the quantitative gas chromatographic determination of metals if the two isomers have slightly different retention times. This is especially true if peak heights are measured instead of peak areas, because very slight fluctuations in operating conditions produce marked chmges in the extent of separation in a broadened but unresolved peak, thus altering the peak height. In the limiting case, if all the isomers have substantially different retention times, twice as many peaks appear in the chromatogram. This would also complicate the quantitative determination because peak overlap and interference would be much more likely. Fortunately, none of these anticipated difficulties were encountered in this quantitative study. Column and instrument conditions were chosen that yielded a single sharp peak for each isomeric pair. However, with a different column we were able to demonstrate that the cis and trans isomers of rhodium(II1) trifluoroacetylacetonate can be separated by gas chromatography. Figure 3 is a chromatogram illustrat-
ing the separation of cis-trans Rh(tfa),. The peak area ratios are 79.8% for the trans isomer and 20.2% for the cis compound, in good agreement with the relative yields reported by Fay and Piper ( 2 ) . This indicates that gas chromatography will be a useful tool in studies of the stereochemistry of metal chelates. For example, equilibrium constants for the reaction trans-M(tfa)s e cis-M(tfa)? can be measured. The kinetics of isomerization, ligand substitution, or ligand exchange reactions can also be followed. Furthermore, preparative scale chromatography may be useful for the purification and isolation of samples of geometrical isomerr, of metal complexes. The technique should be particularly effective for separating mixtures of isomers possessing only subtle differences in structures and properties. LITERATURE CITED
(1) Albert, D. K., ANAL. CHEM. 3 6 , 2034 (1964). ( 2 ) Fay, R. C., Piper, T. S., J . Am. Chem. SOC.8 5 , 500 (1963).
(3) Hill, R. D., Gesser, H., J . Gas Chromatog. 1, 10 (1963). (4) Juvet, R. S., Ilurbin, It. P., Ibid., 1, 12 (1963). (5) hioshier, R.W.,Sievers, R. E., ""a; Chromatography of Metal Chelates, Pergamon Press, Oxford, in press, 1965. (6) Ray, N . H., J . A p p l . Chem. (London) 4, 21 (1954). ( 7 ) Ross, UT.I)., ANAL.CHEM.35, 1596 (1963). (8) Ross, W. I)., Wheeler, G., Ibid., 3 6 , 266 11963). (9) Schwarberg, J. E., Moshier, R. W., Walsh, J. H., Talanta 11, 1213 (1964). (10) SieverP, R. E., 16th Annual Summer Symposium, ACS Division of Analytical Chemistry, Tucson, Ariz., June 1963. i l l ) Sievers. It. E.. hloshier. R. IV.. Morris, 11 L., Inorg Chem. 1, 966 i 1962), \
-
-
-
(12) Sievers, R. E., Ponder, B. W,, hlorrix, hl. L., hloshier, R. W., Inorg. Chem. 2 , 693 (1963). WILLIAM 11. Ross hlonsanto Research Corp. Dayton, Ohio ROBERT E. SIEVERS Aerospace Research Laboratories, ARC Wright-Patterson Air Force Base, Ohio GUTHRIE WHEELER, JR. Monsanto Research Corp. Dayton, Ohio
Dimethyl Sulfoxide as a Solvent in the Pyromellitic Dianhydride Method for Alcohols and Amines SIR: Pyromellitic dianhydride (PMDA) has certain definite advantages when compared with acetic anhydride and phthalic anhydride, the other reagents most commonly used to determine alcohols and amines (2). With PMDA, for example, alcohols can be determined in the presence of phenols, aldehydes do not interfere, and the reaction is much faster than with phthalic anhydride and faster than with acetic anhydride under comparable conditions. A difficulty encountered in the use of PMDA is its sparing solubility in the usual solvents, including pyridine,
Table 1.
from which the THF is permitted to escape, concentrating the reaction mixture. Our plant control laboratory personnel have found that dimethyl sulfoxide (DMSO) is a more satisfactory solvent for use in the routine determination of hydroxyl numbers of polyglycol ethers. At the proper dilutions, bothersome precipitation with the resultant bumping during the esterification reaction is eliminated. DMSO was subsequently tried as a solvent for the reaction of alcohols and amines in general, and this resulted in an improved procedure.
Determination of Alcohols and Amines
Purity, 7G New Original PhIDA PXIDA method methoda 99.8, 99.8 9 9 . 8 Methanol 99.6, 9 9 . 7 99.7 2-Propanol 98 8, 98 8 99 3 1-Butanol 100 0 100 0 3-Pentanol 99 5, 99 7 99 8 1-Heptanol loo 0 Triethylene glycol 99 2 99 6, 100 1 99 9 1,2-Propanediol 99 0, 100 0 99 0 I q o bu t ylamine Ili-isobutylamine 99 7 , 100 1 100 1 98 2 2-Napht hylamine 98 3 Ref. (2).
600
the presence of which is necessary to accelerate the reaction. In the procedure of Siggia, Hanna, and Culmo (2), a solution of PMDA in tetrahydrofuran (THF) is mixed with the sample and then pyridine is added. The addition of the pyridine causes some of the anhydride t o precipitate. Although the reaction proceeds rapidly and quantitatively at the boiling point of the THF (65" C.), vigorous bumping IS common, which can result in a loss of some of the reaction mixture, voiding the determination. The reaction can be performed under reflux, but proceeds much more slowly than in an open flask
ANALYTICAL CHEMISTRY
Table II.
Poly G 3030 PG (Triol molecular weight range 3000) Poly G 4031 PG (Triol molecular weight range 4000) a Ref. (2). bRef. ( 1 ) .
Hydroxyl Values of Polyglycols
Hydroxyl value (mg. KOH/g. sample) New Original Phthalic PMDA method PhlDA method" anhydride method6 54.1, 54.6 54.5 54.3, 54.8
41.4, 41.7
41.5, 41.5
41.3, 41 4
EXPERIMENTAL
Reagent. Pyromellitic dianhydride, 0.5.V. T h e pyromellitic dianhydride, 109 grams, is dissolved in 525 ml. of dimethyl sulfoxide, and then 425 ml. of pyridine are added. Procedure. Fifty milliliters of 0.5M pyromellitic dianhydride solution are pipctted into a glass-stoppered 250-ml. flask. h sample c.ontaining 0.010 to 0.015 equivalent of alcohol or amine 14 weighed and added to the reagent. T h e flask is placed on a steam bath and the .topper is wetted with pyridine and loosely seated in the flask. The contents are heated for 15 to 20 minutes (30 minutes for polyglycols). A 20-ml. portion of water is added and the heating is continued for 2 minutes. The mixture is cooled to room temperature and titrated with 1K sodium hydroxide to the phenol-
phthalein end point. A blank in which only the sample is omitted is treated in the same manner. DISCUSSION A N D RESULTS
The results in Tables I and I1 show that the values obtained using DMSO as the solvent are practically identical to those obtained using the original PMDA method. The reaction proceeds smoothly, and the mixture remains clear. Crystals of pyromellitic acid do appear on cooling after the reaction is complete, but these redissolve on neutralization, and the solution is again clear at the end point of the titration. The hydroxyl values for polyglycol ethers in Table I1 demonstrate that results obtained for these materials by the PMDA and phthalation ( 1 ) methods
are equivalent. The P M D A method has the advantage that it requires significantly less time. The PMDA reaction was complete in 30 minutes, but the phthalation reaction required 2 hours' reflux a t the boiling point of pyridine. LITERATURE CITED
(1) Elving, P. J., Warshowsky, B., ANAL.
CHEM.19, 1006 (1947). ( 2 ) Siggia, S., Hanna, J. G., Culmo, R., Zbid., 33, 900 (1961). ROBERTHARPER Olin Mathieson Chemical Corp. Brandenburg, Ky. SIDNEY SIGGIA J. GORDON HANNA Olin Mathieson Chemical Corp. New Haven, Conn.
Background Corrections in Long Path Atomic Absorption Spectrometry SIR: d method has recently been described for providing a much longer effective absorption path than was previously used in atomic absorption analysis (1). The longer absorption path, which is obtained by passing the hydrogen-oxygen flame from a Beckman burner through a Vycor tube, gives 10 to 100 times greater sensitivity than the more conventional slot or multiple port burner for several elements. We have used the method for the determination of calcium, copper, lead, magnesium, manganese, mercury, silver, thallium, and zinc in a variety of samples of interest in biology and agriculture. The higher sensitivity will frequently allow chemical separations to be avoided and is especially valuable when working with very small samples. Applications of this method to the analysis of tissue ash have been described and sensitivity data given for 13 elements ( 2 ) . At these higher sensitivities, absorption by matrix salts a t the wavelength of a n elemental resonance line can cause significant errors. More detailed information on the absorption by matrix salts and a method for correcting for the effect are given here. EXPERIMENTAL
Apparatus. A Beckman Model D E spectrophotometer with photomultiplier attachment was used. This instrument was modified to reverse the direction of light passage through the monochromator and mounted with a n optical bench t o support t h e external components. h Beckman hydrogm lamp and hydrogen lamp power supply were used for the continuous
ultraviolet source. The Hilger Model FA 41.301 supplied power for the hollow cathode lamps. A simple, demountable hollow cathode lamp, which was described by Werner et al. ( S ) , has been used for atomic absorption work ( 2 ) , and makes a versatile, low cost source. This lamp is operated with continuous pumping as commercial argon is bled into it through a needle valve to maintain the desired pressure. Impurities are removed by the flowing argon so that a tedious cleanup of the lamp is not required. The source for a given element is prepared by placing a small amount (50 to 100 mg.) of the metal or a suitable salt into the water-cooled brass cathode. The lamp is operated at a current of 80 to 100 ma. for a few minutes to sputter the added element over the cathode surface. The current is then reduced to the desired operating level (10 to 40 ma.) and the argon pressure adjusted to give maximum intensity of the line being used. The lamp is ready to use after a 10- to 20-minute stabilization period. A cathode is prepared for each test element and can be used repeatedly without further addition of metal. The brass cathode alone is used as a copper and zinc source. Magnesium is the one element for which the commercial sealed lamps provide a more stable source than the demountable ones which we have been able to prepare. PROCEDURE
Figure 1'4 shows the arrangement of the source, burner, and tube schematically. I t is a simplification of the arrangement used by Fuwa and Vallee ( 1 ) . The source (C) is imaged about
midway through the tube by lens L and again on the spectrometer slit by Lp. Both lenses are diaphragmed to about 6-mm. diameter openings to reduce the effect of reflection and emission from the tube walls. The tube is 40 cm. long (longer tubes cannot be mounted conveniently on our instrument), about 10-mm. i.d., and is constricted to about 6-mm. diameter at the end to reduce the amount of air entrained with the flame. The tube (2') is insulated with asbestos ( I ) to prevent condensation of salts on the walls. One stream of air cools the tube wall a t the point where the flame first strikes, and a second air stream, directed vertically past the end, protects the lens. All components are supported on spectrograph bench riders. The tube rests on asbestos blocks for thermal insulation from the remainder of the support. RESULTS
Emission from the hot tube walls has not been a serious problem, even though there is no electrical discrimination against it. The energy emitted depends on the condition of the tube wall as well as the wavelength. Yew tubes can be used to determine calcium a t 4227 A,, but the emission becomes greater as salts react with the tube walls. Tubes that have been in prolonged use give noticeable emission a t all wavelengths longer than 3000 A. Some emission can be tolerated if the zero setting of the instrument is made with a shutter in front of the lamp while solvent is aspirated into the tube. The use of a light chopper and a.c. amplification would eliminate the problem, but for work a t the short VOL. 37, NO. 4, APRIL 1965
601