Formation and remediation simulation of an in-situ reactive zone with nanoiron for a nitrobenzene-contaminated aquifer

As a type of raw material commonly used in modern agriculture, medicine, and chemistry, nitro-aromatics are accumulating substantially in the environment as industrial development continues (Dubey et al. 2014; Yan et al. 2017). Due to the existence of nitrofunctional groups, the degradation of nitro-aromatics is quite difficult and so they pose a higher toxicity risk to plants and humans than other monocyclic aromatic compounds (Kulkarni & Chaudhari 2007; Rezaei-Vahidian et al. 2017). In recent years, the groundwater pollution resulting from nitro-aromatics has drawn widespread public concern, and so a series of engineering technologies, such as microbial remediation (Kulkarni & Chaudhari 2007; Grube et al. 2008), chemical oxidation/reduction (Yuan et al. 2012; Guria et al. 2016), and permeable reactive barriers (Yuan et al. 2013; Yin et al. 2015), have been tried with the goal of pollution remediation. These methods, however, often need large construction, high costs, and a long period for construction, all of which greatly limit their application in the field (Liu et al. 1981; Higgins & Olson 2009).

The in-situ reactive zone (IRZ) has recently emerged as a new type of in-situ remediation technology. One or more IRZs may be created, in the pollution plume, by injecting chemical reactants or microorganisms underground to intercept, immobilize, or thoroughly degrade the target pollutants (Suthersan 2002). Hence, the key to this technique's success is that a reasonably appropriate reactant is chosen according to the pollutant's characteristics. As a reductive metallic material with a very small particle size, and thus a larger specific surface area, and a higher reactivity and reducibility than iron powder (Elliott & Zhang 2001), nanoscale zero-valent iron (NZVI) can effectively degrade various persistent organic pollutants in the environment, such as chlorinated hydrocarbons (Schlicker et al. 2000; Zhou et al. 2015), heavy metals (Ponder et al. 2000), and pesticides (Han et al. 2016), among others.

However, due to their size effect and high surface free energy, the iron nanoparticles tend to aggregate in an aqueous medium (Adeleye et al. 2013). Therefore, surface modification by some macromolecular materials to enhance the dispersion of nanoiron before its use is necessary. The excellent performance by rare or modified nanoiron in the degradation of nitro-aromatics has been confirmed in several recent studies. Dong et al. (2015) prepared emulsified nanoiron for nitrobenzene (NB) degradation and found that the nanoparticles could provide for significant removal of nitrobenzene through adsorption and reductive reaction, with aniline as the end-product. Similarly, another study used NZVI to remove parachloronitrobenzene, finding that the efficiency could reach 98.8% under the condition of pH 2 in 120 minutes (Dong et al. 2010). Lin et al. (2013) investigated the decomposition of n-nitrodimethylamine (NDMA) in water by using NZVI in the presence of aluminum and iron salts, which removed up to 87.3% of NDMA. Therefore, these studies confirm NZVI could be used as a high-efficiency reactant in the IRZ to remediate groundwater pollution. Generally, however, current study of NZVI tends to mainly focus on its reaction mechanism at the laboratory batch scale (Li et al. 2013), or its transport behavior through columns (Tiraferri & Sethi 2009). Hence, we still lack research on IRZ with NZVI, on the actual removal efficiency of nitro-aromatics in the aquifer, and on the permeability change of IRZ during the remediation process. These research avenues are perhaps more practical and applicable for site engineering and are worthy of study.

Given this context, a two-dimensional simulation experiment of in-situ remediation for nitrobenzene pollution was carried out in this study, which had three objectives: (1) to confirm the remediation ability of nitrobenzene by IRZ with NZVI in a simulated aquifer; (2) to study the formation and development of IRZ according to the degradation of nitrobenzene; and (3) to investigate whether the blockage of IRZ resulting from the reduction reaction will appear. Fulfilling all of these should provide a sound basis for the application of IRZ with NZVI for in-situ remediation.

NZVI particles were prepared by a chemical reduction method in aqueous solution (He & Zhao 2005). To reduce the particle aggregation and improve their dispersibility, sucrose was chosen as the disperser for preparing the modified NZVI (SM-NZVI). In our previous study, transmission electron microscopy (TEM) images showed that the particle size of prepared SM-NZVI was c.100–150 nm with a characteristic diffraction peak at 44.7 °C in the X-ray diffraction (XRD) pattern that represents the α–Fe of the body-centered cubic. The prepared NZVI particles were washed by anhydrous ethanol and deionized water to remove surface impurities, then diluted with water to a fixed concentration for use.

The media materials used for the simulated aquifer were collected from a sandpit in Changchun. The sand was dried in the shade and put through a series of graded sieves ranging in grain size from 0.5–1 mm and 0.1–0.25 mm. Nitrobenzene of analytical grade was purchased from Xinlu Chemical Technology Co., Ltd (Tianjin, China). Several key physical and chemical properties of the sand and the prepared NB-contaminated water are listed in Tables 1 and 2, respectively.

The sand box for the two-dimensional simulation was 130 cm in length, 30 cm in width, and 60 cm in height. There were 60 sample connections on the front of the box, which are expressed by circles in Figure 1. Water distribution boards were installed on the inlet side and outlet side to ensure a uniform media saturation by water. Piezometer tubes (PT) were connected through latex tubing to the sample outlets of the second row and fixed vertically onto the box. Five injection wells, each with a 2 cm internal diameter and all-over holes, were installed in the bottom of the box and c.2 cm away from the left water distribution board (wells were 5 cm apart).

To serve as the media of different simulated aquifers in two boxes, coarse sand (245 kg) and fine sand (256 kg) were filled into the box and both pressed layer-by-layer to a height of 42 cm, which corresponded to a volume weight of 1.62 and 1.69 g/cm³, respectively. To reduce the volatilization loss of NB, some clay was placed on the top of the sand to a depth of 2.5 cm to form an unsaturated zone. Afterwards, NB solution of fixed concentration was slowly introduced into the aquifer via peristaltic pumps until the NB concentration of the outflow was consistent with that of the inflow. Tap water replaced the NB solution as groundwater, which flowed at 0.23 m/d and 0.15 m/d in the coarse and fine sand, respectively; their corresponding permeability coefficients, as calculated through the inlet and outlet water level, were 56.1 m/d and 6.1 m/d. During the next 5 days, prepared 5 L SM-NZVI slurry of 10 g/L was injected into the simulated aquifer via injection wells five times at the same volume. Samples of simulated groundwater were taken regularly to determine the concentrations of NB and aniline (AN) as the end-products.

The changing NB concentration in coarse sand is shown in Figure 2. It is evident that the NB concentration significantly declined at the bottom of the injection wells once all the slurry had been injected, and the result indicates that a portion of NB was effectively degraded by SM-NZVI. The flowing groundwater has a limited attenuation effect on NB, which is likely negligible when considering NB hydrophobicity and the lower flow velocity of groundwater. In subsequent days, the NB concentration decreased rapidly in the area surrounding the injection wells that could be regarded as the IRZ of SM-NZVI. As shown in Figure 2, the reactive zone occurred in the 15–55 cm portion of the box on the 5th day, and it extended in the direction of the flow, thus bringing about an extensive reactive range of 20–100 cm by the 60th day. Furthermore, due to the gradual precipitation of SM-NZVI particles, this extension occurred intensively in the bottom of the zone, which also promoted the reduction of deep-seated NB. In addition, the upstream concentration was obviously higher than that of the downstream reactive area, indicating the dilution effect of inflow was less important as aforesaid, and the reduction of the zone with nanoiron was the principal cause of NB degradation.

The reactivity of nanoiron particles directly determines the effectiveness of the IRZ. It was found that the decrease in the NB concentration had slowed since the 40th day, and that the trailing concentration was steady at more than 100 mg/L. Considering the distribution of the IRZ, evidently the removal of nitrobenzene at the top, and end, of the box happened more slowly than in the bottom of the box. This disparity arose because the transport of NZVI particles has been proved to be quite limited, no greater than several metres, and accompanied by intense sedimentation that comes from aggregation and oxidation (Schrick et al. 2004).

Compared with the removal effect of NB in coarse sand, the results in fine sand were less efficient as shown in Figure 3. The reaction area was in the range of 15–60 cm on the 5th day, and the spreading rate of the zone was also relatively lower for the range of 20–80 cm by the 60th day. Many studies on NZVI transport in different media have demonstrated that the NZVI particles are adsorbed more strongly by media with a smaller size for a larger specific surface area, which naturally leads to their limited transport distances (Vasiliki & Constantinos 2012). Therefore, the NB in the reactive zone with a higher NZVI concentration was degraded substantially, but the NB concentration for the distance range of 0–20 cm and 60–120 cm always stayed above 160 mg/L.

Given the above analyses and results, it may be concluded that the media size of the aquifer has an important influence on the remediation of NB in the IRZ with NZVI; or rather, NB is better removed in the media with larger size due to the wider range of the IRZ for relatively better transport of NZVI particles. On the other hand, the higher concentration of NZVI in a narrow area will lead to severe agglomeration of the nanoparticles, as well as the generation and attachment of oxidation products to the iron surface – both outcomes can further decrease the reactivity of NZVI particles.

Prior studies have shown that nitrobenzene will be reduced to aniline, the end-product, by zero-valent iron (Huang et al. 2015). The spatio-temporal distributions of aniline in the aquifer boxes are shown in Figures 4 and 5. Clearly, after all the slurry was injected, less aniline was generated due to the limited transport of SM-NZVI in the first 5 days; hence, the reduced NB concentration away from the injection wells may be attributed to diluted slurry. As the reaction proceeded, the concentration of aniline in every area gradually increased until the reduction entered the trailing state. As seen in Figure 4, the average concentration of aniline in coarse sand was 24.74, 46.76, 84.89, and 97.02 mg/L on the 10th, 20th, 40th, and 60th day, which corresponded to a production rate – the ratio of an actual vs theoretical amount – of 46.3%, 38.2%, 59.8%, and 25.8%, respectively. Therefore, the area of the IRZ primarily expanded in the first 40 days, thus prolonging the contact time between NB and SM-NZVI, which promoted the transformation of NB into aniline.

The concentration changes of aniline in fine sand are shown in Figure 5. The average concentration of aniline was 10.87, 24.67, 42.77, and 49.40 mg/L, with a corresponding production rate of 29.2%, 30.3%, 31.6%, and 18.4% on the 10th, 20th, 40th, and 60th day, respectively. Over the 2-month period, the average production rate was 1.93 times in coarse than that in fine sand (40.1% and 20.8%, respectively). Hence, for each time period considered, the production rate in fine sand was distinctly lower than that in coarse sand, a result explained by the more effective contact of nanoparticles and the pollutant in the aquifer containing media of a larger size. Some reports about NB degradation by zero-valent iron have drawn a similar conclusion: intermediate products, such as nitrosobenzene and hydroxylaniline, would be generated for inadequate contact, and these pollutants remain more toxic than aniline for the infeasibility of biodegradation. Therefore, not only NB reduction but also aniline production is related to the final remediation effect. How to improve the transport ability and sustainable reactivity of NZVI is the key to applications of IRZ technology.

Based on the above results and their analysis, the formation and development of the IRZ with NZVI may be summed up as the following process: a large area of IRZ in the flow direction is formed due to the transport of injected NZVI, and a small area of IRZ upriver with a lower concentration of NZVI is formed by head or mechanical pressure from the injection process. The IRZ is conspicuously heterogeneous, which is because the NZVI particles will continuously deposit on the surface of the aquifer media during the migration process, as well as settle deeply into the aquifer via gravity. Therefore, the removal efficiency of the target pollutant around the injection wells and deeper in the aquifer will likely exceed that elsewhere, and the efficiency will decrease with the distance from the wells.

The water yield and level in the piezometer tubes were measured regularly, and the changing permeability coefficient was calculated according to the Darcy law. As shown by the curves in Figure 6, the initial water yield of the aquifer box was 18.8 and 11.7 mL/min for coarse and fine sand, respectively, but this quickly declined to 16.2 and 10.3 mL/min as the SM-NZVI slurry was injected. As the IRZ proceeded, the water yield tended to drop slowly, such that the final water yield stayed at c. 15and 9.2 mL/min, for coarse and fine sand respectively, representing a 20.2% and 12.0% decline when compared with the initial values. These results indicate that although the injection of NZVI slurry will reduce the water yield of the aquifer, there is little impact thereafter from the operation of the IRZ or the reduction reaction.

The hydraulic gradient is one of the most important parameters when measuring the permeability change of an aquifer. It was calculated by recording the change of water level in piezometer tubes, named by their column number plus ‘PT’. Take the results of NO.2, NO.4, and NO.8 PT, for example. As shown in Figure 7, the water level of NO.2 PT in both aquifers clearly increased during the first 10 days, indicating that the injection of the SM-NZVI slurry can significantly reduce the permeability of the aquifer around the injection wells. After that, the water level in the coarse sand continued to increase though it remained stable in fine sand apart from NO.2 PT.

The increased water level was bound to change the hydraulic gradient in the aquifer. In the first 20 days, the hydraulic gradient between NO.2 and NO.4 PT gradually increased due to the slurry injection and the degradation of NB by nanoiron, and the corresponding permeability coefficient was reduced, respectively, from 56.12 to 22.94 m/d in coarse sand and from 5.79 to 4.96 m/d in fine sand (Figure 8). After 20 days, the permeability of this area in coarse sand seemed little changed, whereas it continuously decreased to 1.82 m/d in fine sand; this difference suggests that the reduction reaction had a smaller impact on the permeability with the increase of media size. Furthermore, it can be seen that the hydraulic gradient between NO.4 and NO.8 PT was nearly constant (Figure 7), likely because there were fewer NZVI particles transported to the area – which is farther from the injection wells – and so the decrease in permeability during the first 10 days was mainly caused by a change in water yield.

In summary, the injection of NZVI slurry reduced the permeability of the aquifer, though the permeability coefficient always kept in the same order of magnitude; that is, the slurry would not cause obvious changes to the groundwater flow field. As the IRZ or the reaction proceeded, both NB degradation and NZVI oxidation were also not likely to reduce the permeability of the aquifer significantly, let alone block the aquifer.

To investigate the remediation efficiency of an in-situ reactive zone with NZVI for nitrobenzene, a two-dimensional simulated experiment was undertaken in saturated porous media, and the removal efficiency of nitrobenzene and the production rate of aniline were described in detail, as was the permeability change of the aquifer. Based on the properties of NZVI particles and the data maps, the formation and development of an IRZ was easily discerned. The research findings are summarized as follows:

• 1.

  The media of the aquifer has an important effect on the formation of an IRZ. The area of the IRZ will increase with the media size of the aquifer, which enhances the remediation effect. In coarse and fine sand, the IRZ with SM-NZVI entered the trailing period on the 50th and 40th day, respectively, and the corresponding trailing concentration of NB was 87.24 and 170.24 mg/L. After 60 days, the average production rate of aniline had reached 40.1% and 20.8%.

• 2.

  The injection of SM-NZVI reduced the water yield by 13.8% and 11.9% in coarse and find sand, respectively. As the reaction progressed, the water yield basically remained stable, but the hydraulic gradient of the aquifer continued to increase, particularly in the reaction zone.

• 3.

  There was no sign of a blockage in the aquifer, although the permeability declined, by 60.9% and 70.6%, with the injection of NZVI slurry (to 22.94 and 1.82 m/d, respectively).

• 4.

  The formation and remediation ability of the IRZ with NZVI is heterogeneous: the concentration and degradation ability of the IRZ will decrease with distance from the injection wells, and a better removal efficiency for pollutants at the bottom of the zone is predicted because of the precipitation there of the NZVI particles.

This study was funded by the Natural Science Foundation of China (No. 41602269), the Basal Research Fund of the Chinese Academy of Geological Sciences (No.YYWF201629), and the Natural Science Foundation of Hebei Province (No. D2017504004).