Analyst, JanGary, 1979, Vol.104, $$. 47-54 Determination of Ammonia in Low Concentrations 47 with Nessler‘s Reagent by Flow Injection Analysis F. J. Krug,* J. RiliiEka and E. H. Hansen Chemistry Department A Technical University of Denmark Building 207, DK-2800 Lyngby, Denmark A turbidimetric procedure for the determination of ammonia in low concentra- tions with the use of Nessler’s reagent is described. Both natural waters and soil extracts can be analysed a t a rate of up to 120 samples per hour with good precision and accuracy. The effects of reagent composition, flow- rate, temperature and protective colloids in the flow injection system are discussed in detail. Keywords : Flow injection analysis ; ammonia determination ; Nessler’s reagent ; turbidimetric determination ; continuous-flow vneasuvement In 1856 Nessler introduced a reagent consisting of mercury(I1) iodide and potassium iodide in alkaline solution for the qualitative and quantitative determination of ammonia.Known since then as Nessler’s reagent, it has been used extensively as the most sensitive test for ammonia. However, little was known regarding what happens when the components are mixed in aqueous solution until Sarkar and Ghosh1s2 undertook a detailed study over 20 years ago. They concluded that the composition of the colloidal precipitate formed was NH,-,Hg,I,, the value of TZ depending on the concentration of Hg142- and OH- in the reagent mixture and on the amount of ammonia. Thus, at low concentrations of ammonia, Nessler’s reagent was believed to react according to the equation 2Hg142- + NH, + 30H- + O<:i>H21 + 71- + 2H20 forming a brown precipitate, while at increasing ammonia concentrations and a constant OH- concentration the composition tended towards NHHg,I,.H,O (deep brown) and finally NH2Hg,I , (chocolate).Several analytical applications of Nessler’s reagent have been described in the The main advantages of the method are good sensitivity and simplicity. However, the Nessler method is accurate only if a number of conditions are carefully controlled.7 Previous work on the determination of sulphate by continuous flow injection turbidimetry has demonstrated that flow injection analysis can be used with this type of detection with very good control of the experimental parameters, such as precise timing of reaction sequences and reproducible mode of mixing and rate of nucleation, which allow good accuracy to be obtained even at high sampling rates.1° Several papers describing the use of flow injection analysis for the determination of various chemical species by different detection procedures have already been p ~ b l i s h e d .l ~ - ~ ~ The determination of ammonia as total nitrogen in plant digests was previously studied by using both potentiometric (air-gap electrode) and colorimetric (indophenol blue) dete~ti0n.l~ Although the air-gap electrode is suitable for the determination of ammonia in low con- centrations (down to 0.1 p.p,m. of NH4+-N), the method is slow, allowing a sampling rate of only about 60 determinations per hour. The indophenol blue method is reproducible, accurate and fast (120 samples per hour), but not sufficiently sensitive (applicable down to 5 p.p.m.of NH,+-N), and therefore it cannot be applied to soil extracts and natural waters. The aim of this paper is to show the feasibility of continuous-flow injection for the coloured turbidimetric determination of ammonia in low concentrations (0.1 p.p.m. of NH,+-N). A number of variables have been studied, such as optimum reagent concentrations, alkalinity, flow-rates, mixing coil lengths, temperature and presence of protective agents. Further, a practical approach to the determination of optimum reagent composition in a flow injection system is described. * Present address: Centro de Energia Nuclear na Agricultura, Caixa Postal 96, 13400 Piracicaba, ’350 Paulo, Brazil.48 KRUG et al.: DETERMINATION OF AMMONIA IN LOW CONCENTRATIONS Analyst, Vol. 104 Experimental Reagents and Samples All chemicals were of analytical-reagent grade. Nessler’s reagent. A procedure similar to that described by VogeP was used: 14g of potassium iodide are dissolved in 40 ml of distilled water and a 4% solution of mercury(I1) chloride is added while stirring continuously with a magnetic stirrer until a slight red precipitate is formed. Then 100 ml of 10 N sodium hydroxide solution are added and the volume is made up to 500 ml with distilled water. Further small amounts of mercury(I1) chloride are added until there is a permanent turbidity. The mixture is allowed to stand for Id, then decanted and the solution is stored in an amber-glass bottle.If any precipitate is formed during the next 7d, the solution must again be decanted or filtered to avoid deposition of solid particles in the optical path of the flow-through cuvette. A stock solution containing 100 p.p.m. of NH,+-N is prepared by dissolving 0.381 8 g of ammonium chloride in ammonia-free water and diluting to 1000 ml with ammonia-free water. Standard solutions in the concentration range 0.5-8.0 p.p.m. of NH,+-N are prepared by suita’ble dilutions of the stock solution. A 0.02~o stock solution is prepared by suspending 0.2g of poly(viny1 alcohol) in about 100ml of water with continuous stirring, then adding 800 ml of boiling water. After cooling, the solution is made up to 1000 ml with water. Working solutions are prepared by suitable dilutions of the stock solution. Standard nmmonia solutions.PoZy(vinyl alcohol) solutions. Du Pont Elvanol 71-30 was used as a stabiliser. Sodium hydroxide solution, 2.0 M. Water samples. Soil sample extracts. Samples were collected in the Sdlerprd Lake, north of Copenhagen, where the nitrogen balance is currently being investigated in connection with environmental studies. These were obtained from Aarhus University, Denmark, where they were prepared by two different acid digestion procedures and analysed by titration. Apparatus and Procedures The five manifolds (Fig. 1) were made from polyethylene tubing (internal diameter 0.50, 0.75 and 0.86 mm) and plastic toy components (Lego, Billund, Denmark). The experi- ments were carried out using 30-p1 sample volumes, injected by means of a precisely machined rotary valve described previously.18 In all experiments samples and standards were injected at least in triplicate.The coloured turbidity was measured at 410 rim using a Corning, Model 254, colorimeter, equipped with a Hellma flow-through cuvette, Type 0s 178.12 (volume 18p1, light path 10 mm), connected to a Servograph REC recordeir furnished with an REA 112 high-sensitivity unit (Radiometer A/S, Denmark). The reagent streams were pumped by an Isniatec, Model MP 13 GJ-4, peristaltic pump (Ismatec S.A., Switzerland) operated at speed 8, 9 or 10 with suitable pump tubes to obtain the desired flow-rates. Occasionally a slight precipitate might form, which would be deposited on tube walls and in the flow cell.In such an event, usually after a few hundred analyses, the system must be cleaned by pumping through it distilled water for 3 min, 5 M hydrochloric acid for 30 s and again distilled water for 3 min. Results and Discussion Reagent Concentration in Carrier Stream There are numerous methods for the preparation of Nessler’s reagent.3-8 The final choice was a slightly modified version of Nessler’s reagent as described by Vogel.3 The reason for modification is that in manual methods a large volume of sample (about 50 ml) and only 1 or 2 ml of reagent are ~ s e d . ~ ~ ~ ~ ~ ~ ~ In flow injection analysis the ratio between sample and reagent volumes is not as large, and therefore the reagent was further diluted, viz., with 2 N sodium hydroxide solution, so as to ensure the same alkalinity in all sample solu- tions investigated. Therefore, the concentration of Nessler’s reagent in the carrier streams used was 5-500/, V/V of that of the original stock solution.January, 1979 WITH NESSLER’S REAGENT BY FLOW INJECTION ANALYSIS Nessler (’I or NH4+ 49 50 cm 0.75 mm 1.8 12) Nessler S 50 cm -- 1 @ A 0.50 mm - Nessler (3) PVA 0.86 0.86 r - - - - 1 Nessler 15% Water-bath 7 I j 50cm i 4 50cm I I 1.66 , L-!!!-J (5) 4 50cm N essl er 0.86 mm Waste Fig..1. Flow diagrams for the determination of ammonia as used for investigating the various experimental parameters ( 1-4) and for routine analysis ( 5 ) . 1, Effect of reagent composition; 2, effect of pumping rate and alkalinity; 3, effect of using poly(viny1 alcohol) as protective agent; 4, effect of temperature; and 5, confluence system used for analysis of acid sample solutions.S = point of sample injection (30 PI); = pumping rate (ml min-I); the coil lengths are given in centimetres and the internal coil diameters in millimetres. TO find the optimum concentration of Nessler’s reagent two series of experiments were performed : (a) pumping a known ammonia standard solution and injecting reagent solutions of different concentration; and (b) pumping solutions of Nessler’s reagent of different con- centrations and injecting various ammonia standards. In both series of experiments the same manifold was employed (Fig. 1, manifold 1). Using a fixed flow-rate (1.8 ml min-l) and by adjusting the length of the coil and internal diameter of the tube a reaction time of about 10 s was obtained together with a medium dispersion of the sample (or reagent) zone in the reagent (or sample) stream.The resulting sampling rate of about 120 samples per hour was considered to be satisfactory for all practical purposes. The reproducibility of measurement for each point, showed in the family of curves, obtained in both experiments [Fig. 2(a) and ( b ) ] was &l.5y0. Fig. 2(a) shows, for procedure (a), the influence of the concentration of Nessler’s reagent on the peak height for a series of pumped ammonia standard solutions containing 0.0, 1.0 and 2.0 p.p.m. of NH,+-N. The maximum peak height for the 1.0 and 2.0 p.p.m. N samples was obtained by injecting 15y0 V/Y Nessler’s reagent. The linearity of the conventional calibration graphs as obtained with various concentrations of Nessler’s reagent in the carrier stream (10, 15, 20 and 30% V / V ) is shown in Fig.2(b) [procedure (b)]. The reagents were otherwise the same as used in procedure (a). The final choice of adopting the 15% V/V reagent solution concentration was made using Fig. 2(a) and (b). The linearity and slope50 KRUG at?.: DETERMINATION OF AMMONIA I N LOW CONCENTRATIONS AnaZyst, Vd. 104 1.0 - 0.8 - 0.6 - 0.4 - 0.2 r A * 4 0 10 20 30 40 60 Concentration of Nessler reagent, % V/V r A A A I I I 0 2 4 6 8 N concentration, p.p.rn. Fig. 2. Influence of the composition of Nessler’s reagent on the coloured turbidity produced using configuration 1 (Fig. 1). (a), Pumping sample solutions containing (A) 0, (B) 1 and (C) 2 p.p.m.of NH + N and injecting the reagent; and (b), pumping reagent solutions containing 10, 15, 20 and 30% Vl;, i n relation to the originally prepared Nessler s8tock solution) and injecting the samples. of the calibration graphs show that procedure (a) is suitable for selecting the best concentra- tion of reagent. In general, this approach should always be used for developing other methods in which expensive reagents are used. In addition, this method saves time as it allows a number of parameters to be investigated rapidly. An interesting point to consider is the difference between the peak heights for 1 and 2 p.p.m. of NH,+-N, as found by comparing Fig. 2(a) and (b), which represents 0.68 and 0.09 absorbance unit, respectively, for the 20% V/V Nessler’s reagent.This indicates that only a small volume of reagent is required in order to produce a significant coloured turbidity. During the dispersion of the reagent zone [procedure (a)] there is more ammonia available, resulting in a higher peak compared with the original procedure (b), where the sample zone is dispersed in the reagent stream. There- fore, it is possible to consider yet another flow injection system for a high-sensitivity deter- mination. Thus, by using manifold 1 in Fig. 1 and procedure (a), ammonia can be determined down to 0.02 p.p.m. of NH,+-N (theoretical limit of detection for 1% absorbance), but only 30 samples can be analysed per hour. The reagent consumption would be only 3Opl of 15% V/V Nessler’s reagent per determination for a 2-ml sample volume, which, in comparison with the manual methods,3-5 represents a drastic reduction [about 400 times less potassium mercury(I1) iodide].Yet another, recently suggested, technique, the so-called zone merging26 is being investi- gated with the aim of further reducing the consumption of reagents and increasing the sampling rate. This approach involves the use of a new type of valve, which simultaneously injects sample and reagent into a carrier stream of 2 N sodium hydroxide solution, permitting a higher sampling rate of about 100 determinations per hour. Alkali Concentration in Carrier Stream The influence of the concentration of sodium hydroxide on the effectiveness of the Nessler’s reagent is shown in Fig, 3. The experiment was carried out with manifold 2 (Fig. l), by pumping 1.67 ml min-l of 15% VlV Nessler’s reagent with final sodium hydroxide con- centrations of 0.5, 1.0 and 2.0 N.Increasing the alkali concentration increases the sensi- tivity,’~~ but unfortunately also increases the base-line noise owing to the generation of “schlieren patterns.” Also, on increasing the concentration of alkali an increase was noticedJanuary, 1979 WITH NESSLER’S REAGENT BY FLOW INJECTION ANALYSIS 51 in the accumulation of precipitate on the walls of the tube and on the windows of the flow cell, causing a gradual drift of the base line (cj., Fig. 7). Although the high blank values obtained for the reagent containing 2 N sodium hydroxide solution obviously affect the accuracy of the determination of ammonia at concentrations lower than 0.5 p.p.m.of NH,+-N, this alkali concentration was, however, considered to be the optimum. Flow-rate The influence of the flow-rate of the reagent (15% V/V Nessler reagent in 2 N sodium hydroxide solution) on the absorbance signal is shown in Fig. 4. This experiment was carried out using manifold 2 (Fig. 1). By varying the flow-rate from 0.67 to 1.67 ml min-l, the isoconcentration curves obtained for 2.0, 4.0 and 6.0 p.p.m. of NH,+-N indicate that the reaction is very fast. This is surprising as manual methods have been reported to require a considerably greater reaction time697,9 (10-60 min for completion), while a t a flow- rate of 1.67 ml min-l the reaction time (sample residence time) is approximately 6 s. At lower flow-rates (0.67 ml min-l) a double peak is formed owing to the lower dispersion of the sample zone in the reagent stream, which can be explained from the theory of di~persion.~’ A /f N concentration, p.p.m.1.2 al C 42 2 0.8 Q 0.4 Samples per hour 0 0.67 1.00 1.33 1.67 Flow-rate /mi min- ’ Fig. 3. Influence of the concentration of Fig. 4. Influence of flow-rate on the coloured sodium hydroxide on the ammonia calibration turbidity and the sampling rate for 16% V/V graphs, using manifold 2 (Fig. 1) with a pumping Nessler’s reagent in 2 N sodium hydroxide rate of 1.66 ml min-1, the carrier stream being solution,using manifold 2 (Fig. 1) with different 15% V / V Nessler’s reagent containing: (A), pumping rates (4). Isoconcentration curves 2 N ; (B), 1 N ; and (C), 0.5 N sodium hydroxide for: (A), 0.0; (B), 2.0; (C), 4.0; and (D) 6.0 solution.p.p.m. of NH,+-N. Basically, the lower the flow-rate, the lower is the Reynolds number and the lower is the dispersion of the sample zone, which means that the sample in certain regions is not so well mixed in the reagent stream. The high blank values noticed for the isoconcentration curve for 0 p.p.m. of NH,+-N are due to differences in refractive index between the reagent and reagent plus sample, caused by the high content of alkali. The “schliering” decreases on increasing the flow-rate and again the sample dispersion theory explains the differences obtained. As a compromise for performing the ammonia determinations at a high sampling rate and with good accuracy (a high ratio between sample and blank), a flow-rate of 1.67 ml min-l was chosen.52 KRUG et id.: DETERMINATION OF AMMONIA I N LOW CONCENTRATIONS Analyst, Vd. 104 Influence of Temperature The influence of the temperature during the nesslerisation on the ammonia isoconcentration curves is shown in Fig, 5. This experiment was executed with manifold 4 (Fig. l), pumping 15% V/V Nessler's reagent in 2 N sodium hydroxide solution. From the results it can be seen that in the temperature range 2142 "C the coloured turbidity formed is virtually the same for samples containing less than 4 p.p.m. of NH,+-N. Above this level, the iso- concentration curves for 6 and 8 p.p.m. of NH4+-N indicate an anomaly. This might be explained by the fact that at higher concentrations of ammonia, turbidity predominates over colour and the size of the colloidal particles is affected by temperature variations. Again, this observation is in good agreement with results obtained by manual that is, that an increase in temperature leads to an increase in turbidity. This was further confirmed by the drift of the recorded base line above 31 "C, and visually by inspecting the walls of the reaction coil, which showed the deposition of a slight, orange - brown precipitate.Use of a Protective Agent To prevent precipitation on the walls of the tube we tried the use of poly(viny1 alcohol) (PVA), which acts as a protective agent for colloidal suspension^.^^^^ This experiment was carried out with manifold 3 (Fig. 1). Although its use has been recommended for the determination of ~ulphate,~ it was not successful in the determination of ammonia; thus Fig.6 shows that on increasing the PVA concentration the sensitivity decreases. An explanation for this phenomena is the effect of PVA in altering the colour of the complex formed. There was no influence on the slight precipitation on the walls of the tube, but the colour of the precipitate changed from orange - brown to grey - brown. -D 20 30 40 Temperatu re/"C Fig. 5. Influence of temperature on the coloured turbidity produced as recorded with manifold 4 (Fig. 1). Iso- concentration curves for: (A), 0.0; (B), 1.0; (C), 2.0; (D), 4.0; (E), 6.0; and (F), 8.0 p.p.m. of NH,+-N. 1.2 0 0 5 0.8 d 13 L I 0 2 4 6 N concentration, p.p.m. Fig. 6. Influence of the presence of poly(viny1 alcohol) in the carrier stream on the ammonia calibration graphs (manifold 3, Fig.1). Poly- (vinyl alcohol) concentration: (A), 0; (B), 0.005; ( C ) , 0.010; and (D), 0.020%. Practical Applications The accuracy of the proposed method can be judged from the results in Table I and the precision and sampling rate from the recorder output shown in Fig. 7. On comparing the flow injection system with the manual indophenol blue method, a good correlation between the two methods was obtained for water samples. On the other hand, for the soil extracts the results were less satisfactory (comparing flow injection analysis with titration). ForJanuary, 1979 WITH NESSLER’S REAGENT BY FLOW INJECTION ANALYSIS TABLE I COMPARISON OF PROCEDURES FOR THE DETERMINATION OF AMMONIA IN NATURAL WATERS AND SOIL EXTRACTS Values are given in parts per million of NH,+-N.0.4 0 - Sample No. 3 4 5 6 9 11 12 18 19 24 28 29 - Water samples A -l Manual Flow injection indophenol method blue method 3.90 3.90 3.38 3.38 3.10 3.19 2.72 2.70 2.25 2.62 1.95 2.24 1.40 1.45 1.45 1.20 0.60 0.60 2.75 2.35 5.70 5.65 5.80 5.70 7 Sample No. 1.1 1.2 2.1 2.2 3.1 3.2 4.1 4.2 5.1 5.2 6.1 6.2 Soil extracts Flow injection method 4.32 4.24 3.22 3.08 1 .oo 1 .oo 3.56 3.56 2.68 3.28 3.04 3.32 1 Titration method 4.44 3.52 4.08 2.77 1.03 1.49 5.00 3.89 3.67 3.91 2.92 2.53 53 the analysis of soil digests or acidified water samples the confluence manifold (manifold 5, Fig. 1) was used. The advantage of this system is the possibility of injecting acid samples without affecting the coloured turbidity formed. Thus, the samples are introduced into a carrier stream of 2 N sodium hydroxide solution, and the subsequent addition of the Nessler’s reagent, pre-diluted to 25% V/V, ensures an optimum reagent concentration in the last mixing coil.1.2 6, C m 0.8 4 I) Q 10 min I Scan --+ Fig. 7. Routine analysis of digested soil samples recorded with manifold 5 (Fig. 1). From left to right are shown a series of ammonia standard solutions (0.5, 1.0, 2.0, 4.0 and 6.0 p.p.m. of NH,+-N), followed by eight sample solutions and a second set of standards, and finally a blank, all solutions being injected in triplicate.54 KRUG, R ~ Z I C K A AND HANSEN Finally, it should be emphasised that the optical characteristics of the spectrophotometer used have a marked influence on the sensitivity of the method and the blank value. Thus, generally better results were obtained with an instrument that had a more collimated beam (Corning 254) than with an instrument with a wide beam (Beckman DB GT).Conclusion Nessler’s reaction has been adapted successfully to flow injection analysis, making possible the determination of ammonia in water samples and soil extracts in the range 0.5-6.0 p.p.m. of NH,+-N at a rate of 100 samples per hour. In comparison with manual methods the consumption of reagent is significantly reduced and the time required for analysis is shortened. The method would not be applicable if a significant amount of colloidal material is present in the solutions to be analysed. It would be advantageous if a protective agent could be found that would prevent the gradual deposition of precipitate on the walls of the system.The periodic cleaning of the system with acid .would then not be necessary. Thanks are due to Mr. H. Skotte of Lyngby-Taarbaek Town Council Control Laboratory, Denmark, and Dr. Per Nmnberg of Aarhus University, Denmark, for providing water and soil samples, respectively. The authors also extend their appreciation to DANIDA (Danish International Development Agency) for providing a stipend for one of us (F.J.K.) at the Technical University of Denmark, and for partial material support (DANIDA Project No. 104 Dan. 8/241). 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 26. 26. 27. References Sarkar, P. B., and Ghosh, N. N., Analytica Chzm. Acta, 1955, 13, 195.Sarkar, P. B., and Ghosh, N. N., Analytica Chzm. Acta, 1956, 14, 209. Vogel, A. I., “A Textbook of Quantitative Inorganic Analysis Including Elementary Instrumental Marczenko, K., “Spectrophotometric Determin,ation of Elements,” Ellis Horwood, Chichester, 1976, Snell, F. D., and Snell, C. T., “Colorimetric Methods of Analysis Including some Turbidimetric and Snell, F. D., and Snell, C. T., “Colorimetric Methods of Analysis Including Photometric Methods,” Thompson, J. F., and Morrison, G. H., Analyt. Chem., 1951, 23, 1153. Williams, P. C., Analyst, 1964, 89, 276. Massmann, W., 2. Analyt. Chem., 1963, 193, 332. Krug, F. J., Bergamin Filho, H., Zagatto, E. A. G., and Jmgensen, S. S., Analyst, 1977, 102, 503. RbZiCka, J., and Hansen, E. H., Analytica C h i m Acta, 1975, 78, 145. RbiiEka, J., and Stewart, J. W. B., Analytica Chim. Acta, 1975, 79, 79. Stewart, J. W. B., RbiICka, J., Bergamin Filho, H., and Zagatto, E. A. G., Analytica Chim. Acta, RbiiEka, J., Stewart, J. W. B., and Zagatto, E. A. G., Analytica Chim. Acta, 1976, 81, 371. Stewart, J . W. B., and RbiiEka, J., Analytica Chim. Acta, 1976, 82, 137. RbiiEka, J., and Hansen, E. H., Analytica Chiwz. Acta, 1976, 87, 353. RfiiiEka, J., Hansen, E. H., and Zagatto, E. A. G., Analytica Chim. Acta, 1977, 88, 1. Hansen, E. H., RbiiCka, J., and Rietz, B., Anulytica Chim. Acta, 1977, 89, 241. RbiiEka, J., Hansen, E. H., and Mosbaek, H., Analytica Chim. Acta, 1977, 92, 219. Jorgensen, S. S., Bergamin Filho, H., Zagatto, E. A. G., Krug, F. J., and Bringel, S. R. B., Bolm RbiiEka, J., Hansen, E. H., Mosbaek, H., and Krug, F. J., Analyt. Chem., 1977, 49, 1858. Hansen, E. H., Krug, F. J., Chose, A. K., and RbiiCka, J., Analyst, 1977, 102, 714. Hansen, E. H., Chose, A. K., and RbiiCka, J., Analyst, 1977, 102, 705. Betteridge, D., and RbiiCka, J., Talanta, 1976, 23, 409. Bergamin Filho, H., Reis, B. F., and Zagatto, :E. A. G., Analytica Chim. Acta, 1978, 97, 427. Bergamin Filho, H., Zagatto, E. A. G., Krug, F. J., and Reis, B. F., Analytica Chim. Acta, 1978, RbiiEka, J., and Hansen, E. H., Analytica Chim. Acta, 1978, 99, 37. Analysis,” Longmans, London, 1968, p. 847. pp. 18-19. Nephelometric Methods,” Volume 11, Van Nostrand, New York, 1957, pp. 814-818. Volume IIA, Van Nostrand, New York, 1959, p. 705. 1976, 81, 371. C E N A , Piracicaba, 1977, BC 047. 101, 17. Received July loth, 1978 Accepted August 15th, 1978