- Email: [email protected]
Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters
Analytica Chimica Acta xxx (xxxx) xxx
Contents lists available at ScienceDirect
Analytica Chimica Acta journal homepage: www.elsevier.com/locate/aca
Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters a ^s G.S. Almeida a, Lenka O'Connor Sraj Jonathan J. Peters a, M.Ine , Ian D. McKelvie a, b, a, * Spas D. Kolev a b
School of Chemistry, The University of Melbourne, Victoria, 3010, Australia School of Geography, Earth and Environmental Science, Plymouth University, Plymouth, PL4 8AA, United Kingdom
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
A micro-distillation microfluidic paper-based analytical device (mPAD) is proposed. The highly sensitive mPAD is suitable for the screening of low levels of ammonia. The mPAD sample and detection zones are separated by a m-distillation chamber. Detection is based on colour change of bromothymol blue or nitrazine yellow. The mPAD is validated with a certified reference material and spiked water samples.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 9 December 2018 Received in revised form 13 May 2019 Accepted 23 May 2019 Available online xxx
An easy-to-use, portable 3D microfluidic paper-based analytical device (mPAD) for the determination of total ammonia (i.e., ammonia þ ammonium) in freshwaters is described. It consists of two layers of paper patterned with hydrophilic circular zones, one impregnated with sodium hydroxide (sample zone) and another (detection zone) with an acid-base indicator (nitrazine yellow (NY) or bromothymol blue (BTB)), separated by a m-distillation chamber. Ammonium ions present in the water sample are converted into ammonia gas by reaction with sodium hydroxide in the sample zone. Ammonia then diffuses through a headspace and reacts with an acid-base indicator in the detection zone, the reflectance of which can be related to the total ammonia concentration. The analytical signal at 7.8 mg L1 offered by the m-distillation chamber-based mPAD is more than double of that obtained using a gas-permeable membrane. The proposed mPAD is characterised by a limit of detection of 0.32 or 0.47 mg N L-1 and working concentration ranges of 0.5e3.0 mg N L-1 or 2.0e10 mg N L-1 when using NY or BTB indicators, respectively. This is the first mPAD whose working range covers almost the entire trigger value range (0.32 e2.3 mg N L-1) for ammonia nitrogen in freshwater systems which makes it suitable as a field screening tool for ammonia in freshwaters. An inter- and intra-device repeatability of 7.6% and 9.0%, respectively, has been achieved for the NY-based mPAD (3 mg N L-1) and 13 and 2.5% (8 mg N L-1) for the BTB-based
Keywords: Microfluidic paper-based analytical devices (mPADs) Ammonia Distillation Freshwater
* Corresponding author. E-mail address: [email protected] (S.D. Kolev). https://doi.org/10.1016/j.aca.2019.05.050 0003-2670/© 2019 Elsevier B.V. All rights reserved.
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050
2
J.J. Peters et al. / Analytica Chimica Acta xxx (xxxx) xxx
mPAD. The NY-based mPAD is stable under vacuum for at least 12 days at room temperature, and 56 days if stored in a freezer (20 C).
© 2019 Elsevier B.V. All rights reserved.
1. Introduction
2. Materials and methods
Ammonia is one of the most widely produced chemicals in the world, owing to its numerous uses in industry and agriculture; the worldwide production of ammonia was estimated to be 140 million metric tonnes in 2016 alone [1]. As a pollutant, it typically enters water bodies as agricultural runoff and through discharge of poorly treated sewage effluent [2]. This often has far reaching implications for the health of aquatic environments, fuelling the production of algal blooms leading to the eutrophication of waterways [3]. Ammonia has been shown to be toxic to many forms of aquatic life at quite low concentrations [4], and consequently trigger values for ammonia nitrogen in freshwater systems have been established in the range of 0.32e2.3 mg N L1, depending on the level of protection and freshwater pH [5]. The ability to readily and accurately determine ammonia concentrations is therefore important for assessing water quality and implementing appropriate management practices. Existing ammonia monitoring techniques include the use of the ammonia ion-selective electrode, and the spectrophotometric Berthelotphenate method [6]. Flow injection analysis systems employing optical detectors have also been proposed, often using membrane separation of ammonia gas generated from the sample before its spectrophotometric detection using an acid-base indicator [4] or fluorimetric detection [7]. These gas-diffusion techniques are usually restricted to the laboratory, preventing rapid, on-site sample analysis. Samples containing surface active agents may also cause problems because they can promote membrane wetting and leakage. Ammonia test kits and strips are also commercially available but are prone to both inaccuracy due to measurement errors in the field and interference by suspended particulates and natural organic matter. Microfluidic paper-based analytical devices (mPADs) offer a viable alternative to the aforementioned techniques for field monitoring of ammonia in environmental waters. Originally developed for clinical point-of-care assays [8], mPADs have since been employed in a range of different areas of chemical analysis, including environmental monitoring [9,10]. They have several advantages over traditional monitoring techniques in that they are low cost, portable and have high sample throughput. Determination of higher concentrations of ammonia in soil solution and wastewater samples using mPADs with membrane-based [11] or membraneless [12] ammonia separation has been reported, but there are no accounts of robust mPADs suitable for measurement of total ammonia at the concentrations typically found in freshwaters (i.e., <10 mg N L1). This paper describes a mPAD with a sensitivity which makes it suitable for the determination of total ammonia in freshwaters. Enhanced sensitivity was achieved by the use of an appropriately shaped m-distillation chamber which allowed the free diffusion of ammonia gas across the two layers of the mPAD and facilitated enhanced preconcentration of the analyte onto the detection zone. Two different working ranges were also achieved by using different acid-base indicators (i.e., nitrazine yellow or bromothymol blue).
2.1. Reagents and solutions All reagents were of analytical grade and all solutions were prepared in deionized water (18 MU cm, Millipore Synergy 185) unless stated otherwise. Ammonia stock solution (100 mg N L1) was prepared fortnightly from ammonium chloride (Ajax Finechem), and standards were then prepared daily by dilution from this stock solution in the range of 0.5e10 mg N L1. Ammonium chloride salt was dried at 105 C, and periodically weighed until constant mass was observed prior to stock solution preparation. Acid-base indicator solutions were prepared by dissolving 15 mg of bromothymol blue (BTB) (Sigma-Aldrich) or 20 mg of nitrazine yellow disodium salt (NY) (Eastman) in 10 mL of ethanol followed by addition of 90 mL of deionized water. Solutions of NaOH (Chem-supply) were prepared daily by dissolving 800 mg of the solid in 10 mL of deionized water. 2.2. Fabrication of the mPADs The hydrophilic and hydrophobic zones of the mPAD were designed using Adobe Illustrator and printed onto Whatman grade 4 filter paper with a Xerox Colorqube 8870 wax printer. The wax pattern was heated in an oven (Binder) at 150 C for 2 min [13], so as to embed the hydrophobic wax into the cellulose fibre matrix. The proposed mPAD design was the size of a credit card (64 mm 99 mm), and contained two layers of aligned patterned filter paper (Fig. 1 B and D): one with 15 circular hydrophilic sample zones (8 mm diameter, impregnated with 8 mL of 2 M NaOH solution) and the other with 15 circular hydrophilic detection zones (3 mm diameter) with a side-channel (2 mm long); both separated by a m-distillation unit containing 15 chambers (Fig. 1C), which when laminated together formed 15 individual ammonia sensors. The fabrication of the m-distillation chambers is described in Section 2.3. The mPADs were assembled using laminating pouches (GBC) to maintain the alignment of the sensors' layers, provide mechanical strength to the devices and minimise ammonia gas leakage [11,14e16]. Sample and acid-base indicator introduction holes were pre-punched (Japanese screw punch, 2 mm diameter) on either side of the laminated assemblage. 2.3. Fabrication of the m-distillation unit The m-distillation unit (Fig. 1C) was composed of 15 chambers of one of the two shapes shown in Fig. 2. Such configuration was achieved by using two sheets of laser cut plastic: one laminating sheet (GBC) containing 15 holes corresponding to the sample zone size (i.e., 8 mm diameter), and one transparency sheet (cellulose acetate, Lyreco) with 15 holes matching the detection zone size (i.e., 3 mm diameter). The patterns were designed using 3D modelling software (Rhino 3D), and the plastic sheets were cut using a
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050
J.J. Peters et al. / Analytica Chimica Acta xxx (xxxx) xxx
3
Fig. 1. Schematic representation and photographic image of the final mPAD design. A) Layer of laminating sheet with pre-punched sample introduction holes (2 mm diameter). B) Paper layer incorporating 15 sample zones impregnated with NaOH (8 mm diameter). C) m-Distillation unit e containing a two-layer supported m-distillation device with 15 chambers per unit. D) Paper layer with 15 detection zones (3 mm diameter) with side channels (2 mm long) to facilitate the acid-base indicator addition. E) Layer of laminating sheet with pre-punched acid-base indicator introduction holes (2 mm diameter).
precision laser cutter (Epilog Legend 36 EXT). The laser cut sheets were bonded together by using a laminator (GBC Triad). Twochamber configurations were tested; one without supporting ‘pillars’ (Fig. 2A), and another with such ‘pillars’ (Fig. 2B). These supports were introduced within the 8 mm holes (i.e., within the mdistillation chambers) to minimise deflection of paper layers B and D (Fig. 1) during lamination thus resulting in possible contact of the adjacent sample and detection zones.
2.4. Analytical procedure The procedure for measuring total ammonia concentration in samples involved the introduction of 3 mL of indicator solution into each of the side-channels of paper layer D (Fig. 1) through each of the pre-punched holes on laminating sheet E (Fig. 1). The indicator solution was transported to the detection zones along the corresponding side-channels by capillary force. This was followed by the introduction of 8 mL of ammonium standards or samples through the pre-punched holes of laminating sheet A (Fig. 1) into the sample zones of paper layer B (Fig. 1) where upon reaction with the impregnated NaOH, ammonium ions were converted into ammonia gas. Masking tape was then placed over the holes on laminating
sheet A (Fig. 1) to prevent NH3 gas leakage. After a predetermined period of time, an image of the detection zones on paper layer D (Fig. 1) was acquired using a flatbed scanner (Lide 9000F Mark II, Canon). The change in colour intensity of each spot was obtained by processing the image using ImageJ software (National Institute of Health, USA). For each of the 15 detection zones, the average colour intensity was determined within a 2-mm diameter aperture in the centre of the zone. Red colour intensity was selected as it provided the greatest sensitivity when using NY or BTB. Intensity values obtained were converted to pseudo reflectance values: Reflectance ¼ log10 II0 , where I0 refers to the colour intensity of the blank (deionized water), and I refers to the colour intensity of the detection zone when a sample or standard was measured [17]. Blank or sample readings greater than the 90th or less than the 10th percentiles (on the average 4 out of 15 readings) were treated as outliers and were discarded. We established that applying the 90/ 10 percentiles filtering did not affect the mPADs calibration in relation to the case when no filtering was applied, i.e., the two slopes were not statistically significantly different at the 95% confidence level using the t-test for both the BTB- and NY-based mPADs. However, improved repeatability was achieved with such filtering and the two LODs were decreased by approximately 25%. Similar
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050
4
J.J. Peters et al. / Analytica Chimica Acta xxx (xxxx) xxx
Fig. 2. Schematic representation of the unsupported (A) and supported (B) m-distillation chambers viewed from the side and from the top (left and right-hand sides, respectively). a, transparency sheet; b, laminating sheet.
improvements in repeatability when applying percentiles filtering without altering calibration were obtained in earlier mPAD studies [11,13,18]. 2.5. Validation The m-distillation mPADs were validated against the results obtained by a previously reported flow injection analysis (FIA) method [19] (validated using a certified reference material) using spiked real freshwater samples as well as a certified reference material (ERA, A Waters Company, Lot no. 100517, containing 1000 mg L1 ammonia as NH3). Freshwater collected from Princes Park Pond (Melbourne, Australia) was spiked with ammonium (1 and 2 mg N L1) and was analysed using NY as the acid-base indicator (1 mPAD, n ¼ 11), while that from Kalparrin Gardens Lake (Melbourne, Australia), spiked with ammonium (2 and 8 mg N L1), was analysed using BTB as the indicator (1 mPAD, n ¼ 11). Freshwater samples were directly analysed by the mPADs and then were filtered (Millex 0.45 mm syringe filter), stored in a refrigerator (4 C) and analysed by FIA within 12 h from the time of collection. The certified reference material was used to prepare solutions of 1 and 2 mg N L1. These solutions were analysed by mPADs using NY (1 mg N L1) or BTB (2 mg N L1) as indicators (1 mPAD each, n ¼ 11), and by the FIA method mentioned above (in triplicate). 2.6. Stability studies Initial evaluation of the NY-based m-distillation mPADs stability was undertaken over a period of 9 weeks under two separate conditions: (1) unsealed at room temperature; and (2) vacuum sealed and stored at -20 C (freezer). In both cases the mPADs were kept in the dark and masking tape was placed over the sample introduction holes to minimise reagents’ exposure to ambient air. This tape was removed immediately before sample introduction. Ammonia standards were prepared fresh daily, while the NY solution was prepared at the start of the study and used throughout
its entirety. Device stability was evaluated by comparing reflectance values for a 2 mg N L1 standard for both freshly prepared mPADs and those under storage. The stability of the NY-based mPADs was subsequently evaluated over a period of two weeks using the same storage and analysis protocols as described above, except that all mPADs were vacuum sealed and the indicator was prepared fresh daily.
3. Results and discussion 3.1. Comparison between mPADs utilizing a gas-permeable membrane and a m-distillation chamber The efficiency of ammonia gas transport across a gas-permeable hydrophobic membrane or headspace is deemed to be a major factor governing sensitivity in gas-diffusion flow injection analysis systems designed for ammonia detection [20e22]. Ammonia transport efficiencies across PTFE membranes in flow analysis systems have been reported to be as low as < 30% [20], and as high as 62% [23]. Colorimetric ammonia analysis via gas-diffusion has also been reported in a mPAD format and is analogous to the FIA method [11]. Here, ammonia gas is liberated from a sample zone, it diffuses across a gas-permeable membrane before reaching a detection zone containing a suitable detection reagent. By removing the gasdiffusion membrane which serves as a semi-permeable barrier, a headspace can be formed which allows ammonia to diffuse freely into the detection zone, increasing the sensitivity of the ammonia sensing mPAD. This approach has been reported by Phansi et al. [12] where a spacer was used instead of a gas-permeable membrane. However, no sensitivity improvement compared to the membranebased mPAD, reported earlier [11], has been achieved which could have been the result of the drying out of the indicator solution in the detection zones. It should be pointed out that in the membranebased mPAD [11] the detection zones are enclosed between the laminating poach and PTFE tapes and thus do not dry out during the lamination process and storage. Therefore, it was concluded that because of drying of the
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050
J.J. Peters et al. / Analytica Chimica Acta xxx (xxxx) xxx
indicator solution in the detection zone of the m-distillation mPAD during lamination, the indicator solution should be added postlamination. Hence, a side channel was linked to each 8 mm in diameter circular detection zone on paper layer D (Fig. 1) of the newly developed mPAD to facilitate the acid-base indicator solution addition to the detection zones prior to sample introduction. The ammonia mass transfer efficiency of the distillation approach was assessed by comparing the performance of the mPAD with a m-distillation chamber (Fig. 2A), outlined above, with that of a membrane-based mPAD, which was prepared using the same design configuration, i.e., the indicator (BTB 0.24 mM) was added post-lamination through a side-channel (paper layer D, Fig. 1) and a PTFE membrane was used instead of the m-distillation chamber (Fig. 1). This experimental design would allow for the assessment of ammonia mass transport efficiency in a system where the indicator was being utilised wet. It was observed that as the concentration of ammonia standard increased, the reflectance values for both the membrane and mdistillation devices converged and reached identical values for the 78 mg N L1 standard which could have indicated saturation of the detection zones at this relatively high ammonia concentration. The results at low ammonia concentrations indicated that the absence of a membrane barrier in the m-distillation mPAD significantly improved the ammonia mass transport efficiency, and more than doubled the analytical signal at 7.8 mg N L1 (reflectance obtained for the membrane and m-distillation devices was 0.055 ± 0.019 and 0.13 ± 0.02, respectively, n ¼ 11). It should be noted that a comparison below 7.8 mg N L1 was not possible due to the lack of sensitivity of the membrane-based device. Therefore, it was concluded that the m-distillation mPAD was more suitable than the membrane-based mPAD for analysis of water samples where the ammonia concentration is usually below 7.8 mg N L1. 3.2. Optimisation The mPAD design is flexible and customisable, and several factors contribute to the analytical performance of the device. The following sections present the optimisation of the m-distillation mPAD, and Table 1 summarises the results of this optimisation and lists the mPAD parameters studied in the order in which they were optimized. 3.2.1. m-Distillation chamber height The volume of the m-distillation chamber was varied by adding additional transparency sheets to alter its overall height (Fig. 2). Reduction in the height was expected to allow faster transport of ammonia gas from the sample zone to the detection zone. To examine this effect, the m-distillation chamber height was varied (i.e., 250 mm, 500 mm and 750 mm) at 3 and 5 mg N L1 ammonia concentrations, and the corresponding reflectance values are presented in Fig. S1 (Electronic Supplementary Material). There is a clear relationship between the volume of the mdistillation chamber and the reflectance obtained. This observation
5
is consistent with the expectation that by decreasing the separation between the sample and detection zones of the mPAD, the concentration of ammonia in contact with the indicator will be increasing. As the lowest of the three chambers (250 mm) provided the highest reflectance at both concentrations of the ammonia standards, it was used in the subsequent experiments. 3.2.2. Sample zone optimisation By providing larger hydrophilic sample zones, more analyte will be introduced in the mPAD and this is likely to increase sensitivity [24]. The effect of increasing sample zone diameter (and consequently its volume) on sensitivity was investigated using mPADs with sample zone diameters of 6 mm (4 mL), 8 mm (8 mL), and 10 mm (12 mL). Fig. S2 (Electronic Supplementary Material) confirmed the expectation that the larger sample zones provided higher reflectance values. This effect was more pronounced at higher concentrations of ammonia (2 and 5 mg N L1). However, in the lower and more relevant to freshwaters concentration range of 0.5e1 mg N L1 there was little difference between the reflectance values for the 8- and 10-mm sample zone sizes. Furthermore, the mPAD with 8-mm sample zone exhibited less variance in the reflectance values measured, whereas the 10-mm diameter configuration exhibited greater variability due to distortion of the paper as more sample volume was added. For these reasons, the 8mm diameter (8 mL) sample zone was selected for the subsequent experiments. 3.2.3. Cross-contamination study It was noted that the intra-card repeatability of the mPADs at lower ammonia concentrations was as high as 20.8% (at 4 mg N L1). One potential contributor to this poor repeatability was the possibility of cross-contamination between adjacent sensors on the mPAD. Moreover, it was hypothesised that another contributing factor to the poor repeatability could be the premature colour change of the indicator as a result of occasional contacts between sample and detection zones. A range of experiments were thus conducted to explore and remedy these deficiencies with the aim of improving the repeatability and robustness of the device (Electronic Supplementary Material, Cross-contamination study). The results obtained suggested that there was no crosscontamination between adjacent sensors on the mPAD (Electronic Supplementary Material, Cross-contamination study). 3.2.4. m-Distillation chamber improvements The paper and polymer layers of the mPAD are flexible, and physical distortion of the card either during manufacture or use can result in variation of the distillation chamber volume or in the extreme case it may lead to physical contact between sample and detection zones. This was a major source of gross errors when
Table 1 Optimal values and ranges studied of the main parameters of the m-distillation chamber-based mPADs using BTB or NY as the acid-base indicator. Parameter
Approach/range studied
Optimum value BTB mPAD
NY mPAD
m-Distillation chamber height
250e700 mm 6e10 mm (4e12 mL) with, without 0.12e1.2 mM (NY) 0e8 min (BTB) 0e15 min (NY)
250 mm 8 mm (8 mL) supported chamber 0.24 mM [11] 1 min
0.36 mM 5 min
Sample zone diameter (volume) m-Distillation chamber supports Indicator concentration Colour development time
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050
6
J.J. Peters et al. / Analytica Chimica Acta xxx (xxxx) xxx
testing the mPADs. To address this problem, a different m-distillation chamber design was devised, where ‘pillars’ (Fig. 2B) consisting of two 2 mm wide plastic supports were added at the periphery of the m-distillation chambers to minimise the potential for contact between the sample and detection zones. Devices with and without supported chambers were used to measure a 10 mg N L1 ammonia standard, and gave relative standard deviations of 4.86% and 17.4%, respectively (n ¼ 5 mPADs, 11 measurements each). It was concluded that the presence of the supports provided additional structural rigidity to the device, preventing liquid exchange between paper layers, and enhancing the overall robustness of the device. However, it was noted that the reflectance was ca. 12% lower in devices with the supported chamber compared to the unsupported ones, presumably because the added support pillars obstructed the path of ammonia gas diffusion to the detection zone to some extent. However, this modest decrease in sensitivity was considered as acceptable in view of the improved repeatability of the supported structure which was used in all subsequent experiments. 3.2.5. Acid-base indicator screening Firstly, three indicators and one indicator mixture were chosen as potential candidates for ammonia detection: BTB, NY, phenol red, and a cresol red/thymol blue mixture. A detailed explanation of the reasons why these indicators were selected is provided in the Electronic Supplementary Material (Acid-base indicator screening). NY provided the best analytical signal for the two concentrations of ammonia standards studied (i.e., 3 and 5 mg N L1, Electronic Supplementary Material, Fig. S5), and it was thus selected for further studies. The following section (Section 3.2.6) describes the optimisation of its concentration. BTB was also chosen because it had been used successfully in the ammonia membrane-based mPAD [11], and it allowed the determination of ammonia over a wide range of concentrations. The concentration of BTB used in the subsequent experiments was the optimum one reported by Jayawardane et al. [11] (i.e., 0.24 mM in 10% ethanol). Further increasing this concentration proved to be unfeasible, as BTB was present at maximum solubility in the 10% ethanol solution and enhancing the solubility of the compound would have meant increasing the concentration of ethanol present. This would have made the indicator solution more hydrophobic, thereby negatively impacting on its transport along the hydrophilic channels linked to the detection zones, causing it to bleed into the patterned hydrophobic wax areas. 3.2.6. Optimisation of the NY concentration Since the concentration of the BTB indicator solution was optimized previously [11], only the concentration of the NY indicator solution was optimized in the present study. In the optimisation of this parameter, it was taken into account that the optimum indicator concentration should be high enough with respect to the expected analyte concentrations so that saturation can be avoided while at the same time this concentration must not be excessive, as this would impact negatively on the signal-to-noise ratio and thus lead to lower sensitivity and repeatability. Three indicator solutions prepared in deionized water with concentrations 0.12 mM, 0.24 mM and 0.36 mM were tested using two ammonia standards of 3 and 5 mg N L1. The results obtained are shown in Fig. 3A. The 0.36 mM concentration of NY provided the best response at the two analyte concentrations tested. In order to test higher concentrations of the indicator (i.e., 0.60 mM and 1.2 mM), 10% ethanol had to be used instead of deionized water to improve the indicator's solubility. The results obtained are presented in Fig. 3B. As the NY concentration increased, the signal to noise ratio decreased to a point where reflectance measurements
Fig. 3. Optimisation of the NY indicator solution. A) Concentration range between 0.12 and 0.36 mM of NY in deionized water. B) Concentration range between 0.36 and 1.2 mM of NY in 10% ethanol solution. Green, blue and orange bars correspond to 0.5 ( ),1 ( ), 3 ( ) and 5 ( ) mg N L1 ammonia standards, respectively. Error bars correspond to SD (n ¼ 11). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
were either not reliable (i.e., high RSD values in the case of 1.2 mM), or indistinguishable between the standards of 0.5 and 1.0 mg N L1 (i.e., in the case of 0.60 mM). Therefore, and since no difference was observed between the 0.36 mM NY solution prepared in deionized water or 10% ethanol, the former was selected for further use. 3.2.7. Colour development time Finally, indicator colour development time was studied over a 15-min period (in the case of NY) and over an 8-min period (in the case of BTB), at 3 mg N L1 ammonia (Fig. 4). A plateau in the absorbance values was observed between 5 and 13 min for the NY mPADs. To maximize the sampling rate, 5 min was selected as the optimum colour development time. BTB mPADs showed a plateau range between 0 and 2 min, and thus 1 min was selected as the optimum colour development time. The decrease in the reflectance values at longer times was probably due to the drying up of the detection zones. 3.3. Analytical figures of merit The optimal parameters were used to determine the analytical figures of merit of the newly developed m-distillation-based mPAD method using NY or BTB as acid-base indicators (Table 2). The
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050
J.J. Peters et al. / Analytica Chimica Acta xxx (xxxx) xxx
Fig. 4. Colour development times for mPADs with BTB (0.24 mM) and NY (0.36 mM) and a 3 mg N L1 ammonia standard. One mPAD (n ¼ 11) was tested for each indicator. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
calibration curves for two different ammonia working ranges, both suitable for application to freshwaters, are present in the Electronic Supplementary Material (Fig. S6). The limits of detection (LOD) of the NY- and BTB-based mdistillation mPADs, reported in Table 2, were calculated by the following equation:
LOD ¼
3s m
where m is the slope of the calibration equation and s is the standard deviation of the reflectance of the blank calculated on the basis of the calibration data by the method of Miller and Miller [25]. The detection limits of the NY- and BTB-based mPADs are below the trigger value for ammonia in freshwaters at pH 7.0 or 8.0 at a 95% protection level (i.e., 2.2 or 0.9 mg N L1, respectively [5]), making them suitable for the monitoring of ammonia pollution in freshwaters. Regardless of the indicator used, the newly developed m-distillation chamber-based mPAD exhibited significantly better sensitivity in comparison with previously reported mPADs (Table 2). Inter-card repeatability was determined as the standard deviation of the average reflectance value of three mPADs for the same standard. Intra-card repeatability was calculated as the standard deviation of the reflectance values of the detection zones (n ¼ 11) of the same mPAD. 3.4. Validation A selection of freshwater samples and a diluted certified reference material, previously analysed by a FIA method [19], were used to assess the accuracy of the m-distillation chamber-based mPAD.
7
The corresponding results are presented in Tables 3 and 4. When comparing the results obtained for the newly developed mPAD-based method (using either NY or BTB as the indicators) with those obtained by the FIA method, it can be concluded that a good agreement between both methods was achieved because there was no statistically significant differences between the results obtained by the two methods at the 95% confidence level. It should be highlighted that the freshwater samples used for validation purposes were only filtered for the FIA analysis. The original unfiltered samples were analysed directly by the mPADs. Since the mPADs use filter paper as substrate, it removes any particulates present in the freshwater sample as it is introduced into the device's sampling port. It should also be pointed out that the detection zone is separated from the reagent zone by headspace which eliminates possible interferences of non-volatile matrix components of the sample (Fig. 2). Moreover, only species like methylamine could eventually interfere with the detection, although because it is usually present at mg L1 levels in surface waters its interference is negligible [11,26]. The absence of interference was confirmed by the good recoveries obtained for both devices (i.e., 98e113%).
3.5. Stability One of the main reasons for fabricating this mPAD was to provide an inexpensive and reliable screening tool for determining total ammonia concentration of freshwater samples on-site. A device that can provide accurate and reliable data over a long period of time is desirable for this purpose. In a previous study the stability of a BTB-based mPAD, stored inside vacuum sealed zip-lock bags, was found to be stable for more than 3 months at room temperature and for more than 9 months in a freezer (20 C) [11]. Therefore, the stability of only the NY-based mPAD was investigated as part of the current study. Fig. 5 shows the results obtained over a period of 9 weeks of storage at room temperature and in a freezer (20 C) inside vacuum sealed zip-lock bags. About half of the data points obtained for the mPADs stored at room temperature were outside the control limits and a discrepancy between the true and mean concentration values was observed (Fig. 5A). Even though there is no clear trend in terms of decrease or increase in sensitivity, these storage conditions do not seem to be appropriate. On the other hand, the Shewhart chart for the mPADs stored in the freezer and under vacuum (Fig. 5B) shows that the majority of the data points are within the control limits, and that the true and mean concentration values in this case are very close to each other. It can thus be concluded that the NY-based mPAD is stable up to about 56 days when stored at 20 C inside vacuum sealed zip-locked bags. The same batch of indicator solution was used throughout the experiment, which is advantageous as it avoids the need to freshly prepare this solution.
Table 2 Analytical figures of merit of the NY- and BTB-based m-distillation mPAD methods and previously reported mPAD methods for total ammonia determination in aqueous samples. Acid-base indicator
Working range (mg N L1) Calibration curve Coefficient of correlation Limit of detection (mg N L1) Inter-card repeatability (RSD) Intra-card repeatability (RSD)
m-Distillation-based mPADs
Gas-permeable membrane-based mPAD [11]
Membraneless
mPAD [12]
NY
BTB
BTB
1-nitrophenol
0.50e3.0 R ¼ 4.02 103 C2 þ 1.33 102 C - 2.69 103 0.998 0.32 7.6% at 3 mg N L1 9.0% at 3 mg N L1
2.0e10 R ¼ 1.36 102 C þ 1.11 102
10e50 R ¼ 8.10 104 C þ 8.10 104
10e100 I ¼ 0.43 C þ 179
0.999 0.47 13% at 8 mg N L1 2.5% at 8 mg N L1
0.999 1.8 3.7% at 20 mg N L1 3.9% at 20 mg N L1
0.996 9.0 0.97% at 25 mg N L1 N.A.
N.A., Not available; RSD, relative standard deviation; R, reflectance or absorbance; C, total ammonia concentration; I, colour intensity.
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050
8
J.J. Peters et al. / Analytica Chimica Acta xxx (xxxx) xxx
Table 3 Analysis of a diluted certified reference material (1.000 mg N L1) and freshwater samples (spiked with 1 and 2 mg N L1) by the NY-based mPAD and by a FIA method [19]. Student's t-test was used to assess the accuracy of the mPAD in comparison with the validated FIA method at the 95% confidence level. Sample
Total ammonia concentration (mg N L1) determined by the FIA method a [19]
Total ammonia concentration (mg N L1) determined by the NY-based mPAD b
Certified reference material Spiked freshwater from Princes Park Pond
0.913 ± 0.004 1.11 ± 0.01 2.17 ± 0.01
0.836 ± 0.327 1.13 ± 0.24 2.01 ± 0.30
c
(a) n ¼ 3; (b) n ¼ 11; (c) relative error of 16.4% in comparison with the certified concentration of the diluted reference material.
Table 4 Analysis of a diluted certified reference material (2.000 mg N L1) and freshwater samples (spiked with 2 and 8 mg N L1) by the BTB-based mPAD and by the validated FIA method. Student's t-test was used to assess the accuracy of the mPAD in comparison with the FIA method [19] at the 95% confidence level. Sample
Total ammonia concentration (mg N L1) determined by the FIA method a [19]
Total ammonia concentration (mg N L1) determined by the BTB-based mPAD b
Certified reference material Spiked freshwater from Kalparrin Gardens Lake
1.97 ± 0.01 2.20 ± 0.00 7.97 ± 0.04
2.07 ± 0.28 1.96 ± 0.83 8.42 ± 0.83
c
(a) n ¼ 3; (b) n ¼ 11; (c) relative error of 3.5% in comparison with the certified concentration of the diluted reference material.
To further investigate if vacuum sealing of the NY-based mPAD stored at room temperature would improve its stability, a 12-day study was performed (Fig. 6). Results showed that the true and mean concentration values were close to each other, with most of the data points fitting within the warning limits (±2s). Thus, the NY-based mPAD can be considered stable for at least 12 days at room temperature if stored under vacuum. It should be noted that for the mPAD stored in the freezer, similar results were obtained when using the same NY solution or preparing it freshly. 3.6. Suitability for field applications and scalability The present work describes the development of a screening tool for the monitoring of low levels of total ammonia as a proof of concept, although it can easily be applied to field measurements by a wide range of users (e.g., water management technicians, conservation and environmental organisations, citizen scientists) with minor modifications to simplify use. For instance, disposable transfer pipettes can be used instead of micropipettes (the hydrophilic zones will only absorb the volume it was designed to accommodate), and a connected to a laptop portable scanner (retail
Fig. 5. Stability of the NY-based mPAD monitored over 9 weeks under two conditions: (A) room temperature (RT) and unsealed, and (B) 20 C (freezer) and vacuum sealed. UWL and LWL, upper and lower warning limits (±2s); UCL and LCL, upper and lower control limits (±3s); ammonia standard measured periodically for the two storage conditions, 2 mg N L1 (true value); NY concentration 0.36 mM (prepared on day 0 and used throughout the 9-week experiment). Error bars correspond to SD (n ¼ 11).
Fig. 6. Stability of the nitrazine yellow-based mPAD in vacuum sealed zip-lock bags monitored over 12 days at room temperature (RT). UWL and LWL, upper and lower warning limits (±2s); UCL and LCL, upper and lower control limits (±3s); ammonium standard measured periodically, 2 mg N L1 (true value); nitrazine yellow concentration 0.36 mM (prepared daily). Error bars indicate standard deviation for n ¼ 11. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050
J.J. Peters et al. / Analytica Chimica Acta xxx (xxxx) xxx
price under $200) can be utilised instead of the desktop scanner used in the present study. The procedure to use the mPADs (described in detail in Section 2.4) consists of 5 simple steps, namely, addition of acid-base indicator, addition of sample, covering the sample hole with masking tape, waiting 1 or 5 min (depending on the indicator used), scaning, and measuring the colour intensity of the detection zones. The choice of acid-based indicator to use in the mPAD should be made according to the concentration range of interest, and this decision can be done onsite. If the aim is to assess if the sampled freshwater is above/ below the ammonia trigger values, then NY should be used. Nevertheless, the BTB-based mPAD covers a wider range of concentrations which can be useful when sampling more contaminated waters. The mPADs can easily be stored in the freezer under vacuum and when needed for ammonia monitoring they can be transported to the field at room temperature (i.e. 20e24 C) without losing performance for the day of use. In terms of the manufacturing of the proposed m-distillation devices on a larger scale, this can be readily achieved since even under laboratory conditions several mPADs can be prepared simultaneously. Overall, we were able to fabricate 20 mPADs per hour which involved wax printing and melting, paper cutting, laser cutting of the chambers, adding and drying NaOH reagent, and assembling the devices and laminating them. However, if the mPAD fabrication is to be scaled up for mass production a higher fabrication rate can be achieved by mechanizing the individual steps of this process. 4. Conclusions The development and validation of portable, easy to use and inexpensive mPADs based on m-distillation for the screening of total ammonia in environmental freshwaters have been described. A 2.4-fold increase in analytical signal at 7.8 mg N L1 was achieved by eliminating the gas-permeable membrane barrier between the two paper layers and replacing it with a m-distillation chamber. The inclusion of supporting ‘pillars’ in the m-distillation chamber (Fig. 2B) has increased the robustness and precision of the device. Two working ranges could be achieved by using either nitrazine yellow (0.5e3.0 mg N L1) or bromothymol blue (2.0e10 mg N L1) as acid-base indicators. Both working ranges overlap with the trigger value range (0.32e2.3 mg N L1) for ammonia nitrogen in freshwater systems, although only the NYbased mPAD covers almost the entire ammonia trigger value range. Its low fabrication cost (30 US cents per determination) and relatively good repeatability (7.6% at 3 mg N L1 e nitrazine yellow, 13% at 8 mg N L1 e bromothymol blue) further enhance the attractiveness of the newly developed mPADs as in-field monitoring tools for routine environmental total ammonia analysis. Acknowledgements The authors are grateful to Purnendu (Sandy) Dasgupta from the University of Texas (Arlington) for valuable suggestions regarding this study. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.aca.2019.05.050.
9
References [1] J.A. Ober, Mineral Commodity Summaries 2017, Mineral Commodity SummariesReston, VA, 2017, p. 202. [2] A. Bouwman, D. Lee, W. Asman, F. Dentener, K. Van Der Hoek, J. Olivier, A global high-resolution emission inventory for ammonia, Glob. Biogeochem. Cycles 11 (1997) 561e587. [3] H.W. Paerl, Coastal eutrophication and harmful algal blooms: importance of atmospheric deposition and groundwater as “new” nitrogen and other nutrient sources, Limnol. Oceanogr. 42 (1997) 1154e1165. [4] L.O.C. Sraj, M.I.G.S. Almeida, S.E. Swearer, S.D. Kolev, I.D. McKelvie, Analytical challenges and advantages of using flow-based methodologies for ammonia determination in estuarine and marine waters, Trac. Trends Anal. Chem. 59 (2014) 83e92. [5] Australian and New Zealand Guidelines for Fresh and Marine Water Quality, vol. 1, 2000. The Guidelines/Australian and New Zealand Environment and Conservation Council, Agriculture and Resource Management Council of Australia and New Zealand, Australian and New Zealand Environment and Conservation Council and Agriculture and Resource Management Council of Australia and New Zealand, Canberra. [6] E.W. Rice, L. Bridgewater, A.P.H. Association, A.W.W. Association, W.E. Federation, Standard Methods for the Examination of Water and Wastewater, American Public Health Association, 2012. [7] R.J. Watson, E.C.V. Butler, L.A. Clementson, K.M. Berry, Flow-injection analysis with fluorescence detection for the determination of trace levels of ammonium in seawater, J. Environ. Monit. 7 (2005) 37e42. [8] A.W. Martinez, S.T. Phillips, M.J. Butte, G.M. Whitesides, Patterned paper as a platform for inexpensive, low volume, portable bioassays, Angew. Chem. 46 (2007) 1318e1320. [9] N. Meredith, C. Quinn, D. Cate, T. Reilly, J. Volckens, C. Henry, Paper-based analytical devices for environmental analysis, Analyst 141 (2016) 1874e1887. [10] M.I.G.S. Almeida, B.M. Jayawardane, S.D. Kolev, I.D. McKelvie, Developments of microfluidic paper-based analytical devices (mPADs) for water analysis: a review, Talanta 177 (2018) 176e190. [11] B.M. Jayawardane, I.D. McKelvie, S.D. Kolev, Development of a gas-diffusion microfluidic paper-based analytical device (muPAD) for the determination of ammonia in wastewater samples, Anal. Chem. 87 (2015) 4621e4626. [12] P. Phansi, S. Sumantakul, T. Wongpakdee, N. Fukana, N. Ratanawimarnwong, J. Sitanurak, D. Nacapricha, Membraneless gas-separation microfluidic paperbased analytical devices for direct quantitation of volatile and nonvolatile compounds, Anal. Chem. 88 (2016) 8749e8756. [13] A.N. Ramdzan, M.I.G.S. Almeida, M.J. McCullough, S.D. Kolev, Development of a microfluidic paper-based analytical device for the determination of salivary aldehydes, Anal. Chim. Acta 919 (2016) 47e54. [14] B.M. Jayawardane, S. Wei, I.D. McKelvie, S.D. Kolev, Microfluidic paper-based analytical device for the determination of nitrite and nitrate, Anal. Chem. 86 (2014) 7274e7279. [15] B.M. Jayawardane, L. Coo, R.W. Cattrall, S.D. Kolev, The use of a polymer inclusion membrane in a paper-based sensor for the selective determination of Cu(II), Anal. Chim. Acta 803 (2013) 106e112. [16] B.M. Jayawardane, I.D. McKelvie, S.D. Kolev, A paper-based device for measurement of reactive phosphate in water, Talanta 100 (2012) 454e460. [17] N.C. Birch, D.F. Stickle, Example of use of a desktop scanner for data acquisition in a colorimetric assay, Clin. Chim. Acta 333 (2003) 95e96. [18] B.M. Jayawardane, W. Wongwilai, K. Grudpan, S.D. Kolev, M.W. Heaven, D.M. Nash, I.D. McKelvie, Evaluation and application of a paper-based device for the determination of reactive phosphate in soil solution, J. Environ. Qual. 43 (2014) 1081e1085. [19] L.O.C. Sraj, M.I.G.S. Almeida, I.D. McKelvie, S.D. Kolev, Determination of trace levels of ammonia in marine waters using a simple environmentally-friendly ammonia (SEA) analyser, Mar. Chem. 194 (2017) 133e145. [20] W.E. Van Der Linden, Membrane separation in flow injection analysis: gas diffusion, Anal. Chim. Acta 151 (1983) 359e369. [21] N. Choengchan, T. Mantim, P. Wilairat, P.K. Dasgupta, S. Motomizu, D. Nacapricha, A membraneless gas diffusion unit: design and its application to determination of ethanol in liquors by spectrophotometric flow injection, Anal. Chim. Acta 579 (2006) 33e37. [22] P. Mornane, J. van den Haak, T.J. Cardwell, R.W. Cattrall, P.K. Dasgupta, S.D. Kolev, Thin layer distillation for matrix isolation in flow analysis, Talanta 72 (2007) 741e746. [23] S.W. Willason, K.S. Johnson, A rapid, highly sensitive technique for the determination of ammonia in seawater, Mar. Biol. 91 (1986) 285e290. [24] T. Akyazi, L. Basabe-Desmonts, F. Benito-Lopez, Review on microfluidic paperbased analytical devices towards commercialisation, Anal. Chim. Acta 1001 (2018) 1e17. [25] J.N. Miller, J.C. Miller, Statistics and Chemometrics for Analytical Chemistry, Pearson/Prentice Hall, 2005. [26] A.E. Poste, M. Grung, R.F. Wright, Amines and amine-related compounds in surface waters: a review of sources, concentrations and aquatic toxicity, Sci. Total Environ. 481 (2014) 274e279.
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050
Contents lists available at ScienceDirect
Analytica Chimica Acta journal homepage: www.elsevier.com/locate/aca
Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters a ^s G.S. Almeida a, Lenka O'Connor Sraj Jonathan J. Peters a, M.Ine , Ian D. McKelvie a, b, a, * Spas D. Kolev a b
School of Chemistry, The University of Melbourne, Victoria, 3010, Australia School of Geography, Earth and Environmental Science, Plymouth University, Plymouth, PL4 8AA, United Kingdom
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
A micro-distillation microfluidic paper-based analytical device (mPAD) is proposed. The highly sensitive mPAD is suitable for the screening of low levels of ammonia. The mPAD sample and detection zones are separated by a m-distillation chamber. Detection is based on colour change of bromothymol blue or nitrazine yellow. The mPAD is validated with a certified reference material and spiked water samples.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 9 December 2018 Received in revised form 13 May 2019 Accepted 23 May 2019 Available online xxx
An easy-to-use, portable 3D microfluidic paper-based analytical device (mPAD) for the determination of total ammonia (i.e., ammonia þ ammonium) in freshwaters is described. It consists of two layers of paper patterned with hydrophilic circular zones, one impregnated with sodium hydroxide (sample zone) and another (detection zone) with an acid-base indicator (nitrazine yellow (NY) or bromothymol blue (BTB)), separated by a m-distillation chamber. Ammonium ions present in the water sample are converted into ammonia gas by reaction with sodium hydroxide in the sample zone. Ammonia then diffuses through a headspace and reacts with an acid-base indicator in the detection zone, the reflectance of which can be related to the total ammonia concentration. The analytical signal at 7.8 mg L1 offered by the m-distillation chamber-based mPAD is more than double of that obtained using a gas-permeable membrane. The proposed mPAD is characterised by a limit of detection of 0.32 or 0.47 mg N L-1 and working concentration ranges of 0.5e3.0 mg N L-1 or 2.0e10 mg N L-1 when using NY or BTB indicators, respectively. This is the first mPAD whose working range covers almost the entire trigger value range (0.32 e2.3 mg N L-1) for ammonia nitrogen in freshwater systems which makes it suitable as a field screening tool for ammonia in freshwaters. An inter- and intra-device repeatability of 7.6% and 9.0%, respectively, has been achieved for the NY-based mPAD (3 mg N L-1) and 13 and 2.5% (8 mg N L-1) for the BTB-based
Keywords: Microfluidic paper-based analytical devices (mPADs) Ammonia Distillation Freshwater
* Corresponding author. E-mail address: [email protected] (S.D. Kolev). https://doi.org/10.1016/j.aca.2019.05.050 0003-2670/© 2019 Elsevier B.V. All rights reserved.
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050
2
J.J. Peters et al. / Analytica Chimica Acta xxx (xxxx) xxx
mPAD. The NY-based mPAD is stable under vacuum for at least 12 days at room temperature, and 56 days if stored in a freezer (20 C).
© 2019 Elsevier B.V. All rights reserved.
1. Introduction
2. Materials and methods
Ammonia is one of the most widely produced chemicals in the world, owing to its numerous uses in industry and agriculture; the worldwide production of ammonia was estimated to be 140 million metric tonnes in 2016 alone [1]. As a pollutant, it typically enters water bodies as agricultural runoff and through discharge of poorly treated sewage effluent [2]. This often has far reaching implications for the health of aquatic environments, fuelling the production of algal blooms leading to the eutrophication of waterways [3]. Ammonia has been shown to be toxic to many forms of aquatic life at quite low concentrations [4], and consequently trigger values for ammonia nitrogen in freshwater systems have been established in the range of 0.32e2.3 mg N L1, depending on the level of protection and freshwater pH [5]. The ability to readily and accurately determine ammonia concentrations is therefore important for assessing water quality and implementing appropriate management practices. Existing ammonia monitoring techniques include the use of the ammonia ion-selective electrode, and the spectrophotometric Berthelotphenate method [6]. Flow injection analysis systems employing optical detectors have also been proposed, often using membrane separation of ammonia gas generated from the sample before its spectrophotometric detection using an acid-base indicator [4] or fluorimetric detection [7]. These gas-diffusion techniques are usually restricted to the laboratory, preventing rapid, on-site sample analysis. Samples containing surface active agents may also cause problems because they can promote membrane wetting and leakage. Ammonia test kits and strips are also commercially available but are prone to both inaccuracy due to measurement errors in the field and interference by suspended particulates and natural organic matter. Microfluidic paper-based analytical devices (mPADs) offer a viable alternative to the aforementioned techniques for field monitoring of ammonia in environmental waters. Originally developed for clinical point-of-care assays [8], mPADs have since been employed in a range of different areas of chemical analysis, including environmental monitoring [9,10]. They have several advantages over traditional monitoring techniques in that they are low cost, portable and have high sample throughput. Determination of higher concentrations of ammonia in soil solution and wastewater samples using mPADs with membrane-based [11] or membraneless [12] ammonia separation has been reported, but there are no accounts of robust mPADs suitable for measurement of total ammonia at the concentrations typically found in freshwaters (i.e., <10 mg N L1). This paper describes a mPAD with a sensitivity which makes it suitable for the determination of total ammonia in freshwaters. Enhanced sensitivity was achieved by the use of an appropriately shaped m-distillation chamber which allowed the free diffusion of ammonia gas across the two layers of the mPAD and facilitated enhanced preconcentration of the analyte onto the detection zone. Two different working ranges were also achieved by using different acid-base indicators (i.e., nitrazine yellow or bromothymol blue).
2.1. Reagents and solutions All reagents were of analytical grade and all solutions were prepared in deionized water (18 MU cm, Millipore Synergy 185) unless stated otherwise. Ammonia stock solution (100 mg N L1) was prepared fortnightly from ammonium chloride (Ajax Finechem), and standards were then prepared daily by dilution from this stock solution in the range of 0.5e10 mg N L1. Ammonium chloride salt was dried at 105 C, and periodically weighed until constant mass was observed prior to stock solution preparation. Acid-base indicator solutions were prepared by dissolving 15 mg of bromothymol blue (BTB) (Sigma-Aldrich) or 20 mg of nitrazine yellow disodium salt (NY) (Eastman) in 10 mL of ethanol followed by addition of 90 mL of deionized water. Solutions of NaOH (Chem-supply) were prepared daily by dissolving 800 mg of the solid in 10 mL of deionized water. 2.2. Fabrication of the mPADs The hydrophilic and hydrophobic zones of the mPAD were designed using Adobe Illustrator and printed onto Whatman grade 4 filter paper with a Xerox Colorqube 8870 wax printer. The wax pattern was heated in an oven (Binder) at 150 C for 2 min [13], so as to embed the hydrophobic wax into the cellulose fibre matrix. The proposed mPAD design was the size of a credit card (64 mm 99 mm), and contained two layers of aligned patterned filter paper (Fig. 1 B and D): one with 15 circular hydrophilic sample zones (8 mm diameter, impregnated with 8 mL of 2 M NaOH solution) and the other with 15 circular hydrophilic detection zones (3 mm diameter) with a side-channel (2 mm long); both separated by a m-distillation unit containing 15 chambers (Fig. 1C), which when laminated together formed 15 individual ammonia sensors. The fabrication of the m-distillation chambers is described in Section 2.3. The mPADs were assembled using laminating pouches (GBC) to maintain the alignment of the sensors' layers, provide mechanical strength to the devices and minimise ammonia gas leakage [11,14e16]. Sample and acid-base indicator introduction holes were pre-punched (Japanese screw punch, 2 mm diameter) on either side of the laminated assemblage. 2.3. Fabrication of the m-distillation unit The m-distillation unit (Fig. 1C) was composed of 15 chambers of one of the two shapes shown in Fig. 2. Such configuration was achieved by using two sheets of laser cut plastic: one laminating sheet (GBC) containing 15 holes corresponding to the sample zone size (i.e., 8 mm diameter), and one transparency sheet (cellulose acetate, Lyreco) with 15 holes matching the detection zone size (i.e., 3 mm diameter). The patterns were designed using 3D modelling software (Rhino 3D), and the plastic sheets were cut using a
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050
J.J. Peters et al. / Analytica Chimica Acta xxx (xxxx) xxx
3
Fig. 1. Schematic representation and photographic image of the final mPAD design. A) Layer of laminating sheet with pre-punched sample introduction holes (2 mm diameter). B) Paper layer incorporating 15 sample zones impregnated with NaOH (8 mm diameter). C) m-Distillation unit e containing a two-layer supported m-distillation device with 15 chambers per unit. D) Paper layer with 15 detection zones (3 mm diameter) with side channels (2 mm long) to facilitate the acid-base indicator addition. E) Layer of laminating sheet with pre-punched acid-base indicator introduction holes (2 mm diameter).
precision laser cutter (Epilog Legend 36 EXT). The laser cut sheets were bonded together by using a laminator (GBC Triad). Twochamber configurations were tested; one without supporting ‘pillars’ (Fig. 2A), and another with such ‘pillars’ (Fig. 2B). These supports were introduced within the 8 mm holes (i.e., within the mdistillation chambers) to minimise deflection of paper layers B and D (Fig. 1) during lamination thus resulting in possible contact of the adjacent sample and detection zones.
2.4. Analytical procedure The procedure for measuring total ammonia concentration in samples involved the introduction of 3 mL of indicator solution into each of the side-channels of paper layer D (Fig. 1) through each of the pre-punched holes on laminating sheet E (Fig. 1). The indicator solution was transported to the detection zones along the corresponding side-channels by capillary force. This was followed by the introduction of 8 mL of ammonium standards or samples through the pre-punched holes of laminating sheet A (Fig. 1) into the sample zones of paper layer B (Fig. 1) where upon reaction with the impregnated NaOH, ammonium ions were converted into ammonia gas. Masking tape was then placed over the holes on laminating
sheet A (Fig. 1) to prevent NH3 gas leakage. After a predetermined period of time, an image of the detection zones on paper layer D (Fig. 1) was acquired using a flatbed scanner (Lide 9000F Mark II, Canon). The change in colour intensity of each spot was obtained by processing the image using ImageJ software (National Institute of Health, USA). For each of the 15 detection zones, the average colour intensity was determined within a 2-mm diameter aperture in the centre of the zone. Red colour intensity was selected as it provided the greatest sensitivity when using NY or BTB. Intensity values obtained were converted to pseudo reflectance values: Reflectance ¼ log10 II0 , where I0 refers to the colour intensity of the blank (deionized water), and I refers to the colour intensity of the detection zone when a sample or standard was measured [17]. Blank or sample readings greater than the 90th or less than the 10th percentiles (on the average 4 out of 15 readings) were treated as outliers and were discarded. We established that applying the 90/ 10 percentiles filtering did not affect the mPADs calibration in relation to the case when no filtering was applied, i.e., the two slopes were not statistically significantly different at the 95% confidence level using the t-test for both the BTB- and NY-based mPADs. However, improved repeatability was achieved with such filtering and the two LODs were decreased by approximately 25%. Similar
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050
4
J.J. Peters et al. / Analytica Chimica Acta xxx (xxxx) xxx
Fig. 2. Schematic representation of the unsupported (A) and supported (B) m-distillation chambers viewed from the side and from the top (left and right-hand sides, respectively). a, transparency sheet; b, laminating sheet.
improvements in repeatability when applying percentiles filtering without altering calibration were obtained in earlier mPAD studies [11,13,18]. 2.5. Validation The m-distillation mPADs were validated against the results obtained by a previously reported flow injection analysis (FIA) method [19] (validated using a certified reference material) using spiked real freshwater samples as well as a certified reference material (ERA, A Waters Company, Lot no. 100517, containing 1000 mg L1 ammonia as NH3). Freshwater collected from Princes Park Pond (Melbourne, Australia) was spiked with ammonium (1 and 2 mg N L1) and was analysed using NY as the acid-base indicator (1 mPAD, n ¼ 11), while that from Kalparrin Gardens Lake (Melbourne, Australia), spiked with ammonium (2 and 8 mg N L1), was analysed using BTB as the indicator (1 mPAD, n ¼ 11). Freshwater samples were directly analysed by the mPADs and then were filtered (Millex 0.45 mm syringe filter), stored in a refrigerator (4 C) and analysed by FIA within 12 h from the time of collection. The certified reference material was used to prepare solutions of 1 and 2 mg N L1. These solutions were analysed by mPADs using NY (1 mg N L1) or BTB (2 mg N L1) as indicators (1 mPAD each, n ¼ 11), and by the FIA method mentioned above (in triplicate). 2.6. Stability studies Initial evaluation of the NY-based m-distillation mPADs stability was undertaken over a period of 9 weeks under two separate conditions: (1) unsealed at room temperature; and (2) vacuum sealed and stored at -20 C (freezer). In both cases the mPADs were kept in the dark and masking tape was placed over the sample introduction holes to minimise reagents’ exposure to ambient air. This tape was removed immediately before sample introduction. Ammonia standards were prepared fresh daily, while the NY solution was prepared at the start of the study and used throughout
its entirety. Device stability was evaluated by comparing reflectance values for a 2 mg N L1 standard for both freshly prepared mPADs and those under storage. The stability of the NY-based mPADs was subsequently evaluated over a period of two weeks using the same storage and analysis protocols as described above, except that all mPADs were vacuum sealed and the indicator was prepared fresh daily.
3. Results and discussion 3.1. Comparison between mPADs utilizing a gas-permeable membrane and a m-distillation chamber The efficiency of ammonia gas transport across a gas-permeable hydrophobic membrane or headspace is deemed to be a major factor governing sensitivity in gas-diffusion flow injection analysis systems designed for ammonia detection [20e22]. Ammonia transport efficiencies across PTFE membranes in flow analysis systems have been reported to be as low as < 30% [20], and as high as 62% [23]. Colorimetric ammonia analysis via gas-diffusion has also been reported in a mPAD format and is analogous to the FIA method [11]. Here, ammonia gas is liberated from a sample zone, it diffuses across a gas-permeable membrane before reaching a detection zone containing a suitable detection reagent. By removing the gasdiffusion membrane which serves as a semi-permeable barrier, a headspace can be formed which allows ammonia to diffuse freely into the detection zone, increasing the sensitivity of the ammonia sensing mPAD. This approach has been reported by Phansi et al. [12] where a spacer was used instead of a gas-permeable membrane. However, no sensitivity improvement compared to the membranebased mPAD, reported earlier [11], has been achieved which could have been the result of the drying out of the indicator solution in the detection zones. It should be pointed out that in the membranebased mPAD [11] the detection zones are enclosed between the laminating poach and PTFE tapes and thus do not dry out during the lamination process and storage. Therefore, it was concluded that because of drying of the
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050
J.J. Peters et al. / Analytica Chimica Acta xxx (xxxx) xxx
indicator solution in the detection zone of the m-distillation mPAD during lamination, the indicator solution should be added postlamination. Hence, a side channel was linked to each 8 mm in diameter circular detection zone on paper layer D (Fig. 1) of the newly developed mPAD to facilitate the acid-base indicator solution addition to the detection zones prior to sample introduction. The ammonia mass transfer efficiency of the distillation approach was assessed by comparing the performance of the mPAD with a m-distillation chamber (Fig. 2A), outlined above, with that of a membrane-based mPAD, which was prepared using the same design configuration, i.e., the indicator (BTB 0.24 mM) was added post-lamination through a side-channel (paper layer D, Fig. 1) and a PTFE membrane was used instead of the m-distillation chamber (Fig. 1). This experimental design would allow for the assessment of ammonia mass transport efficiency in a system where the indicator was being utilised wet. It was observed that as the concentration of ammonia standard increased, the reflectance values for both the membrane and mdistillation devices converged and reached identical values for the 78 mg N L1 standard which could have indicated saturation of the detection zones at this relatively high ammonia concentration. The results at low ammonia concentrations indicated that the absence of a membrane barrier in the m-distillation mPAD significantly improved the ammonia mass transport efficiency, and more than doubled the analytical signal at 7.8 mg N L1 (reflectance obtained for the membrane and m-distillation devices was 0.055 ± 0.019 and 0.13 ± 0.02, respectively, n ¼ 11). It should be noted that a comparison below 7.8 mg N L1 was not possible due to the lack of sensitivity of the membrane-based device. Therefore, it was concluded that the m-distillation mPAD was more suitable than the membrane-based mPAD for analysis of water samples where the ammonia concentration is usually below 7.8 mg N L1. 3.2. Optimisation The mPAD design is flexible and customisable, and several factors contribute to the analytical performance of the device. The following sections present the optimisation of the m-distillation mPAD, and Table 1 summarises the results of this optimisation and lists the mPAD parameters studied in the order in which they were optimized. 3.2.1. m-Distillation chamber height The volume of the m-distillation chamber was varied by adding additional transparency sheets to alter its overall height (Fig. 2). Reduction in the height was expected to allow faster transport of ammonia gas from the sample zone to the detection zone. To examine this effect, the m-distillation chamber height was varied (i.e., 250 mm, 500 mm and 750 mm) at 3 and 5 mg N L1 ammonia concentrations, and the corresponding reflectance values are presented in Fig. S1 (Electronic Supplementary Material). There is a clear relationship between the volume of the mdistillation chamber and the reflectance obtained. This observation
5
is consistent with the expectation that by decreasing the separation between the sample and detection zones of the mPAD, the concentration of ammonia in contact with the indicator will be increasing. As the lowest of the three chambers (250 mm) provided the highest reflectance at both concentrations of the ammonia standards, it was used in the subsequent experiments. 3.2.2. Sample zone optimisation By providing larger hydrophilic sample zones, more analyte will be introduced in the mPAD and this is likely to increase sensitivity [24]. The effect of increasing sample zone diameter (and consequently its volume) on sensitivity was investigated using mPADs with sample zone diameters of 6 mm (4 mL), 8 mm (8 mL), and 10 mm (12 mL). Fig. S2 (Electronic Supplementary Material) confirmed the expectation that the larger sample zones provided higher reflectance values. This effect was more pronounced at higher concentrations of ammonia (2 and 5 mg N L1). However, in the lower and more relevant to freshwaters concentration range of 0.5e1 mg N L1 there was little difference between the reflectance values for the 8- and 10-mm sample zone sizes. Furthermore, the mPAD with 8-mm sample zone exhibited less variance in the reflectance values measured, whereas the 10-mm diameter configuration exhibited greater variability due to distortion of the paper as more sample volume was added. For these reasons, the 8mm diameter (8 mL) sample zone was selected for the subsequent experiments. 3.2.3. Cross-contamination study It was noted that the intra-card repeatability of the mPADs at lower ammonia concentrations was as high as 20.8% (at 4 mg N L1). One potential contributor to this poor repeatability was the possibility of cross-contamination between adjacent sensors on the mPAD. Moreover, it was hypothesised that another contributing factor to the poor repeatability could be the premature colour change of the indicator as a result of occasional contacts between sample and detection zones. A range of experiments were thus conducted to explore and remedy these deficiencies with the aim of improving the repeatability and robustness of the device (Electronic Supplementary Material, Cross-contamination study). The results obtained suggested that there was no crosscontamination between adjacent sensors on the mPAD (Electronic Supplementary Material, Cross-contamination study). 3.2.4. m-Distillation chamber improvements The paper and polymer layers of the mPAD are flexible, and physical distortion of the card either during manufacture or use can result in variation of the distillation chamber volume or in the extreme case it may lead to physical contact between sample and detection zones. This was a major source of gross errors when
Table 1 Optimal values and ranges studied of the main parameters of the m-distillation chamber-based mPADs using BTB or NY as the acid-base indicator. Parameter
Approach/range studied
Optimum value BTB mPAD
NY mPAD
m-Distillation chamber height
250e700 mm 6e10 mm (4e12 mL) with, without 0.12e1.2 mM (NY) 0e8 min (BTB) 0e15 min (NY)
250 mm 8 mm (8 mL) supported chamber 0.24 mM [11] 1 min
0.36 mM 5 min
Sample zone diameter (volume) m-Distillation chamber supports Indicator concentration Colour development time
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050
6
J.J. Peters et al. / Analytica Chimica Acta xxx (xxxx) xxx
testing the mPADs. To address this problem, a different m-distillation chamber design was devised, where ‘pillars’ (Fig. 2B) consisting of two 2 mm wide plastic supports were added at the periphery of the m-distillation chambers to minimise the potential for contact between the sample and detection zones. Devices with and without supported chambers were used to measure a 10 mg N L1 ammonia standard, and gave relative standard deviations of 4.86% and 17.4%, respectively (n ¼ 5 mPADs, 11 measurements each). It was concluded that the presence of the supports provided additional structural rigidity to the device, preventing liquid exchange between paper layers, and enhancing the overall robustness of the device. However, it was noted that the reflectance was ca. 12% lower in devices with the supported chamber compared to the unsupported ones, presumably because the added support pillars obstructed the path of ammonia gas diffusion to the detection zone to some extent. However, this modest decrease in sensitivity was considered as acceptable in view of the improved repeatability of the supported structure which was used in all subsequent experiments. 3.2.5. Acid-base indicator screening Firstly, three indicators and one indicator mixture were chosen as potential candidates for ammonia detection: BTB, NY, phenol red, and a cresol red/thymol blue mixture. A detailed explanation of the reasons why these indicators were selected is provided in the Electronic Supplementary Material (Acid-base indicator screening). NY provided the best analytical signal for the two concentrations of ammonia standards studied (i.e., 3 and 5 mg N L1, Electronic Supplementary Material, Fig. S5), and it was thus selected for further studies. The following section (Section 3.2.6) describes the optimisation of its concentration. BTB was also chosen because it had been used successfully in the ammonia membrane-based mPAD [11], and it allowed the determination of ammonia over a wide range of concentrations. The concentration of BTB used in the subsequent experiments was the optimum one reported by Jayawardane et al. [11] (i.e., 0.24 mM in 10% ethanol). Further increasing this concentration proved to be unfeasible, as BTB was present at maximum solubility in the 10% ethanol solution and enhancing the solubility of the compound would have meant increasing the concentration of ethanol present. This would have made the indicator solution more hydrophobic, thereby negatively impacting on its transport along the hydrophilic channels linked to the detection zones, causing it to bleed into the patterned hydrophobic wax areas. 3.2.6. Optimisation of the NY concentration Since the concentration of the BTB indicator solution was optimized previously [11], only the concentration of the NY indicator solution was optimized in the present study. In the optimisation of this parameter, it was taken into account that the optimum indicator concentration should be high enough with respect to the expected analyte concentrations so that saturation can be avoided while at the same time this concentration must not be excessive, as this would impact negatively on the signal-to-noise ratio and thus lead to lower sensitivity and repeatability. Three indicator solutions prepared in deionized water with concentrations 0.12 mM, 0.24 mM and 0.36 mM were tested using two ammonia standards of 3 and 5 mg N L1. The results obtained are shown in Fig. 3A. The 0.36 mM concentration of NY provided the best response at the two analyte concentrations tested. In order to test higher concentrations of the indicator (i.e., 0.60 mM and 1.2 mM), 10% ethanol had to be used instead of deionized water to improve the indicator's solubility. The results obtained are presented in Fig. 3B. As the NY concentration increased, the signal to noise ratio decreased to a point where reflectance measurements
Fig. 3. Optimisation of the NY indicator solution. A) Concentration range between 0.12 and 0.36 mM of NY in deionized water. B) Concentration range between 0.36 and 1.2 mM of NY in 10% ethanol solution. Green, blue and orange bars correspond to 0.5 ( ),1 ( ), 3 ( ) and 5 ( ) mg N L1 ammonia standards, respectively. Error bars correspond to SD (n ¼ 11). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
were either not reliable (i.e., high RSD values in the case of 1.2 mM), or indistinguishable between the standards of 0.5 and 1.0 mg N L1 (i.e., in the case of 0.60 mM). Therefore, and since no difference was observed between the 0.36 mM NY solution prepared in deionized water or 10% ethanol, the former was selected for further use. 3.2.7. Colour development time Finally, indicator colour development time was studied over a 15-min period (in the case of NY) and over an 8-min period (in the case of BTB), at 3 mg N L1 ammonia (Fig. 4). A plateau in the absorbance values was observed between 5 and 13 min for the NY mPADs. To maximize the sampling rate, 5 min was selected as the optimum colour development time. BTB mPADs showed a plateau range between 0 and 2 min, and thus 1 min was selected as the optimum colour development time. The decrease in the reflectance values at longer times was probably due to the drying up of the detection zones. 3.3. Analytical figures of merit The optimal parameters were used to determine the analytical figures of merit of the newly developed m-distillation-based mPAD method using NY or BTB as acid-base indicators (Table 2). The
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050
J.J. Peters et al. / Analytica Chimica Acta xxx (xxxx) xxx
Fig. 4. Colour development times for mPADs with BTB (0.24 mM) and NY (0.36 mM) and a 3 mg N L1 ammonia standard. One mPAD (n ¼ 11) was tested for each indicator. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
calibration curves for two different ammonia working ranges, both suitable for application to freshwaters, are present in the Electronic Supplementary Material (Fig. S6). The limits of detection (LOD) of the NY- and BTB-based mdistillation mPADs, reported in Table 2, were calculated by the following equation:
LOD ¼
3s m
where m is the slope of the calibration equation and s is the standard deviation of the reflectance of the blank calculated on the basis of the calibration data by the method of Miller and Miller [25]. The detection limits of the NY- and BTB-based mPADs are below the trigger value for ammonia in freshwaters at pH 7.0 or 8.0 at a 95% protection level (i.e., 2.2 or 0.9 mg N L1, respectively [5]), making them suitable for the monitoring of ammonia pollution in freshwaters. Regardless of the indicator used, the newly developed m-distillation chamber-based mPAD exhibited significantly better sensitivity in comparison with previously reported mPADs (Table 2). Inter-card repeatability was determined as the standard deviation of the average reflectance value of three mPADs for the same standard. Intra-card repeatability was calculated as the standard deviation of the reflectance values of the detection zones (n ¼ 11) of the same mPAD. 3.4. Validation A selection of freshwater samples and a diluted certified reference material, previously analysed by a FIA method [19], were used to assess the accuracy of the m-distillation chamber-based mPAD.
7
The corresponding results are presented in Tables 3 and 4. When comparing the results obtained for the newly developed mPAD-based method (using either NY or BTB as the indicators) with those obtained by the FIA method, it can be concluded that a good agreement between both methods was achieved because there was no statistically significant differences between the results obtained by the two methods at the 95% confidence level. It should be highlighted that the freshwater samples used for validation purposes were only filtered for the FIA analysis. The original unfiltered samples were analysed directly by the mPADs. Since the mPADs use filter paper as substrate, it removes any particulates present in the freshwater sample as it is introduced into the device's sampling port. It should also be pointed out that the detection zone is separated from the reagent zone by headspace which eliminates possible interferences of non-volatile matrix components of the sample (Fig. 2). Moreover, only species like methylamine could eventually interfere with the detection, although because it is usually present at mg L1 levels in surface waters its interference is negligible [11,26]. The absence of interference was confirmed by the good recoveries obtained for both devices (i.e., 98e113%).
3.5. Stability One of the main reasons for fabricating this mPAD was to provide an inexpensive and reliable screening tool for determining total ammonia concentration of freshwater samples on-site. A device that can provide accurate and reliable data over a long period of time is desirable for this purpose. In a previous study the stability of a BTB-based mPAD, stored inside vacuum sealed zip-lock bags, was found to be stable for more than 3 months at room temperature and for more than 9 months in a freezer (20 C) [11]. Therefore, the stability of only the NY-based mPAD was investigated as part of the current study. Fig. 5 shows the results obtained over a period of 9 weeks of storage at room temperature and in a freezer (20 C) inside vacuum sealed zip-lock bags. About half of the data points obtained for the mPADs stored at room temperature were outside the control limits and a discrepancy between the true and mean concentration values was observed (Fig. 5A). Even though there is no clear trend in terms of decrease or increase in sensitivity, these storage conditions do not seem to be appropriate. On the other hand, the Shewhart chart for the mPADs stored in the freezer and under vacuum (Fig. 5B) shows that the majority of the data points are within the control limits, and that the true and mean concentration values in this case are very close to each other. It can thus be concluded that the NY-based mPAD is stable up to about 56 days when stored at 20 C inside vacuum sealed zip-locked bags. The same batch of indicator solution was used throughout the experiment, which is advantageous as it avoids the need to freshly prepare this solution.
Table 2 Analytical figures of merit of the NY- and BTB-based m-distillation mPAD methods and previously reported mPAD methods for total ammonia determination in aqueous samples. Acid-base indicator
Working range (mg N L1) Calibration curve Coefficient of correlation Limit of detection (mg N L1) Inter-card repeatability (RSD) Intra-card repeatability (RSD)
m-Distillation-based mPADs
Gas-permeable membrane-based mPAD [11]
Membraneless
mPAD [12]
NY
BTB
BTB
1-nitrophenol
0.50e3.0 R ¼ 4.02 103 C2 þ 1.33 102 C - 2.69 103 0.998 0.32 7.6% at 3 mg N L1 9.0% at 3 mg N L1
2.0e10 R ¼ 1.36 102 C þ 1.11 102
10e50 R ¼ 8.10 104 C þ 8.10 104
10e100 I ¼ 0.43 C þ 179
0.999 0.47 13% at 8 mg N L1 2.5% at 8 mg N L1
0.999 1.8 3.7% at 20 mg N L1 3.9% at 20 mg N L1
0.996 9.0 0.97% at 25 mg N L1 N.A.
N.A., Not available; RSD, relative standard deviation; R, reflectance or absorbance; C, total ammonia concentration; I, colour intensity.
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050
8
J.J. Peters et al. / Analytica Chimica Acta xxx (xxxx) xxx
Table 3 Analysis of a diluted certified reference material (1.000 mg N L1) and freshwater samples (spiked with 1 and 2 mg N L1) by the NY-based mPAD and by a FIA method [19]. Student's t-test was used to assess the accuracy of the mPAD in comparison with the validated FIA method at the 95% confidence level. Sample
Total ammonia concentration (mg N L1) determined by the FIA method a [19]
Total ammonia concentration (mg N L1) determined by the NY-based mPAD b
Certified reference material Spiked freshwater from Princes Park Pond
0.913 ± 0.004 1.11 ± 0.01 2.17 ± 0.01
0.836 ± 0.327 1.13 ± 0.24 2.01 ± 0.30
c
(a) n ¼ 3; (b) n ¼ 11; (c) relative error of 16.4% in comparison with the certified concentration of the diluted reference material.
Table 4 Analysis of a diluted certified reference material (2.000 mg N L1) and freshwater samples (spiked with 2 and 8 mg N L1) by the BTB-based mPAD and by the validated FIA method. Student's t-test was used to assess the accuracy of the mPAD in comparison with the FIA method [19] at the 95% confidence level. Sample
Total ammonia concentration (mg N L1) determined by the FIA method a [19]
Total ammonia concentration (mg N L1) determined by the BTB-based mPAD b
Certified reference material Spiked freshwater from Kalparrin Gardens Lake
1.97 ± 0.01 2.20 ± 0.00 7.97 ± 0.04
2.07 ± 0.28 1.96 ± 0.83 8.42 ± 0.83
c
(a) n ¼ 3; (b) n ¼ 11; (c) relative error of 3.5% in comparison with the certified concentration of the diluted reference material.
To further investigate if vacuum sealing of the NY-based mPAD stored at room temperature would improve its stability, a 12-day study was performed (Fig. 6). Results showed that the true and mean concentration values were close to each other, with most of the data points fitting within the warning limits (±2s). Thus, the NY-based mPAD can be considered stable for at least 12 days at room temperature if stored under vacuum. It should be noted that for the mPAD stored in the freezer, similar results were obtained when using the same NY solution or preparing it freshly. 3.6. Suitability for field applications and scalability The present work describes the development of a screening tool for the monitoring of low levels of total ammonia as a proof of concept, although it can easily be applied to field measurements by a wide range of users (e.g., water management technicians, conservation and environmental organisations, citizen scientists) with minor modifications to simplify use. For instance, disposable transfer pipettes can be used instead of micropipettes (the hydrophilic zones will only absorb the volume it was designed to accommodate), and a connected to a laptop portable scanner (retail
Fig. 5. Stability of the NY-based mPAD monitored over 9 weeks under two conditions: (A) room temperature (RT) and unsealed, and (B) 20 C (freezer) and vacuum sealed. UWL and LWL, upper and lower warning limits (±2s); UCL and LCL, upper and lower control limits (±3s); ammonia standard measured periodically for the two storage conditions, 2 mg N L1 (true value); NY concentration 0.36 mM (prepared on day 0 and used throughout the 9-week experiment). Error bars correspond to SD (n ¼ 11).
Fig. 6. Stability of the nitrazine yellow-based mPAD in vacuum sealed zip-lock bags monitored over 12 days at room temperature (RT). UWL and LWL, upper and lower warning limits (±2s); UCL and LCL, upper and lower control limits (±3s); ammonium standard measured periodically, 2 mg N L1 (true value); nitrazine yellow concentration 0.36 mM (prepared daily). Error bars indicate standard deviation for n ¼ 11. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050
J.J. Peters et al. / Analytica Chimica Acta xxx (xxxx) xxx
price under $200) can be utilised instead of the desktop scanner used in the present study. The procedure to use the mPADs (described in detail in Section 2.4) consists of 5 simple steps, namely, addition of acid-base indicator, addition of sample, covering the sample hole with masking tape, waiting 1 or 5 min (depending on the indicator used), scaning, and measuring the colour intensity of the detection zones. The choice of acid-based indicator to use in the mPAD should be made according to the concentration range of interest, and this decision can be done onsite. If the aim is to assess if the sampled freshwater is above/ below the ammonia trigger values, then NY should be used. Nevertheless, the BTB-based mPAD covers a wider range of concentrations which can be useful when sampling more contaminated waters. The mPADs can easily be stored in the freezer under vacuum and when needed for ammonia monitoring they can be transported to the field at room temperature (i.e. 20e24 C) without losing performance for the day of use. In terms of the manufacturing of the proposed m-distillation devices on a larger scale, this can be readily achieved since even under laboratory conditions several mPADs can be prepared simultaneously. Overall, we were able to fabricate 20 mPADs per hour which involved wax printing and melting, paper cutting, laser cutting of the chambers, adding and drying NaOH reagent, and assembling the devices and laminating them. However, if the mPAD fabrication is to be scaled up for mass production a higher fabrication rate can be achieved by mechanizing the individual steps of this process. 4. Conclusions The development and validation of portable, easy to use and inexpensive mPADs based on m-distillation for the screening of total ammonia in environmental freshwaters have been described. A 2.4-fold increase in analytical signal at 7.8 mg N L1 was achieved by eliminating the gas-permeable membrane barrier between the two paper layers and replacing it with a m-distillation chamber. The inclusion of supporting ‘pillars’ in the m-distillation chamber (Fig. 2B) has increased the robustness and precision of the device. Two working ranges could be achieved by using either nitrazine yellow (0.5e3.0 mg N L1) or bromothymol blue (2.0e10 mg N L1) as acid-base indicators. Both working ranges overlap with the trigger value range (0.32e2.3 mg N L1) for ammonia nitrogen in freshwater systems, although only the NYbased mPAD covers almost the entire ammonia trigger value range. Its low fabrication cost (30 US cents per determination) and relatively good repeatability (7.6% at 3 mg N L1 e nitrazine yellow, 13% at 8 mg N L1 e bromothymol blue) further enhance the attractiveness of the newly developed mPADs as in-field monitoring tools for routine environmental total ammonia analysis. Acknowledgements The authors are grateful to Purnendu (Sandy) Dasgupta from the University of Texas (Arlington) for valuable suggestions regarding this study. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.aca.2019.05.050.
9
References [1] J.A. Ober, Mineral Commodity Summaries 2017, Mineral Commodity SummariesReston, VA, 2017, p. 202. [2] A. Bouwman, D. Lee, W. Asman, F. Dentener, K. Van Der Hoek, J. Olivier, A global high-resolution emission inventory for ammonia, Glob. Biogeochem. Cycles 11 (1997) 561e587. [3] H.W. Paerl, Coastal eutrophication and harmful algal blooms: importance of atmospheric deposition and groundwater as “new” nitrogen and other nutrient sources, Limnol. Oceanogr. 42 (1997) 1154e1165. [4] L.O.C. Sraj, M.I.G.S. Almeida, S.E. Swearer, S.D. Kolev, I.D. McKelvie, Analytical challenges and advantages of using flow-based methodologies for ammonia determination in estuarine and marine waters, Trac. Trends Anal. Chem. 59 (2014) 83e92. [5] Australian and New Zealand Guidelines for Fresh and Marine Water Quality, vol. 1, 2000. The Guidelines/Australian and New Zealand Environment and Conservation Council, Agriculture and Resource Management Council of Australia and New Zealand, Australian and New Zealand Environment and Conservation Council and Agriculture and Resource Management Council of Australia and New Zealand, Canberra. [6] E.W. Rice, L. Bridgewater, A.P.H. Association, A.W.W. Association, W.E. Federation, Standard Methods for the Examination of Water and Wastewater, American Public Health Association, 2012. [7] R.J. Watson, E.C.V. Butler, L.A. Clementson, K.M. Berry, Flow-injection analysis with fluorescence detection for the determination of trace levels of ammonium in seawater, J. Environ. Monit. 7 (2005) 37e42. [8] A.W. Martinez, S.T. Phillips, M.J. Butte, G.M. Whitesides, Patterned paper as a platform for inexpensive, low volume, portable bioassays, Angew. Chem. 46 (2007) 1318e1320. [9] N. Meredith, C. Quinn, D. Cate, T. Reilly, J. Volckens, C. Henry, Paper-based analytical devices for environmental analysis, Analyst 141 (2016) 1874e1887. [10] M.I.G.S. Almeida, B.M. Jayawardane, S.D. Kolev, I.D. McKelvie, Developments of microfluidic paper-based analytical devices (mPADs) for water analysis: a review, Talanta 177 (2018) 176e190. [11] B.M. Jayawardane, I.D. McKelvie, S.D. Kolev, Development of a gas-diffusion microfluidic paper-based analytical device (muPAD) for the determination of ammonia in wastewater samples, Anal. Chem. 87 (2015) 4621e4626. [12] P. Phansi, S. Sumantakul, T. Wongpakdee, N. Fukana, N. Ratanawimarnwong, J. Sitanurak, D. Nacapricha, Membraneless gas-separation microfluidic paperbased analytical devices for direct quantitation of volatile and nonvolatile compounds, Anal. Chem. 88 (2016) 8749e8756. [13] A.N. Ramdzan, M.I.G.S. Almeida, M.J. McCullough, S.D. Kolev, Development of a microfluidic paper-based analytical device for the determination of salivary aldehydes, Anal. Chim. Acta 919 (2016) 47e54. [14] B.M. Jayawardane, S. Wei, I.D. McKelvie, S.D. Kolev, Microfluidic paper-based analytical device for the determination of nitrite and nitrate, Anal. Chem. 86 (2014) 7274e7279. [15] B.M. Jayawardane, L. Coo, R.W. Cattrall, S.D. Kolev, The use of a polymer inclusion membrane in a paper-based sensor for the selective determination of Cu(II), Anal. Chim. Acta 803 (2013) 106e112. [16] B.M. Jayawardane, I.D. McKelvie, S.D. Kolev, A paper-based device for measurement of reactive phosphate in water, Talanta 100 (2012) 454e460. [17] N.C. Birch, D.F. Stickle, Example of use of a desktop scanner for data acquisition in a colorimetric assay, Clin. Chim. Acta 333 (2003) 95e96. [18] B.M. Jayawardane, W. Wongwilai, K. Grudpan, S.D. Kolev, M.W. Heaven, D.M. Nash, I.D. McKelvie, Evaluation and application of a paper-based device for the determination of reactive phosphate in soil solution, J. Environ. Qual. 43 (2014) 1081e1085. [19] L.O.C. Sraj, M.I.G.S. Almeida, I.D. McKelvie, S.D. Kolev, Determination of trace levels of ammonia in marine waters using a simple environmentally-friendly ammonia (SEA) analyser, Mar. Chem. 194 (2017) 133e145. [20] W.E. Van Der Linden, Membrane separation in flow injection analysis: gas diffusion, Anal. Chim. Acta 151 (1983) 359e369. [21] N. Choengchan, T. Mantim, P. Wilairat, P.K. Dasgupta, S. Motomizu, D. Nacapricha, A membraneless gas diffusion unit: design and its application to determination of ethanol in liquors by spectrophotometric flow injection, Anal. Chim. Acta 579 (2006) 33e37. [22] P. Mornane, J. van den Haak, T.J. Cardwell, R.W. Cattrall, P.K. Dasgupta, S.D. Kolev, Thin layer distillation for matrix isolation in flow analysis, Talanta 72 (2007) 741e746. [23] S.W. Willason, K.S. Johnson, A rapid, highly sensitive technique for the determination of ammonia in seawater, Mar. Biol. 91 (1986) 285e290. [24] T. Akyazi, L. Basabe-Desmonts, F. Benito-Lopez, Review on microfluidic paperbased analytical devices towards commercialisation, Anal. Chim. Acta 1001 (2018) 1e17. [25] J.N. Miller, J.C. Miller, Statistics and Chemometrics for Analytical Chemistry, Pearson/Prentice Hall, 2005. [26] A.E. Poste, M. Grung, R.F. Wright, Amines and amine-related compounds in surface waters: a review of sources, concentrations and aquatic toxicity, Sci. Total Environ. 481 (2014) 274e279.
Please cite this article as: J.J. Peters et al., Development of a micro-distillation microfluidic paper-based analytical device as a screening tool for total ammonia monitoring in freshwaters, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2019.05.050