ANALYST, JANUARY 1984, VOL. 109 19 Sequential Flow Injection Voltammetric Determination of Phosphate and Nitrite by Injection of Reagents into a Sample Stream Arnold G. Fogg and Nuri K. Bsebsu Chemistry Department, Loughborough University of Technology, Loughborough, Leicestershire, LEI I 3TU, UK For the intermittent analysis of large-volume water samples, such as a hydroponic fluid, a modified use of flow injection analysis is proposed. In this modification the water sample is made the eluent in the flow injection system and reagents for the determination of individual components are injected into the sample stream sequentially. As an illustration of this principle a sample solution containing phosphate and nitrite has been analysed in this way using acidic molybdate and acidic bromide reagents to determine phosphate and nitrite voltammetrically at a glassy carbon electrode.Phosphate was determined at the 0.5-50 x 10-5 M level and nitrite at the 0.55 x 10-4 M level without mutual interference and with good precision. Keywords: Flow injection analysis; phosphate determination; nitrite determination; reagent injection In the usual application of flow injection analysis (FIA), aliquots of sample are injected into a stream of reagent and the derivative that is formed is determined at a suitable detector.’ Previously, flow injection voltammetric procedures were developed in this laboratory for determination of phosphate2.3 and nitrite4 using this normal procedure. Early in these studies the determination of phosphate and other nutrient ions in hydroponic fluids using this approach was being considered.3 In this and similar applications the requirement is not for a rapid throughput of samples but for intermittent analyses at infrequent intervals of a single sample of changing composi- tion.Further in these applications the sample solution is in plentiful supply and is inexpensive. At that time in unpub- lished work the possibility of making determinations by injection of reagent into a sample stream was demonstrated. Johnson and Petty5 have reported a visible spectrophoto- metric method for the determination of phosphate using FIA with injection of reagent. They point out that whereas with normal FIA, in which the sample is injected into a reagent stream, the sample is diluted by dispersion as it passes to the detector, in reverse FIA, where the reagent is injected into a sample stream, the concentration of “sample” in the injected reagent plug increases as the plug passes to the detector.For this reason, assuming that a sufficiently concentrated reagent solution is injected, the sensitivity of the reverse technique should be greater than that of the normal technique. Johnson and Petty5 obtained a 5-fold increase in sensitivity for the reverse technique over that reported by R6iiCka and Hansenl using the analogous normal technique. In this work the application of the reverse technique with voltammetric detection has been demonstrated. Further, the sequential injection of different reagents to determine several constituents of a sample stream has been illustrated by the determination of phosphate and nitrite.Experimental Flow-injection analysis was applied as before2-4 except that the sample solution was used as the eluent. The flow of sample stream was usually produced by means of a Metrohm pressure bottle system (EA 1101) working at 0.8 bar, although a peristaltic pump (Gilford Minipuls 2) was used also. Injections were made by means of a Rheodyne injection valve (5020). The sample stream was presented to a glassy carbon electrode (Metrohm EA 286) in the wall-jet configuration. A Metrohm detector cell (EA 1096) was used, but without inserting the counter and reference electrodes and with partial immersion of the cell in electrolyte (0.01 M sulphuric acid). Contact between the electrolyte and the counter and reference electrodes was made by means of salt bridges.The glassy carbon electrode was cleaned with 1 M sodium hydroxide solution daily, or as required. It was found to be unnecessary to de-gas the sample stream. The potential of the glassy carbon electrode was held at +0.22V (versus S.C.E.) in phosphate determination and at +0.30 V in nitrite determina- tion using a PAR-174 polarographic analyser (Princeton Applied Research). Current signals were monitored on a Tarkan 600 Y - t recorder. All the results reported here were obtained with the Metrohm pressure vessel and a Metrohm detector cell. The same glassy carbon electrode was used throughout. Repeat injections of a reagent into a sample solution gave reproducible signals; coefficients of variation (eight injec- tions) were typically <1 YO.Reagents Standard orthophosphate solution, 3 X 10-3 M (285 pg ml-1 of POq3-). Dissolve 0.408 g of analytical-reagent grade potas- sium dihydrogen orthophosphate in water and dilute to 11 in a calibrated flask. This solution is 3 x l o - 3 ~ in phosphate. Prepare less concentrated standard solutions by dilution. Acidic molybdate solution, 0.5%mlV. Add 0.6 ml of analytical-reagent grade concentrated sulphuric acid to 60 ml of water. Dissolve 0.5g of ammonium molybdate in the resulting solution and dilute to 100 ml with water. Standard sodium nitrite solution, approximately 1 X 10-2 M. Dissolve approximately 0.172 g of analytical-reagent grade sodium nitrite, accurately weighed, in water and dilute to 250ml in a calibrated flask.This solution is 1 X 10-*M in nitrite. Prepare more dilute standard solutions from this solution. Acidic bromide solution, 20% mlV in potassium bromide and 3 . 2 ~ in hydrochloric acid. Dissolve 20g of potassium bromide in 70ml of water, add 27.6ml of concentrated hydrochloric acid, cool the solution, dilute to 100 ml and mix. Procedures Direct injection of acidic molybdate reagent into a stream of phosphate solution Inject 100 pl of 0.6% mlV acidic molybdate reagent into a stream of phosphate solution (0.5 x 10-5-50 x 10-5 M). Use a 3-m delay coil (0.58 mm bore).20 ANALYST, JANUARY 1984, VOL. 109 Direct injection of acidic bromide reagent into a stream of nitrite solution Inject 100 1-11 of 20% mlV acidic bromide reagent into a stream of nitrite solution (0.5 x 10-4-5 x 1 0 - 4 ~ ) .Use a 3-m delay coil (0.58 mm bore). Results Injection of the 2% mlV acidic molybdate reagent (1001-11) used previously2.3 in the normal FIA method into a sample stream of 2.0 x l o - 4 ~ phosphate was studied initially. The phosphate sample stream was prepared as required by a 10-fold dilution of the stock solution. The effect of the length of the delay coil between the injection valve and the detector cell at this concentration of phosphate was studied. The results given in Table 1 indicate that the optimum length for the delay coil is 3 m. Injections into a sample stream with a lower phosphate concentration (2 x l o - 5 ~ ) were made next using the 3-m delay coil and the same reagent. A proportionately much lower peak current value of 0.27yA was obtained and a double peak was apparent. The formation of a double peak would be expected when dispersion occurred only at each end of the sample bolus.Experience suggests, however, that they may be arising here also owing to an electroanalytical phenomenon or artefact that we have been unable so far to identify. The formation of a double peak was clearly unsatis- factory and attention was directed to studying the effect of the reagent composition. Reagent solutions were prepared with different concentrations of ammonium molybdate while the sulphuric acid concentration was kept constant at 0.6% V/V. Table 2 shows the effect of the concentration of ammonium molybdate used on the peak current obtained for a sample solution 2 x l o - 4 ~ in phosphate.The use of an ammonium molybdate concentration of 0.5% mlV was adopted. The optimum concentration of sulphuric acid used was found to be the same as previously, i.e., 0.6% VlV. The effect of the length of the delay coil on the signal obtained at the new optimum concentration of ammonium molybdate (0.5% mlv) was studied (see Table 3). The optimum length was found to be 3m, as before. The signals obtained when the optimised reagent (0.5% mlV in ammonium molybdate and 0.6% V/V in concentrated sulphuric acid) was injected into a sample stream of phosphate at the 2 X M levels were 0.61 and 6.18 PA, respectively. A rectilinear calibration graph was obtained using signals obtained with eluents 0, 1,2,3,4,10,20,30 and 40 X 10-5 M in phosphate. Determination of phosphate at the 1 X 1 0 - 6 ~ level was not possible, however, as the signal obtained was similar to the blank obtained by injecting reagent into distilled water.Attention was directed at the determination of nitrite by injection of reagent into sample solution. Injection of acidic bromide reagent that is 3 . 2 ~ in hydrochloric acid and 20% mlV potassium bromide into the sample stream of 2 x 10-4 M nitrite was studied initially. The nitrite sample stream was prepared as required by a 10-fold dilution of the stock solution. The effect of the length of the delay coil between the injection valve and the detector cell at this concentration of nitrite was studied. The peak heights at different delay-coil lengths are given in Table 4. The 3-m delay coil gave the largest signal.The effect of the reagent composition was studied next. Reagent solutions were prepared with different concentra- tions of potassium bromide while the hydrochloric acid concentration was kept constant at 3.2 M. The results in Table 5 show the effect of the concentration of potassium bromide used on the peak current obtained for a sample solution 2 X l o - 4 ~ in nitrite. Clearly the highest concentration of and 2 x bromide gives the greatest signal. The effect of the concentra- tion of hydrochloric acid used was then studied (see Table 6). Again the highest concentration gives the largest signal. Precipitation occurs at potassium bromide concentrations of >20% mlV in 3.2 M hydrochloric acid, and at hydrochloric acid concentrations >3.2 M in 20% mlV potassium chloride solution.Determination of nitrite at the 1 x 1 0 - 5 ~ level was not possible as the signal obtained was similar to that obtained for the blank, i.e., injection of reagent into distilled water. Nitrite can be determined at the 2 x l o - 3 ~ level but the signal is disproportionately less than that obtained at 2 x 10-4 M. Determinations of phosphate and nitrite were made on a series of solutions 0.5 x 10-5-50 x 10-5 M in phosphate and 0.5 X 10-4-5 x 10-4 M in nitrite. No mutual interference was observed within these limits. Typical signals are shown in Fig. 1. Table 1. Direct injection of acidic molybdate reagent: effect of delay coil length on peak current. Phosphate concentration of eluent = 2 X M. Molybdate concentration = 2% m/V Delaycoillength/m . .. . 1.0 2.0 3.0 4.0 5.0 Peakcurrent/pA . . . . . . 2.50 4.70 5.80 5.00 5.00* * Reproducibility bad. Table 2. Direct injection of acidic molybdate: effect of amount of ammonium molybdate on the peak current. Delay coil length = 3 m. Phosphate concentration of eluent = 2 x l o - 4 ~ Concentration of ammonium molybdateinreagent, YO m/V . . 0.5 1.0 1.5 2.0 Peakcurrent/pA . . . . . . . . 6.20 6.00 5.20 4.65 Table 3. Direct injection of acidic molybdate reagent: effect of delay coil length at optimised ammonium molybdate concentration. Phosphate concentration of eluent = 2 x 10-4 M. Molybdate concen- tration = 0.5% m/V Delaycoillength/m . . . . 1 2 3 4 5 Peakcurrent/pA . . . . . . 2.0 3.8 6.15 6.25* 5.75 * Reproducibility bad. Table 4. Direct injection of acidic bromide reagent: effect of delay coil length on peak current.Nitrite concentration in eluent = 2 X 10-4 M. Bromide concentration = 20% m/V. Hydrochloric acid concentration = 3.2 M Delaycoillength/m . . . . . . 1.0 2.0 3.0 4.0 Peakcurrent/pA . . . . . . . . 2.80 3.05 3.20 2.80 Table 5. Direct injection of acidic bromide reagent: effect of potassium bromide concentration on peak current. Nitrite concentra- tion in eluent = 2 X M. Hydrochloric acid concentration = 3.2 M. Delay coil length = 3 m Potassium bromide concentration, %m/V . . . . . . . . . , 5.0 10.0 15.0 20.0 Peakcurrent/pA . . . . . . . . 0.6 1.30 2.30 3.30 ~ Table 6. Direct injection of acidic bromide reagent: effect of hydrochloric acid concentration on peak current. Nitrite concentration in eluent = 2 x 10-4 M.Potassium bromide concentration = 20% m/V. Delay coil length =; 3 m Hydrochloric acidconcentrationh . . 1.0 2.0 3.0 3.2 Peakcurrent/pA . . . . . . . , 1.2 1.98 2.60 3.05ANALYST, JANUARY 1984, VOL. 109 21 1 C i A i Time 4 Fig. 1. Typical signals obtained for the determination of phosphate in the presence of nitrite using the reverse FIA procedure. Phosphate concentration in eluent: A, 2 x 10-5; B, 4 x 10-5; C, 6 x 10-5 and D, 6 x 10-5 M. Nitrite concentration in eluent: A-C, 1 X M; and D, zero Discussion PhosphatezJ and nitrite4 were determined previously by injection of sample solution into eluents consisting of acidic molybdate and acidic bromide reagents and monitoring at a glassy carbon electrode the molybdophosphate and nitrosyl bromide formed.Johnson and Petty5 studied the inversion of the roles of reagent and sample solution [which they termed reverse FIA (rFIA)] for the visible spectrophotometric determination of phosphate, and showed that increased sensitivity was obtained. This work has been concerned with assessing the reverse procedure for the determination of phosphate and nitrite voltammetrically. The voltammetric signals obtained in the reverse procedure were found to be of a similar magnitude to those obtained with the normal proce- dure, but it was not possible to make determinations at as low a level with the reverse procedure. This appears to be due to the particular characteristics of the voltammetric detection method. Thus, for successful voltammetry a reasonably conducting solution is required but unless additional elec- trolyte is added to the eluent, the blank is essentially pure water. Further, there is a background current associated with the reagents,2 and the signal associated with changes in this on injection appear to be more effectively minimised in the normal FIA method.A brief study of the addition of inert electrolyte to sample solutions has indicated that this does not readily solve the problem, but further studies are being made. Thus, whereas usable signals were obtained at the 1 x 1 0 - 6 ~ level of phosphate and nitrite by injection of sample into reagent, the minimum amounts to produce usable signals by injection of reagent into sample solution are 5 x 10-6 and 5 X 10-5 M , respectively. The determination of phosphate by the reverse FIA method has a useful rectilinear range and might be applied with advantage to the determina- tion of phosphate in samples such as hydroponic fluids.An advantage of using the sample as eluent is that contamination of the glassy carbon electrode occurs more slowly than it does in the normal FIA method (particularly for phosphate) in which reagent is passed over the electrode continuously during the determination. The satisfactory determination of nitrite by this reverse FIA procedure, except within a very narrow range of concentrations, has not proved possible so far. No mutual interference was found in the determination of phosphate and nitrite in the same eluent and the determina- tion of these ions by sequential injection of acidic molybdate and acidic bromide into the sample stream has been demon- strated.The sample mixture is somewhat artificial but the principle of sequential determination of constituents in a sample stream is clearly illustrated. The ions which are required to be determined in hydroponic fluids include phosphate, nitrate, potassium and ammonium. Calcium, sodium and certain heavy metals are also determined fre- quently. Computer-controlled systems for monitoring conduc- tivity, pH and temperature are currently in use in research institutes and elsewhere, and control systems for the monitor- ing of the ions mentioned above are being sought actively. Some progress has been made in this laboratory for on-line reduction of nitrate to nitrite prior to voltammetric determina- tion.6 We envisage the construction of a reverse FIA system in which both voltammetric and potentiometric detectors are used. Valve switching could be used to direct the sample stream to the required detector, to inject the required reagent for voltammetric determination and to change the potential of the voltammetric electrode as required. The authors thank Mr. G. S. Weaving of the National Institute of Agricultural Engineering for information and advice and for his interest in this work. N. K. B. thanks the people of the Socialist People’s Libyan Arab Jamahiriya for financial support and .leave of absence. 1. 2. 3. 4. 5. 6. References RfiiiEka, J., and Hansen, E. H., “Flow Injection Analysis,” John Wiley, New York, 1981. Fogg, A. G., and Bsebsu, N. K., Analyst, 1981, 106, 1288. Fogg, A. G., and Bsebsu, N. K., Analysf, 1982, 107, 566. Fogg, A. G., Bsebsu, N. K., and Abdalla, M. A., Analyst, 1982, 107, 1040. Johnson, K. S., and Petty, R. L., Anal. Chem., 1982,54,1185. Fogg, A. G., Chamsi, A. Y., and Abdalla, M. A., Analyst, 1983, 108, 464. Paper A3t142 Received May 19th, 1983 Accepted August 23rd, 1983