Applications. The method described has been applied routinely to the determination of bismuth in over one hundred samples, with concentrations ranging from 0.1 to 10 ppm of bismuth. The convenient operating range is from the detection limit to 1.0 ppm. For higher concentrations of bismuth, conventional direct spraying AA may be used. The hydride evolution AA technique described has proved useful as an independent method of analysis for low bismuth levels existing in commercial copper. Moreover, its high sensitivity is of interest for measuring bismuth in ultra-pure copper a t levels not easily measured previously. The method described has since been extended to determine arsenic, selenium, tellurium, and tin a t low ppm levels in electrolytic copper samples ( 1 4 ) . ACKNOWLEDGMENT The advice and encouragement of F. M. Kimmerle, Universit6 de Sherbrooke, and the facilities provided by Noranda Research Centre, are gratefully acknowledged.
LITERATURE CITED (1) "Methods for Emission Spectrochemical Analysis", 6th ed., American Society for Testing and Materials, Philadelphia, PA, 1971, p 299. (2) R. J . Lacoste, M. H. Earing, and S. E. Wiberley, Anal. Chern., 23, 871 (1951). (3) E. B. Sandell, "Colorimetric Determination of Traces of Metals", 3rd ed., lnterscience Publishers, New York, 1959, p 332. (4) W. Reichel and E. G. Bleakley. Anal. Chern., 46, 59 (1974). (5) E. N. Pollock and S. J. West, At. Absorpt. Newsl., 11, 104 (1972). (6) E. N. Pollock and S. J . West, At. Absorpt. News/., 12, 6 (1973). (7) F. J. Schmidt and J . L. Royer, Anal. Lett., 6, 17 (1973). (8) F. J . Fernandez, A t . Absorpt. News/., 12, 93 (1973). (9) R. S. Braman, L. J . Justen, and C. C. Foreback. Anal. Chem., 44, 2195 (1972). 10) D. C. Manning, At. Absorpt. Newsl., 10, 123 (1971). 11) H. L. Kahn and J. E. Schalis, At. Absorpt. Newsl., 7, 5 (1968). 12) E. J. Knudson and G. D. Christian, Anal. Lett., 6, 1039 (1973). 13) R. E. Dean and W. J . Dixon, Anal. Chern., 23, 636 (1951). 14) M. Bedard and J. D. Kerbyson, to be published. 15) M. Bedard, M.Sc. Thesis, Universite de Sherbrooke, November 1974.
RECEIVEDfor review November 20, 1974. Accepted February 27, 1975. The work described formed part of a university-industry cooperative research project and has been partly described in an internal M.Sc. thesis (15).
Phosphorus-31 Fourier Transform Nuclear Magnetic Resonance Spectrometry as a Trace Analysis Tool for the Determination of Inorganic Phosphates Thomas W. Gurley and William M. Ritchey Department of Chemistry, Case Western Reserve University, Cleveland, OH 44 106
31PNMR is well established as a powerful tool of structure elucidation of phosphorus compounds. The quantitative capabilities have been employed mainly to determine phosphorus a t relatively moderate concentrations 1%and above ( I ) . Recent studies employing signal averaging capabilities report that the low ppm (milligrams/liter) level can be detected for orthophosphates, phosphonates, and cyclic phosphates but the experimental time is of the order of days and no quantitative data are given ( 2 , 3 ) . Therefore, it is the goal of this work to determine the capability of the NMR technique a t the lower pprn concentration range in a relatively short period of time. This research interest has developed from the need for better methods of analysis of environmental pollutants in wastewater and related aqueous systems. Our studies of the most abundant phosphorus compounds in the aqueous environment, namely, ortho-, pyro-, and tripolyphosphate, are reported in this paper. The advantages of using 3lP NMR include specificity based on the chemical shift (phosphorus nuclei resonate over a range of 500 ppm). There are also no interfering ion problems as is the case with a colorimetric determination. The major drawback of NMR is the inherent low sensitivity as phosphorus is only about 6% as sensitive as hydrogen. Therefore, sensitivity enhancement is necessary to enable one to observe phosphorus a t low concentrations. Among the signal enhancement techniques employed are pulsed Fourier transform capabilities, 12-mm sample tube, and the addition of a paramagnetic "relaxagent". The relaxagent can be considered effectively a signal-noise-time enhancer since its function is to reduce the relaxation time of the 31Pnuclei. The unpaired electrons of a paramagnetic species augment the normal spin-lattice relaxation ( T I ) 1444
ANALYTICAL CHEMISTRY, VOL. 47, NO. 8, JULY 1975
pathway which results in a more rapid and efficient relaxation process. This allows more pulses to be applied per unit time without saturating the nuclei and, hence, greater signal-to-noise (S/N).
EXPERIMENTAL Apparatus. NMR spectra were taken a t ambient probe temperature (30 "C) on a Varian XL-100-15 spectrometer. An external 19F lock was employed and proton decoupling was not used. An external reference standard of approximately 0.3M tetraethylammonium phosphate (TEAP04) was placed in a precision capillary tube and used in the quantitative work. The chemical shift of TEAP04 is -2.1 ppm with respect t o 85% H3P04 (downfield from H3PO.d. Reagents. All reagents were reagent grade. No special handling was required. The tetraethylammonium phosphate [(Et)4N]2HP04 was prepared by titrating a given amount of tetraethylammonium hydroxide with phosphoric acid to a pH of 10. The solution is jellylike and was diluted before being placed in a precision capillary. Procedure. Standard solutions of 100 to 500 ppm phosphorus (a mixture of each specified phosphate) were prepared and incrementally diluted to obtain lower concentrations. The iron(II1) was added as the hydrated nitrate salt to a known volume of a phosphate mixture for the HC1 and EDTA systems. Then, after the addition of the iron, known amounts of HC1 or excess molar quantities of EDTA were added and thoroughly mixed to ensure equilibrium. Iron(II1) was added directly to a known volume of acetylacetone (acac), thoroughly mixed, and then given volumes of the chelate solution were added to the phosphate solutions. Fresh samples were prepared to ensure little or no hydrolysis of the condensed phosphates was occurring. The tripolyphosphate NMR signal used for analysis was the doublet due to the end phosphate groups and therefore was representative of only two-thirds of the phosphorus in the actual compound. The triplet due to the central phosphate group was too weak to be useful.
Table I. Variation of S I N of Phosphorus vs. Ligand a n d Time Concenbation, pprn
SIN
Ortho
Pyro
Iron
Ligand
30
30 0 24 20
0 50 20 20
chloride acac EDTA
25 24 20
...
Transients
Ortho
Pyro
8000 8000 3000 8500
4 5
0
Acquisition time, sec
1.6 0.3 0.5 0.5
0 0 7
7 10
Pulse delay, sec
1.4 0 0 0
Accumulation time, min
4 06 40 25 67
0 PYRO 8 T R I P O L I
Table 11.Relaxation Time D a t a
A
T I ,sec
ORTHO PYRO
0 TRIPOLI
Relaxagent
Ortho
Pyro
None Ir on(1II-HC1 Ir on(II1)acac Iron(II1)-EDTA
-4 0.05 0.5 0.5
-2 1.o 1 0.5
150
-
RESULTS AND DISCUSSION Inorganic phosphate mixtures containing various ironchelate “relaxagents” were observed with NMR to determine an optimum system for low level analysis. The phosphates were first observed without any impurities present. A typical sample containing 30 ppm phosphorus ( 10-3M) as ortho- and as pyrophosphate produced a signal only for the orthophosphate with a S/N of 4 after approximately 7 hours of accumulation (Table I). Since soluble iron would be a component of most polluted systems, it was believed to be the best paramagnetic ion to employ. Iron(II1) was then added to the mixtures and severe broadening of the phosphate signals was observed. This was due to the close proximity of the unpaired electrons of the iron(II1) ions to the phosphorus nuclei. Since iron forms relatively strong complexes with oxygen-containing compounds and especially condensed phosphates, it was assumed that iron phosphate complexes were formed. This caused not only a reduction of the T I relaxation time, which was desired, but apparently also of the T2, spin-spin relaxation time. As T2 is decreased, line width is increased. Upon addition of 3 drops (0.15 ml) of concentrated HC1 to a 25-ml solution containing 1000 ppm orthophosphate and iron, the solution turned yellow, indicative of the iron(111) chloride complex, and a sharp NMR signal was observed. A 25-ppm solution of orthophosphate containing 50 ppm iron(III), acidified with HCl, showed a S/N of 5 after 8000 transients (Table I). This was a reduction in time by a factor of ten compared to the solution containing no iron. A T Istudy (inversion-recovery method) ( 4 ) of the ironHCl system (1000 ppm P/1000 ppm Fe-HCl) indicated that the orthophosphate phosphorus was relaxing in less than 0.1 second. Whereas the relaxation time for orthophosphate with no iron present was estimated (progressivesaturation method) ( 5 )to be >4 seconds (Table 11). Although the HC1 system was excellent for observing orthophosphate, its major limitation was its inability to allow the observation of pyro- and tripolyphosphate. There were two possible reasons why this was the case. The pyrophosphate ions form very strong complexes with iron (Kf 10”) (6) whereas the chloride complexes of iron are not as strong (Kf los) (7). If all of the pyrophosphate ions are complexed directly to iron, then the signal will be severely broadened. This was observed experimentally at higher concentrations (1000 ppm P). The other possible reason was that, in an acidic solution, condensed phosphates hy-
-
-
P CONCENTRATION
plot of phosphorus concentration in parts-per-million vs. area of the NMR signal. T h e reference peak, TEAP04, was assigned the value of 40
Figure 1. A t h e relative
drolyze much more rapidly than in a neutral or basic solution. The experimental data indicated that no noticeable hydrolysis was occurring. According to Levy (8) the Fe(acac)s and Cr(acac)s complexes are effective relaxagents for 13C NMR. Therefore the iron(II1) complex was examined to determine its feasibility. With this system, it was again found that orthophosphate could be observed at 24 ppm P with 20 ppm iron. In 25 minutes, a S/N of 7 was achieved (Table I). However, pyrophosphate and tripolyphosphate were not observed, possibly again because of the equilibrium constant of ironacac, los. The acac appeared to be as good a ligand as the HC1 except for its limited solubility in water. A study was conducted to determine the optimum P:Fe ratio using the iron(II1)-acac system. A 24 ppm P (orthoand pyro-) was observed in a range of iron concentrations from 10 to 60 ppm. The optimum iron concentration was 20 ppm and this sample produced a S/N of 7 for orthophosphate in 25 minutes (3000 transients). Enough iron needs to be present to enable the P nuclei to relax rapidly by coupling of the field of the P nuclei with the field of the unpaired electrons of the iron(II1) but, when excess iron is present, broadening due to a reduction in T2 occurs. The optimum conditions appear to be achieved when the paramagnetic ion is not in the first coordination sphere causing a long range perturbation of the P nuclei by the unpaired electrons. The final system reported herein is an iron(111)-EDTA complex. Iron-EDTA was chosen because of its large equiANALYTICALCHEMISTRY, VOL. 47, NO. 8, JULY 1975
1445
Table 111. Linear Regression D a t a
Phosphates
Ortho Pyro Tripoly P y r o and t r i p o l y
y-inter-
x-
Slope,
cept,
intercept,a
arealppm
area
ppm
area
coefficient
1.0 -15 -32 41
-1 +15 +48 +20
3.4 3.0 0.2 7.3
0.99 0.99 1.0 0.99
0.99 1.0 0.67 2.1
Std Correlation
a T h i s value is the lower detection limit for a 25-minute accumul a t i o n t i m e . Standard deviation is computed by
4 -
B
2
0 PPM
Figure 2. 31P NMR spectra. Spectrum (A) is a 100-ppm phosphorus concentration of ortho-(- 1.1 pprn) and tripolyphosphate (+4.4 ppm) containing 100 ppm iron-EDTA relaxagent. Spectrum (6)is a 20-ppm sample of ortho-(- 1.1 ppm) and pyro- and tripoiyphosphate (+4.3 ppm) containing 20 ppm iron-EDTA. The spectra were taken in 25 minutes (3000 transients) and the reference peak, TEAP04, is shown at -2.1 ppm
librium constant (Kf of and solubility characteristics in water. A solution containing 20 ppm P and 20 ppm ironEDTA produced a S/N of 10 for orthophosphate and 7 for pyrophosphate after 8500 transients or 1 hour (Table I). The S/N for orthophosphate in the acac and EDTA systems is almost identical, compared on an equal time basis. The EDTA was thought to be complexing all of the iron and, therefore, the pyrophosphate was uncomplexed and observed. A quantitative study was attempted on the phosphate mixtures using iron-EDTA as the relaxagent and TEAPOd as a reference. Three mixtures of equivalent amounts of iron(II1)-EDTA and phosphates (ortho-, pyro-, and tripolyphosphate; ortho- and pyrophosphate; and ortho- and tripolyphosphate) ranging from 10 to 100 ppm ( 3 x to 3 X 10-3M) were studied. The results of this study are shown in Figure 1. The standard set of NMR parameters used were spectral width, 2000 Hz; acquisition time, 0.5 second; pulse width, 30 microseconds (90' pulse); pulse delay, 0 seconds; and transients, 3000 (25 minutes). The areas of the peaks were determined by computer integration of the spectra and normalized by the area of the
reference peak. The orthophosphate line in Figure 1 is an average of all three solutions. The linear regression data for Figure 1 are given in Table 111. These data indicate that the NMR signal is linear with concentration a t low concentrations which has not been previously demonstrated at this level for phosphorus (9). Two typical spectra are given in Figure 2. CONCLUSION The described 31PNMR technique for determination of inorganic phosphates shows promise as a low level analysis tool. The lowest level achieved was 10 ppm orthophosphate in 25 minutes of instrument time. It is certainly feasible to assume that the 1 ppm level (3 X 10-5M) could be realized with an extended time experiment (12 hours) under the same conditions. Further work needs to be done to enable one to observe tripolyphosphate a t the 20-ppm level in a relatively short time. This new technique could find application not only in environmental studies but also in biochemically related areas. LITERATURE CITED (1) J. G. Colson and D. H. Marr, Anal. Chem., 45,370 (1973). (2) T. Glonek, J. R. Van Wazer, R. A. Kleps, and T. C. Myers, Inorg. Chem., 13, 2337 (1974). (3) T. 0. Henderson, T. Glonek, R. L. Hilderbrand, and T. C. Myers, Arch. Biochem. Biophys., 149, 484 (1972). (4) R. L. Vold, J. S. Waugh, M.P. Klein, and D. E. Phelps, J. Chem. Phys., 48, 3831 (1968). (5) R. Freeman, H. D. W. Hill, and R. Kaptein, J. Magn. Reson., 7, 82 (1972). (6) Water Quality Division Committee on Nutrients in Water, J. Am. Water Works Assoc., 82, 127 (1970). (7) L. G. Sillen and A. E. Martell, "Stability Constants of Metal-ion Complexes", Spec. Publ. No. 17, The Chemical Society, London, 1964. (8) G. C. Levy and J. D. Cargioli. J. Magn. Reson., 10, 231 (1973). (9) W. F. White and W. C. Easley, NASA Technical Note D-5726, Interpretation of Microwave Spectral Absorption Intensity Measurements in Partially Saturated Lines, March 1970.
RECEIVEDfor review January 13, 1975. Accepted March 27, 1975.
Determination of Benzyl Chloroformate by Nuclear Magnetic Resonance Julian G. Michels Research and Development Department, Norwich Pharmacal Company, Division of Morton-Norwich Products, lnc., Norwich, NY 138 15
Benzyl chloroformate (carbobenzoxy chloride) finds widespread use as a reagent for the introduction of protecting groups in peptide syntheses. Although it is desirable to be able to control accurately the excess of this re1446
ANALYTICAL CHEMISTRY, VOL. 47, NO. 8, JULY 1975
agent used, few good, specific methods for determining its purity are to be found in the literature. Most methods found (1, 2 ) were titrimetric and have the disadvantage of being either differential in nature or subject to interference
