Estuarine, Coastal and Shelf Science (1983) 17, 219-224

Optimization of the indophenol for the automated determination in estuarine waters

blue method of ammonia

R. F. C. Mantoura

and E. M. S. Woodward
Research, FVospect Place, The Hoe, Plymouth,

Institutefor Marine Environmental Devon PLI 3DH, U.K.

Received 16 July 1982 and in revised form 9 November 1982

Keywords:

ammonia; analysis; estuarine in natural waters pH, ionic strength indophenol blue 8% over the entire

Existing automated methods for the determination of ammonia suffer from serious salt error in estuaries because of changes in and optical properties with salinity. A modified automated method is described which minimizes the salt error to less than salinity of estuaries.

Introduction Ammonia is an important constituent of the nitrogen cycle in natural waters and its involvement in the biogeochemical processesof estuariesis receiving increased attention (Wollast, 1981; Knox et al., 1981). Most of the methods developed for the analysis of ammoniaare basedon the spectrophotometric determination of the indophenol blue (IPB) complex formed by the reaction of ammoniawith phenol and hypochlorite, in alkaline pH (Berthelot reaction; Solorzano, 1969; Riley, 1975; Krom, 1980). There are several automated IPB procedures for the analysis of ammonia in seawater (Head, 1971; Grasshoff & Johannsen, 1972; Benesch & Mangelsdorf, 1972; Le Corre & Triguer, 1978; Loder & Glibert, 1976; Reusch-Berg & Abdullah, 1977; Folkard, 1978), but these are not entirely suitable for estuarine usebecausethey suffer interferences from changesin salinity (Sasaki & Sawada, 1980), pH (Harwood & Huyser, 1970; Krom, 1980) and alkalinity commonly encountered in estuarine waters. The salinity dependenciesare also inconsistent: Liddicoat et al. (1975) and Loder & Glibert (1976) have reported higher sensitivity in freshwater relative to seawater, whereas the opposite was noted by Head (1971) and Benesch & Mangelsdorf (1972). In other modifications, nonlinear salinity dependencewas observed, with maximum sensitivity varying between 8%0 (Le Corre & Treguer, 1976), and 15%0
(Grasshoff & Johannsen, 1972). In this paper we describe an automated method for ammonia which overcomes most of the salt errors by use of a highly pH-buffered formulation.

Reagents
Reagent No. 1: dissolve separately 10 g of phenol (AnalaR, BDH) in 40 ml ethanol (95%)

and 0.16 g sodium nitroprusside catalyst (AnalaR, BDH) in 100ml deionized water and combine. Store in amber bottle and prepare fresh daily.
219
0272-7714/83/080219+06$03.00/0 0 1983 Academic Press Inc. (London) Limited

220

R F. C. Mantouta 6 E. M. S. Woodward

Reagent No. 2: dissolve 30.0 g tri-sodium citrate dihydrate (Na,C,H,0,.2H,O; AnalaR, BDH), 0.20 g DTT (dichloro-s-triazine-2,4,6-( IH, 3H, 5H)-trione sodium salt dihydrate; Koch-Light) and 20 *O ml of 4.3 M NaOH and make up to 100 ml with ammonia-free water. Prepare fresh daily. Ammonia standards: dissolve 0.099 g of ammonium sulphate (AnalaR, BDH) in 1.0 1 deionized water; add 5 ml chloroform preservative. Store up to a month in refrigerator. This ammonia stock standard is 1500 pg-at. NH,-N 1-l. Working standards may be prepared by volumetric dilution into ammonia-free water or by standard addition into GFX-filtered estuary water. Ammonia-free water is prepared by passing deionized water through a column of Amberlite IR-120 (hydrogen form) and used immediately as the blank, reagent diluent or wash in the automatic analyser.
Manifold

The reaction manifold describing the automated determination of ammonia is shown in Figure 1. Two alternative modes of sampling are shown, discrete and continuous. Discrete 5 ml samples contained in ashed (450 C) gl ass vials are sampled from an autosampler (Hook & Tucker model A40-II; 1.5 min sample/wash). For high resolution work in the estuary, the continuous sampling mode is preferred. We use a custom-built filtration block (Morris et al., 1978) fabricated from stainless steel and supporting a 47 mm Whatman GF/C filter. Sample and reagent streams are pumped through Technicon Tygon flow-rated tubes fitted onto an Ismatek pump (model MP-13, Switzerland) with the exception of the reagent 1 which requires solvent-resistant Solvaflex tubing. Glass transmission tubes are used throughout. The sample stream is segmented with acid-scrubbed air. After mixing of sample and reagents, the IPB complex is developed in a delay coil (4 *6 min; coil diam. 35 mm, length 4.4 m, 40 turns) immersed in an oil bath at 50 C. Following cooling to room

630

Colorimeter nm 50 mm f/c

Icy-

I o-10 I

I
1

R2

ML-

,I..
Waste 4 I I

1 042 1

./

340 ----

Figure 1. Manifold for the automatic cates Solvaflex tubing).

determination

of ammonia

(#

on pump tube Rl indi-

Automated determination of ammonia

221

TABLE 1. Analytical

performance

of the automated 0.2-18

NH, B-at

analyser N 1-1

Linear detection range Reproducibility (% SD of 10 replicates at 3 pg-at N L-l) Detection limit (S/N = 2) Delay time Response time (95%) Sample/wash times Sample throughput

kl,O% 0.02 B-at N L-1 11.7 minutes 2.5 minutes 1 5 minutes 20 h-

temperature, the IPB complex is measuredat 630 nm with a Chemlab Colorimeter (model Mk III, Hornchurch, U.K.) and the absorbancerecorded on a chart recorder. On occasion, a Technicon II Auto Analyser@ calorimeter (model SCIC-AA II) was also deployed. The entire system is palletized for easeof transport and operation in the field. The analytical performance figures basedon the Chemlab Colorimeter are summarizedin Table 1. Results and discussion Magnesiumprecipitation Since estuarine waters bear a greater chemical resemblanceto seawaterthan to river water, we evolved our method by careful consideration of the automated IPB methods developed for the analysis of ammonia in seawater (Head, 1971; Grasshoff & Johannsen, 1972; Benesch & Mangelsdorf, 1972; Grasshoff, 1976; Le Corre & Treguer, 1978; Loder & Glibert, 1976; Reusch-Berg & Abdullah, 1977; Folkard, 1978). These show an inconsistency in the formulation of reagents which may in part explain the variations in salt error. Although all the methods employ tri-sodium citrate chelator to avoid precipitation of magnesiumhfdroxide at alkaline pH, there is a stoichiometric deficiency of citrate with respect to magnesiumin two of these reports (20% in Grasshoff & Johannsen, 1972; and 46% in Le Corre & Treguer, 1978). We overcame the precipitation problem commonly encountered in the automated methods (Head, 1971) by ensuring a stoichiometric excess of citrate (~120%). Since surfactants are not used, the entire system requires occasional wash (every N 10 h) with 1 M HCl followed by 1 M NaOH. Salt error Experiments were conducted with waters obtained from the Tamar Estuary (U.K.) and in mixtures of River Tamar water with seawater. Most of the salt error in estuarine samplesoriginates from poor pH buffering (Sasaki & Sawada, 1980) rather than ionic strength. Despite the river water being more acidic (pH = 7.5) than seawater (pH = 8. l), the final pH of the reaction mixture was higher in the freshwater samples.This arisesfrom the lower alkalinity (- O-8 meq 1-i) and hence buffering capacity of river water relative to sea water (alkalinity = 2.3 meq 1-i). Since the IPB reaction is sensitive to the pH of the medium (Harwood & Huyser, 1970; Riley, 1975; Krom, 1980), pH variations in estuarine waters must be minimized by the use of buffered reagents operating about the optimum pH of cv 10.6. At this pH, citrate (pK, = 3.13, 4.76, 6.40) is ineffective as buffer. Grasshoff & Johannsen (1972) utilized unspecified amounts of borate (pK, = 9.23), but we experienced solubility problems which may explain why these authors omitted it in a subsequent report (Grasshoff, 1976). Amino sulphonate buffers such as CAPS (cyclohexl-amino propane sulphonic acid; pK, = 10.4) were also unsuitable becauseof limited solubility and contamination with ammonia.

222

R. F. C. Mantoura

& E. M. S. Woodward

140 (a) 120 -

c-------__ c------___ c,... --** . . . . .,*/ +..* 4cc -* . ...* *... .+ ** 4cc . ;.....*

-I.*.-. .a.-.

--

-----__
l

ioo(~++*

-*-,-*L1*-T. -*-,-*L1*-T -.. -... -.. -... a... A a... a... A a... *... *... -.. -..

* * *... *... a..... a..... *--* . . . ...** *--* . . . ...**

80 80

-...., -...., -I

60 601 0.5 0.5

.(b)

04

0.3

Refractwe

index

blank

IO

15 Sallnlty

20 %I

25

30

35

Figure 2. (a) The effect of salinity on the sensitivity of standard additions of ammonia in laboratory mixed waters (0) and in waters from the Tamar estuary (A) expressed as 96 of response in river water. For comparison, the salt error curves reported by Grasshoff & Johannsen (1972) and Loder & Glibert (1976) are also shown (. . . and ---, respectively). (b) Contribution of refractive index and organic absorbance to the optical blanks in the Chemlab Colorimeter. River water-seawater mixture (0) de-ionized water-seawater mixture (0).

Since phenol has a pK, = 100, then the IPB reaction could be made self-buffering provided the unreacted phenol is in excessof the alkalinity. The concentration of phenol in the final mix with seawaterreported in the literature varies between 0.0014 M (Grasshoff & Johannsen, 1972) and 0.050 M (Reusch-Berg & Abdullah, 1977). A fIna phenol concentration of 0.06 M was sufficient, since even in the presence of 1008 ug- at NH,-N l-1, the IPB reaction will consume only 3% of the phenol leaving most of the phenol to act as a pH buffer. Ethanol was used to solubilixe the high concentration of phenol used in our system. The salt error of our method, asdetermined by standard addition of ammoniainto waters of different salinities, is shown in Figure 2(a). When compared with other methods, our method displays minimal salt error (-8%) even though the final pH of the river water mixture (pH 10.9) was greater than seawater (pH 9.9). Although Liddicoat et al. (1975) reported that the IPB reaction is light sensitive, we found that varying the ambient light levels had no effect. The optimum (highest sensitivity) temperature was 50 C, which according to Benesch & Mangelsdorf (1972), should not cause interference from amino acids. DTT was used in preference to commercial hypochlorite as the chlorinating agent becauseof its greater stability in solution (Grasshoff & Johannsen,1972; Krom, 1980).

Automated determination

of ammonia

223

1-

3-

2-

I-

I. 3
Distance (km )

Figure 3. Axial distribution of ammonia and salinity in the Tamar Estuary, 25 August 1981. Corrections due to salt error are apparent in the ammonia peaks (- - - ) in the more saline waters, whereas the optical blank corrections (. .) are linearly related to salinity.

Optical

blanks

In addition to the chemical effects of varying salinity, there are optical interferences in calorimetric analysiswhich are peculiar to estuarine samples.Saline waters and river waters have, in the absenceof calorimetric reagents, an apparent absorbancearising from: (1) refractive bending of light beamsby sea salts- refractive index blank (Atlas et al., 1971; Loder & Glibert, 1976); (2) Background absorbanceby dissolved organics of riverine origin; the former is a function of the optical geometry of the light beam and the flow cell, and the latter is related to the organic loading of river water. As shown by Figure 2(b), both are linearly related to salinity, which makesoptical blank corrections easy to apply to estuarine samples. Although the Chemlab Colorimeter performed satisfactorily during continuous analysis of estuarine waters, it suffered from serious optical interferences during discrete analysis. The problem lies with the flow cell geometry which gives poor flushing between dense saline samples,and deionized water wash. This gives rise to Schlieren effects and a noisy absorbancetrace. For discrete analysis, the more costly Technicon Colorimeter with its superior flow cell geometry (dead volume w 150 l.rl) is recommended, since it does not suffer from Schlieren effects, and has a lower refractive index blank. The axial concentration of ammonia in the Tamar Estuary, shown in Figure 3, varies markedly and this emphasizesthe importance of continuous analysis in chemical studies of estuaries. The contribution of the optical blank and effects of salt error are also shown. Acknowledgements We thank Mrs C. M. Goodchild for assistance the early phaseof this work. This work in

224

R F. C. Mantoura

& E. M.

S. Woodward

forms part of the Estuarine Ecology Programme of the Institute for Marine Environmental Research, a component of the Natural Environment Research Council, and was partly supported by the Department of the Environment under Contract DGR 4801684.
References
Atlas, E. L., Hager, S. W., Gordon, L. I. & Park, P. K. 1971 A practical manual for use of the Technicon@ AutoAnalyser@ in sea water nutrient analyses: revised. Technical Report 215, Ref. 71-22, Oregon State University, 47 pp. Benesch, R. 81 Mangelsdorf, P. 1972 Eine Methode zur colorimetrischen Bestimmung von Ammoniak in Meerwasser Helgol&der w&enschaftliche Meeresuntersuchungen 23, 365-375. Folkard, A. R 1978 Automatic analysis of sea water nutrients. Fisheries Technical Report No. 46. Ministry of Agriculture, Fisheries and Food, Lowestoft, 23 pp. Grasshoff, K. 1976 Methods of Seawater Analysis. Verlag Chemie, Weiheim. 317 pp. Grasshoff, K. & Johannsen, H. 1972 A new sensitive and direct method for the automatic determination of ammonia in sea water. yournal du Gntseil Internatimal pour IExploratbn de la Mer 34, 516-521. Harwood, J. E. & Huyser, D. J. 1970 Automated analysis of ammonia in water. Water Research 4, 695-704. Head, P. C 1971 An automated phenolhypochlorite method for the determination of ammonia in sea water. Deep Sea Research 18, 531-532. Knox, S., Turner, D. R., Dickson, A. G., Liddicoat, M. I., Whitfield, M. % Butler, E. I. 1981 Statistical analysis of estuarine profiles: Application to manganese and ammonium in the Tamar Estuary. Estuarine, Coastal and shelf Science 13,357-371. Krom, M. D. 1980 Spectrophotometric determination of ammonia: a study of a modified Berthelot reaction using salicylate and dichloroisocyanurate. Analyst 105,305-316. Le Corre, P. & Treguer, P. 1978 Dosage de lammonium darts Ieau de mer: comparaison entre deux methodes danalyse automatique. journal du Conseil Internatioml pour IExploration de la Mer 38, 147-153. Liddicoat, M. I., Tibbitts, S. & Butler, E. I. 1975 The determination of ammonia in sea water. Limndogy and aeanography 20, 131-132. Loder, T. C. & Glibert, P. M. 1976 Blank and salinity corrections for automated nutrient analysis of estuarine and sea waters. University of New Hampshire contribution UNH-SG-JR-101 to Technicon International Congress, December 13-15 1976. Morris, A. W., Howland, R. J. M. & Bale, A. J. 1978 A filtration unit for use with continuous autoanalytical systems applied to highly turbid waters. Estuarine and Coastal Marine Science 6, 105-109. Reusch-Berg, B. & Abdullah, M. I. 1977 An automatic method for the determination of ammonia in sea water. Watm Research 11,637-638. Riley, J. P. 1975 The analytical chemistry of sea water. In Chemical Oceanography (Riley, J. P. & Skirrow, G., eds) Vol. 3, 2nd edition, Academic Press, London, pp. 193-477. Sasaki, K. & Sawada, Y. 1980 Determination of ammonia in an estuary. Bulletin ofthe Japanese Society of Scimrific fiheriks 46,319-321. Soi&zano, L. 1969 Determination of ammonia in natural waters by the phenol hypochlorite method. Limnologv and Oceanography 14, 799-801. Wollast, R. 1981 Redox processes in estuaries. In Riwer inputs to Ocean S~tenas (Martin, J.-M., Burton, J. D. & Eisma, P., eds) SCOR/UNESCO/IOC Workshop, 26-U) March, 1979, Rome. 384 pp.