were added as ammonium; potassium, or sodium salts. Table I11 summarizes the results of this study. The interference of phosphate, arsenate, and iron(II1) is an especially serious limitation of the method. KOattempts were made t o remove phosphate or arsenate by a preliminary extraction with ethyl acetate by the DeSesa and Rogers method (4), or t o circumvent the deleterious effect of iron(II1). -1Ithoug.h less than 100 p.p.m. of either the fluoborate or the borate ion exhibits negligible absorbance a t 230 mp, the effect of these ions on the specificity and efficiency of extraction of molybdosilicic acid was not studied. Therefore, their possible deleterious effect on the extraction process would ha1 e to be ascertained before this method could be applied to samples in which the solubilization of silicon had been achieved by the fluoride method.
Reproducibility. An indication of t h e precision of this procedure was ascertained from the results of 12 samples, each containing 0.2 p.p.m. of silicon. These samples gave mean absorbance values of 0.392 and 0.792 at 230 and 210 mp, respectively. The standard deviations were 0.006 absorbance unit, or a relative standard deviation of 1.5%upport in the form of an SSF Summer Re;;earch
Fellowship for High School Chemistry Teachers. LITERATURE CITED
(1) Boltz. D. F.. Mellon. RI. G.. IXD. ’ ESG.CHEU.,AXAL. ED.i9,8T3 (1947). (2) Case, 0. P., Ibzd., 16,300 11944). (3) De Sesa, 11.A , Rogers, L B., .4h.4~. CHEY.26,1278 (1954) (4) Ibid., p. 1381. ( 5 ) Jolles, A4.,Seurath. F.. Z . anaew. Chem. 11,315 (1898). 16) Judav. C.. Meloche. \ , Iinudson. H. IT.. V. IV.! ISD:ESG.CHEII.;hs.;~. ED. 12; 270 (1950). (7) h e c k , C. H., Boltz, D. F., A 4 s a ~ . CHEU.30,183 (1958). (8) Milton, R. F., dnaly.?t 76,431 (1951). 19) Strickland. J. D. H.. J . din. Chern. SOC.74.872 il952) ~
RECEITED for review Soveiiiber 21) 1962, Accepted September 13, 1963. Division of Analytical Chemistr)., l 4 k d Meeting, ACS, Atlantic City, S . J . Septeniberjl062.
Visible Absorption Characteristics of the Bis-(2,9dimet hy I- 1,lO- phena nt hroline)- a nd Bis- (4,4’,6,6‘tetramethyl-2,2‘-bipyridine)-Copper(I) Ions J.
R. HALL, M. R. LITZOW, and R. A. PLOWMAN
Chemisfry Department, University o f Queensland, Queensland, Australia
b Spectrophotometric studies have indicated that the absorbing species in solution for the determination of copper using 2,9-dimethyl- 1 , I 0-phenanthroline (dmp) and 4,4’,6,6’-tetramethy1-2,2’bipyridine (tmb) are the bis complex ions, [Cu(dmp)~]+ and [Cu(tmb)n] +, respectively. Solutions of the pure compounds, [Cu(ligand)s]X where X = CI, Br, I, Nos, and CI04, have spectral characteristics in agreement with the earlier studies. In general, however, the solutions conform to Beer’s law only when a large excess of the ligand i s added. Deviations from Beer’s law in the absence of excess ligand are attributed to dissociation of the [Cu(ligand)*]X complexes to the corresponding monochelate species. These ligands coordinate to many transition metals and their apparent specificity for copper in extraction procedures i s probably due to complexes of other metals having wavelengths of maximum absorption well removed from the A,, values for the copper complexes, or much lower molar absorptivities.
R
have been published about the use of 2,9-dimethyl-l,lO-phenanthroline (dmp; trivial name, neocuproine) for the determination of EPORTS
2124
ANALYTICAL CHEMISTRY
copper (8,12). n’hile the reagent is regarded as specific for copper, i t is not \Tell known that the ligand forms complexes with a number of transition metal ions. An analogous reagent, 4,4‘,6,6’tetramethyl-2,2‘-bipy~idine(tmb), has also been suggested for the determination of copper (9). This article describes the absorption behavior of solutions of pure compounds containing the bis(dmp)- and bis(tmb)-copper(1) ions and reports that some complexes may be found that interfere with the determination of copper using these reagents. EXPERIMENTAL
Apparatus. A Unicani SP500 spectrophotometer was used for the absorption measurement.. The wavelength and absorbance *calm were checked against standard solutions, d a t a for which are available from the IT. S. Department of Commerce ( 1 1 ) . Reagents. The compounds CuX t m b , where X = C1 and Rr. and [ C ~ ( t n i b ) ~ ] nSh, e r e X = S O n and Clod, were prepared by rrduction of an aqueous solution of the corresponding copper(I1) complex with hydrazine sulfate. The copprr(I1) compleves were isolated by method5 similar to those previously described for the analogous dmp complekeq ( 6 ) . CuItmb was obtained by the reaction of Cu-
Cltmb with excess sodium iodide dissolved in acetone. The compounds [Cu(tmb)n]X.HzO, where X = C1 and Br, n-ere prepared,by refluxing an ethanol solution of the corresponding mono(tmb) complex and tmb, while [ C ~ ( t n i b ) ~was ] I formed by adding cuprous iodide to an ethanol solution of excess tmb. The compounds lICl,tmb, where M = Fe, Co, and S i , n-ere prepared by addition of tmb to exce;s metal chloride dissolved in methanol (for FeC12) or ethanol (for CoClzand SiCI,). All the compounds were analyzed for metal, carbon, hydrogen, and nitrogen. The samples used for the absorption experiments were analytically pure. The solvents were Baker analyzed reagent chloroform and l n a l a r isoamyl alcohol. The solution, n-ere examined in matched, stoppered I-cm. cell.?. RESULTS A N D DISCUSSION
The procedure generally adopted for the determination of copper is to reduce an aqueous solution of copper(I1) to copper(I), t,hen add esceij neocuproine, and finally extract n i t h i;oamyl alcohol (I+$) or chloroform (5). The extracts show maximum absorption in the visible a t about 455 m p (molar absorptivity E = 7950 i 100) and the absorbing species has been determined spect’rophotometrically ( 1 4 ) to lie [Cu(dmp),]+.
Wavelength ( m y ) Figure 1. cess tmb
Absorption curves of [Cu(tmb)z]NO3 in isoamyl alcohol containing ex-
Complexes of the type [ C ~ ( d m p ) ~ ] X , where X = so$,Clod, C1, Br, and I and 2 X = SO,, have been isolated (6); these compounds in isoamyl alcohol and e q i a l t o 454 and chloroform show, , ,A 457 mp, respectively, in agreement with the earlier work (5, 14). We found that isoamyl alcohol was the better solvent for the pure compounds, since their solutions conformed to 13eer's law and yielded E values aboLt 7900. The chloroform solutions on i,he other hand conformed to Beer's law only when the anion TT-as nitrate or perchlorate. The molar absorptivities obtained for chloroform solutions of the halides were very low, decreasing with both age of the solutions and dilution. However, the addition of a sixfold excess of dmp elevated E to i900 and the solutions were stable and conformed to Beer's law in the range studied (1 to 12 p.p.m.). The low e values in the absenw of excess dmp were explained (6) by the dissociation of the bis complexes to the mono complexes, CuXdmp. One of the often repori ed advantages of the use of dnip for copper analysis is its alleged specificity for that metal. This conclusion has been drawn after tests which have shown t i a t addition of ions of other metals such as iron, nickel, mercury, etc., even in large excess to the copper present, causes no apparent interference (8). While i t is well known that 1,lO-phenanthrolinc: readily coordinates to metal ion:> (I)-for example, iron(I1) salts yield tris(che1ate) complexes n.hich are intensely colored (ferroin reaction)-the coordinating ability of dmp has been little investigated. Dmp nhen added to solutions of iron(I1) salts does not produce intensely colored qolutions (2, 7 ) and it has been inferred that t h k molecule does not form complexes with iron(I1). However, mono- and klis(dmp) com-
plexes of iron(I1) have been isolated and characterized (4). Dmp coordinates to a variety of metal ions, besides Fe+", forming both mono and bis complexes. Some of these substances have relatively intense colors and are capable of extraction into organic solvents. The apparent specificity of this reagent for copper is probably the result of other metal complexes having wavelengths of maximum absorption well removed from 455 mp, with relatively low E values. For example, chloroform solutions of CoC12dmp and NiBr2dmp have, , ,A ( E ) equal to 660 mp (493) and 514 mp (215), respectively. Bis(che1ate) complexes of these metals have lower absorptivities (about 50) and are unlikely to interfere (3) The analytical method for the determination of copper, using the reagent 4,4',6,6 - tetramethyl - 2,2' - bipyridine (tmb), depends on the measurement of absorption a t 455 mp which, according to spectrophotometric studies (IO), is due to the [Cu(trnb)~]+ion. T m b has a sensitivity toward copper similar to that of dmp, but its usefulness has not been studied to the same extent. Complexes of the type [ C ~ ( t m b ) ~ ] X where , X = Koa, Clod, C1, Br, and I, have been prepared. Table I summarizes the ab-
Table 1.
sorption characteristics of these compounds in the solvents chloroform and isoamyl alcohol and the effect of the addition of excess tmb. The chloroform solutions of the halide compounds required the addition of a very large excess of t m b (200 times the molar concentration of the complex) before the molar absorptivities reached the maximum values shown in the table. I n all other cases the excess t m b amounted to 12 times the concentration of the complex. Figure 1 illustrates the visible absorption curve of the [Cu(tmb)*]+ ion at various concentrations. T m b behaves like dmp as a reagent for copper(1). Isoamyl alcohol is the better solvent for the pure compounds. Only when excess t m b is present do the solutions in general conform to Beer's law (range studied, 1 to 10 p.p.m.) and produce molar absorptivities approaching the value 6800 reported in the literature (IS). We have not recorded a t any time a value as high as 8100, which was reported by Linnell and -Vanfredi (IO). We repeated their extraction experiment and obtained an average value of 6590 in isoamyl alcohol (6510 in CHC1,). The low values for solutions of the halide compounds in the absence of excess tmb are doubtless due to dissociation to the CuXtmb mono complexes. The substances CuXtmb, where X = C1, Br, and I, have been prepared and characterized. CuCltmb was shown by freezing-point depression measurements to be dimeric in camphor. Low solubilities of the compounds in organic solvents did not permit other determinations to be made. I n support of the suggested dissociation 2 [ C ~ ( t r n b ) ~ ]e X [CuXtmb]2
+
2tmb
we found that a chloroform solution of the compound [CuCltmbIz absorbed weakly a t 454 mp but on addition of a large excess of t m b the yellow solution became orange (A,, = 454 mk) and, assuming complete conversion to the bis complex, yielded an E value of 5710 at that wavelength. The CuNO3tmb and CuCIOltmb complexes could not be prepared. Attempts to do so always resulted in the production of the bis complexes. This negative result would indicate that S O 3 - and C104- anions, unlike the halide ions, are reluctant to
Visible Absorption Characteristics of [ Cu(tmb)z]X Compounds
Chloroform
(Amax = 454 mu)
Compound [Cu(tmb)2?$03 [ C ~ ( t m b Clod )~] [Cu(tmb)rlC1.H20
[ C ~ ( t m b Br ) ~.]H20 [Cu(tmbMI Do not conform to Beer's law.
E t
6180 5780 1100" 1850" 1310"
(excess tmb) 6300 6330 6000 6100 6230
Isoamyl alcohol (A,,=
=
453 mu)E
E
6550 6600 6180" 5630a 5830"
VOL 35, NO. 13, DECEMBER 1963
(excess tmb ) 6700 6600 6600 6550 6630
2125
Table II. Visible Absorption Characteristics of Some tmb-Metal Complexes in Chloroform Solution Compound A=, mp e FeClztmb 417, 445 67, 71 CoClztmb 573, 653 254, 486 SiClztmb 489, 610, 830, 144, 12, 32, 975 75
form covalent bonds with Cu+. If this is so, the normal B values observed for solutions of [Cu(tmb)a]NOa and [Cu(tmb)z]C1O, are to be expected, since the equilibrium between the complex and its dissociation products must lie predominantly in favor of the bis complex. I n these cases the positions of the equilibria are little changed by the mass action effect of excess tmb (see Table I). T m b shows a negative ferroin reaction (9, 13) and is reported not to complex with iron(I1). Our experiments indicate that complexes with iron(I1) are
formed. As for the dmp complexes of iron(II), these are pale yellow solids and their solutions are noticeably colored only when concentrated. It appears that t m b is able to complex with the transition metals generally. Table I1 lists some of the compounds that have been isolated and their spectral characteristics in CHCb solution. The suitability of t m b for the determination of copper in the presence of other transition metals will depend upon the Amax values of their complexes and the intensities of the solutions. Results indicate that other metals will not interfere because of either small e values for the complexes or remoteness of regions of maximum absorption from 454 mp of the [ C ~ ( t m b ) ~ion, ] + just as in the case of dmp. LITERATURE CITED
(1) Brandt, W. W., Dwyer, F. P., Gyarfas, E. C., Chem. Revs. 54,959 (1954). (2) Case, F. H., J . Am. Chem. Soc. 70, 3994 (1948).
(3) Fox, D. B., Ph.D. thesis, University of Queensland, 1963. (4) Fox, D. B., Hall, J. R., Plowman, R. A,, Australian J. Chem. 15, 235 (1962). (5) Gahler, A. R., ANAL.CHEM.26, 577 (1954). (6) Hall, J. R., Marchant, N. K., Plowman, R. A,, Australian J . Chem. 15, 480 (1962). ’ ( 7 ) Irving, H. M., Cabell, M. J., Mellor, D. H., J. Chem. SOC.1953,341i. (8) Jones, P. D., Newman, E. J., Analyst 87,637 (1962). (9) . , Linnell. R. H., J . Ow. Chem 22, 1691 (1957). ‘ (10) Linnell, R. H., Manfredi, D., J . Phys. Chem. 64,497 (1960 ). (11) Natl. Bur. Standards, Letter Circ. LC-1017 (1955). (12) Kewman, E. J., Peters, G., “2,9Dimethyl-1,lO-phenanthroline: Reagent for Copper,’’ Monograph 40,
Hopkin and Williams. Chadwell Heath, Essex. 1960. (13) Schilt, A. A., Smith, G. F., Anal. Chim.Acta 16, 401 (1957). (14) Smith, G. F., McCurdy, W. H., Jr., ANAL.CHEU. 24,371 (1%i2).
RECEIVEDfor review March 29, 1963. Accepted August 26, 1963.
Purity Examination of Silicon and Germanium Halides by Long-Path Infrared Spectrophotometry MYRON J. RAND Bell Telephone Laboratories, Inc., Allenfown, Pa.
b The infrared absorption spectrum of 10-cm. liquid layers of silicon and germanium tetrahalides is capable of detecting many impurities in the 1 - to 200-p.p.m. range. Dissolved gases and volatile compounds, and especially organic contaminants, are seen readily. The technique is also useful for following the progress of purification procedures. Typical long-path spectra of liquid SiC14, GeC14, SiBr4, and GeBrc from 2 to 15 microns are shown, and methods are given for distinguishing impurity absorptions from weak bands of the matrix material. A compilation of common impurities, with absorption maxima, approximate absorpiivities, and detection limits, is also presented. Batches of supposedly hyper pure “semiconductor grade” halides from several suppliers show significant differences. Atmospheric moisture introduces hydroxyl groups and such compounds as Si2OC16 so that exposure is a source of oxygen contamination, even if all the absorbed water reacts with the halide.
T
of the Group IV metals are now used extensively for t h e production of the pure elements by various reduction processes and for the HE HALIDES
2126
ANALYTICAL CHEMISTRY
preparation of pure oxides and pigments by hydrolysis or burning. SiCl,, GeCl,, and T i c & are particularly important, and the Si and Ge chlorides and bromides are of interest in semiconductor technology not only for production of the bulk metals but for epitaxial deposition from the gas phase. For all these uses the purity of the halides is of paramount importance, but these compounds are good solvents for many gases and organic materials, and they are powerful dehydrating agents, Hydrolysis by atmospheric moisture contaminates them with hydrogen halides and oxygen-bearing compounds; the latter are particularly unwelcome, since small amounts of oxygen may seriously degrade the physical and electrical properties of Group IV metals. Little has been published on the detection of impurities in the halides, aside from conventional spectrochemical techniques for foreign metals. It is the purpose of the study reported here to point out the wealth of information obtainable from the infrared absorption spectrum of deep liquid layers of Si and Ge halides. Liquid sample thicknesses of more than 1 mm. are rarely used in infrared work, since most materials would be virtually opaque over most of the 2- to 25-micron analytical
range. The Group IV liquid halides, however, are tetrahedral molecules with quite simple spectra; eren more important, they contain no light atoms, so that all the fundamental absorptions lie beyond 15 microns. Even the first overtones and simple combinations usually lie beyond 10 microns. Consequently these compounds are quite transparent in the 2- to 10-micron range even in path lengths of many centimeters. Organic compounds and other volatile impurities which do have strong absorptions in this range may then be detected in concentrations as low as 1 p.p.m. Halides from various sources, all represented to be of the highest purity-e.g., “semiconductor grade”-have given spectra showing significant differences. Often the exact impurities may be identified with fair confidence. The success of attempted purification procedures may be followed readily. Only one compound, titanium tetrachloride, has been examined extensively by the long-liquid-path technique (4, 15,,24,.%%%9. It is evident from these results, as well as from the present work, that absorption maxima of contaminants ordinarily shift very little upon solution in the halide, generally 0 to 0.1 micron; the same now seems true of the variation
