Table I. Sodium Analysis by NMR Concentration, .N
Solute
Solient
Known
KMR
Line width
'/T2, sec-l
10 22 3.00b same NaCla 76 168 12.00c 12.06 NaOH 9.72 108 240 9.65d NaC10, 3 . 4 l C 3.37 280 630 NaOCH, a Standard for analysis by ?;MR. b By weight. By titration with acid. d From density.
H,O H20 H20 CH,OH
Figure 1. Peak height of the sodium NMR resonance in aqueous solution
( 0 )NaCI: (El, M) NaOH; (El) amplitude of y H1 optimized:
constant y H1. The peak height is in arbitrary units (12.3 mm), chosen to make the slope equal to one
lutions, as 1/Tz increases with concentration for these solutions, resulting in a curved calibration plot which has a maximum (8),as shown in Figure 1. Limitations. The method is suitable for solutions only, including nonaqueous solutions. In viscous solutions, the resonance will be broadened; but as long as sufficient RF power is available to reach the maximum peak height, the analysis will be correct. Viscous solutions occur in biological tissues. If the sodium ion is complexed (11-13), the resonance will also be broadened, again requiring more R F power. For extremely high RF power, it is difficult to balance the bridge of a single coil spectrometer perfectly, resulting in increased noise and lower sensitivity. For a cross coil instrument, the same problem obtains in that the adjustment of the paddles becomes very critical. The sensitivity is rather poor. While we have detected solutions as dilute as 0.01M or 230 Mglml (ppm), much higher concentrations are needed if good precision is re-
quired. We prefer solutions with a concentration over 0.3M. However, the sensitivity can be improved by more than an order of magnitude by using larger sample tubes, time averaging, and stronger magnetic fields. A principal advantage of the method is that it is essentially free from interferences. Although a standard is used, it is an absolute method in that urnaxis always directly proportional to the sodium ion concentration. Calibration curves should not be needed except to check the procedure. Furthermore, the method is non-destructive and rather fast, once the spectrometer is adjusted.
LITERATURE CITED (1) A. L. Van Geet and G. J. Templeman in "Abstracts of Papers", 167th National Meeting of the American Chemical Society, Los Angeles, Calif., April, 1974. Abstract No. Anal. 161. (2) G. J. Templeman, Ph.D. dissertation, State University of New York at Buffalo, 1970; Diss. Abstr. lnt. 6,31, 5301 (1971). (3) G. J. Templeman and A. L. Van Geet, J. Am. Chem. Soc., 94, 5578 (1972). (4)M. Eisenstadt and H. L. Friedman, J. Chem. Phys., 44, 1407 (1966). (5) A. L. Van Geet and D. N. Hume, Anal. Chem., 37, 979 (1965). (6) J. A. Pople, W. G. Schneider, and H. J. Bernstein, "High Resolution Nuclear Magnetic Resonance", McGraw-Hill. New York, 1959, p 35. (7) J. D. Roberts, "Nuclear Magnetic Resonance. Applications to Problems in Organic Chemistry", McGraw-Hill, New York. 1959, p 88. (8)0. Jardetzky and J. E. Wertz, J. Am. Chem. Soc., 82, 318 (1960). (9) G. 2 . Mal'tser, G. V. Malinin, V. P. Mashorets, and V. A. Shcherbakor. Zh. Strukt. Khim., 6,353 (1965) (English). (10) C. A. Rotunno, V. Kowalewski. and M. Cereijido, Biochim. Biophys. Acta., 135, 170 (1967). (11) D. H. Haynes, B. C. Pressman, and A. Kowalsky, Biochemistry, IO, 852 (1971). (12) E. Shchori, J. Jagur-Grodzinski, Z. Luz. and M. Shporer, J. Am. Chem. SOC.,93, 7133 (1971). (13) A. M. Grotens. J. Smid, and E. de Boer, Chem. Commun., 759 (1971).
RECEIVEDfor review April 18, 1974. Accepted March 3, 1975.
Determination of Parts per Billion Phosphate in Natural Waters Using X-Ray Fluorescence Spectrometry Donald E. Leyden,' William K. Nonidez, and Peter W. Carr Department of Chemistry, University of Georgia, Athens, GA 30602
The importance of excess phosphate in environmental waters has been of much concern and discussion. The determination of orthophosphate or t$al phosphate converted to orthophosphate is frequently performed by environmental monitoring laboratories. A standard method for the Author to whom all correspondence should be addressed.
determination of orthophosphoric acid involves the formation of 12-molybdophosphoric acid (12-MPA) in a strongly acidic solution. Reduction of the 12-MPA with ascorbic acid or stannous chloride forms the heteropolyblue complex which is measured spectrophotometrically. Because germanium, silicon, and arsenic also form heteropolyacids with molybdate anion in acidic solution, a number of techANALYTICALCHEMISTRY,
VOL. 47, NO. 8, JULY 1975
1449
niques have been developed which allow the separation of these acids by extracting them into an organic solvent with subsequent analysis by the spectrophotometric method or atomic absorption determination of molybdenum (1-6). The extraction step not only allows separation of the 12MPA from contaminants but also preconcentrates the analyte in the organic phase which increases the sensitivity of the analysis. X-Ray fluorescence is a versatile analytical tool which may be applied to the determination of elements of atomic number greater than 11. However in the case of elements of low atomic number, such as phosphorus, the fluorescence yield is low, in part because of the competing Auger process. Also these elements emit X-rays of relatively low energy which are severely absorbed by the sample matrix. Moreover, even in the best cases, X-ray fluorescence is not considered a technique for trace determinations; a lower limit of detection of 0.1-1 pg is commonly quoted. However, because the technique can be applied to many elements with good precision and, if used properly, with good accuracy, it should be useful in many cases if some chemical advantage is used. Attempts have been made to improve low level X-ray determinations of phosphorus by precipitating the orthophosphate salt of a more efficient X-ray fluorescent emitter. Stork and Jung (7) have analyzed orthophosphate in the l- to 10-mg range by precipitating it as Ag2TlP04 and measuring the fluorescence of silver. G. S. Smith (8) suggested that orthophosphate could be precipitated as bismuth phosphate and the bismuth concentration determined. Both these methods have three distinct advantages-the analyte is preconcentrated by precipitation, the element measured is more efficient for X-ray analyses than phosphorus, and the element measured in the precipitate exists in a relatively high ratio to phosphorus. The main disadvantage to these techniques is that, in each case, a precipitate must be formed and recovered. Care must be taken to ensure that no other species is present which will precipitate salts of the reagents. For example, chloride ion could interfere with the use of a silver salt. This paper describes a method for the determination of orthophosphate which takes advantage of a selective extraction of phosphomolybdic acid into ethylacetate and subsequent adsorption upon silica gel which contains N substituted diamine functional groups. The silica gel is then pressed into pellets and the Kn line of the molybdenum is measured by X-ray fluorescence. As in the methods of Stork and Smith, this technique also allows the preconcentration of the analyte as well as measuring an efficient X-ray fluorescence emitter which exists in a 12:l ratio to the phosphate.
EXPERIMENTAL X-Ray Fluorescence Measurements. The X-ray determinations were performed using a Philips-Norelco PW-1410 X-ray spectrograph. The source of excitation was a tungsten target tube operated a t 90 kV and 35 mA. A LiF-220 analyzing crystal was used. T h e detection of the fluorescence X-rays was accomplished using a scintillation counter. The optimum parameters for the detector and the pulse height analyzer were determined visually by means of a Philips-Norelco pha-2100 scope. The samples were placed in a circular aluminum holder equipped with an 0,00025-inch mylar window and rotated. The optimum goniometer setting was determined by scanning the 20 angle between 27.00' and 29.00'. The 28 setting used was 28.87 for the Kcu lines of molybdenum. All counting times were adjusted so that the expected relative standard deviation in the counts was less than I%, but never exceeded 200 seconds. Reagents a n d Materials. All chemicals used in this investigation were reagent grade and obtained from Fisher Scientific Company and Mallinckrodt Chemical Works. All solutions were made using distilled deionized water. All glassware was soaked for at 1450
ANALYTICAL CHEMISTRY, VOL. 47, NO. 8, JULY 1975
~
Table I. Effect of Washing the Extract with 10 ml of 1M HCl KO. of washings
Relative X-ray intensity of .Mo
1
1.oo
2 3
0.88 0.89 0.89
4
least I hour in 6M HC1 and was never exposed to phosphate-containing detergents. The phosphate standard solutions were prepared daily by diluting a stock solution of KH2P04 prepared by dissolving a known amount of dry KH2P04 and diluting to 1000 ml. A 0.313M MOO^^solution was prepared by dissolving 75.6 g of Na2Mo04.2H20 in 1000 ml of HzO and aging for 24 hours. The acid wash solution was prepared by diluting 90 ml of concentrated HC1 to 1 liter with H20. The silica gel containing N-substituted diamine functional groups was prepared by reacting N-P-aminoethyl-y-aminopropyltriethoxysilane (Dow-Corning 2-6020) (9) with silica gel type G (E. M. Reagents) as described by Leyden, Luttrell, and Patterson (10). The silylation solution was prepared by mixing 10 ml of Dow-Corning 2-6020 with 18 ml of 0.1% aqueous acetic acid and diluting with deionized water to a volume of 100 ml. Optimum silylation resulted when 100 ml of this solution were stirred with 50 g of silica gel for about 10 minutes. This mixture was then allowed to dry overnight in an oven a t 80 "C. The silylated material was then washed with deionized water, allowed to air dry, and stored. The material is air stable and may be kept at room temperature. The only precaution necessary is to prevent exposure of the material to strongly acidic or basic solutions for long periods of time, as the silyl bonds will hydrolyze slowly under these conditions. Procedure. A 50.00-ml portion of sample is pipetted into a 60-ml separatory funnel and is made approximately 1M in HC1 by the addition of 4.50 ml of concentrated HCI. Four ml of 0.313M NazMo04.2H20 is then added to make the solution approximately 0.022M in molybdate. The solution is shaken thoroughly and 10 ml of ethyl acetate is added. The mixture is then shaken for 30 seconds and allowed to stand undisturbed for 10 minutes. The aqueous layer (lower layer) is then removed and 10 ml of 1M HC1 is added which has been equilibrated with ethyl acetate. The mixture is then shaken briefly and allowed to stand until the aqueous layer is clear. The washing procedure is repeated and the HCl removed. The remaining ethyl acetate layer is emptied into a 50-ml beaker and the separatory funnel washed with 5 ml of ethyl acetate which is added t o the ethyl acetate in the beaker. Ten ml of HzO is then added and the two phases are stirred a t high speed to thoroughly mix the two layers. Then 100 mg (&I mg) of the prepared silica gel is added and the p H adjusted to 2.8. The mixture is stirred for five minutes and the silica gel is collected with a Hirsch funnel on E. H. Sargent and Co. No. 500 filter paper (3 cm). A retainer ring should be used to ensure that no silica gel is lost around the sides of the paper. After drying for 10 minutes in an oven set at 100 "C, the silica gel is carefully scraped into a Fisher 5-ml disposable microbeaker which contains 100 mg of cellulose powder (Whatman CF 11).The two powders are mixed by manual stirring with a spatula and two drops of H20 are added. After stirring again until the mixture becomes a semi-paste, a pellet is pressed in a 'h-in. diameter hardened stainless steel die under 10,000 pounds of pressure. The Mo is then determined by X-ray fluorescence. Standards may be prepared exactly as unknowns by using 50-ml aliquots of standard phosphate solutions instead of unknown solutions and proceeding as above.
RESULTS AND DISCUSSION The extraction of 12-MPA from aqueous solution is well known (1-6). We have adopted a method suggested by Simon and Boltz ( 5 ) but have used ethyl acetate rather than diethyl ether as.&he organic phase because we have observed a more complete extraction (96%) of the 12-MPA and the solvent is more convenient. We did not study extraction schemes in detail. However, a number of different schemes may be applicable to this technique.
I
IO
a
I
I
I
I
I
I
30
w
yl
fin
in
en
i 30.
P(PWx10-2)
PH
Figure 1. Relative count rate for Mo vs. pH to procedure
of
extraction according
30.5 63.3 126 95.1 318 470 160 67.7 66.6 (1
\-Ray
32 .O 66.4 150 100 303 478 168 68.7 68.1
count rate vs. ppb phosphorus
Table 111. Effect of Silicate and Arsenate on the Determination of Phosphate
Table 11. Comparison of X-Ray and AutoAnalyzer Resultsa >utoAl,Jl, 'Lr
Figure 2. Calibration curve for
Silicate Krlativr . diiicrmce, .
4.7 4.7 16.0 4.9 -5 .O 1.7 4.8 1.5 2.2
hlolariv P
Sample
found x 10''
1
10.5 10.5 10.5 11.1 10.4 10.5
2 3 4 5 6
lslOi?-l Llolarity S1o3'-
0 1.01 x 1.01 x 5.05 x 7.07 x 10.1 x
Ratio
I P O !3-1
... 10-4 10-1 lo-' 10-3 10-3
9.6:l 96 :1 454:l 740:l 962:l
Arsenate
All values in ppb as phosphorus. 1
After the initial extraction of the 12-MPA, the effect of washing the organic layer with 10-ml portions of 1M HCl was studied. Table I shows that two 10-ml washings are sufficient to rid the organic layer of excess molybdate species which would cause high results if not removed. More than two washings have no effect on the amount of molybdate found in the organic layer. The mechanism by which the treated silica gel adsorbs the molybdenum is still under investigation. Untreated silica gel does not adsorb molybdenum from molybdate solutions or 12-MPA solutions. There is a significant p H dependence of the phenomenon which may be related to the pK, values of the acid (e.g., molybdic acid) and the diamine. The ability of this material to recover the molybdenum from solution appears quantitative down t o the X-ray detection limits. The species which is adsorbed on the silated silica gel is not known to us a t this time. Since we have detected phosphorus on the gel by X-ray fluorescence, we cannot rule out the possibility of the adsorbtion of an anionic form of the 12-MPA. This hypothesis is strengthened by the fact that 12-MPA is known to be quite stable a t the p H of maximum adsorption onto the gel ( 1 1 ) . However, it is quite possible that phosphate anion would be adsorbed on the gel after the dissociation of the 12-MPA anion. ESCA data have shown that the molybdenum is present in the f 6 oxidation state indicating no redox reaction between the ethylenediamine groups on the gel and the molybdate species. The optimum pH for the extraction of the molybdate (as 12-MPA) was studied by adsorbing equal aliquots of extracted 12-MPA on the solid substrate at various pH values. Figure 1 shows that a p H of approximately 3 is optimum, and that it is important to control the p H of the ad-
2 3 4
5
10.04 10.33 12.21 16.42 17.93
0 1 x 10-5 1 x 10-1 2.5 x lo-'
5.0 x lo-'
... 0.968 :1 8.20:l 20.5:l 41:l
sorption process to ensure that the extraction of the phosphomolybdate species is complete. Figure 2 shows an example of a standard working curve for this method. I t is linear to 3000 ppb P. I t should be noted that, when working with higher concentrations of phosphates, the linear range may be adjusted by using less sample in the initial extraction. We have performed analyses on samples containing as little as 15 ppb P with a relative standard deviation of 3%. I t is also possible to analyze a t lower levels by using more sample in the original extraction. For the linear portion of the curve shown in Figure 2, the correlation coefficient is 0.998 and the Student's t value is 39.4. Table I1 gives a comparison of the results from ten water samples prepared in our laboratory. The samples in the first column were analyzed using an AutoAnalyzer method developed by the Environmental Protection Agency (12). The same samples were analyzed by our method. In order to test for a systematic difference which might be present, a paired t-test was carried out on the two methods ( 1 3 ) .A comparison of the data by a t-test indicates that the mean difference is not statistically different from zero a t the 95% confidence interval. Because silicate and arsenate anions form heteropoly acids with molybdate, we analyzed a solution containing a constant amount of orthophosphate and varying molar ratios of silicate or arsenate to orthophosphate. I t may be seen from Table I11 that silicate does not interfere with this ANALYTICALCHEMISTRY, VOL.
47, NO. 8, JULY 1975
1451
method at molar ratios as high as 1000:1, whereas arsenate interferes significantly at molar ratios of 8.2:l. It is worth noting that a mole ratio of 8.2:1 hopefully reflects a larger arsenate-to-phosphate ratio than would be encountered normally in water samples. It should be pointed out that, in the present study, no effort was made to optimize the extraction to eliminate these interferences. The method presented here provides a simple X-ray fluorescence procedure for the determination of aqueous phosphate with high precision in the 20-ppb level. Lower detection limits may be achieved by the extraction of a larger aliquot of sample than used here. The use of an X-ray tube more efficient for the excitation of molybdenum such as rhodium, would significantly reduce the detection limit. The standard pellets may be stored for long periods of time, providing rapid calibration of the X-ray instrumentation. Although the method does not lend itself to automated analysis, it provides an alternate method convenient for a small member of samples requiring high accuracy. Such would be the case when confirmation was desired of samples which had been through a less precise screening analysis.
ACKNOWLEDGMENT The authors thank Dennis Revel1 of the Environmental Protection Agency, Athens, Ga, for his assistance in acquiring the data using the AutoAnalyzer.
LITERATURE CITED R. I. Alakseyev, Zavod. Lab., 11, 123 (1945). J. Paul, Anal. Chim. Acta, 35, 200 (1966). S. V. Eisenreich and J. E. Going, Anal. Chim. Acta, 71, 2 (1974). A. Halosk, E. Pungor, and K. Polyak, Talanta, 18, 577 (1971). S. J. Simon and D. F. Boltz, Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Cleveland, OH, 1974. (6) W. S. Zaugg and R . J. Knox, Anal. Chem., 38, 12 (1966). (7) G. Stork and H. Jeng, Fresenius'Z. Anal. Chem., 249, 161 (1970). ( 8 ) G. S. Smith, Chem. lnd. (London). 22, 907 (1963). (9) Dow-Corning Bulletin No. 03-023. (10) D. E. Leyden, G. H. Luttrell, and T. A. Patterson, Anal. Left., 8, 51 (1975). (11) A. Duca and T. Budiu, Rev. Roum. Chim., 11, 585 (1966). (12) "Methods for Chemical Analysis of Water and Wastes", Environmental Protection Agency, 1971, p 235. (13) M. G. Natrella. "Experimental Statistics", U.S. Government Handbook 91. Aug. 1, 1963.
(1) (2) (3) (4) (5)
RECEIVEDfor review December 16, 1974. Accepted February 20, 1975. This work was supported in part by Research Grant GP-38396X from the National Science Foundation.
Stereospecific Reactivity by Ion Cyclotron Resonance Spectrometry: Optimization of Reactivity Differences in Two Isomeric Esters Maurice M. Bursey' Venable and Kenan Chemical Laboratories, The University of North Carolina, Chapel Hill, NC 27514
J. Ronald Hass National Institute of Environmental Health Sciences, P.O. Box 12233, Research Triangle Park, NC 27709
Robert L. Stern Department of Chemistry, Oakland University, Rochester, MI 48063
Ion cyclotron resonance (ICR) spectrometry has been suggested as an analytical technique for the identification of isomers difficult to distinguish by conventional mass spectrometry (1-4). Differences do exist in the mass spectra of geometrical isomers ( 5 ) , but they are sometimes small and, in general, not predictable from the general theory of organic cracking patterns. The ion-molecule reactions of these isomers at low energies frequently do show significant differences. We now have optimized a set of conditions for distinguishing a pair of epimeric esters such that one undergoes a certain ion-molecule reaction to an easily measured extent while the other isomer does not. The reaction seems to be generally applicable, and we anticipate that it will be the prototype of many useful applications. We view our reactions in the ICR spectrometer as analogous to chemical ionization in which we ionize not by transfer of a proton but by transfer of the bulkier CH&O+ group, as for example in the reactions 1 and 2 (6). (CH,CO),*+ + M (CH,CO),' + M
CH,COM+ + CH,COCHsCOM' + (CH,C0)2
+
--+
(CH,C0)2*++ (CH,CO)?
-
(CH,CO),'
+ CH,CO*
0
Dcoco
(2)
ANALYTICAL CHEMISTRY, VOL. 47, NO. 8, JULY 1975
(3)
Using this sort of relative rate determination, it has been possible to show very clearly that steric factors are important in ion-molecule reaction rates. In a series of substituted cyclohexanones 1 taken as M in Equation 1, the rate of reaction decreases in a predictable fashion as R becomes larger ( 3 ) .
(1)
A u t h o r t o w h o m correspondence should be addressed.
1452
In Equations 1 and 2, the reagent ion is derived from biacetyl. The pathway is confirmed by the double resonance technique and, in all cases which we have examined so far, the product ion has not decomposed to other products. In these cases, then, it is possible to relate the amount of product formed to the rate of formation. In particular, we have often used a competitive technique in which the acylation of the substrate M in Equation 1 is compared with the acylation of the standard, biacetyl itself (Equation 3).
1
2
a
