2218
Anal. Chem. 1984, 56, 2218-2223 Palframan, J. F.; MacNab, J.; Crosby, N. T. J . Chromatogr. 1973, 76,
307-319. Cox, G. B. J . Chromatogr. 1973, 33, 471-481. Fan, T. Y.; KrYk I.S.; Ross, R . D.; Wolf, M. H.; Fine, D. H. "Environmental Aspects of N-Nitroso Compounds"; Walker, Castegnaro, Griciute, Lyle, Eds.; International Agency for Research on Cancer: Lyon, France, 1978;Sci. Publ. No. 19, p 3. Fine, D. H.; Huffman, F.; Rounbehler, D. P.; Belcher, N. M.; "Environmental N-Nltroso Compounds Analysis and Formatlon"; Waiker, Begovski, Griciute, Eds.; International Agency for Research on Cancer: Lyon, France, 1976;Sci. Publ. No. 14,p 43. Chang, S. K.; Harrington, G. H. Anal. Chem. 1975, 47, 1857-1860. Application Note C-4, EG&G Princeton Applied Research, Princeton, NJ, 1980. Vohra, S. K.; Harrington, G. W. J . Chromatogr. Sci. 1980, 78,
379-383. Hasebe, K.; Osteryoung, J. Anal. Chem. 1975, 47,2412-2418. Althorp, J.; Goddard, D. A.; Sissons, D. J.; Telling, G. M. J . ChromatOQr. 1970, 53, 371-373. Ishibashi, T.; Kawabata, T. J . Chromatogr. 1980, 795,416-420. Young, J. C. J . Chromatogr. 1978, 757, 215-221. Gough, T. A.; Sugden, K.; Webb, K. S. Anal. Chem. 1975, 47,
509-512.
(19) Flne, D. H.; Rounbehler, D. P. J . Chromatogr. 1975, 709,271-279. (20) Oettinger, P. E.; Huffman, F.; Fine, D. H.; Lieb, D. Anal. Lett. 1975, 8 , 41 1-414. (21) Flne, D. H.; Lieb, D.; Rufeh, E. J . Chromatogr. 1975, 707, 351-357. (22) Fazio, T.; Howard, J. W.; White, R. "N-Nitroso Compounds; Analysis and Formation"; Bogovski, Preussmann, Walker, Eds.; International Agency for Research on Cancer: Lyon, France, 1972;Sci. Publ. No.
3. 16. (23) Hoichkiss, J. H.; Libbey, L. M.; Scanlan, R. A. J . Assoc. Off. Anal. Chem. 1980. 63. 74-79. (24) Havery, D.; Hotchkiss, J. H.; Fazio, I.J . Food Sci. 1981, 4 6 , 501-505. (25) Scanlan, R. A.; Barbour, J. F.; Hotchkiss, J. H. Food Cosmet. Toxicol. 1980, 78, 27-29. (26) Popovich, D. J.; Dixon, J. 6.; Ehrlich, B. J. J . Chromatogr. Sci. 1979, 77,643-650. (27) Walters, S. M. J . Chromatogr. 1983, 259, 227-242. (28) Havery, D.; Fazio, T.; Howard, J. W. J . Assoc. Off. Anal. Chem. 1978, 61,1374-1378.
RECEIVED for review March 16,1984. Accepted June 11,1984.
Determination of Orthophosphate by Flow Injection Analysis with Amperometric Detection Sara M. Harden' and William K. Nonidez*
Department of Chemistry, University of Alabama in Birmingham, Birmingham, Alabama 35294
A slmple flow lnjecton system for determlnatlon of orthophosphate Is descrlbed. The system utlllzes a comrnerclally available thln-layer amperometrlc detector for measurement of 12-molybdophosphorlc acld reductlon currents In an aqueous solvent system 0.10 F nltrlc acld, 1.95 X lo-' M sodium molybdate, and 30% (v/v) methanol. The system Is capable of analysls of 70 samples per hour at a flow rate of 2.2 mUmln. The linear dynamlc range extends to a concentratlon of 5.0 X lo-$ M and the detection limit at a slgnal to nolse level of 2:l Is 2 X lo-' M orthophosphate. The relatlve standard devlatlon of the technique at the 1.0 X lo-' M orthophosphate level Is 2.6%.
The determination of phosphorus by a variety of analytical methods has been carried out in laboratories throughout the world in order to diagnose a multiplicity of industrial, environmental, agricultural and physiological problems. In almost all cases this analysis is carried out spectrophotometricly by a variation of the time-honored method of Fisk and SubbaRow (1). This method involves the formation and reduction of 12-molybdophosphoric acid (12-MPA) to produce a species which absorbs a t approximately 700 nm. The determination may be performed using equilibrium methods or methods based on reaction rates of 12-MPA reduction or formation (2). For large numbers of analyses, automated methods have been devised which involve the spectrophotometric measurement of 12-MPA or its reduction product in a moving stream. The most common of these methods involves the use of the Technicon AutoAnalyzer which can analyze 20-30 samples per hour in the 3.2 X lo-* to 3.2 X M P range in a segmented flowing stream (3). More recent developments 'Present address: United States Air Force Armament Laboratory, Eglin Air Force Base, FL 32542.
in flow injection analysis (FIA) have allowed the analysis of phosphate at the rate of between 90 and 100 samples per hour with detection limits as low as 5.0 x M P using a novel method recently described by Johnson and Petty ( 4 ) . More typical detection limits in the lo-' M range have been reported by others (5). These methods also utilize the formation and reduction of 12-MPA with spectrophotometric measurement. Electrochemical detection of 12-MPA in flow injection analysis has been reported by Fogg and Bsebsu (6) who used organic solvents to increase 12-MPA formation. Analysis of phosphate as low as M was reported by the above researchers by injecting preformed 12-MPA into a flowing stream of reagent blank with detection a t a glassy carbon electrode by differential-pulse voltammetry. Fujinaga et al. (7) also report phosphate analysis by FIA in aqueous-organic solvent system with voltammetric detection. This paper describes a simple electrochemical FIA system developed in our laboratory for the analysis of orthophosphate. 12-MPA is preformed by diluting the sample with carrier stream which is an aqueous solution 0.10 F nitric acid, 1.95 X F sodium molybdate, and 30% (v/v) methanol. This mixture is then injected into the flow system with detection by 12-MPA reduction at a platinum electrode using a commercially available amperometric thin-layer detector. Solvent composition effect on reducing current is studied by cyclic voltammetry in order to ascertain the effect of varying methanol and nitric acid concentration. System response and reproducibility are studied as a function of pumping rate. Actual analyses are performed on urine and the result is compared with spectrophotometric analysis of the same samples. Effects of interferences are also investigated. Orthophosphate can be analyzed by this technique at the maximum rate of 90 samples per hour with a detection limit of 2.0 X lo-@M orthophoshate in the original sample. Due to the low potential at which 12-MPA is reduced, no deoxygenation of solvent or samples is necessary. Since no reagents must be mixed in the flow system nor time expended for color
0003-2700/84/0356-2218$01.50/00 1984 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 56, NO. 12, OCTOBER 1984
818
x
Figure 1. Block dhgram of flow injection system: (A) soivent reservoir; (B) peristaltic pump; (C) injection valve; (D) electrochemical detector; (E) waste; (F) amperometric controller; (G) recorder; (H) integrator.
development, the flow system is uncomplicated and relatively inexpensive to assemble and operate. EXPERIMENTAL SECTION Reagents. All solutions were prepared from high purity (greater than 99.7 % ) reagent grade chemicals. Deionized water was used for solution preparation in both the spectrophotometric and the cyclic voltammetry studies. HPLC grade water (Baker Chemical Co., Phillipsburg, NJ) was used in solutions involved in the flow injection analysis studies. All organic solvents used were HPLC grade with the exception of Spectrophotometricgrade methanol, which was used in the spectrophotometric studies. Glassware used in solution preparation was cleaned with a 50% (v/v) solution of HCl followed by multiple rinsings with deionized water and was isolated from other laboratory glassware to prevent contamination from laboratory detergents. Cyclic Voltammetry. A Model RDE3 potentiostat (Pine Instruments Co., Grove City, PA) combined with a Bascom-Turner Model 4120 X-Y recorder was used for the cyclic voltammetry studies. The working electrode was a platinum disk electrode 7.6 mm in diameter also purchased from Pine Instruments Co. The reference electrode was a Fisher Scientific saturated calomel electrode and the auxiliary electrode was a platinum basket type purchased from Fisher Scientific Co. The working electrode was polished with a solution of ethanol and a O.l-pm polishing alumina (Gamal, Fisher Scientific Co.) on an Alpha-A-Polishing Cloth (Fisher Scientific Co.). All cyclic voltammetric scans in the solvent composition studies were run reductively starting from 0.513 V to a switchingpotential of 0.275 V vs. a SCE. Scan rates of 92 mV/s were used in all cases. Each data point representing a CV peak current is an average of eight separate scans between which the working electrode was cleaned as above. The relative standard deviation of any set of replicates was no greater than 5%. Base line currents were established by the "potential hold" method as described by Adams (8). Flow Injection Analysis System. The electrochemical flow cell was purchased from Bioanalytical Systems and consisted of a C-19 flowcell and a C-4A amperometric controller. A planar platinum electrode with a radius of 3.0 mm was purchased from Bioanalytical Systems for use in the C-19 flowcell. A block diagram of the flow injection system appears in Figure 1. The system was constructed using a single channel Rainin Rabbit peristaltic pump and a Rheodyne Model 7010 injection valve. Poly(viny1chloride) manifold tubing of 0.056 mm internal diameter and 36 cm length was used in the pumping head. This tubing was connected to the first tee joint with a 37 cm length of 2.29 mm i.d. PVC tubing and a 18 cm length of 1.42 mm i.d. Viton tubing. The injection valve bypass tube was a 12.7 cm PVC tube with 3 mm i.d. This tubing was also used to connect the second tee joint and the flow cell. The tee joints were connected to the sample valve with 0.8 mm i.d. Teflon tubing. Both tee joints were Swagelok union tees (Catalog no. 100-3). The sample loop was constructed of 0.8 mm i.d. tubing and held a volume of 45 KL.
Continuous Variation Studies. All spectrophotometric measurements were performed with a Beckman Model 26 spectrophotometer and a matched pair of quartz cuvettes purchased from Fisher Scientific Co. 12-MPA absorbance measurements were made at 430 nm as in the studies of Kircher and Crouch (9).
2219
All continuous variations studies were carried out at a constant total phosphate (C,) p h s molybdate concentration (C,) of 2.0 x 10" F. The mole ratio of molybdate was varied between 0 and 0.985 for each plot. Flow Rate Studies. A 28.0-fiLportion of a 3.23 mM phosphate stock solution was diluted with carrier solution in a 25-mL volumetric flask. The final concentration of 12-MPA formed was 9.00 KM. After electrode equilibration at a potential of 0.28 V, five injections were made at flow rates between 0.70 and 3.25 mL/min. The integrated response for each peak was obtained with a Hewlett-Packard Model 3390A integrater and the average of five replicate injections was used for every data point. Relative standard deviations for each set of replicates never exceeded 3.1%. Sample Preparation, Samples whose orthophosphate conM may be M and 5.0 X centrations are between 1.00 X diluted 1OO:l with carrier solution and injected into the flow system. Samples containing less than 1.00 X loW6M orthophosphate must be diluted in such a manner that the final orthophosphate concentration in the diluted samples be between 1.00 x IO-? M (ten times the detection limit) and 5.00 X lo-' M while maintaining the same concentration of nitric acid and molybdate as is present in the carrier stream. To produce such a solution the sodium molybdate and nitric acid must be added separately due to the insolubility of sodium molybdate in highly acidic media. Since the addition of these reagents dilutes the aqueous sample appreciably, the least convenient amount of dilution possible is approximately 2:1, making the lowest level of practical analysis in the original sample 2.00 X lo-? M orthophosphate. The relative standard deviation of five replicates at the 7.5 X lo-' M orthophosphate level is 9.5% and 2.6% at the 1.0 X lo4 M orthophosphate level. Urine Analysis. Thirty minutes before sample injections were begun, the carrier stream in the FIA system was turned on and the pumping rate adjusted to approximately 2.2 mL/min. Potential (0.28 V vs. a Ag/AgCl electrode) is then applied to the electrochemical cell and the base line current allowed to stabilize to a constant value. The carrier stream can be recirculated during this procedure for the purpose of conservation. Urine samples which were collected and stored with a small amount of nitric acid at 0 "C were diluted 1 : l O with deionized water. A 1OO-wL portion of each diluted sample was placed in a 10-mLvolumetric flask and futher diluted with carrier stream. Each solution was immediately injected into the flow system and the integral response related to phosphate concentration by means of a calibration curve. Spectrophotometricanalysis of urine samples was accomplished by means of an inorganic phosphorus test set purchased from Stanbio Laboratory, Inc. The directions supplied in this kit were followed in performing the analysis. The phosphate standard supplied in this kit was used in calibrating both the spectrophotometric and FIA analysis in order to eliminate any determinate error that might exist between separately produced standards. RESULTS AND DISCUSSION Continuous Variation Studies. The stabilizing effect of water soluble organic solvents upon 12-MPA has been known for sometime (10, 11). Since methanol is available in high purity at re.asonable prices and is capable of dissolving relatively large amounts of sodium molybdate and other electrolytic species, preliminary spectrophotometric studies were undertaken to access its suitability for use as a stabilizing agent for 18-MPA in this analysis. To this end a series of continuous variation experiments were performed and Job plots constructed for solutions with a constant nitric acid concentration of 0.5 F and methanol concentrations varying between 0% and 60% (v/v) (Figure 2). Because the absorbance band of acidic molybdate overlaps the absorbance band of 12-MPA, a wavelength of 430 nm was chosen which resides on the high wavelength slope of 12-MPA absorbance. Previous work by Kircher and Crouch (9) suggests the presence of other molybdophosphate species in equilibrium with 18-MPA in strong acid solutions. Evidence for this was
ANALYTICAL CHEMISTRY, VOL. 56, NO. 12, OCTOBER 1984
2220 I .o
/
----, /
0 8
/
w
,'-\, \
T \
I
/\
\
500 mii
0 8
4 z (L
P 3
0 4
0 2
'\ 00 0 0
0 2
0 4
0 8
0 8
I O
MOLE RATIO MOLYBDATE
+
Flgure 2. Continuous variation plots for 12-MPA: Cp Cm = 2.0 X lo3 F, HNO, concentration 0.50 F; absorbance measured at 430 nm; MeOH concentration varied between 0 % and 60 % (v/v); (0)0 % , (A) l o % , (0) 2 0 % , (0) 30%, (*) 4 0 % (V)60% methanol by volume.
0.5
0.4
0.3
0.2
0.1
0.0
' * r (VOLTS v s . SCE)
Figure 4. Cyclic voltammagrams of solutions 0% and 60.0% in methanol: (-) 0%, (- - -) 60% methanol by volume. Both solutions are 2.5 X F in orthophosphate, 9.75 X lo-, F in sodium moiybdate, and 0.50 F in nitric acid. Scan rate = 29 mV/s, platinum electrode area = 0.45 cm2.
I O
0 8 U W
2 0 8 12:
51 4
0 4
0 2
00 0 0
0 2
0 4
0 8
0 8
I O
M O L E R A T I O MOLYBDATE
+
Figure 3. Continuous variation plots for 12-MPA: Cp Cm = 2.0 X los3 F, methanol concentration 60% (v/v), absorbance measured at 430 nm; "0, concentration varied between 0.20 and 3.0 F; (V)0.2 F, (0) 0.5 F, (A)1.0 F, (0) 2.0 F, (0)3.0 F nitric acid.
found in the upward and downward concavity of the experimental Job plots produced as well as a shift in the maxima of these plots from a value indicative of 12-MPA. The Job plots in Figure 2 demonstrate a gradual loss of the above mentioned concavities as well as a continuing movement toward an ideal maximum value for 12-MPA of 0.923 as methanol concentration is increased. These trends are indicative of the increasing stability of 12-MPA in solutions of increasing methanol concentrations with the largest gains in stability achieved with increasing methanol concentrations between 0 and 40% (v/v). Additonal methanol concentration increases after this point lead to only minimal increases in 12-MPA concentration. Figure 3 contains Job plots obtained from a second series of continuous variation experiments in which methanol concentration was held constant at 60% (v/v) and nitric acid concentration varied between 0.20 F and 3.0 F. These plots indicate that 12-MPA stability is decreased by increasing acid concentrations as has been demonstrated by others (9). The concavity at a molybdate mole ratio of 0.88 which appears at higher concentrations of nitric acid suggests the appearance of a second absorbing species whose molybdate to othophosphate ratio is smaller than that of 12-MPA. Cyclic Voltammetry. Since 12-MPA stability can apparently be enhanced by appropriate choices of solvent
methanol and nitric acid concentration, attempts were made to demonstrate this effect electrochemically so that optimum solvent conditions could be chosen for othophosphate analysis in a FIA system. Two series of cyclic voltammograms were recorded between potentials which were deemed practical for uncomplicated amperometric analysis, Le., 0.5 V to 0.0 V vs. an SCE. Both a platinum and a glassy carbon electrode were utilized initially in this work with the platinum electrode chosen for futher work due to the higher reduction currents obtainable through its use. Two consecutive quasi-reversible reductions are observed in this range (Figure 4). These have been reported as two electron reductions in similar solvent systems (12). In purely acidic aqueous solutions the first reduction peak is substantially smaller than the second. Addition of methanol causes the first quasi-reversible system to increase in size with concomitant but smaller increase in the second quasi-reversible system. The second system also shows a substantial shift to more oxidizing potentials with increasing methanol concentrations. Since the reduction peak at the more reducing potential (about 0.175 V) was difficult to reproduce due to an apparent adsorption phenomenon, a systematic study of solvent composition on only the first reduction peak (0.32 V) was attempted. Figure 5 is a plot of the first reduction peak current as a function of methanol concentration as well as absorbance of the same solutions at 430 nm. The absorbance values and peak currents were plotted in relative units. In all cases the molybdate to phosphate ratios were held at a constant ratio of 391 which ensures complete reaction of the orthophosphate with molybdate as is revealed by the Job plots in Figures 2 and 3. The nitric acid concentration was held constant at 0.50 F. No significant peak potential shifts were observed for changes in solvent systems or changes in cycle rates up to 920 mV/s. Since the absorbance of 12-MPA under these conditions increases with reduction current, it is assumed that 12-MPA is the major, but probably not the only, species being reduced at this potential. Substantial increases in response cannot be made at methanol concentrations greater than approxi-
ANALYTICAL CHEMISTRY, VOL. 56,NO. 12, OCTOBER 1984
2221
THOUSANDS
I .2
400 N
0 R M A L
300
I
z
E
P
E R E
K A
S
P
200
R
0 N S
A
0 0
&,
IO0
IO
I
I
20
30
X v / v METHANOL
ABSORBANCE
0
0
0
0 PEAK CURRENT
I 40 50 eo 0
Flgure 5. Normalized cyclic voltammetric peak current and absorbance of solutions which are 2.5 X lo-' F in orthophosphate, 9.75 X M in sodium molybdate, and 0.50 F in nitric acid: (0) absorbance, (0)peak current. CV rate = 29 mV/s, platinum electrode area = 0.45 cm2. Methanol concentration varied between 0% and 60.0% methanol.
0 5
1
K
IO5
3
1
C
n
\
2
3 0
Figure 7. Integral response of FIA system as a function of pumping
E
I 0
2 5
speed.
'75
!
2 0
I S
CML/MIN)
5
4
1 0
I
0.20
0
0.24
0 28 (VOLTS
0.0
0.2
0.4
0.6
0.8
I
.o
N I T R I C ACIO CONCENTRATION (FORMAL)
Figure 6. Cyclic voltammetric peak currents of 12-MPA, 12-MSA, and l 2 M A A reductions at E = 0.34 V vs. SCE: (0)phosphate, (A)silicate, (0) arsenate. Solutions are 2.5 X F in elther orthophosphate, silicate, or arsenate anion and 25% v/v in methanol. Nitric acid concentrations vary between 0.10 and 0.70 F. mately 30%. This methanol concentration was considered optimum for later use in the FIA system since the dilution effect of additional methanol causes 12-MPA response to diminish rather than increase after this point. Figure 6 is a plot of peak current as a function of nitric acid concentration for the heteropoly acids of phosphate, silicate, and arsenate a t a constant methanol concentration of 25% (v/v). It is evident from this plot that increasing nitric acid concentration between 0.10 F and 0.50 F has little effect on 12-MPA peak current. A nitric acid concentration of 0.10 F was chosen for use in the FIA system due to the lower and more stable background current present at this concentration. The remaining data of Figure 6 are discussed in a later section. Flow Rate Studies. Figure 7 is a plot of current-time peak area vs. flow rate for the system previously described. The response here behaves roughly as would be predicted by the integrated form of the equation for the limiting current of a single species to the surface of a channel electrode with fully developed laminar flow given by Meyer et al. (23). The irregularities which exist in the flat portion of the curve were traceable to mixing irregularities in the tee-joint used to rejoin the bypass carrier stream with the portion of the carrier stream flowing through the sampling valve. The tee-joint finally employed was selected on the basis of its simplicity and its
0 32 V.
0.38
0.40
Ag/AgCI)
Figure 8. Hydrodynamic voltammoagram for 9.0 X lo-' F orthophosphate at a platinum electrode: flow rate, 2.2 mL/min; carrier solution, 30% methanol (v/v), 0.10 F HNO,, 9.75 X lo4 F (NH4),Mo0,. ability to minimize the irregularities present in the flow rate response cure. Since system response and reproducibility do not change drastically for flow rates values between 1.02 mL/min and 2.80 mL/min, a flow rate of 2.2 mL/min was chosen for further work as a value which allows a reasonably high analysis rate (70 samples per hour) as well as reasonable conservation of carrier stream. For analysis where faster through-put takes precedence over solvent conservation, flow speeds up to 3 mL/min are practical which allow sampling rates as large as 90 samples per hour. Hydrodynamic Voltammagram. The dependence on potential of the electrochemical chemical detector was evaluated by plotting amperometric system response vs. applied potential at a constant pumping rate of 2.2 mL/min (Figure 8). Since the reduction of 1ZMPA takes place in two separate steps at overlaping potentials, the diffusion limited plateau for the final peak is very small. A potential of 0.28 V vs. a Ag/AgC1 electrode was chosen on the basis of this plot for futher experiments and analysis. Analysis at lower potentials was not attempted due to the electrode fouling problems found earlier with cyclic voltammetric studies. Interference Studies. Cyclic voltammograms run on solutions of uric acid and ascorbic acid in 30% (v/v) methanol water solutions which were also 0.1 F in nitric acid indicate no electroactivity of these species at potentials which would interfere with this analysis. A number of inorganic species such as arsenate and silicate, which form heteropoly acids with
2222
ANALYTICAL CHEMISTRY, VOL. 56, NO. 12, OCTOBER 1984
Table I interfering species Fe(II1) Cu(I1) silicate arsenate glucose
% signal enhancement or attenuationa O.lb
1.0b
lob
lOOb
6.02 f 2.43 -0.15 f 2.06 1.38 f 1.56 2.54 f 2.59 -0.34 f 1.82
27.01 f 1.71 -6.16 f 1.85 1.84 f 1.86 0.25 f 1.71 -0.50 f 2.40
177.88 f 2.51 -7.30 f 2.45 16.62 f 1.58 2.94 f 2.93 -1.66 f 1.76
14.02 f 1.97 101.90 f 4.16 12.66 f 1.95 -2.73 f 2.23
(+) signal enhancement; (-) signal attenuation. *Ratios of interfering species concentration to POa” concentration.
molybdate, as well as copper(I1) and iron(III), however, were found to be possible interferences by these studies. To further test the effect of the presence of these on the analysis, solutions of arsenate, silicate, copper(II), and iron(II1) as well as glucose were prepared with ratios of interference to orthophosphate between 0.1 and 100. These solutions were analyzed in the flow injection system and the signals compared with the signal produced by a standard phosphate solution. The percent enhancement or attenuation achieved for each species at the various ratios appears in Table I, as well as the standard deviation of each. It can be seen from this table that silicate caused substantial interference at ratios of 1O:l in this method. Iron(II1) also can be seen to cause a serious positive interference even at iron to phosphate ratios of 0.1. Arsenate can be seen to interfere only at ratios of 1O:l or greater. Copper(I1) can be seen to have a mild attenuating effect at ratios of 1:lor greater and a substantial enhancing effect at a ratio of 1OO:l. No significant effect can be found for glucose even at ratios of 1OO:l. Since it might be desirable to apply this method to analysis of phosphorus in natural waters, an attempt was made to determine conditions which would inhibit the formation of silicomolybdic acid (12-MSA). Since Chalmers and Sinclair (10) report that acid concentrations affect the stability of the molybdenum heteropolyacids of silicate, arsenate, and phosphate, cyclic voltammetric currents were measured for each of these heteropolyacids as a function of nitric acid concentration. These currents were measured at 0.34 V (vs. SCE) which was the peak potential for 12-MPA. Plots of these data appear in Figure 6. These data indicate that the interference from silicate may be drastically lowered without degrading phosphate response by increasing the nitric acid concentration to values between 0.4 F and 0.5 F. Attempts to perform analysis and interference studies in the flow system itself at these higher acid concentrations were not attempted due to difficulties encountered in maintaining constant base line currents and high noise levels. Since the peak potential of 12-MSA was 0.30 V (vs. SCE), it is only slightly more difficult to reduce than 12-MPA making it impossible to increase the selectivity of the technique for phosphate analysis by potential adjustment. The low current response for 12-MAA at all of the above acid concentrations indicates only minor interference from arsenate as seen in Table I. Large concentrations of arsenate are not normally found in natural samples and hence would not be expected to be a commonly occurring interference. Calibration Curves. Plots of peak area as a function of orthophosphate concentration were found to be linear up to a concentration of 5.0 X M. At higher concentrations of orthophosphate the calibration curve deviates negatively from linearity since the ratio of molybdate to phosphate is too low at this point to force complete formation of 12-MPA. Although it is possible to extend the linear dynamic range by increasing molybdate concentration, attempts to accomplish this led to large increases in background current which caused currents produced by samples in higher concentration ranges
Flgure 9. Replicate res onse peaks for solutions (left to right) 1.00 X M, 7.50 X 10- M, 5.00 X lo-’ M, and 2.5 X lo-’ M in orthophosphate: sensitivity, 500 nA full scale; chart speed, 0.1 in./min; pumping speed, 2.2 mL/min.
f
to exceed the current capacity of the measurement system. Linear regression analysis of calibration curves produced correlation coefficients typically of 0.997. Predicted intercepts of the integral response axis were less than one standard deviation from zero. The sensitivity of the technique is approximately 4 mA L/mol. The detection limit calculated by determining the concentration of orthophosphate corresponding to twice the noise level was found to be 1.00 X M in the diluted injection mixture. This corresponds to a total of 0.45 pmol of orthophosphate. Figure 9 contains replicate response peaks for solutions of 1.00 X M to 2.50 X lo4 M orthophosphate. The slightly increased peak height of the initial peak in each series indicates a small memory effect which was observed at high sample injection rates. No special advantage except convenience was derived by measuring integral response. Urine Analysis. In order to test for any systematic difference between the electrochemical FIA method and the standard spectrophotometric method, ten urine samples were analyzed by both methods. The results of these analysis appear in Table I1 as well as the differences between the results obtained by each method. The average of these differences was found to lie within the 95% confidence interval estimate of the average difference in results between the two methods. This leads to the conclusion that the two methods do not differ in average results at the above confidence level.
CONCLUSION The ability to analyze for orthophosphate by flow injection analysis using a commercially available amperometric detector
2223
Anal. Chem. 1904, 56,2223-2228 ~~
Table 11
sample
phosphate concn M FIA analysis spectrophotometric (Xa) analysis (Xb)
1
2 3 4 5 6 7 8 9
10
X
Xa - Xb
0.002 56 0.005 47 0.007 37 0.002 80 0.002 377 0.00301
0.001 98
0.000 99
0.001 12 0.011 06
-0.000 14
0.011 42
+O.OOO 61
0.002 82
-0.000 29
0.01062 0.012 03 0.002 53
0.005 37
0.008 26 0.002 71 0.002 371 0.00334
+O.OOO 58 +o.ooo 10 -0.00 89 +O.OOO 08 +O.OOO 006
-0.000 33
Interference from silicate ion is a more difficult problem which can be partially solved by increasing the nitric acid concentration in the sample and mobile phase. This could not be done in our system however without greatly decreasing base line stability and noise characteristics. It is planned to continue work on this problem in our laboratory in order to make this method more competitive with present spectrophotometric techniques. Registry No. Orthophosphate, 14265-44-2; iron, 7439-89-6; silicate, 12627-13-3; 12-MPA, 12026-57-2; 12-MSA, 12027-12-2; 12-MAA, 12005-91-3.
LITERATURE CITED
-0.000 44
= 0.000071
Confidence limits = 0.000071 f 0.00033 at 95% confidence level ~~
was assessed. This assessment indicates that systems of this sort have potential for analysis in a large number of sample types at cost equal to or lower than the FIA system employing spectrophotometric detection. The added perquisites of low detection limits, simplicity, and fast sampling times increase this competitiveness. The negative aspects of this method center around the problem of iron(II1) and silicate interferences since both of these ions are found in a large number of natural and biological substances which are routinely analyzed for orthophosphate. The problem of iron removal can probably be solved by pretreating samples either in the flow system or before injection by passing them through an ion exchange column or a column filled with a chelating resin. This may be accomplished simply and should serve to remove other possibly interfering cations from the sample.
Fisk, C.; SubbaRow, Y. J . Biol. Chem. 1925, 66, 374. Crouch, S. R.; Malmstadt, H. V. Anal. Chem. 1967, 39, 1090. "Methods for Chemical Analysis of Water and Wastes"; Environmental Monitoring and Support Laboratory, United States Environmental Protection Agency: Clnclnnati, OH. Johnson, K. S.; Petty, R. L. Anal. Chem. W62, 5 4 , 1185. Ruzlcka, J.; Hansen, E. H. "Flow Injection Analysis"; Wiley-Interscience: New York, 1981. Fogg, A. G.; Bsebsu, N. K. Analyst(London) 1981, 106, 1288-1295. Fujinag, T.; Okazaki, S.; Hari, T. Bunsekl Kagaku 1960, 29, 367. Adams, R. N. "Electrochemistry at Solid Electrodes"; Marcel Dekker: New York, 1969. Kircher, C. C.; Crouch, S. R. Anal. Chem. 1982, 5 4 , 879-884. Chalmers, R. A.; Sinclair, A. Anal. Chim. Acta 1965, 33, 384. Chalmers, R. A.; Sinclair, A. G. Anal. Chim. Acta 1966, 3 4 , 412. Tsigdinos, G. A.; Hallada, C. J. J . Less-Common Met. 1974, 36, 79-93, and references therein. Meyer, R. E.; Banta, M. C. Lantz, P. M.; Posey, F. A. J . Electroanel. Chem. 1971, 30, 345-358.
RECEIVED for review January 17,1984. Accepted May 30,1984. The authors gratefully acknowledge funding for this project by the Graduate School of the University of Alabama in Birmingham through a Graduate School Faculty Research Grant.
Comparison of Mass Spectrometric Methods for Trace Level Screening of Hexachlorobenzene and Trichlorophenol in Human Blood Serum and Urine Richard A. Yost* and Dean D. Fetterolf' Department of Chemistry, University of Florida, Gainesville, Florida 32611
J. Ronald Hass and Donald J. Harvan Laboratory of Molecular Biophysics, National Institute of Environmental Health Sciences, P.O. Box 12233, Research Triangle Park, North Carolina 27709 Alan F. Weston, Peggy A. Skotnicki, and Nannette M. Simon Occidental Chemical Corporation, Research Center, Grand Island, New York 14072 The comblnatlon of more seiectlve mass spectrometric technlques (high-resolutlon mass spectrometry and tandem mass spectrometry-both triple quadrupole MS/MS and MIKES) with short retention time gas chromatography Is compared with conventional capillary GC/MS for the screening of human serum and urine for hexachiorobenreneand trlchiorophenoi. The various techniques are evaluated In terms of detection ilmlts, ability to obtain zero blanks, reproducibility, Ilnearity, and speed of analysis. GC/MS/MS makes possible screening at sub-part-per-billion levels 20 times more rapidly than by HRGWMS.
Current address: Forensic Science Research and Training Center, FBI Academy, Quantico, VA 22135.
Remedial construction a t chemical landfill sites often requires that excavation and earthmoving occur in areas of suspected chemical contamination. Dust generated by the construction activities may contain chemicals from the site; also chemical vapors may be released if the landfill itself is penetrated. Site workers and local area residents therefore have a potential for exposure to chemicals from the landfill. The potential for exposure also exists in other chemical waste handling activities such as waste treatment and transportation. Analysis of blood and urine has been used to assess human exposure to chemicals (1,2). This study compares the utility of several mass spectrometric techniques for the screening of chlorinated organic compounds in human blood serum and urine.
0003-2700/84/0356-2223$01.50/00 1984 Amerlcan Chemical Society
