an inert gas fusion technique similar to that described by Banks, O’Laughlin, and Kamin (1). The innovations are in sample preparation, elimination of platinum flux, and use of a water and an acid trap to remove volatilized fluorine compounds. EXPERIMENTAL
The basic instrument used was a Leco oxygen analyzer consisting of a n induction furnace and conductometric analyzer. Argon of 99.99575 purity, further purified by gettering and drying, was used to sweep the evolved gases out of the reaction chamber. The carbon monoxide liberated by the reaction of the sample with the graphite crucible was carried by the argon through ascarite to remove any acidic gases, and through heated iodine pentoxide for conversion to carbon dioxide. The COzwas passed into a Ba(OH)2 solution causing a change in conductance, which was measured by a Wheatstone bridge. A detailed description, including a schematic diagram of the Leco unit, accessory equipment, and reagents used, is given in the literature (1). The only addition made to the Leco system was a series of traps, to remove respectiveljvolatilized fluorides and any entrained moisture from the gas stream. A coarser frit was used in the conductometric cell t o decrease the possibility of frothing. A mixer mill, with an alundum vial and balls, was used to pulverize the fluorides. A 15-mm. diameter, evacuable, stainless steel die and a hydraulic press, capable of developing 20,000 pounds pressure, were used to produce the desired pellets. A graphite reacApparatus.
tion crucible of the following dimensions: 7/g inch o.d., 6 / * inch i.d., 21/16 inches long, and 17/18 inches deep was held in a silica cup packed with a special grade of small, granular carbon black. An optical pyrometer having a n accuracy of h 2 5 ” C. was used to measure the furnace temperature. Platinum sheet 0.002 inch thick, containing less than 10 p.p.m. oxygen, was cut into squares of the desired size and shaped to hold the sample. Outgassing and Calibration. To outgas the equipment, the furnace temperature was raised rapidly to 2300” C., maintained for 30 minutes, and then lowered to the operating temperature of 2050“ C. T h e oxygen blank was checked by observing the change in conductance t h a t occurred in a 4-minute reaction period. About 15 to 30 minutes was required to attain an acceptable blank equivalent to 10 t o 30 pg, of oxygen. Aliquots of standard potassium acid phthalate solution were p i y t t e d into tin capsules, dried a t 110 C., and analyzed. A calibration curve was prepared by plotting resistance us. oxygen concentration. This fast, simple, calibration technique eliminated the use of microbalance to prepare accurate metal oxide standards. A similar calibration, based on the use of a fluoride mixture containing added amounts of oxygen, gave essentially the same calibration curve. PROCEDURE
Approximately 5 t o 10 grams of fluoride electrolyte is placed in an alundum vial and pulverized for 10 minutes in a mixer mill. The powder is dried in a vacuum oven a t 110” C. for about 2 hours, and approximately 2 grams is placed in a die, ~ h i c his then VOL. 34, NO. 12, NOVEMBER 1962
e
1635
evacuated to a pressure of about 10 microns. The sample is compacted at a pressure of 20,000 pounds, and the pellet i. stored in a desiccator until analyzed. The furnace, after being outgassed, is set a t the operating temperature of 2050" C. The crucible and glassware are conditioned by introducing 300 to 500 mg. of solid fluoride into the hot crucible. The evolved fluorides apparently form a film on the glass apparatus to prevent surface absorption of carbon monoxide. A fluoride sample, having a known oxygen content, is then run to check the calibration. The pelletized sample is cut to obtain the desired weight of 25 t o 100 mg. for samples containing 0.05 to 0.5% oxygen. The weighed sample is dropped into the reaction crucible, and the evolved gases are swept through the traps into the conductance cell by argon flowing a t a rate of 400 to 500 ml. per minute. The change in conductance caused by the CO, being precipitated in the barium hydroxide solution is balanced a i t h the Wheatstone bridge. ilfter a h i i n u t e reaction period, the observed change in conductance is converted to niicrograms of oxygen by use of the cnlihration curve. DISCUSSION
Temperature. Thermodynamic d a t a were used to determine the approximate temperature required for the carbothermal reduction of metal oxides in a fluoride medium. The basic mechanism was assumed to be the reaction of metal oxide with carbon to form carbon monoxide and metal. Sloman and Harvey (10) sholyed this reaction was more feasible than one resulting in the formation of metal carbide and carbon monoxide. The possibility of a metal oxyfluoride reaction with carbon was not considered because this type compound is presumably less stable than the respective metal oxide. Shown in Table I are free energy equations for several oxide reduction reactions presumed to be associated with the indicated three-component
Table 1.
Thermodynamic Data for Selected Fluoride Electrolytes
Proposed oxide reduction reactions
Electrolytp CeFt-I,iF-BaFz YF8-LiF-BaF2 CF4-LiF-RaF, UF1-BaF?-CaF2
fluoride systems. The thermochemical data were obtained from Kubaschewski and Evans (7) and Glassner ( 2 ) . By use of Sloman and Harvey's (IO) technique, in conjunction with these free energy equations, the reaction temperatures shown in column A of Table I were calculated for the listed oxide reduction reactions. HoIvever, this approach does not consider the possibility of interaction effects owing to the presence of a small amount of metal oxide in a predominantly metal fluoride flux. To compensate for this factor, consider the activities of the reactants. *lssurning the oxygen content of the fluoride to be a few tenths per cent, the metal ovide concentration would be about 1% or 0.01 mole fraction. By Henry's law, the activity of the metal oxide would also be approximately 0.01 because the activity coefficient n ould presumably approach unity a t high temperatures ( 3 ) . Calculated reaction temperatures, based on 0.01 rather than unit activity, are shown in column B of Table I. These data indicated that a temperature of 1800" to 2100" C. n-ould be required to remove oyygen completely from the fluorides. Several three-component fluorides, containing knov n added amounts of oxide, were run ten times each at varions temperatures ranging from 1800" to 2200" C. to determine the effect of temperature on oxygen recovery. The associated factor of reaction time was also evaluated by making two series of runs a t 4 to 8 minutes over the same temperature range. A reaction time of 4 minutes a t an operating temperature of 2050" C. was adequate for any of the described fluoride systems. The operating temperature could be lowered somewhat by increasing the reaction time. For example, a t a reaction time of 8 minutes, a n operating temperature of about 1950" C. would be adequate. As the reaction temperature increased, the oxygen blank also increased to give values of 12, 30, and 6 0 g. ~ a t 1900", 2050", and 2100" C., respectively. This
+ 1/2(Ce02)s 1/,Ce, + (CO), C, + (BaO), Ra, + CO), C, + 1/3(Y203)a ] Y1 + (COh C, + 1/2(U02)s= l/zU1 +
C,
=
=
2/3
(COL C, + (CaO),
=
Ca1 + (CO),
Calcd. reaction temp., ' C.
Free energy equations"
izb
Bb
103,100 - 437'
1630 1800
AF = 109,300 - 48T AF = 125,300 - 487'
1580 1920 1840 1950
AF
AF
=
=
105,XOO - 437'
A F = 163,400
-
687'
1670
1850
1800 2070
Free energy is expressed in calories and "K. * Temperatures in columns A and B were calculated on the basis of a metal oxide activity of 1 and 0.01, respectively.
1636
ANALYTICAL CHEMISTRY
effect v a s apparently due to the reaction of carbon black with the quartz glassware, according to the proposed reaction Si02
+ 2C
-+
Si
+ 2C0
At temperatures above 2100" C. the blank increased rapidly, causing high and erratic oxygen results. Fluoride Trap. h t high teniperatures t h e fluorides volatilized rapidly and reacted t o some extent with t h e glassware according to the proposed reaction 4MeF3 f 3SiOz
+
3SiF4
+ 2RIe201
The SiF4 reacted to form a BaF2 y e cipitate in the barium hydroxide solution, which changed the conductance and caused high, erratic results. -41~0, the SiF, attacked and clogged the glass frit in the conductometric cell. -4fter a series of runs, without a fluoride trap, fluorine was detected chemically in the barium hydroxide solution. Small amounts of hydrofluoric acid pipetted directly into the conductometric cell caused the resistance to change linearly Tyith fluorine concentratioii. &in addition of 0 1 mg. of fluorine g a ~ ea resistance increasr of 2.0 ohms, corresponding to about 0.057, oxygen for a 100-mg. sample. In analyzing oxygen in YF3, Banks, O'Laughlin. and Kamin (1) uqed a mixture of niagnesiuni oxide and ascarite as a trap to remove the volatilized fluorides. -1 similar system proved successful, but when many samples I\ ere analyzed, i t was necesqary t o change the trap daily. .Ilso, the fluoride gases were not completelretained if the magnesium oxide naq not carefully packed. -4system was required to remove fluorides effectively, but which would not affect the CO being evolved or need frequent attention and replacement. -4trap consisting of two interconnected 500-ml. flasks was installed between the furnace outlet and the ascarite tube. Water in the first flask dissolved or decomposed the volatilized fluoride, and concentrated sulfuric acid in the second flask removed any moisture from the argon gas stream. X safety valve was installed between the furnace outlet and water flask to prevent water from being accidentally sucked into the hot furnace. Standard potassium acid phthalate samples analyzed, with and without the trap, gave identical oxygen results. The mater and acid trap could be used for several weeks without being changed. Even after prolonged use, fluorine could not be detected chemically in the barium hydroxide, but considerable Si02 accumulated in the water trap. Pelletizing. T h e samples of fluoride electrolyte were frequently hetero-
g('ii(v~usowing to a laycr formation t h a t occurred when tlw molten fluoride coiitaining escws nietal oxide was cooled and solidificd. The osygen content of t,he layc>rs varied widely, as shown in Table 11. a n d there were also considerable intralayer variations, as sho\vn by the relative standard dt>i-iritions for thcw anal: C,'liuriks of electrolyte w r e placed in :ti1 alundum vial cont,aining alundum h:dk anti pulverized in a Miser Mill for 10 minutes t o ~irodueea homogeneous fluoride 1)owder having a particle size of about 100 microns. The pox-der Tvas dried in :t vacuum oven for about 2 hours at 110" C . to remove any moisture absorbed during pulverization. I'on-derrrl samples ciiiilti not be convenic>ntly introduced into tlie reaction chamber n-ithout, first heing encapsuEsperiments lat,ctl or pelletized. shoived that if the fluoride po\rder was compressed in a I-inch die a t pressures up to 30,000 pounds, the pellet \vas thin and very easily broken. Pellets produced in .a '/?-inch die with 20,000 pounds pressure were sufficiently durable and homogeneous, as demonstrated by the relative standard deviations listed in Table 11. I n one experiment, samples \\-ere nithdrawn from the various niolten electrolytes by a special filter stick procedure ( 9 ) in an attempt t o obtain homogeneous samples. Analyses of the electrolytes, given in Table 11, did not shon. significant differences in oxygen content betneen the solid and pelletized samples. However, the high deviation in analyses of the solid electrolyte shon-ed that the samples were not homogeneous. Platinum Flux. The advantages of a platinum flux in osygen analysis of metals have been presented by Hansen and coworkers ( 5 ) . Its use has also been reconimeiided for determining oxygen in yttrium fluoride (1, 6j. The platinum functions as a reaction medium. thus providing an efiectivc contact betxeen the crucible and the oxide in the sample. In initial studies, a platinum sheet 0.002 inch thick n-as cut into squares wigliirig about 0.5 gram and formed int,o capsules. The fluoride samples were weighed directly into the capsules, n-hich were then crimped shut and introduced into the reaction chamber. Comparative analyses of fluoride powder arid pellets gave results shown in Table 111. The consistently lower oxygen values obtained for t8he pcllcts were presumably due to the decrease in contamination by surface adsorption and physical entrapment of oxygen. I h n k s and colvorkers (I) reported that considerably less plat'inuni was required to analyze ytt'riuni fluoride thmi yttrium metal because most of the fluoridr was distilled from the crucible. In tlie three-component fluoride elec-
II.
Table
Effect of Pelletizing on Oxygen Analysis of Fluoride Electrolyte
Solid chunks Re1 std. Oxjgen, m,5 dev., % b
Electrolj te C ~ F ~ - B ~ F L - I , IconF taining CeO? Top Layer Middle Layer Bottom Layer Homogeneous electrolyte LaFa-RaF2-LiF UFd-BaF?-CaF, CeF3-Ba$2-LiJf a
Pellets Oxygen,
0 6S 5 20 9 73
4 72 4 so 3 60
10 11
0 25 0 3?l
8 76: 5 00 :3 34
0 24 0 35 0 56
0 5s
0 62 4
0 40 2 22 2 00
50
0 83 1 82 0 50
Average of SIX determinations. (Standard deviation X 100)/me:in
Table 111.
Comparison of Fluoride Powders and Pellets Using a Platinum Flux
I'order Electrol3 te l;Fd-RaF9-I,iF - ~" CeFa-BaFs-LiF YF3-BaF2-LiF ~
a
Rel.tddev., ycb
A v o\jgen,
Pellets Rel. std. dev , sc
Av. osygen, 9;"
Ilel. std. dev., c &
1 40 3 33 6 25
0 256 0 333 0 134
2 73 1 70 2 58
0 2%
0 384 0 168
Average of six determinations.
Table IV.
Comparison of Platinum Flux and Dry Crucible Methods of Analysis
Electrolyte UFI-BaFz-CaF? LaFa-BaFz-LiF CeFs-BaFz-LiF YF3-BaF?-LiF
Platinum flux Rel. std. .Iv. oxygen, r c dev., % 2 80 0 281 0 257 0 260 0 13G
trolytes studied, nearly complete volatilization occurred rapidly a t the selected operating temperature of 2050" C. The solid chunk of fluoride, nhen dropped into the hot reaction chamber, was observed to form inimediately a molten ball that vibrated vigorously and appeared to explode into a cloud of vapor. Consequently, it seemed reasonable t h a t the brief solid-liquid-gas transition might, of itself, serve the same function as the platinum fluu in providing an adequate reaction medium for conversion of the sample oxide to carbon monoxide. Several electrolytes were run six times each, with and TT ithout a platinum flux. The osygen recoveries as shown in Table IV established that the use of a flus v a s not necessary. Elimination of the platinum afforded several advantages: The dry crucible could be used for much longer periods without change because there was no buildup of platinum residues, the inconvenience and the cost of reclaiming the platinum could bc eliniinated, and considerable time could be saved in the sample preparation procedure.
3 54 2 61 4 71
Ilry crucible
.4v.oxygen, 70
0 288-1 0 254 0 263 0 132
Rel. s t z dev . 'I, 1 96 2 19
2 08 4 29
RESULTS
Seven different ternary fluoride electrolytes nere analyzed ten times each t o measure the precision of the described procedure. The oxygen results ranged from 0.040 to 0.324%, and those values viere considered to be the rcsidual oxygen contents of the respective electrolytes. The relative standard deviations ranged from 1.30 to 2.84Yc showing that good precision was obtained for all of the elcctrolytes. Knon-n amounts of nietal ouide, corresponding to the major metal fluoride, were added to the various ternary electrolytes to obtain standards for an accuracy evaluation. The standards were pulverized, dried, pelletized, and analyzed ten times each in the prescribed manner. The results in Table V shoi\ that a similar accuracy v-as achieved for each electrolyte, and the relative error varied from - 5 to +6%. This method of preparing standards was satisfactory for all electrolytes except where lanthanum fluoride \+-as the major component. Because of the VOL. 34, NO. 12, NOVEMBER 1962
1637
~~
Table V.
Electrolyte UFd-BaFg-CaFz UF,-B aF2-C aFz UFa-BaFn-SrF? UFi-BaF;SrFi UF4-BaF2-LiF UF4-BaF2-LiF CeF3-LiF-BaF2 CeF3-LiF-BaF2 YFX-LiF-BaFv YFlLiF-BaF2LaFs-LiF-BaF2 LaF8-LiF-BaFz
Accuracy Tests on Mixed Fluoride Standards
Residual 0.189 0,189 0.321 0.321 0,227 0.227 0.180 0 243 0 128 0 128 0 324 0 324
Oxygen, % Added 0.100 0.500 0.100
0,500 0,100 0.500 0.050 0.300 0 100 0 500 0 100 0 500
Rel. error Recovereda 0.095 0.500 0.106
0.iii
%b
-5.0
0.0
+6.0 i2.2
0.101 $1.0 0.504 $0.8 0.052 +4.0 0.290 -3.3 0 102 +2.0 0 508 t-1 6 0 104 $4 0 0 504 +O 8 Values have been corrected for residual oxygen, average of 10 determinations. Relative error = 100 (mean - known)/known.
high affinity of La203for mater, it was necessary to prepare the lanthanum standards in a dry atmosphere. The weighed oxide was ignited a t 1000° C. and, while red hot, was placed in a controlled atmosphere chamber. The standards were mixed and fused a t 1000” C. in this chamber. Even after long exposure a t room atmosphere, the fused standards did not show any detectable increase in oxygen content. The oxygen content of most ternary electrolytes, as well as the individual metal fluorides, usually exceeded O.lY0. Consequently, no special handling techniques were required t o achieve a suitable degree of accuracy. To analyze
fluorides containing lesser amountct of oxygen, i t would be necessary to grind, weigh, and encapsulate the powder in platinum before removing the sample from the inert atmosphere chamber. One of the ternary fluorides was subjected to special electrolysis conditions to deplete deliberately the oxygen content. Because of the extreme hardness of the sample, contamination from the alundum mixing vial occurred during pulverization causing the oxygen results to be abnormally high. =Ifter this problem was eliminated by using a stainless steel vial, an oxygen analysis of 0.04y0 was obtained. Prolonged
pulverization in alundum and stainless steel vials showed that, for samples pulverized in excess of about 15 minutes, there was a gradual increase in oxygen owing t o the alundum contamination whereas the stainless steel had no effect. However, in both cases, pulverization increased the oxygen content unless the samples were oven dried prior to analysis. LITERATURE CITED
(1) Banks, C.
V., O’Laughlin, J. W., Kamin, G. J., -49.4~.CHEM.32, 1613 (1960). (2) Glassner, A , U. S. Atomic Energy Commission ANL-5750 (1959). (3) Glasstone, S., “Textboo_k of Physical Chemistry,” 2nd ed., p. (13, Van Nostrand, New York, 1948. (4) Goldberg, G., hleyer, A. S., Jr., White, J. C., ANAL.CHEW32, 314 (1960). (5) Hansen, W. R., Mallett, M. W., Trzeciak, M. J., Ibid., 31, 1237 (1959). (6) Horrigan, V. M., Fassel, V. .4., Goetzinger, J. W., Ibzd., 32, i87 (1960). ( 7 ) Kubmchewski, O., Evans, E. L., “Metallurgical Thermochemistry,” pp. 36, 336, Pergamon, New York, 1958. (8) Morrice, E., Darrah, J., Brown, E , Wyche, C., Hedrick, W., Williams, R., Knickerbocker, R. G., Bur. Mznes Rept. Invest. 5549 (1960). (9) Porter, B., Brown, E. A., Ibzd., 5878 (1961). (10) Sloman, H. A , , Harvey, C. A,, J . Inst. Metals 80, 392 (1951). RECEIVEDfor review June 8, 1962. hccepted September 1962. Reference to specific makes or models of equipment is made to facilitate understanding and does not imply endorsement by the Bureau of Mines.
Action of Perchloric Acid and Perchloric Acid Plus Periodic Acid on Ammonia and Amino Nitrogen FRANCIS B. MOORE’ and HARVEY DIEHL Department of Chemistry, lowa State University, Ames, lowa Ammonia is not oxidized b y boiling perchloric acid (the 203” C. boiling azeotrope, HC104.2HzO) nor is it oxidized by boiling perchloric acid plus periodic acid plus vanadic acid. Earlier troubles in the application of perchloric acid to the Kjeldahl digestion have been traced to the action of chlorine (hypochlorous acid) on the ammonia after dilution with water. By making provision for the prompt removal of chlorine, quantitative recovery of ammonia can b e obtained. When the perchloric acid-periodic acid digestion was applied to the determination of nitrogen, good results were obtained on acetanilide, trishydroxymethylaminomethane, and wheat, and with suitable modifications, on Nbromosuccinimide and azobenrene. Good results were obtained on urea
1638
ANALYTICAL CHEMISTRY
and chloroacetamide only when the periodic acid was omitted. Results on 8-quinolino1, linseed oil meal, and nylon were unsatisfactory. The perchloric acid-periodic acid digestion is an excellent method for the wet oxidation of organic materials but it does not furnish a generally applicable method for the determination of nitrogen.
T
HE VERY NUMBER of the papers dealing with the Kjeldahl determination of amino nitrogen attests the difficulties inherent to its application and the qualified precision and accuracy of the results. Recent success ( 7 ) in the destruction of organic matter with a mixture of periodic acid and perchloric acid led us to adapt this mixture to
the determination of amino nitrogen, presuming as in the classical Kjeldahl procedure that all of the nitrogen is retained in the minus three state and yields ammonia on treatment with fixed alkali. Perchloric acid has been used previously in the Kjeldahl method, both alone and mixed with sulfuric acid t o effect the digestion (Q), and also with questionable success when added dropwise during the last stage of the conventional digestion n-ith sulfuric acid as an aid in completing the destruction of organic matter (3, 8); a review of the subject is presented by Bradstreet ( 2 ) . The action of periodic acid on aliphatic amino compounds yields ammonia al1 Dept. of Chemistry, University of Minnesota, Duluth, Minn. On sabbatical leave 1961-62.
