Determination of nutrients in seawater by segmented-flow analysis with higher analysis rate and reduced interference on ammonia.

A microbore continuous-flow analyser was used to determine ammonia, nitrite, nitrate and phosphate in seawater. Ninety samples per hour were measured by optimizing the hydraulic conditions and using a cadmium coil for nitrate reduction. The analysis range was extended and the detection limit was reduced by using two analytical ranges with automatic range-changing. Interference from magnesium on ammonia was investigated and reduced by using a different complexing agent. The results correlated well with a Japanese reference method.


Introduction
Since the introduction of air-segmented continuous-flow analysis (CFA) in 1958 there have been continual improvements in system performance, mainly in analysis rate and the automation of activities subordinate to the analysis, such as automatic data handling, sample dilution and reagent change-over.
The analysis rate in segmented-flow systems is inversely dependent on dispersion. Reductions in dispersion have centred around reductions in the internal diameter of the reaction tubing, from 2"4 mm in the original Technicon AutoAnalyzer I to 2 rnrn in the AutoAnalyzer II and mm in modern instruments. The latest systems have been designed with a proper understanding of the way in which dispersion, tubing diameter and flow rate are interrelated, following Snyder's definitive description ]. As Snyder has also explained, a further reduction in dispersion in commercial systems is unlikely for practical reasons [2]. Segmented-flow systems run faster than flowinjection (FIA) systems operating at the highest possible sensitivity with slow reactions such as colorimetric ammonia and phosphate, due to the time needed to introduce a large injection volume into the manifold and the time required for the signal to return to near zero in a medium-dispersion FIA system. A major source of dispersion in the determination of nitrate in seawater is the cadmium reduction stage, which Author to whom correspondence should be addressed. commonly uses a column packed with cadmium granules. Patton [3] first described the use of a cadmium coil for analysing nitrate in fresh water. The coil allows the segmented stream to flow uninterrupted through the reduction stage. This more than doubled the analysis rate for seawater samples compared to methods using a packed column.
Previous experience using the AutoAnalyzer ammonia method [4] showed negative values in surface samples far from the coast. The method uses citrate and tartrate buffers to reduce magnesium interference similar to the method described by Hansen and Grasshof [5]. Experiments showed that this was not due to impurities in the standards and base-line reference solution, but, rather, to a suppression of the ammonia colour reaction due to magnesium. The complexing agents tartarte, citrate and EDTA were found to be effective against calcium but did not completely eliminate interference from magnesium. As described below, 1,2 cyclohexane diamine tetraacetic acid (CyDTA) proved to be a more effective complexing agent for magnesium.
To attain low detection limits in automated methods, it is necessary to use a high sample-to-reagent ratio, a flowcell with a long path-length and high signal amplification. This restricts the analytical range so that high samples have to be diluted. The range-expanding facility of the analyser is used to maintain a wide analytical range and achieve lower detection limits.
After optimizing the analysis conditions for nitrite, nitrate, ammonia and phosphate, the results from the automatic analyser were compared to those from a standard manual method [6]. The correlation (R) was greater than 0"99 for each parameter.

Materials and methods
Analyser A four-channel TRAACS 800 with random-access sampler and automatic dilution facility was used (Bran + Luebbe, Norderstedt, Germany).

Reagents
All chemicals were of analytical grade.

Analytical conditions
The system was run at a sampling rate of 90/h and a sample wash ratio of 4 1. Oxygen was removed from the    Table 3. Between-run reproducibility (N 30). air used for bubble segmentation on the nitrate channel by passing it through an oxygen absorber (activated iron, commercial hand-warmer available in Japan). An alternative would be to use nitrogen. This is necessary to avoid premature degradation of the reduction performance of the cadmium coil [3].

Manifolds
The flow diagram for all four channels is shown in figure   1. The cadmium reduction coil (Bran + Luebbe part No.  Each manifold has two sample pump tubes of different flow rates. In routine operation, samples were first analysed using the large pump tubes. If any samples fell off scale, the following sample was automatically resampled at the end of the analysis, to eliminate any effect of carry-over from the off-scale sample. At the end of the run, the off-scale samples, and high standards, were reanalysed with the small sample pump tubes, giving a higher dilution. This allowed samples falling into both ranges to be placed randomly into the sample tray without risk of inaccuracy. By splitting the analytical range like this a lower detection limit on the low range was attained. Results and discussion The effects of reagent composition on magnesium interference in the ammonia determination were investigated.  The sensitivity of each method in the low-level analysis range, measured by the absorbance produced by the top standard, was as follows: ammonia 0"1 at 10 Bmol/1, nitrate 0"23 at 10Bmol/1, nitrite 0"23 at 2 tmol/1, phosphate 0" 19 at 4 tmol/1. Reproducibility was measured by analysing five different concentrations of each analyte 10 times, in random order, on three separate days. There were thus 30 results at each concentration. The day-to-day relative standard deviation was less than 1% for all parameters at levels above 5 Bmol/1. These results are summarized in table 3.
The detection limit, defined as three times the standard deviation of the zero samples in the above test, was 0"12 tmol/1 for ammonia, 0"024 tmol/1 for phosphate, 0-1 tmol/1 for nitrate and 0.052 Bmol/1 for nitrite. Based on past experience, determining the detection limit by the EPA procedure for Method Detection Limit, which measures 10 or more consecutive low-level samples, would produce figures about half those above.
To validate the results from the automated methods, about 100 samples were collected from coastal areas around Japan and analysed according to the manual methods in the Manual for Oceanographic Observation [6], and on the automated system. The results showed good agreement, as shown in figures 2 to 5.

Conclusions
By applying a flow-through cadmium reductor to a micro-bore segmented-flow analyser, reliable and reproducible results were achieved for nutrient analysis at a rate of 90 samples/h. CyDTA proved to be an effective complexing agent for magnesium, and reduced the negative effect on the ammonia determination. It also improved the robustness of the method to varying NaOH concentration..A disadvantage of CyDTA is its high cost. By operating the analyser in the dual-range mode it was possible to reduce the detection limit and increase the analytical range by reducing the effect of carry-over on very low samples. This also improved the convenience of daily operation.