ANALYST, MARCH 1993, VOL. 118 245 Determination of Total Phosphate in Waste Waters by On-line Microwave Digestion Incorporating Colorimetric Detection Kathleen E. Williams and Stephen J. Haswell" School of Chemistry, University of Hull, Hull, UK HU6 7RX David A. Barclay CEM Microwave Technology Ltd., 2 Middle Slade, Buckingham Industrial Park, Buckingham, UK MK18 1 WA Gaynor Preston Severn Trent Laboratories, 156-1 70 Newhall Street, Birmingham, UK B3 ISE A flow injection (FI) system for on-line microwave digestion of waste water samples with determination of total phosphate by means of colorimetric detection is described. Acidified samples are introduced into a water carrier stream and digested under continuous-flow conditions in thin-bore tubing. Following digestion, the samples are cooled on-line and then combined with further reagent streams for subsequent colorimetric detection based on the reduction of heteromolybdophosphoric acid to Molybdenum Blue as an FI peak at 690 nm.Signals from the flow-through detector are recorded as peak height on a chart recorder. Optimization of parameters such as digestion tube length, digestion tube diameter and reagent concentrations is discussed. Calibration was found to be linear up to 20 ppm of phosphate with a limit of detection of 0.10 ppm. Samples of waste water can be analysed at a rate of approximately 2 min per sample with typical sample relative standard deviations of <5% being achieved. Results for a range of samples were found to agree with those obtained by a conventional 'block' digestion autoanalyser method.The pre-treatment of samples with pyrophosphate phosphohydrolase ensured that the determination of total phosphate as orthophosphate could be achieved for samples containing pyrophosphate species. Keywords: Microwave digestion; on-line digestion; colorimetric detection; waste water; phosphate determination Waste water samples, which often contain high levels of suspended organic solids, are routinely analysed by the water industry for phosphate concentration. 1 This form of routine analysis requires rapid, simple and robust methodology in order to handle the large numbers and varying concentrations of phosphate species found in such samples. Most procedures for the determination of trace elements require the digestion of sample prior to instrumental analysis.2 Commonly, this preliminary step proves to be both laborious and time consuming, particularly when using traditional digestion methods such as wet and dry ashing.3 However, the develop- ment of microwave sample dissolution has proved to be of great advantage in that it dramatically reduces digestion time, and reduces both volatile analyte loss and sample contamina- tion from the atmosphere.4 The development of an on-line microwave digestion method would offer a more attractive approach to sample preparation over the discrete open or bomb digcstion of samples.There have been a limited number of reports in the literature of systems for flow injection (FI) based methodologies5 that will accommodate on-line sample preparation.&' 1 This paper describes the development of an FI system with on-line microwave digestion and subsequent colorimetric detection of phosphate in waste water samples.Experimental Reagents All chemicals were of analytical-reagent grade, and distilled, de-ionized water was used throughout. Ammonium heptamolybdate (0.005 mol 1-1) (BDH, Poole, Dorset, U K ) in 0.4 moll-' nitric acid (Fisons, Loughborough, Leicestershire, U K ) . * To whom correspondence should be addressed. Ascorbic acid (2% mlv) (Aldrich, Gillinghum, Dorset, U K ) in 1 1 distilled water with 10 ml of glycerol [May & Raker (now Rhone-Youlenc), Dagenham, Essex, UK]. A few drops of Triton X-100 (Sigma, Poole, Dorset, UK) were added to each of the above reagents. Phosphate stock standard solution, 100 ppm.Prepared from potassium dihydrogen orthophosphate (KH2P04) (BDH) in 5% vlv nitric acid. Nitric acid (5% vlv) in distilled water. Stock solution o f sodium tetrametuphosphate [(Na- Stock solution of anhydrous trisodium trimetaphosphate Inorganic enzyme pyrophosphatase (pyrophosphate phos- P02)4-4H20] (100 ppm) (Albright & Wilson, London, U K ) . C(NaP~li)liI (100 PPm) (Sigma). phohydrolase; EC 3.6.2.1.) (Sigma). Apparatus The on-line system is illustrated in Fig. 1 and consisted of the following: a water carrier stream in 0.5 mm i.d. poly- (tetrafluoroethylene) (PTFE) tubing; an Tsmatec MV-Z pump (Glattbrugg, Zurich, Switzerland); a Rheodyne injection valvc (Model 5020 Anachem, Luton, Bedfordshire, UK) with a 1 ml sample loop; a Model MDS81 microwave oven (CEM, Buckingham, UK) containing 7.2 m of 0.5 mm i.d.PTFE tubing wrapped around a PTFE box; a 7 m cooling loop in an anti-freeze bath cooled by Peltier devices (M11069T-O3AC, Marlow Industries, Tadworth, Surrey, UK); a Rheodyne injection valve (Anachem 5020) with an on-line back-flush filter (made in-house) fitted instead of a sample loop; a CEM pressure sensor; a 516.75 kPa (75 psi) back-pressure regulator (Anachem P736); and a de-bubbler fitted to the outlet from the back-pressure regulator. The outlet from the back-pressure regulator was coupled to a colorimetric detection system consisting of: two colorimetric reagent streams for molyhdate and ascorbic acid reagents; two Gilson Minipuls 2 peristaltic pumps (Anachem); a 100 cm mixing coil; a 4 m reaction coil in a thermostatically controlled246 Carrier ANALYST, MARCH 1993, VOL.118 11.8 I Cooler Pressure sensor Back-pressure regulator V Waste 150 cm mi coil Reaction Xing coil I ( 4 m ) Waste Fig. 1 Schematic diagram of the on-line digestion FI system water-bath; a Cecil spectrophotometer (S & M Products, Didsbury, Greater Manchester, UK); and a Kipp & Zonen (Delft, Holland) BD112 flat-bed chart recorder. Poly(tetrafluoroethy1ene) tubing of 0.5 mm i.d. was used throughout the whole system. Procedure Standards covering the range 0-20 ppm were prepared by taking appropriate dilutions of the stock standard phosphate solution and making up to volume in 5% v/v nitric acid. A 0.5 mi aliquot of concentrated nitric acid was added to 10 ml of waste water sample, resulting in a 5% v/v acidic solution.Standards and samples were analysed by injecting 2 ml aliquots into the system. Absorbance signals were recorded on a chart recorder and peak height measurements taken. Total analysis time for each sample was approximately 2 min. Replicate measurements were performed for each standard and sample. Study on phosphate species present in samples Samples and standards were prepared as described in the above procedure but without acidification. Sufficient units of enzyme activity were added to ensure hydrolysis of all pyrophosphate present in the standards and samples to orthophosphate. [Note: 1 U of enzyme activity will liberate 0.1 ppm min-1 of orthophosphate at pH 7.5 and 25"C.I For example, 100 U of enzyme activity (500-700 U mg-I) were added to 10 ml of a 10 ppm standard and the solution was left to stand at room temperature for 1 min before being acidified and analysed as detailed above.Results and Discussion Optimization of Digestion Conditions It has been previously determined that the optimum micro- wave power and acid strength for sample digestion are 90 and 5% v/v, respectively.:! The remaining variables controlling the _ _ 3.4 3.8 4.2 4.6 5.0 5.4 5.8 6.2 Carrier flow rate/ml min Effect of carrier flow rate through the FI system on Fig. 2 response 2o I 18 16 . c-' L .: 14 JZ - / signal 10 "y 8 1 I I I 0.5 1.5 2.5 3.5 Effect of sample injcction volume on signal response Sample injection volume/mI Fig. 3 rate of digestion of sample in the microwave cavity are, therefore, flow rate, sample volume tube length and internal tube diameter. Signal sensitivity was evaluated by comparing the absorbance response for a standard 5 pprn phosphate solution in 5% v/v nitric acid.Having already selected a tube length of 7.2 m and an internal tube diameter of 0.5 mrn,lO the effect of flow rate on signal response was investigated. The colorimetric reagent flow rate was kept constant throughout this experiment. Fig. 2 shows a plot of absorbance (measured as peak height) as a function of carrier flow rate through the F1 system. At lower flow rates, dispersion will tend to increase with the sample slug being observed as a broad peak. However, as the carrier flow rate increases, the peaks become narrower while maintaining the same peak area; hence an increase in peak height is observed.It can be observed from the graph that an optimum signal response occurs when the carrier flow rate is set at 4.5 ml min-1, indicating that on reaching the colorimetric reagents, an optimum reaction time and hence colour development has been achieved. At the higher flow rates (4.5-6 ml min-1) there is insufficient reaction time to achieve optimum colour development and so the peak response rapidly begins to tail off. Fig. 3 illustrates the effect of sample injection volume on signal response and clearly indicates that an optimum peak height for absorbance is achieved when a sample injection volume of 2 ml or more is used. Hence an optimum carrier flow rate of 4.5 ml min-1 and an injection volume of 2 ml were used. Optimization of Phosphate Detection by FI In order to obtain optimum signal sensitivity in the detection system it was necessary to minimize the dispersion of the sample slug as it passed through the detector.The degree of dispersion of the sample slug as it passed through the detectorANALYST, MARCH 1993. VOL. 118 L i - - - I I 1 I I 1 247 11.1 I I I I 0 0.008 0.016 0.024 0.032 Effect of molybdate concentration on signal response Moly bda t e co ncen t rat ionim o I I 1 Fig. 4 1 - 10 I 5 4 - A 0 2 4 Ascorbic acid concentration (% m/v) Fig. 5 Effcct of ascorbic acid conccntration on signal response was controlled in effect by the flow rate, and the intensity of colour produced was found to be dependent on reagent concentration. Therefore, the two variables that most affect the signal sensivity were flow rate and reagent concentration.Signal sensitivity was evaluated by observing the absorbance response as peak height while holding the sample carrier flow rate constant at the pre-determined optimum. Figs. 4 and 5 show the effect of molybdate concentration and ascorbic acid concentration on signal response, respectively. Although the graph in Fig. 4 shows an optimum molybdate concentration being approached at 0.03 mol 1-1, it was found to be more practical to operate at a concentration of 0.015 mol 1-1 molybdate, as solutions above this concentration were found to be unstable and began to change colour and re-precipitate after several hours. Fig. 5 simply shows that an optimum signal response is achieved at an ascorbic acid concentration of 2.5% m/v .Hence the optimum reagent concentrations were chosen as 0.015 mol 1-1 molybdate and 2.5% m/v ascorbic acid. A study of the effect of reagent flow rate on peak height was carried out and the results are summarized in Fig. 6. As the flow rate increases, an increase in peak height is observed owing to a reduction in dispersion while maintaining sufficient reaction time for colour development to occur. At high flow rates, a reduction in peak response is attributed to inadequate time for reaction and colour development to occur. Hence an optimum peak response is observed at a flow rate of 2.4 ml min-1. The graph shown in Fig. 7 illustrates the effect of changing the reaction coil length for colour development on signal response. As the coil length increases, the time allowed for colour development increases and reaches an optimum at approximately 3 m, at which point the peak response begins to decrease owing to an increase of dispersion in the longer coils.Hence it can be concluded that at reagent concentrations of 0.015 rnol 1-1 (molybdate) and 2.5% m/v (ascorbic acid), a flow rate of 2.4 ml min-1 and a coil length of 3 rn, an optimum 7 6 E g 5 0, c Y a .- 0 4 3 3 2 1 Fig. 6 Effcct of reagent flow rate on signal response 11.9 11.8 A I! 11.2 11.3 1 \ b- 2.0 2.4 2.8 3.2 3.6 4.0 4.4 4.8 5.2 Reaction coil length/m Effect of changing reaction coil length for colour development Fig. 7 on signal response signal response was obtained in the FI detection system for a given fixed carrier flow rate of 4.5 ml min- 1 .Calibration Calibration using standards over the range 0-20 ppm phos- phate in 5% v/v nitric acid was carried out and proved to be linear up to 20 ppm with relative standard deviations (RSDs) 4% ( n = 10). The equation lor the linear calibration graph was: y = 1854~ - 0.152 (Y = 0.9983). A limit of detection of 0.1 ppm was obtained with a samplc throughput time of 30 h-1. Analysis of Samples Table 1 shows a comparison of results for the determination of total phosphate in waste waters by a conventional 'block' batch digestion method with those obtained by the proposed FI methodology. The results, in general, were in good agreement with those obtained with the standard 'block' method and analysis was completed much more rapidly. The FI method was found to have a precision (RSD) of 1.4% (n = 10) at the 5 pprn level and a sampling rate of 25 h-1 (including sample preparation). No precision data (RSDs) were avail- able for the batch method but they were expected from performance data to be in the range 1-5%.No particulates were observed in the flow system on leaving the microwave, indicating that complete digestion of suspended solids in the sample had taken place. The above results indicate that the proposed FI method represents a rapid and effective analytical methodology comparable in total phosphate recoveries to the traditional batch technique. Phosphate Species Study Previous studies have shown that only three classes of phosphate are stable to any extent in aqueous solution:" (i) orthophosphate (monophosphate); (ii) straight-chain poly- phosphates (including pyro- and tripolyphosphate); and (iii) ring met aphospha tes.248 ANALYST.MARCH 1993, VOL. 118 Table 1 Analysis of waste water samples by batch and FI methods for the determination of total phosphate Sample number 1 2 3 4 5 6 7 8 9 10 11 Total phosphatc Total phosphatc by batch method by on-line FI (PPm) method (PPm) 10.7 10.5 14.9 14.4 17.0 16.0 11.1 11.8 2.2 2.4 4.2 4.4 5.8 6.3 20.9 21.1 14.3 15.6 6.4 6.0 31.7 30.5 Total analysis time: 3 h 25 min Precision ( n = 10) 1.4 1.9 4.8 1.4 1.6 1.8 2.9 2.3 3.1 2.6 1.4 (RSD) (%) Several studies have also shown that after complete hydrolysis, all of the ring and chain phosphates are converted into orthophosphate. 13 It was considered important to determine whether or not all forms of phosphates present in waste water samples are converted into orthophosphate prior to detection and hence the total phosphate content is determined.Hence, several inorganic phosphates were chosen to be analysed and their percentage recoveries determined as orthophosphate. The phosphates analysed were as follows: tetrasodium pyrophos- phate (Na4P407- 10H,O); sodium tetrametaphosphate [(Na- P02)4.4Hz0]; and trisodium trimetaphosphate [(NaPO&] (anhydrous). By using the system and methodology described previously the results showed that good recoveries were obtained for the tetra- (101%) and trimetaphosphate (99%) but a fairly poor rccovery was observed for the pyrophosphate (67%). This low percentage recovery is attributed to the rate of hydrolysis of pyrophosphate to orthophosphate.The three main factors affecting the rate of hydrolytic degradation of ring and chain phosphates in aqueous solution are temperature, pH and enzymes.13 As the system was already operating at optimum temperature and pH (strong acid), it was necessary to examine the effects of enzymes on the rate of hydrolysis of pyrophos- phate. The inorganic enzyme pyrophosphatase (pyrophosphate phosphohydrolase; EC 3.6.1.1) was considered to be a suitable enzyme for this experiment. Standard solutions were analysed using the methodology described and the results are reported in Table 2. Good recoveries were observed for ortho-, trimeta- and tetrameta- phosphate, but a reduction of signal was found for the pyrophosphate standard. However, after treatment of the standard with the inorganic enzyme pyrophosphate phos- phohydrolase, a recovery of 99.1% was achieved.On analys- ing waste water samples, the same reduction in signal was observed as before; however, the addition of enzyme en- hanced the signal to give a spiked recovery of 99%. It can, therefore, be concluded that pre-treatment of samples with enzyme prior to acidification and analysis results in good percentage recoveries of total phosphate in samples where pyrophosphate might be present. Although pyrophosphates were not thought be be present in the waste water samples analysed, the proposed method is recommended where the form of phosphate in a sample is not known. Table 2 Pcrcentage recovery of samples and standards for the determination of phosphate using the proposcd FI method Rccovcry (as orthophosphate) Sample ("/.I RSD ( Y o ) Orthophosphatc (10 ppm standard) 100 3.4 Trimetaphosphate (10 ppm standard) 101 2.4 Tctrametaphosphate (10 ppm standard) 99.3 3.I Tctrapyrophosphate (10 ppm standard) 66.4 4.6 Pyrophosphatc (10 ppm standard) + enzyme 99.1 2.7 Waste water sample + Wastc water sample + 2 ppm pyrophosphate spike (2 ppm) 68.1 3.1 pyrophosphate spike + enzyme 99.5 2.8 Conclusion The proposed on-line sample digestion method with colori- metric detection was found to have a limit of detection of 0.1 ppm, a precision (RSD) of <5% ( n = 10) and a sample throughput rate of 25 h- 1 (including sample enzyme prepara- tion). No particulates were observed in the flow system on leaving the microwave, indicating that complete digestion of suspended solids in the samples had taken place.Results for a range of waste water samples agreed with those obtained by a conventional 'block' digestion autoanalyser method. The pre-treatment of samples with the inorganic enzyme pyro- phosphatase ensured that the determination of total phos- phate as orthophosphate could be achieved for samples containing pyrophosphate species. 1 2 3 4 5 6 7 8 9 1 0 11 12 13 14 References Total Nitrogen and Total Phosphorus in Sewage Sludge. Section C. Methods for the Examination of Waters and Associafed Materials, HM Stationery Office, London, 1985. Matusiewicz, H., and Sturgeon, R. E., frog. Anal. At. Spectrosc., 1989, 12, 21. Minczewski, J., Chwastowska. J., and Dybezynski, R., Separa- tion and Preconcentration Method3 in Inorganic Trace Analysis , Halsted Press, Chichester, 1983. Gorsuch, T. T., The Destruction of Organic Matter, Pcrgamon Press, Oxford, 1970. RfiiiCka, J . , and Hansen, E. H., Flow Injection AnalyJis, Wiley, New York, 2nd edn.. 1988. Burguera, M., Burguera, J . L.. and Alarcon, 0. M., Anal. Chim. Acta, 1086, 179. 351. Petersen, C., New Sci., 1989, September, 44. Burgucra, J. L., de la Guardia, M., and Salvador, A . , 1. Flow Inject. Anal., 1988, 5 , 121. Haswell, S. J., and Barclay, D., Analyst, 1992, 117, 117. Williams, K. E., M.Sc. Thesis, University of Hull, 1991. Hinkamp, S., and Schwedt, G.. Anal. Chim. Acta, 1990, 236. 34s. van Wazer, J. R., in Encyclopedia of Chemicul Technology, eds. Kirk, R. C., and Othmcr, D . F . , Interscience, New York, 1953, van Wazer, J. R., Anal. Chem., 1954, 26, 1755. van Wazer, J. R., Phosphorus and its Compounds. Volume I . Chemistry, Wiley-Interscience, New York, 1958. V O ~ . X, pp. 403-510. Paper 210551 7A Received October 15, 1992 Accepted November 3, 1992