Temperature is one of the most important factors that can affect the inorganic phosphorus analytical method because it affects the reaction sensitivity and hence the calibration gradient. Several studies have reported a rate increase of the formation of the molybdenum blue complex with increasing temperature [8] [9] [10] [11]. Two reagents are needed in the phosphomolybdenum blue method to measure DIP: the ammonium molybdate solution and the reducing agent

(ascorbic acid or tin (II) chloride). Both reagents are very sensitive to light and temperature. Therefore, the reagents should be stored at 4˚C when not in use. It is also important that the reagent solutions are equilibrated to room temperature prior to use.

Figure 3 shows instrument baseline when using cold reagents stored at 4˚C and equilibrated at 25˚C, respectively. The baseline increased from 2434 at 4˚C to 2572 as the reagents being warmed up to 25˚C. The baseline difference between 4˚C and 25˚C was 138, which is roughly equivalent to a 15 µg∙L⁻¹ P concentration difference. Clearly, the regent temperature is one of many errors that could occur in the analytical method, which can be avoided by equilibrating the reagents at room temperature for approximately 2 h prior to use.

The phosphomolybdenum blue method is subject to potential interference from certain matrix ions, e.g., fluoride, nitrate, silicate, arsenate and trace metals (Fe (III), Al (III), Mg (II)). These interferences are more severe in polluted waters and wastewater samples. For example, fluoride slows down the development of the phosphomolybdenum blue complex, but does not affect the maximum absorbance reached at steady state [11]. Nitrate only interferes at very high concentrations (>10 mg∙L⁻¹) [13]. Chromium (III) only interferes with this method when the Chromium (III) concentration above 1 mg∙L⁻¹ [14], which is only found in polluted waters or waste water samples. Silicate can have a major interference effect by reacting with ammonium molybdate to form molybdosilicic acid.

Arsenic and phosphorus are both group V elements, therefore arsenate ( AsO 4 3 − ) and phosphate ( PO 4 3 − ) are chemically similar. It has been shown that arsenate can positively interfere in the determination of inorganic phosphate

[15] [16] [17] [18]. Arsenic exists in different forms in the environment depending on the pH values. Under strong acidic conditions, it exists in the As(III) oxidation state as arsenate (H₃AsO₄). Under weak acidic conditions, arsenic ion exists as the dihydrogen arsenate ion H 2 AsO 4 − . However, under weak basic conditions, the hydrogen arsenate ion H 2 AsO 4 − is formed, while under strong basic conditions the arsenate ion AsO 4 3 − is predominant [19]. Previous study showed that under typical pH and Eh conditions for the River Tamar (pH = 6.1 - 8.2; Eh > 600 mV) [20] [21], arsenate ( H 2 AsO 4 − / H 2 AsO 4 2 − ) was a major component of the total arsenic concentration, which is typically ranged from 2 to 16 µg∙As∙L⁻¹, but can be as high as 40 As∙µg∙L⁻¹ due to the impact of historical arsenic mining [16].

In this study, the impact of arsenate on P measurement under different temperature conditions (i.e. 30˚C, 40˚C, and 50˚C) was investigated. An ANOVA test (p = 0.876) showed that there was no significant difference in P concentration determined between the P standards and the standards spiked with As up to 0.10 mg∙As∙L⁻¹ at 30˚C, 40˚C, and 50˚C. This result demonstrated that P determinations using the segmented flow analyser are insensitive to arsenate concentrations up to 0.10 mg As∙L⁻¹ at temperatures up to 50˚C.

At higher arsenate concentrations, the significant impact on P measurement was observed. Figure 4 shows the experimental results with the addition of 1 mg∙As∙L⁻¹ to P standards (as phosphate) suggesting a significant interference of As and this interference effect was increased with higher temperatures. An increase of temperature from 30˚C to 50˚C had no effect on the non-spiked P standards as the slopes remained same. However, the slopes for the arsenate spiked P standards were significantly different under different temperatures. Figure 5 also shows that the instrument is more sensitive to arsenate at higher temperatures (50˚C). At a lower temperature (30˚C) arsenate concentrations ranged from 0 to 1 mg∙As∙L⁻¹ demonstrated a much less interference, equivalent to 0.002 - 0.004 mg∙P∙L⁻¹. Based on this study, 30˚C is the optimum temperature for minimising the arsenate interference. Consequently, it is important to understand the arsenate concentration in samples before the P is determined, particularly for samples from impacted catchments. If the arsenate concentration is larger than 0.10 mg∙As∙L⁻¹, sample pre-treatment is required [22].

The molybdenum blue method only measures the concentration of molybdate-reactive phosphorus species in water samples, which are predominantly orthophosphate species but also include some condensed IP species and labile organic phosphorus species. There is no established method that can directly measure DOP. Determination of DOP and TDP are generally based on the conversion of all P species to orthophosphate. This is carried out by the hydrolysis and oxidation of molecules containing essentially P-O-P, C-O-P and C-P (see Figure 1 for examples) bonds [23]. In natural waters, DOP includes nucleic

acids, phospholipids, inositol phosphates, phosphoamides, phosphonic acids, organophosphate pesticides, humic associated organic phosphorus compounds [24] and organic condensed phosphates in dissolved, colloidal and particle-associated forms [25].

An acid peroxydisulphate method was developed by Espinosa et al. [33] and an autoclave digestion protocol has been reported by Haygarth et al. [26]. Ridal and Moore [27] suggested that adjustment of the peroxydisulphate concentration is the most important parameter for improving recoveries. The USEPA Persulfate Digestion Method 365.1 uses a lower mass of peroxydisulphate (0.2 g per 20 mL sample) [28]. In this study, several parameters in the autoclave digestion were investigated including pH values of test samples and oxidizing reagent concentration.

1) Peroxydisulphate concentration

Figure 6 shows the measured P concentration in the blank and different oxidizing reagents with and without autoclave process. DOP signal can be observed even without autoclave procedure. For example, a signal equivalent to 3.7 µg∙P∙L⁻¹ was observed with 0.8 g K₂S₂O₈. Approximately 4.7 µg∙P∙L⁻¹ was observed with the addition of both K₂S₂O₈ and sulphuric acid. This result suggested that approximately 80% of the signal came from the K₂S₂O₈ and 20% from the sulphuric acid. Clearly, the high blank signal could largely come from the oxidizing reagent. According to the USEPA Method 365.1 and the results observed in current study, the amount of K₂S₂O₈ was reduced from 0.8 to 0.2 g.

The recoveries of all 8 organic P compounds were > 95% (n = 9) using this method (0.2 g peroxydisulphate), and the limit of detection (LOD) was 5 µg∙P∙L⁻¹. The LOD was calculated from the blank signal plus three times the standard deviation of triplicate blank sample measurement.

2) Stability of autoclaved samples

Generally, “sample stability” means the DOP concentration of the sample remains constant or less than 5% variation through times. Autoclaved samples normally have a pH in the range 0.9 - 1.2. Theoretically, at this pH, the nutrient concentrations in the samples should be stable for at least 5 days for storage at room temperature [29]. Figure 7 shows the results of the stability of autoclaved samples over 3 weeks. A batch of 6 orthophosphate standards (0, 10, 20, 30, 40 and 60 μg∙P∙L⁻¹, n = 3 for each concentration) were autoclaved using the optimized autoclave procedure. All standards were stored at room temperature and analysed on day 1, 3, 7, 14, and 21. Day 1 sample means samples collected from autoclaved samples stored for 24 hr. There was no significant concentration change after 7 days (p = 0.004). However, from the day 14, 6% of loss at 30 and 60 μg∙P∙L⁻¹ standards were observed. This suggests that autoclaved samples can be stable for up to 7 days at room temperature prior to analysis.

The rate of molybdenum blue formation is very pH dependent and therefore the effect of pH on the detection response was investigated. The results showed that

increasing of the pH of the autoclaved samples did not affect the calculated IP concentrations. However, in the USEPA Persulfate Digestion Method 365.1, an acidified wash, (pH 1.0) is used in order to match the carrier stream with the pH of the samples after autoclaving. This was used to reduce the baseline noise and the blank response.

Figure 8 shows the comparison of calibration graphs for 6 autoclaved standards analysed using the normal wash stream (water) and an acidified wash stream. Using the non-acid wash, the blank signal was significantly higher than those with the acidified carrier, resulting in a lower LOD. However, the acid carrier only resulted in a slightly better sensitively compared with the non-acid carrier, as shown by the gradient of the calibration graph.

The sensitivity and reproducibility over time of the acidified carrier stream method for determining IP in autoclaved standards were conducted. Table 1

shows the results from 9 calibrations over a 10 week period (Jun.-Aug., 2015). The analytical variability for the 40 µg∙P∙L⁻¹ standards was < 3.5%, the gradients (sensitivities) in all the equations were within 5% of each other and R² values were all > 0.98. The LOD was 3 µg∙P∙L⁻¹, which is comparable with the LOD of the USEPA Persulfate Digestion Method 365.1 of 4 µg∙P∙L⁻¹. Table 2 shows typical calibration data for autoclaved inorganic P standards

The model compounds selected to examine the efficiency of the autoclaved method include a variety of naturally occurring molecules, most of which are widely distributed in natural waters [30]. Phytic acid is one of the P compounds most resistant to hydrolysis and is also one of the most refractory organic phosphorus compounds found in soils [31] [32]. β-D-glucose-1-phosphate contains a labile C-O-P bond and is a sugar arising from bacterial decomposition that is

widely distributed in soil leachates and marine waters [30]. Adenosine-5’-triphosphate (ATP) and adenosine-5’-diphosphate (ADP) are ubiquitous in nature [33] and contain C-O-P and P-O-P bonds. Methyltriphenylphosphonium bromide is a refractory compound containing a C-P bond. Sodium tripolyphosphate is a condensed inorganic compound containing P-O-P bonds. Cocarboxylase is also called thiamine pyrophosphate and is a coenzyme important in respiration in the Krebs cycle.

Figure 9 shows the results for 8 model compounds and orthophosphate (for comparison) at 50 µg∙P∙L⁻¹ using the optimized autoclave method. Quantitative recovery was achieved for all model organic P compounds, with the only exception of phytic acid (C-P bond), which has the lowest recovery at 90% ± 5%. This is not surprising as phytic acid is one of the most resistant compounds to hydrolysis. Sodium tripolyphosphate (P-O-P bonds) had 95% ± 3% recoveries. Adenosine-5'-triphosphate (C-O-P and P-O-P bonds) recovery was 100% ± 10%. D-glucose-1-phosphate (labile C-O-P bond) gave 100% ± 1% recoveries. P-nitrophenyl phosphate and cocarboxylase had larger standard errors compared to the other compounds, presumably due to contamination of one test sample. Orthophosphate gave 100% ± 3.1% recovery. These results clearly show that organic compounds containing C-P bonds are the most resistant to hydrolysis, whereas C-O-P bonds are relatively labile, and that the autoclave method is suitable for most natural organic phosphorus compounds.

Although the acid peroxydisulphate method had a very good recovery for most of the organic P compounds in fresh water, it is not the ideal method for sea water samples. The matrix absorbs much of the energy and there is a potential risk that salts will precipitate out during digestion [34], making the method much more difficult to perform [35] [36] [37]. It can be seen from Figure 10 that the recoveries of DOP model compounds in low nutrient sea water were low. The recovery of phytic acid was < 20% and G-6-P and ATP were < 80%, possibly because the peroxydisulphate reagent was consumed by the oxidation of chloride ions to free chlorine [27]. One way to avoid this effect is to dilute sea water samples [36] by a factor of five to give a final salinity of 7. As shown in

Figure 10, phytic acid recovery in salinity 7 water was 70%, and A-T-P and G-6-P were >90%.