The current study combines two materials, i.e. the membrane filters and the nanomaterials, to exploit their inherent characteristics for the removal of toxic materials from water. For this purpose, membranes of various pore sizes are used along with silver nanoparticles with various average particle sizes.
Figure 2 shows the transmission electron micrograph of the silver nanoparticles, which were prepared using three distinct combinations of the bottom-up nanoparticles synthesis process for obtaining nanoparticles of three different average nanoparticles sizes i.e. 10 nm (Fig. 2a), 40 nm (Fig. 2b) and 70 nm (Fig. 2c). Silver nanoparticles with average size of 10 nm were obtained with 10 °C synthesis temperature, Polyvinyl Pyrolidone (PVP) as the dispersing and stabilizing agent and hydrazine as the reducing agent; 40 nm were obtained with 40 °C synthesis temperature, PVP and hydrazine; and 70 nm were obtained with 40 °C synthesis temperature, oligo condensate of naphthalene sulfonic acid (OCNS) and D-glucose combinations. It can be observed from the transmission electron micrographs (Fig. 2) that the particles have been formed with proper geometrical shape, almost spherical for most particles. It can be observed that the particles prepared with PVP have less variation in the particle size distribution (narrow) while those prepared with OCNS comparatively have a wide particle size distribution (broad). The mechanism of such particles formation with varying average particle size and distribution are discussed elsewhere .
Nanoparticle assisted nano filtration (NANF) effect on phenol removal from water
The work started with adsorption of phenols using silver nanoparticles. Since three distinct average size particle samples were prepared, the three particles batches are used in equal proportion, with the total particle concentrations at 60 ppm. Gulrajani et al.  have shown that at 60 ppm particle concentration good particle performance was obtained in their study. The effects of nanoparticle size and membrane pore size on filtration of the phenol are presented in Table 1. It can be observed from Table 1 that for all particle sizes the presence of the nanoparticles show a significant improvement in filtering the phenol with reference to the respective counterpart, i.e. in the absence of the particles for the respective membrane. The results clearly show the effectiveness of the NANF process for phenol removal. As high as 675 % improvement is observed for NANF process (membrane filtration with nanoparticles), compared to the process without nanoparticles. These results can be explained based on the mechanism of phenol adsorption followed by agglomeration of the nanoparticles, which results in a net increase in the particle size, which then cannot pass through the membrane pores and hence are filtered. Such effect is schematically shown in Fig. 3. In the absence of nanoparticles (Fig. 3a) the molecular phenolic compounds can easily pass through the pores of the membranes by diffusion. However, in the presence of nanoparticles (Fig. 3b) whose surface area is very large, there exists a spontaneous attraction of phenolic compounds towards the surface of the particles due to i) electrostatic charge attraction between the nanoparticles and the phenolic compounds (Fig. 3c) and ii) aggregate formation to obtain a lower energy state of the solute and the squeezing out of solvent in support of solute separation .
The effect of membrane pore size on NANF is observed to be significant. Filtration effects are large for the 80 nm pore size membranes (M4), i.e. up to a 675 % improvement (obtained from the phenol removal % with the presence and absence of nanoparticles [((31–4) × 100)/4]). It is surprising to note that such % improvement is low for 10 nm pore size membranes (M1), i.e. only a 151 % improvement. This may be explained to the fact that for large pore size almost all the phenolic compounds pass through the pores in the absence of the particles and the phenol removal % is very low; but with the low value in the denominator, for the case of the presence of nanoaparticles, the % improvement becomes large, as shown in the calculation above. However, in contrast with the small pore size membrane, i.e. 10 nm, there is a considerable amount of phenol removal even in the absence of particles, i.e. 27 %. Hence, even though the actual amount of phenol removal is high in the combined nanoparticles and membrane approach, i.e. 68 %, the % improvement in the presence of nanoparticles is only 151 %. It can be noticed that the best results are obtained for the small pore size membrane (10 nm) and the nanoparticles combinations. This can be explained to the fact that the nanoparticles have enormous surface area particularly below the 100 nm size. The specific surface area of the particles increases exponentially with decreasing size (Theodore, 2005). Hence, availability of large surface areas contributes to surface adsorption of the toxic compounds, which are then filtered by membrane filtration. The low phenol removal %, i.e. 31 % with the 80 nm membrane, even in the presence of the nanoparticles, is due to the presence of nanoparticles whose size is less than 80 nm. These particles pass through the pores resulting in lower nanoparticles concentration and a consequent decrease in phenol removal. This indicates that the smaller particles play an important role in adsorption and hence the phenol removal %. The effectiveness of the hybrid process is also reported by Geckeier et al. . They studied polymer enhanced ultrafiltration (PEUF), which was helpful in removing metal from wastewater by complexing of polymers with the metals and subsequent removal from the wastewater. Similar study was reported by Li et al. .
Effect of concentration and particle size of silver nanoparticles
Silver nanoparticles with three different average particle sizes, i.e. 10, 40 and 70 nm, each with three different particle concentrations, 60, 120 and 180 ppm, are used for phenol removal study on membranes (M1 – M5). The results are depicted in Fig. 4 and in Fig. 5. Figure 4a shows the effects of the three particle concentrations, for 10 nm silver nanoparticles. Similarly, Figs. 4b and c show the effects of the three particle concentrations for 40 and 70 nm nanoparticles, respectively. Inversely, Fig. 5 shows the effect of particle size for a given concentration, i.e. Figure 5a shows the effect of the three nanoparticles sizes for the 60 ppm particle concentration. Similarly, Figs. 5b and c show the effect of the three nanoparticles size for the 120 and 180 ppm concentrations, respectively.
From Figs. 4-5, it can be inferred that for the 10 nm pore size membrane that about 100 % phenol removal is obtained for 120 and 180 ppm concentrations for all the three nanoparticles sizes. Further, it can be observed that for the 60 ppm particle concentration that, for almost for all particle sizes (Fig. 5a), the results are less than 100 % phenol removal on the 10 nm pore size membrane. Further, it can be observed that the phenol removal % decreases with increase in pore size of the membrane for almost all particle sizes, but with a varying slopes (Fig. 4a to c). The decrease is sharp after the 30 nm membrane and is well noticed for 60 ppm particle concentration (Fig. 4a to c and Fig. 5a). However, for higher particle concentrations, the decrease in phenol removal % with increasing membrane pore size is low. For the 10 nm size nanoparticles, a notable decrease is observed both for 120 (Fig. 5b) and 180 ppm (Fig. 5c) above the 30 nm membrane size. However, the decrease is less for the particle size of 40 nm both for 120 (Fig. 5b) and 180 ppm (Fig. 5c) after the 30 nm membrane size and is even less, almost reaching 100 % for the particle size of 70 nm both for 120 (Fig. 5b) and 180 ppm (Fig. 5c) above the 30 nm membrane size (Fig. 4c).
These observations can be explained based on nanoparticles size, membrane pore size and the surface area available of the nanoparticles for adsorption . The surface area of the individual nanoparticles, total number of nanoparticles available for a given particle size and concentration and hence the total surface area (TSA) available for the adsorption studies have been discussed previously and the values are presented in Table 2. It can be observed from Table 2 that the highest TSA is obtained for the 10 nm particle size at 180 ppm concentration, i.e. 4710 × 1012 nm2, and the lowest TSA of 2461 × 1011 nm2 for the particle size of 70 nm at 60 ppm concentration with the total % difference of 1813 [(4710 × 1012 - 2461 × 1011) × 100/2461 × 1011]. For the other particle sizes and concentrations, the TSAs of the particles lie in between these two limits. Additionally, it can be observed from Table 2 that for the 60 ppm concentration the TSA for 10, 40 and 70 nm particles are 1570 × 1012, 4521 × 1011 and 2461 × 1011, respectively, and so on for the other concentrations. The percentage increases in TSA between the particles with various concentrations are shown separately in Table 3. The general formula for calculating % increase in TSA between various samples is shown as a foot note under Table 3. It can be observed from Table 3 that between samples 1& 4 (between10 and 40 nm both at 60 ppm conc.) the TSA increase is 247 % while between samples 4 & 7 (between 40 and 70 nm both at 60 ppm conc.) the TSA increase is only 84 %. However, the TSA increase between samples 1 & 7 (between 10 and 70 nm both at 60 ppm conc.) is a maximum of 537 %. Further it could be carefully noted that similar comparisons for the higher concentrations, i.e. 120 ppm and 180 ppm, also result in similar TSA % increases i.e. 247 %, 84 % and 537 %. However, in such trials when compared for the smallest particle (10 nm) with highest concentration (180 ppm) (sample 3) and largest particle (70 nm) at the lowest concentration (60 ppm) (sample 7) the TSA increase is the highest of all, about 1813 %.
It can be observed from Figs. 4-6 that the 10 nm particles at 180 ppm concentration with the 10 nm pore membrane result in 100 % phenol removal (Fig. 4a), while the 70 nm at 60 ppm with the 10 nm pore size membrane results in phenol removal of only about 50 % (Fig. 4c), which can be attributed to the large difference in the total surface areas available for adsorption, between these two samples, i.e. 1813 % difference. However, the phenol removal % for 10 nm particles on the larger pore size membranes, shows less than 100 % and is dramatic above the 30 nm pore size for all three concentrations (Fig. 4a), which could be because of the loss of nanoparticles through the large pore size membranes. Contrastingly, with 70 nm particle such decrease is observed to be low for 60 ppm and very low for the higher concentrations i.e. 120 and 180 ppm (Fig. 4c). In fact, at these concentrations almost 100 % phenol removal is observed for 70 nm particles except with the 80 nm pore size membrane, which may be due to some loss of the 70 nm particles with the 80 nm pore size membrane. The observations of phenol removal by the other particle sizes and concentrations lie in between these two limits of smallest particle and highest concentrations with the smallest membrane pore size (Fig. 4a & Fig. 5c) and the largest particle and lowest concentration with the largest membrane pore size (Fig. 4c & Fig. 5a). The results can be attributed to the corresponding TSA along with the loss of particles through the large pore size membrane, i.e., small particles of high concentrations might have large TSA but fail to show 100 % phenol removal with the large pore size membranes, which might be by virtue of their small size because they pass through the membrane pores and are lost (Fig. 4a & Fig. 5a to Fig. 5c).
The effect of the particle size upon phenol removal can be realized at 60 ppm concentration and with the 10 nm pore size membrane (Fig. 5a) where the loss of small particles should be very low. It can be observed from Fig. 4a that the 10, 40 and 70 nm size particles give about 82 %, 68 % and 52 % phenol removal which might be due to the difference in the TSA i.e. 1570 × 1012 nm2 (sample 1), 4521 × 1011 nm2 (sample 4) and 2461 × 1011 nm2 (sample 7) for 10, 40 and 70 nm size particles, respectively. The corresponding % TSA increase between the samples are found to be 247 %, 84 % and 537 % for sample 1 & 4, sample 4 & 7 and sample 1 & 7, respectively. Though % increase between sample 1 & 7 is very large, the % phenol removal is only about 82 % for the 10 nm particles as against 52 % for the 70 nm particles. This is because the 10 nm particle is only the average size and so there may be particle loss for those whose size is smaller than 10 nm. This is also reported by Sarkar and Acarya . They observed that phenol removal was maximum at lower particle size. Furthermore, Roostaei and Tezel , noted that adsorption capacity decreased by increasing the particle size. It can however, be concluded that the particle loss through the membrane impacts the phenol removal even though small size particle yields high TSA.
Effect of nanoparticles mixtures
In the individual particle study discussed in the previous section, either there was a particle loss effect for small particles or there was a low total surface area (TSA) effect for large particles which were particularly noticed at 60 ppm concentration. Hence we studied the effect of particle mixtures in various proportions for three concentrations, 60 ppm, 120 ppm and 180 ppm, for phenol removal. Results of the studies are shown in Fig. 6. Figure 6a depicts the effect of the three different particles i.e. 10 nm, 40 nm and 70 nm mixtures in equal proportions (1:1:1) for the three different concentrations. Accordingly 20 ppm of 10 nm particles, 20 ppm of 40 nm particles and 20 ppm of 70 nm particles were taken for forming a silver nanoparticles mixture solution. Similarly this was done for the other two concentrations i.e. for 120 ppm 40 ppm of each particle solutions and for 180 ppm 60 ppm of each particle solutions were taken. The effect of equal proportions (1:1:1) of the three nanoparticles (triparticle mixture) for three concentrations on various membrane are shown in Fig. 6a. Similarly, the other possible proportions of particles i.e. diparticle mixtures such as 1:1 (10 nm, 40 nm), 1:1 (10 nm, 70 nm,) and 1:1 (40 nm, 70 nm) were carried out. The results of such diparticle mixtures are shown in Fig. 6b, c and d, respectively.
It can be observed from Fig. 6a, b, c, d that the triparticle mixture and all the diparticle mixtures at 60 ppm result in less than 100 % phenol removal ranging from 70 % for triparticle mixture (Fig. 6a) to 58 % for diparticle (40 nm, 70 nm) (Fig. 6d) for the 10 nm pore membrane. The phenol removal % decreases drastically on higher pore size membranes (M2-M4) for the triparticle and all the diparticle combinations for 60 ppm concentrations. This clearly brings out the effect of particle loss, particle size and the TSA effect. The effect of particle size can be observed from Fig. 6a in that the phenol removal of the triparticle mixture for the 10 nm pore membrane is 70 % however the same is about 80 % for pure 10 nm particles for 60 ppm (Fig. 4a). This indicates mixing the 40 nm & 70 nm particles results in a net decrease in TSA and so a decrease in phenol removal. The particle size and the mixture effects can be perceived from Figs. 6b to c, in that for the diparticle mixture of 10 nm and 40 nm the phenol removal is again 80 % (Fig. 6b). This result can be explained that without 40 nm particles, there would be some loss of 10 nm particles and in this case that would be minimal as there would be 40 nm particles which would prevent the loss of particles. Hence, the decrease in TSA of pure 60 ppm 10 nm particle is countered by the 40 nm particles to prevent their loss and, additionally, the TSA contribution from the 40 nm particles together might contribut for the equivalent phenol removal % to that of pure 10 nm at 60 ppm silver nanoparticles. The particle loss factor and particle size can also be perceived at high concentrations of 180 ppm from Fig. 6b and c. It can be observed from these Figures that for the diparticle mixture of 10 nm and 40 nm the phenol removal is only about 90 % whereas in the other diparticle mixture of 10 nm and 70 nm it is 100 % for the 80 nm pore membrane. In the former case there would be notable loss of particles, though there might be some agglomeration effect, resulting in less than 100 % phenol removal. However, in the latter case, though there might be loss of 10 nm particles, the contribution of TSA by 70 nm particles coupled with the TSA of 10 nm particles (which are retained by the 70 nm particles and by agglomeration effects) is sufficiently high to attain 100 % phenol retention. Likewise, the combined particle based adsorption effect would be techno-economically and ecologically better for efficient removal of the toxic contents rather than using the individual small and large particles for a given pore size membrane. For example, it can be seen in Figs. 4 and 5 that 100 % phenol removal was not observed even at 180 ppm concentration for 70 nm particles. However, in the case of diparticle mixtures of 10 nm and 70 nm and 40 nm and 70 nm for the same 180 ppm concentration, 100 % phenol removal has been obtained for all pore size membranes including the 80 nm pore size. Though the phenol removal % could be increased with individual particles just by increasing the concentration above 180 ppm, it would not be economical. Similarly, low particle size would be preferred because, generally, it is difficult to synthesize smaller particles and for smaller the particles there are chances that these might interfere with the ecological system. Further, although small particles lead to higher surface areas, it also creates operational issues as small particles would not be stable in a treatment system. They could be easily swept away by the feed, thus posing difficulties with the maintenance of the system. Hence, for a given set of membranes, a combined particle mixture would yield a better toxic content removal due to the control of the loss of small particles by larger particles by a blocking effect and also by a particle agglomeration effect. This allows for exploiting the large total surface area (TSA) of small particles.
COD removal & toxicity study
As a further examination to the phenol removal % study, the filtrates were subjected to chemical oxygen demand (COD) studies as described in the experimental section. For this study, only four selected combinations were chosen from inference of the detailed phenol removal study done in the previous sections. The effect on COD removal by low and high particle concentrations i.e. 60 ppm and 180 ppm were studied for both individual particle (Fig. 7a & b) and for particle mixtures (Fig. 7c & d). It can be observed from these Figures that the COD removal % almost reflects the phenol removal %. It can be observed from Fig. 7a and b that the 10 nm particle is effective even at 60 ppm particle concentration, but only for the 10 nm & 30 nm membrane pore sizes. Beyond the 30 nm membrane pore size, the COD removal % decreases notable due to particle loss through the larger size pores. With high particle size, i.e. 70 nm, drastic change is not noticed with membrane pore size, there is only a steady decrease. However, with the 180 ppm concentration, such effects are unnoticed due to the high concentration, which results in 100 % COD removal with only a slight decrease for the 50 nm and 80 nm membrane pore size due to phenol and particle loss effects. For the same study particle mixtures, it is interesting to note that the mixture yields a better effect than the results for individual particle sizes. It can be observed from Fig. 6c that the best result is achieved for a 10 nm and 40 nm diparticle mixture. The particle loss effect becomes dominant above the 30 nm membrane pore size. However, it can be observed from the same Figure that the 40 nm and 70 nm diparticle mixture performs better than the other particle mixtures for the 80 nm membrane pore size due to low particle loss and agglomeration effects. With high concentration, i.e. 180 ppm, almost all the diparticle mixtures give close to 100 % COD removal for most membrane pore sizes whereas the triparticle mixture has shows good results only up to the 30 nm membrane pore size and then shows a notable decrease due to the particle loss effects.
To further elucidate the mineralization potential of NANF, toxicity studies with resazurin were conducted for the four combinations for COD removal: the effect of low and high concentrations i.e. 60 ppm and 180 ppm were studied for both the individual particle study (Fig. 7a & b) and also for the particle mixture study (Fig. 7c & d). The results of this study are similar to those observed for the COD reductions and hence the results are not separately presented.
Thus it can be inferred that the small particles perform well even at low concentration for small membrane pore size while the large particles perform well only at high particle concentration for all membrane pore sizes. The diparticle mixture is most effective in obtaining good filtration results at low particle concentrations for all the membrane pore sizes.