Arsenic Monitoring in Water by Colorimetry Using an Optimized Leucomalachite Green Method

Arsenic contamination of drinking water is a global concern. Standard laboratory methods that are commonly used for arsenic detection in water, such as atomic absorption spectroscopy and mass spectroscopy, are not suitable for mass monitoring purposes. Autonomous microfluidic detection systems combined with a suitable colorimetric reagent could provide an alternative to standard methods. Moreover, microfluidic detection systems would enable rapid and cost efficient in situ monitoring of water sources without the requirement of laborious sampling. The aim of this study is to optimize a colorimetric method based on leucomalachite green dye for integration into a microfluidic detection system. The colorimetric method is based on the reaction of arsenic (III) with potassium iodate in acid medium to liberate iodine, which oxidizes leucomalachite green to malachite green. A rapid colour development was observed after the addition of the dye. Beer’s law was obeyed in the range between 0.07–3 µg mL−1. The detection limit and quantitation limit were found to be 0.19 and 0.64 µg mL−1, respectively.


Introduction
Arsenic contamination of groundwater and surface water is a major issue in certain regions of the world [1]. Arsenic and its compounds are toxic and can cause serious health effects [2]. Human exposure to arsenic arises through consumption of arsenic contaminated food and water. Chronic exposure to high concentrations of arsenic can cause severe health implications, collectively known as arsenicosis [3]. Some of the symptoms of arsenicosis include skin lesions, nervous system disorders, gastrointestinal problems and various types of cancers [4,5].
Inorganic arsenic is naturally present at high levels in the groundwater of several countries, including India [6,7] Pakistan [8], China [9], Vietnam [10] and several parts of the United States [11][12][13]. Arsenic contamination of ground water in Bangladesh is one of the most serious examples of chronic arsenic exposure [14]. Arsenic concentrations in drinking water in Bangladesh far exceed the World Health Organization's (WHO) maximum permissible limit of 10 µg L −1 [15]. In some tube wells arsenic concentrations as high as 2500 µg L −1 have been detected [16].
In order to improve and monitor the environmental quality of water, reliable and good quality information is needed. Analytical methods capable of detecting arsenic at low concentrations are usually based on sophisticated laboratory instrumentation [17]. Atomic absorption spectroscopy (AAS), induced coupled plasma atomic emission spectroscopy (ICP-AES), X-ray fluorescence, and atomic fluorescence spectroscopy are examples of sensitive and selective methods used for arsenic detection and analysis [18]. While high quality data is obtained from these methods, the cost of with broader linear ranges. For example, Lin et al. developed an arsenic detection assay based on G-quadropole complex with LOD of 4.5 ng mL −1 and linear range of 0.74-14.98 µg mL −1 [43]. Dai and Compton used electrodes coated with gold nanoparticles for determination of arsenic and achieved LOD of 5 ng mL −1 and a linear range of 1-180 ng mL −1 [44].
quadropole complex with LOD of 4.5 ng mL −1 and linear range of 0.74-14.98 μg mL −1 [43]. Dai and Compton used electrodes coated with gold nanoparticles for determination of arsenic and achieved LOD of 5 ng mL −1 and a linear range of 1-180 ng mL −1 [44]. However, colorimetric methods should be considered for arsenic monitoring purposes as they have several advantages over electrochemical methods in terms of cost effectiveness, portability, and lower susceptibility to fouling.

Path Length
The absorbance values for 1 mm quartz cuvette measurements was 10 times lower than for standard cuvette measurements, as expected (Table 1 and Figure 2). As it can be seen from the calibration graphs ( Figure 3) the analytical response was strong for samples measured in microcuvettes. The linearity compared to standard quartz cuvette measurements was good. The response signal and linearity obtained from the microcuvettes was strong which indicates that the leucomalachite green method is suitable for use in microfluidic detection systems.  Table 1. Average absorbance values for arsenic samples (0.2-1 μg mL −1 ) analyzed in two different types of cuvettes. Abs 1, SD 1 and % RSD 1 show the data for measurements carried out with 10 mm quartz cuvettes. Abs 2, SD 2 and % RSD 2 show the data for measurements in 1 mm quartz cuvettes. All measurements were carried out in triplicate (n = 30). However, colorimetric methods should be considered for arsenic monitoring purposes as they have several advantages over electrochemical methods in terms of cost effectiveness, portability, and lower susceptibility to fouling.

Path Length
The absorbance values for 1 mm quartz cuvette measurements was 10 times lower than for standard cuvette measurements, as expected (Table 1 and Figure 2). As it can be seen from the calibration graphs ( Figure 3) the analytical response was strong for samples measured in microcuvettes. The linearity compared to standard quartz cuvette measurements was good. The response signal and linearity obtained from the microcuvettes was strong which indicates that the leucomalachite green method is suitable for use in microfluidic detection systems.  [43]. Dai and Compton used electrodes coated with gold nanoparticles for determination of arsenic and achieved LOD of 5 ng mL −1 and a linear range of 1-180 ng mL −1 [44]. However, colorimetric methods should be considered for arsenic monitoring purposes as they have several advantages over electrochemical methods in terms of cost effectiveness, portability, and lower susceptibility to fouling.

Path Length
The absorbance values for 1 mm quartz cuvette measurements was 10 times lower than for standard cuvette measurements, as expected (Table 1 and Figure 2). As it can be seen from the calibration graphs ( Figure 3) the analytical response was strong for samples measured in microcuvettes. The linearity compared to standard quartz cuvette measurements was good. The response signal and linearity obtained from the microcuvettes was strong which indicates that the leucomalachite green method is suitable for use in microfluidic detection systems. Absorption spectra of a sample containing 1μg mL −1 arsenic with reagents measured in 10 mm cuvettes (A) and 1 mm quartz cuvettes (B) against reagent blank. Table 1. Average absorbance values for arsenic samples (0.2-1 μg mL −1 ) analyzed in two different types of cuvettes. Abs 1, SD 1 and % RSD 1 show the data for measurements carried out with 10 mm quartz cuvettes. Abs 2, SD 2 and % RSD 2 show the data for measurements in 1 mm quartz cuvettes. All measurements were carried out in triplicate (n = 30).   Figure 3. Comparison of arsenic standards (1-10 μg mL −1 ) measured in quartz cuvettes with 10 mm and 1mm path lengths. Left vertical axes represents the absorbance of standards analyzed in 10 mm quartz cuvettes (blue markers). Right vertical axes represents the absorbance of standards analyzed in 1mm quartz microcuvettes (red markers). All measurements were carried out in triplicate.

Time
The maximum absorbance was reached 73 min after the addition of the dye, however, the absorbance reached 95% of the maximum value within approximately 5 min. At this time the absorbance was sufficiently stable to allow a measurement to be taken, and a 5 min reaction time was used in subsequent experiments. After the maximum absorbance was observed, there was a gradual decrease in absorbance up to a time of 600 min ( Figure 4). Overall, the colour stability was good and suitable for measurements in a microfluidic detection system.

Interference
Among the different species investigated. Fe (II) interfered with the leucomalachite green method ( Table 2). Masking agents such as EDTA, citric acid, ascorbic acid and trimethylethanolamine were tested to overcome the interference, but it was found, that EDTA trimethylethanolamine, and citric acid themselves interfered with the leucomalachite green method. Iron interference was therefore masked by 1% ascorbic acid. Table 2. Effect of foreign species on the determination of arsenic (III) (1 μg mL −1 ). Figure 3. Comparison of arsenic standards (1-10 µg mL −1 ) measured in quartz cuvettes with 10 mm and 1 mm path lengths. Left vertical axes represents the absorbance of standards analyzed in 10 mm quartz cuvettes (blue markers). Right vertical axes represents the absorbance of standards analyzed in 1 mm quartz microcuvettes (red markers). All measurements were carried out in triplicate.

Time
The maximum absorbance was reached 73 min after the addition of the dye, however, the absorbance reached 95% of the maximum value within approximately 5 min. At this time the absorbance was sufficiently stable to allow a measurement to be taken, and a 5 min reaction time was used in subsequent experiments. After the maximum absorbance was observed, there was a gradual decrease in absorbance up to a time of 600 min ( Figure 4). Overall, the colour stability was good and suitable for measurements in a microfluidic detection system.  Figure 3. Comparison of arsenic standards (1-10 μg mL −1 ) measured in quartz cuvettes with 10 mm and 1mm path lengths. Left vertical axes represents the absorbance of standards analyzed in 10 mm quartz cuvettes (blue markers). Right vertical axes represents the absorbance of standards analyzed in 1mm quartz microcuvettes (red markers). All measurements were carried out in triplicate.

Time
The maximum absorbance was reached 73 min after the addition of the dye, however, the absorbance reached 95% of the maximum value within approximately 5 min. At this time the absorbance was sufficiently stable to allow a measurement to be taken, and a 5 min reaction time was used in subsequent experiments. After the maximum absorbance was observed, there was a gradual decrease in absorbance up to a time of 600 min ( Figure 4). Overall, the colour stability was good and suitable for measurements in a microfluidic detection system.

Interference
Among the different species investigated. Fe (II) interfered with the leucomalachite green method ( Table 2). Masking agents such as EDTA, citric acid, ascorbic acid and trimethylethanolamine were tested to overcome the interference, but it was found, that EDTA trimethylethanolamine, and citric acid themselves interfered with the leucomalachite green method. Iron interference was therefore masked by 1% ascorbic acid. Table 2. Effect of foreign species on the determination of arsenic (III) (1 μg mL −1 ).

Interference
Among the different species investigated. Fe (II) interfered with the leucomalachite green method ( Table 2). Masking agents such as EDTA, citric acid, ascorbic acid and trimethylethanolamine were tested to overcome the interference, but it was found, that EDTA trimethylethanolamine, and citric acid themselves interfered with the leucomalachite green method. Iron interference was therefore masked by 1% ascorbic acid. The method performed best at 50 • C ( Figure 5). For practical applications, carrying out the method at high temperatures would add to the cost and overall complexity of the method. The slope and linearity of 4 • C incubation temperature was low compared to the other temperatures (Table 3). It can, however, be concluded that the method has the potential to be applied in low temperature environments, and further examination of the kinetics of the reaction at low temperatures will be carried out. Table 3. Average absorbance values of arsenic samples (0.2-1 µg mL −1 ) analyzed at various incubation temperatures (4, 10, 18, 30, 40, 50 and 60 • C). All measurements were carried out in triplicate (n = 105). The method performed best at 50 °C ( Figure 5). For practical applications, carrying out the method at high temperatures would add to the cost and overall complexity of the method. The slope and linearity of 4 °C incubation temperature was low compared to the other temperatures (Table 3). It can, however, be concluded that the method has the potential to be applied in low temperature environments, and further examination of the kinetics of the reaction at low temperatures will be carried out.  Table 4 shows the results obtained when arsenic samples were analyzed using different buffer pH values. The highest absorbance values were obtained when using pH 5.5 buffer ( Table 4). Because of the analytical response buffer pH of 5.5 was found to be the optimum pH for the procedure ( Figure  6) and used in subsequent experiments. Using ANOVA analysis, significant difference was found between the absorbances at different buffer pH (p < 0.05).  Table 4 shows the results obtained when arsenic samples were analyzed using different buffer pH values. The highest absorbance values were obtained when using pH 5.5 buffer ( Table 4). Because of the analytical response buffer pH of 5.5 was found to be the optimum pH for the procedure ( Figure 6) and used in subsequent experiments. Using ANOVA analysis, significant difference was found between the absorbances at different buffer pH (p < 0.05).   6. Comparison of arsenic samples (0.2-1 μg mL −1 ) analyzed using various sodium triacetate buffers (pH 3.5, 3.9, 4, 4.7, 5, 5.5, 5.8, 6). All measurements were carried out in triplicate (n = 120).

Reagent Ratio
The statistical analysis showed that there was a significant difference between the absorbances obtained using different reagent ratios (p < 0.05). Reagent ratio A gave the best response as it had the highest slope (Figure 7) from all the reagent ratios tested and also the highest absorbance values (Table 5). However, for microfluidic detection system use reagent ratio D was chosen as it has the simplest reagent ratio. Small number of reagents is desirable for colorimetric method's incorporation into microfluidic chip, as this reduces the fabrication costs. Therefore, this would simplify the design of the microfluidic detection system and overall device.

Reagent Ratio
The statistical analysis showed that there was a significant difference between the absorbances obtained using different reagent ratios (p < 0.05). Reagent ratio A gave the best response as it had the highest slope ( Figure 7) from all the reagent ratios tested and also the highest absorbance values (Table 5). However, for microfluidic detection system use reagent ratio D was chosen as it has the simplest reagent ratio. Small number of reagents is desirable for colorimetric method's incorporation into microfluidic chip, as this reduces the fabrication costs. Therefore, this would simplify the design of the microfluidic detection system and overall device.

Reagent Stability
Decreasing absorbance values over time were noted for KIO3 and 0.4 M HCl mixture (Table 6 and Figure 8), however, the slope of the calibration line was relatively consistent in each case. The method yielded analytically useful calibration data over the time period studied and for implementation in a microfluidic device, the change in absolute absorbance values can be corrected for using a regular calibration protocol. Table 6. Average absorbance values for arsenic samples (0.2-1 μg mL −1 ) analyzed periodically on day 0, 1, 2, 3, 4 and 5 under the same conditions with the same potassium iodate and hydrochloric acid mix. All measurements were carried out in triplicate (n = 90).

Reagent Stability
Decreasing absorbance values over time were noted for KIO 3 and 0.4 M HCl mixture (Table 6 and Figure 8), however, the slope of the calibration line was relatively consistent in each case. The method yielded analytically useful calibration data over the time period studied and for implementation in a microfluidic device, the change in absolute absorbance values can be corrected for using a regular calibration protocol. Table 6. Average absorbance values for arsenic samples (0.2-1 µg mL −1 ) analyzed periodically on day 0, 1, 2, 3, 4 and 5 under the same conditions with the same potassium iodate and hydrochloric acid mix. All measurements were carried out in triplicate (n = 90).

Reagent Stability
Decreasing absorbance values over time were noted for KIO3 and 0.4 M HCl mixture (Table 6 and Figure 8), however, the slope of the calibration line was relatively consistent in each case. The method yielded analytically useful calibration data over the time period studied and for implementation in a microfluidic device, the change in absolute absorbance values can be corrected for using a regular calibration protocol. Table 6. Average absorbance values for arsenic samples (0.2-1 μg mL −1 ) analyzed periodically on day 0, 1, 2, 3, 4 and 5 under the same conditions with the same potassium iodate and hydrochloric acid mix. All measurements were carried out in triplicate (n = 90).  A decrease in absorbance for sodium triacetate buffer and LMG dye mix was observed over a five day period (Table 7 and Figure 9), however, the slope of the calibration line was relatively consistent in each case. Table 7. Average absorbance for arsenic samples (0.2-1 µg mL −1 ) analyzed periodically on day 0, 1, 2, 3, 4 and 5 under the same conditions with the same dye and buffer mix. All measurements were carried out in triplicate (n = 75).

Time (Days)
Conc (µg mL − A decrease in absorbance for sodium triacetate buffer and LMG dye mix was observed over a five day period (Table 7 and Figure 9), however, the slope of the calibration line was relatively consistent in each case. Table 7. Average absorbance for arsenic samples (0.2-1 μg mL −1 ) analyzed periodically on day 0, 1, 2, 3, 4 and 5 under the same conditions with the same dye and buffer mix. All measurements were carried out in triplicate (n = 75).

Environmental Samples
The highest absorbance values were observed in samples collected from the St Mullins site. The lowest absorbance was obtained from Bog Lake samples (Table 8). Water samples collected and analyzed from Killeshin reservoir and the River Barrow Carlow site had similar response to control samples ( Figure 10).  All measurements were carried out in triplicate (n = 75).

Environmental Samples
The highest absorbance values were observed in samples collected from the St Mullins site. The lowest absorbance was obtained from Bog Lake samples (Table 8). Water samples collected and analyzed from Killeshin reservoir and the River Barrow Carlow site had similar response to control samples ( Figure 10).  This would indicate that these water samples matrices did not contain high amounts of interfering substances. Statistical analysis revealed significant difference between Bog Lake and all the other water samples (p < 0.05). Also, significant difference was found between Barrow 2 and the other water sample matrices, except for control. The difference in absorbance values could be explained by factors such as sample colour and chemical composition. The Bog Lake sample was strongly coloured due to the presence of humic substances. From this it can be concluded that different water matrices have the potential to affect the result of the leucomalachite green method and this should be taken into account when designing the calibration protocol in any future analytical device.

Apparatus
A 1800 UV-visible spectrometer (Shimadzu, Canby, OR, USA) was used with Hellma (Mullheim, Germany) 10 mm and 1 mm quartz cuvettes for the absorbance measurements. A pH 20 pH meter (Hanna, Nusfalau, Romania) was used for pH measurements.

Sample Preparation
Arsenic (As) sample (6 mL) was transferred to a glass vial. Potassium iodate (1%, 1 mL) and hydrochloric acid (1 M, 0.5 mL) were added, and the mixture was gently shaken and left for 2 min. Leuco malachite green dye was added (0.05%, 0.5 mL), followed by sodium triacetate buffer (13.6%, This would indicate that these water samples matrices did not contain high amounts of interfering substances. Statistical analysis revealed significant difference between Bog Lake and all the other water samples (p < 0.05). Also, significant difference was found between Barrow 2 and the other water sample matrices, except for control. The difference in absorbance values could be explained by factors such as sample colour and chemical composition. The Bog Lake sample was strongly coloured due to the presence of humic substances. From this it can be concluded that different water matrices have the potential to affect the result of the leucomalachite green method and this should be taken into account when designing the calibration protocol in any future analytical device.

Apparatus
A 1800 UV-visible spectrometer (Shimadzu, Canby, OR, USA) was used with Hellma (Mullheim, Germany) 10 mm and 1 mm quartz cuvettes for the absorbance measurements. A pH 20 pH meter (Hanna, Nusfalau, Romania) was used for pH measurements.

Sample Preparation
Arsenic (As) sample (6 mL) was transferred to a glass vial. Potassium iodate (1%, 1 mL) and hydrochloric acid (1 M, 0.5 mL) were added, and the mixture was gently shaken and left for 2 min. Leuco malachite green dye was added (0.05%, 0.5 mL), followed by sodium triacetate buffer (13.6%, 2 mL). The mixture was gently shaken and left for 5 min. The absorbance was measured at 617 nm against reagent blank.

Path Length
Effect of cuvette light path on absorbance was investigated. The procedure was carried out in standard 10 mm quartz cuvettes and micro cuvettes with 1 mm light path for 1-10 µg mL −1 arsenic concentration range. The experiment was carried out in triplicate. The average absorbance was calculated and showed in the result tables and calibration curves were plotted.

Time
The stability of the colour of the sample was tested over time. One µg mL −1 arsenic sample was analyzed for a time period of 600 min. The absorbance measurement was started after the addition of the dye. The procedure was carried out in triplicate.

Interference
The effect of various foreign species at µg mL −1 level on the determination of arsenic was examined. Various foreign ions with concentrations ranging from 100 to 200 µg mL −1 were introduced to 1 µg mL −1 arsenic sample. Tolerance limits of interfering agents were established at concentrations that do not cause more than 5% error in the absorbance values of arsenic at 1 µg mL −1 .

Temperature
A range of different incubation temperatures (4-60 • C) were analyzed. Low temperatures were used in order to determine the viability of the method in low-temperature environments.

pH
The effect of sodium triacetate buffer pH was studied using a range of different pH (3.7-7.3). Also, one-way analysis of variance (ANOVA, Single Factor) was applied to analyze the results.

Reagent Stability
The effect of reagent stability on the arsenic determination was investigated. 1% KIO 3 and 0.4 M HCl reagent mixture was prepared and used for arsenic determination over 5 day time period. Over the course of the experiment fresh arsenic standards and dye and buffer reagent mixture was prepared daily.
The sodium triacetate buffer and LMG dye mixture was prepared and used for arsenic analysis over a 4 day time period. During this fresh arsenic standards and KIO 3 and 0.4 M HCl mixture was prepared daily.

Environmental Samples
Water samples were collected from Bog Lake, Co. Laois (pH 8.39), Killeshin water reservoir, Co. Laois (pH 7. 93), groundwater well Co. Laois (pH 7.4), St. Mullins, Co. Carlow (Barrow 1) (pH 7.31) and the River Barrow Carlow (Barrow 2) (pH 7. 27). All water samples were analyzed in triplicate. The sample matrices were analyzed using the leucomalachite green method in order to determine whether or not arsenic was present in concentrations detectable by the method. The different water matrices were then spiked with arsenic (0.03-20 µg mL −1 ) and appropriate dilutions were made. Prior to the analysis the water samples were filtered firstly using Whatman grade 1 filter paper and secondly with sterile 0.2 µm syringe filters. The pH of the water samples was adjusted to 5.5. The absorbances between different water matrices were compared. In addition, one-way analysis of variance (ANOVA, Single Factor) was used to analyze the results for the different water matrices.

Conclusions
The leucomalachite green method proved to be a good candidate for deployment in microfluidic detection systems and arsenic detection in water. The method was optimized for integration into small scale detection systems. The optimum reaction conditions and other analytical parameters were evaluated. The method was found to be simple, reproducible, fast and robust. The reagent mixtures yielded the optimum results on the day of their preparation, with gradual decrease in absorbance noted over five days. Strong analytical response was obtained from 1 mm light path cuvettes indicating that the method would be suitable for use in small dimension microfluidic detection system. The optimized method was also cost effective as only a small number of reagents were required. The method yielded good results with simple 1 to 1 sample to reagent ratio which would be ideal for microfluidic detection applications. Following an investigation of method's performance in different water samples, it was shown that the method is capable to determine arsenic in various water matrices. There is a potential for method's application in waste water monitoring as well as arsenic detection in areas with particularly high arsenic levels.