3.1 Comparison of PVA-assisted and regular sensors in Ni2+ analysis
Dimethylglyoxime has commonly been used as a chelating agent to Ni2+ to form the nickel dymethylglyoxime square planar complex (Minster 1946). As a colorless bidentate ligand, two dimethylglyoxime molecules coordinate one Ni2+, forming a square planar geometry and inducing a pink color development that is visible to the naked eye. This color development is a sign of the detection procedure, and its intensity is directly proportional to the concentration of Ni2+ in analyte samples (Fig. 2a). This reaction can be carried out on a paper substrate which is low-cost, readily available, and degradable material. However, regular paper-based colorimetric sensors are only suitable for qualitative or semi-quantitative analysis. This is mainly due to the color signal loss at the effect of the infiltration (Pack et al. 2015) and uneven mass flow distribution with nonuniform evaporation flux conditions.
PVA matrix has been widely reported as a non-toxic and low-cost immobilization media in various fields (Takei et al. 2011). In this study, cross-linked by BA, PVA matrix immobilized the DMG molecules within its polymer network on the paper, mitigating the infiltration effect and further suppressing CRE by enhanced hydrogen bonding with the sample droplet, therefore assisting the Ni2+ detection.
To identify the role of PVA matrix in the signals enhancement process, we deposited the same concentration and volume of DMG on the test zone (1 wt%, 4 µL) before the resulting color signals related to various Ni2+ spiked sample was collected, analyzed and compared at the diameter and at global view of the test zone in terms of the surface morphology of blank filter, regular, and PVA-assisted sensors.
As shown in Fig. 2b, a stronger color intensity was observed at PVA-assisted sensors from 0.5 ppm to 1000 ppm compared to that at the regular ones, which became more and more obvious as Ni2+ concentration increased in the sample analyte, suggesting that PVA can reduce the infiltration effect by immobilizing DMG molecules. It could also be seen from Fig. 2c and d that PVA-assisted sensor exhibited higher intensity with much smaller fluctuation along the diameter and across the whole test zone even at 1000 ppm, while the regular one presented a sharp color change of over 60 from the center to the edge, implying that PVA can suppress the CRE.
As explained in the introduction, DMG molecules, due to the infiltration effect of paper substrate, are imbibed and trapped into the fiber network, failing to chelate Ni2+ and thus to be observed by the naked eye (Fig. 3, left), which could be proved by the surface morphology in Fig. 4d that there was no visible color change after adding DMG to the test zone in regular sensor compared with the blank filter paper showed in Fig. 4a. However, in the PVA-assisted sensor (Fig. 3, right), PVA entangled with the cellulose fiber, leaving most DMG on the surface to capture Ni2+, as seen in Fig. 4g that PVA polymer chain interweaved with the cellulose fiber. When Ni2+ sample droplet was added, some particles with a size less than 50 µm, presumably the DMG-Ni2+ complex, were observed on the detection zones of both regular (Fig. 4e) and PVA-assisted sensors (Fig. 4h). However, it has been shown that the numbers of these particles are much higher in the PVA-assisted sensor which can be attributed to the large amount of anchored DMG on the surface by PVA matrix, suggesting that cross-linked PVA can suppress the infiltration effect.
To illustrate the CRE, the edge of regular and PVA-assisted sensors detecting Ni2+ with the concentration of 100 ppm were examined. As shown in Fig. 4c and e, it is obvious that there are a large number of chelated precipitants gathering at the edge of the regular sensor compared to that of few at the center, which explained the stronger color intensity at the edge (Fig. 2c and d). Indeed, this phenomenon has been profoundly reported by a previously published work (Deegan et al. 2000), caused by a strong evaporation flux at the edge of the droplet with a pinned contact line (Deegan et al. 1997). As demonstrated in Fig. 3 (left), the Ni2+ cations were moved by the strong outward flow called capillary toward the edge of the sample droplet while complexing with DMG and forming a ring-like pink spot. As Ni2+ concentration increased, more Ni2+ cations were moved outward, resulting in a more significant fluctuation of color signal and an increasingly larger gap from the center to the edge. Notably in the PVA-assisted sensor, a much more uniform morphology was observed. It could be ascribed to the abundant hydroxyl groups on the PVA chains that can form more strong intermolecular hydrogen bond with the water molecules of the sample droplet as Wang et al. found at their work (Wang et al. 2003), which slowed down the evaporation flux and weaken the outward capillary flow. This would allow sufficient complexation between the Ni2+ cation and DMG molecule that is previously immobilized inside the network made by PVA chain and the cellulose fiber (Fig. 3 right).
Therefore, the PVA matrix could enhance the intensity of the color signals as well as their uniformity during the detection of Ni2+ on the paper substrate by suppressing the infiltration and coffee ring effects with anchored DMG and increased hydrogen bonds.
3.2 Effect of different fabricating conditions of the PVA-assisted Ni2+ sensor
3.2.1 Effect of the presence and adding order of BA
BA is one of the most used cross-linking agents with PVA. This is due to the presence of vacant d-orbital in boron, which results in BA forming strong bonds with hydroxyl groups of PVA chain, and leading to a rapid complexation formation (Fig. 1). This would further impact the morphology and hydrophilicity of the resulting network and thus the distribution of the color signals after capturing Ni2+. To optimize this procedure, the adding order of BA and its weight ratio to PVA were subsequently investigated. As shown in Fig. 5a, whether the BA was added or not, the PVA-assisted sensors presented stronger color intensity in the presence of Ni2+ at the concentrations of 10 ppm and even higher at 100 ppm. Therefore, it was confirmed that even without being cross-linked, the PVA matrix possessed particles with the diameter of 50 µm which are larger than the pores of the filter paper (less than 25 µm, Table S1). Taking the viscosity of PVA into account, DMG aggregated with PVA matrix retained merely at the superficial part of the filter paper. This could be testified by the debris distributed in the center of the test zone (Figure S1b) and an increased WCA of 30.1° (Figure S2c) resulting from the weakened hydrophilicity of the substrate. The debris, aggregates of PVA and DMG, were stuck due to their size which is over 50 µm, which exceeded the aperture of the filter paper (#1 CHR).
And in the cases when PVA cross-linked by BA, there was no distinction in the color intensity between the sensor with BA added before the PVA/DMG addition (BA-pre) and the sensor with BA added right after the PVA/DMG solution (BA-post). However, the BA-pre sensor exhibited a smaller standard deviation of color intensity than that of the BA-post sensor, whose PVA matrix was cross-linked more thoroughly, generated a denser network with less free hydroxyl groups exposed (Figure S1c) and bond with water molecules in the sample droplet for evaporation retarding (Wang et al. 2016). Compared to the WCA of BA-pre (37.5°), the WCA of BA-post sensor reached 72.7°( Figure S2e), implying a drastic decrease in hydrophilicity. It can be speculated that the dense PVA matrix may block the imbibition of the Ni2+ sample droplet, whose dried color pattern was determined by the nonuniform evaporation flux. Therefore, the resulting color intensity became inconsistent, decreasing the accuracy and reliability of the sensor.
Based on these findings, it is important to regulate the density of PVA matrix by using BA to balance the imbibition and evaporation. Obviously, the sensor with BA added before the PVA/DMG addition showed the strongest color intensity and outputted most stable color signal.
3.2.2 Effect of BA concentration
To find the optimal cross-linkage level of the PVA matrix, the weight ratio between PVA and BA was investigated by tuning the concentration of BA aqueous solution (2 µL) dropped on the test zone from 0 wt% to 5 wt% with an increment of 1% while the concentration of PVA was set as 5 wt% (4 µL).
As shown in Fig. 5b, the color intensity of the reaction was increased to its highest level when 2 wt% BA used, and then decreased as the BA weight increased further, despite the concentration of Ni2+ in the analyte sample. Also, as seen in the SEM images, when the BA was absent, coffee ring effect took placed and pushed the debris to at the edge of the test zone, leaving very less derbies at the center (Figure S3a). Unlike this, in the presence of BA, PVA matrix was cross-linked while being moved toward the edge of the test zone, forming larger debris, which was harder to be imbibed or transported to the edge by the capillary flow, and thus distributed more evenly at the center (Figure S3b). By those anchored PVA/DMG debris, the added Ni2+ was caught effectively, leaving more colored precipitation to scan. When BA weight ratio increased to 2 wt%, the cellulose fiber started being fully covered by the cross-linked PVA matrix while most of the pores left unblocked as well as no debris was observed either at the center of the test zone or at the edge (Figure S3c). Attached to the evenly-spread PVA matrix, DMG precipitated the Ni2+ cations in the sample droplet subjected to a reduced evaporation flux by increased hydrogen bonds with the free hydroxyl groups on PVA chain. Thus, a stronger and more uniform color signal was observed and collected by the scanner. However, when more BA was deposited (Figure S3d), the more tightly the cellulose fibers were wrapped by the denser PVA/fiber network, and fewer pores were exposed to impair CRE (Pack et al. 2015). Moreover, an enlarging gap between the PVA/fiber network and the edge could be observed in the Figure S3e and f, implying that the PVA matrix was completely cross-linked before spreading across the whole test zone. It can be speculated that the PVA matrix may be over cross-linked, leaving hydroxyl groups to bond with water molecules for the suppression of the evaporation. These findings have been confirmed by SEM images of the correspondent morphology at the center and edge of the test zones (Figure S3).
With a BA solution at a weight ratio of 2 wt%, PVA matrix was properly cross-linked to form a network with cellulose microfiber anchoring DMG as well as reserving sufficient pores and hydroxyl groups against the nonuniform evaporation flux and consequently endowing the sensor with strongest color signals. Hence, we utilized BA solution of 2 wt% with PVA solution at the concentration of 5 wt% to fabricate the sensor. Apart from the cross-linkage, the increased viscosity of the PVA/DMG compound enabled the droplet to permeate the very superficial part of the substrate causing a slower infiltration (Hapgood et al. 2002).
3.2.3 Effect of the paper substrate
To find the best paper substrate for the colorimetric quantification of Ni2+ in water, we compared the analytical performance of five commercial filter papers complexed with a layer of PVA matrix. As shown in Fig. 5c, the color intensity on the test zone increased as the thickness of the filter paper decreased. #I F and #I CHR, with a thickness of 180 µm (Table S1) produced higher color intensity due to their less diffuse reflection within the thinner cellulose fiber network that distracted the reflection captured by the scanner for colorimetric calculation. At the same time, a smaller error of color intensity and a more uniform color distribution was seen in the #1 CHR compared to the #I F, which could be ascribed to the lower flow rate of filter paper to weaken the infiltration effect. Although not many filter papers are tested in this study, it is predictable that the thickness of paper can significantly influence the performance of the sensor in Ni2+ detection.
3.2.4 Effect of adding volume of PVA/DMG droplet
The concentration of DMG ethanol solution used in this study (80 mM) is aligned with that applied in previous research detecting Ni2+ by DMG on the paper substrate (Zhang et al. 2019). A drop of BA aqueous solution (2 wt%) was first added to wet the test zone with a volume that is half of the subsequently-added PVA/DMG solution. This would ensure the proper cross-linkage of the PVA matrix. To find the best loading volume of the PVA/DMG solution for the colorimetric detection of Ni2+, we evaluated the analytical performance of the sensor under five different adding volumes ranging from 1 to 5 µL, contributing to an overall volume of liquid deposited on the test zone and varying from 1.5 to 7.5 µL. The corresponding analytical result was presented in Fig. 5d. The real and SEM images can be found in Figure S4.
As shown in Fig. 5d, there was a clear trend of increasing color intensity as more PVA/DMG solution was added until 4 µL in the presence of Ni2+ of 10 ppm and 100 ppm, followed by a drastic decrease at 5 µL, implying the saturation of loaded liquid on paper. Further results in Figure S4 revealed that the color signal was distributed evenly with more uniform particles observed at the center of the sensor. A possible explanation for this might be that there were more DMG molecules anchored to the network for complexation with Ni2+ as more PVA/DMG solution was deposited. However, when the added volume exceeded 4 µL, the paper test zone was over-saturated, leaving some of PVA/DMG solution overflowing outside of the paper substrate, and thus reducing the intensity of the colored ferrous-bathophen complex.
3.3 Analytical performance of the PVA-assisted Ni2+ sensors
After determining the optimum fabricating conditions, the sensitivity and accuracy of PVA-assisted senor was validated with a wide range of Ni2+ concentration from 0.5 to 1000 ppm in the analyte sample, with comparison to the regular sensor in the following features; the LOD, linear range and linear coefficient, as shown in Fig. 6 and summarized in Table 1.
As expected, Fig. 6 revealed a clear improvement of color signal that the PVA-assisted sensor exhibited stronger color intensity at all concentrations than the regular ones, accounting for a greater slope of 0.86 than 0.50 of the regular one at a linear range of 0.5–50 ppm. Within this range, a notable linear behavior (R2 > 0.97) was also accomplished by the PVA-assisted senor, indicating a greater accuracy achieved by the PVA matrix. Regarding the sensitivity, the detection limit of the PVA-assisted Ni2+ was calculated as low as 0.92 ppm which is below the regular paper-based sensor with the detection limit of 2.04 ppm. More importantly, the detection limit of the PVA-assisted sensor successfully met the quality standard of wastewater discharged directly from the existing facilities (GB25467-2010) in China, showing great potential in the on-site monitoring of Ni2+ in industrial discharge water.
3.4 Selectivity and interference analysis
The selectivity of the PVA-assisted sensor for Ni2+ detection was validated by the addition of heavy metal ions into the sample analyte, individually and as a mixture in the presence of masking reagents (NaF/NA2S2O3) (Wu et al. 2019). As shown in Fig. 7 left, all individually added ions with 25 times the concentration of Ni2+, showed no pronounced colorimetric response in the sensor — only less than 10% of that from Ni2+ (0.2 mM). For the experiment with a mixture of ions, the colorimetric response of the sensor to Ni2+ varied at a maximum of 5% of the normalized response, indicating the good tolerance to interfering ions including Al3+, Fe2+, Cu2+, Co2+, Zn2+, Pb2+, Ca2+, Mn2+, Mg2+, Li+, K+, and Na+ (Fig. 7 right). Though many other heavy metals were not tested in this study, the high sensitivity of PVA-assisted Ni2+ sensor can presumably be achieved by the addition of a proper masking reagent.
3.6 Robustness and shelf-life analysis
Limited shelf-life has been an issue with paper-based sensors due to lack of stability of reagents on the paper substrate. Herein, the robustness and shelf-life of the PVA-assisted sensor were explored to understand the change of the colored precipitate over time. The color intensity of the sensor responding to the Ni2+ at the concentration of 100 ppm was recorded initially and every 24 hours by the scanner with the sensor sealed in a plastic bag and stored in the dark. As shown in Fig. 9a, the sensor retained nearly 100% within a week, implying the excellent robustness and the negligible impact from the addition of PVA and masking reagents. Furthermore, we fabricated the PVA-assisted sensors at optimum conditions at the intervals of 7, 14, 30, 60, and 90 days, which were then sealed in a plastic bag and stored in the dark. These sensors were disclosed simultaneously to conduct analysis at the same ambient conditions. And by their response to the Ni2+ at the concentration of 100 ppm, the shelf life of the sensor was assessed. Figure 9b revealed that the PVA-assisted sensor retained over 95% of the normalized color response after 90 days and no pronounced change of the colorimetric detection performance within a month, suggesting the excellent storage stability of the sensor. It can be speculated that a longer shelf life than 90 days can be obtained by the PVA-assisted sensor as a further shelf-life experiments extend.
Table 1
Comparison of the analytical performance of PVA-assisted and the regular sensors in the detection of Ni2+.
Method
|
LOD
(ppm)
|
Range of linearity (ppm)
|
Linear coefficient
(ppm)
|
Slope
(a.u.)
|
Regular
|
2.04
|
0.5–50
|
0.95261
|
0.50
|
PVA-assisted
|
0.92
|
0.5–50
|
0.97963
|
0.86
|