Heavy metal pollution and the role of inorganic nanomaterials in environmental remediation

Contamination of water and soil with toxic heavy metals is a major threat to human health. Although extensive work has been performed on reporting heavy metal pollutions globally, there are limited review articles on addressing this pernicious phenomenon. This paper reviews inorganic nanoparticles and provides a framework for their qualities required as good nanoadsorbents for efficient removal of heavy metals from water. Different inorganic nanoparticles including metals, metal oxides and metal sulfides nanoparticles have been applied as nanoadsorbents to successfully treat water with high contaminations of heavy metals at concentrations greater than 100 mg l−1, achieving high adsorption capacities up to 3449 mg g−1. It has been identified that the synthesis method, selectivity, stability, regeneration and reusability, and adsorbent separation from solution are critical parameters in deciding on the quality of inorganic nanoadsorbents. Surface functionalized nanoadsorbents were found to possess high selectivity and capacity for heavy metals removal from water even at a very low adsorbent dosage of less than 2 g l−1, which makes them better than conventional adsorbents in environmental remediation.


Introduction
Pollution of water and soil is a global concern. Heavy metals such as mercury (Hg), lead (Pb), cadmium (Cd) and arsenic (As) are highly poisonous and their pollution even at low concentrations in water and soil poses a serious threat to the health of humans and other terrestrial animals. Mercury has deleterious effects on the human kidney, brain and central nervous system, whereas adsorbents possess (i) high surface area to volume ratio, (ii) surface functional groups tailored for a specific application, (iii) short diffusion route or absence of internal diffusion resistance, (iv) high porosity, (v) enhanced structural properties, and (vi) catalytic properties [8,72,73,[77][78][79]. These properties make nanoadsorbents highly selective, and with the capacity to achieve efficient adsorption of a variety of water pollutants including heavy metals. Examples of useful nanoadsorbents are zerovalent iron (n-ZVI), iron oxide (Fe 3 O 4 ), Fe 3 O 4 @SiO 2 -SH, iron sulfide (FeS), amorphous silica (SiO 2 ) and aluminium-silicate-mixed oxides nanomaterials. Interestingly, blends of inorganic nanomaterials with large surface area metal-organic frameworks (MOFs) and covalent organic frameworks (COFs) have shown high adsorption capacity for heavy metals and radionuclides [80,81].
This review covers several key areas including the different types of inorganic nanoparticles useful as adsorbent materials and the potential qualities of nanoparticles required for efficient removal of heavy metals from solution. This knowledge is essential for the design of efficient nanoadsorbents with capacities to remove heavy metals at levels prevailing in the environment and subsequently providing improvement to water treatment strategies, especially in mining areas.

Inorganic nanomaterials used as adsorbents
Nanomaterials are nanoscale materials with improved performance relative to the bulk material, often called emergent properties. Nanoscale materials due to the size effect emanating from the quantum confinement can be used as efficient adsorbents for environmental remediation processes. Nanomaterials span a variety of materials including inorganic, carbon-based, polymer, nanocomposites and biomaterials. In environmental remediation, inorganic nanomaterials have been widely applied [82]. Table 1 shows the characteristics and adsorption capacities of some inorganic nanoadsorbents for removal of Hg, Pb, Cd and As ions from solution.

Metal nanoparticles
Metal nanoparticles employed for the removal of heavy metals include Au, Ag and nano-zero-valent iron (n-ZVI) nanoparticles [7,104,105]. To improve the selectivity of some common adsorbents such as silica, alumina and activated carbon, Au and Ag metal nanoparticles have been used together with these materials to form composites or core-shell nanostructured materials [9,105,106]. Solis et al. [9] compared the adsorption capacities of a 3.19 nm size Au metal nanoparticle coated on silica and sand and found that the former showed a high affinity towards adsorption of Hg from water with K D (partition coefficient, which is a measure of sorbent affinity towards adsorbate) value of 9.96 l g −1 and adsorption efficiency of 96%. This confirmed the successful deposition of 6.9 mg of Au nanoparticles on a gram of silica compared with the sand which had only 1.5 mg deposition. The high selectivity was possible because Au forms a stable amalgam with Hg in the form of Au 3 Hg. The Au 3 Hg alloy is thermodynamically favourable at room temperature, and thus, the mercury adsorbed was directly proportional to the Au content in the silica [104]. Other metals like Ag, Al and Cu also form amalgams with Hg [6,7,104]. The adsorption capacities of Au and Ag metal nanoparticles can be less than or equal to 800 mg g −1 which is comparable to metal oxides but may be relatively expensive [9,105].

Nano-zero-valent iron
Nano-zero-valent iron (n-ZVI) is a very promising metal nanoparticle that is gaining much interest as nanoadsorbents for a wide range of water pollutants including heavy metals [7]. The n-ZVI consists of an inner zero-valent iron Fe 0 core and an outer mixed iron oxides shell layer [7]. The core acts on the source of contaminants through electrostatic interaction and surface complexation [7]. The n-ZVI with a surface area of 26.3 m 2 g −1 and a wide range of particle sizes between 20 and 200 nm ensured the adsorption of Cd 2+ ions from an aqueous solution up to the tune of 769.2 mg g −1 [94]. Making a similar composite with activated carbon led to a low adsorption capacity of 142.8 mg g −1 for Cd 2+ [107]. Compositing 10 nm n-ZVI particle with graphene also ensured the removal of Pb 2+ from aqueous solution with an adsorption capacity of 585 mg g −1 [108]. In addition to the adsorption of Pb 2+ , the n-ZVI reduces Pb 2+ to Pb 0 , and this ensures the easy magnetic separation from solution [108]. However, there are possible handling challenges with n-ZVI, since it is highly reactive and may be toxic [109].
Amorphous silica is non-toxic, has a high specific surface area and regular pore structures, making it a very promising adsorbent material [73]. Aminopropyl-functionalized silica nanoparticles with an average size of 7 nm possessed a large BET surface area of 390 ± 40 m 2 g −1 and could adsorb 0.44 (49.46) and 0.195 (40.40) mmol g −1 (mg g −1 ) of Cd 2+ and Pb 2+ , respectively [90]. Arce et al. [90] confirmed using DFT quantum computational calculations the formation of Pb-N and Pb-C bond distances which strongly agree with the extended X-ray absorption fine structure data. Alumina with a particle size of approximately, 6-13 nm adsorbed Pb 2+ at 47.08 mg g −1 capacity [124]. Mixed oxides of silica and alumina nanoparticles functionalized with L-cysteine methyl ester dendrimer demonstrated an incredible adsorption capacity of 3232 mg g −1 for Hg 2+ [71]. Figure 1 shows the morphology and performance of the L-cysteine methyl ester dendrimer-coated mixed oxides of silica and alumina nanoparticles. Alumina-silicate-mixed oxides are resistant to oxidation, chemically and thermally stable and have high conductivity, low dielectric constant, creep resistant and low thermal expansion [71].
Nanoparticles of CuO and ZnO also have high adsorption capacities for Hg 2+ . Fakhri [85] and Sheela et al. [126] reported CuO and ZnO adsorption capacities of approximately 825.21 and 714 mg g −1 , respectively, for Hg 2+ . The high adsorption of CuO may be that Cu forms a stable amalgam with Hg 2+ [104]. SnO nanoparticles are conducting, transparent and gas-sensitive, and thus, are gaining much interest as nanoadsorbents and as catalysts. SnO with a surface area of 24.48 m 2 g −1 as measured by BET isotherms demonstrated adsorption of Pb 2+ and Cd 2+ at 1265.8 and 1275.5 mg g −1 , respectively [93]. Flowerlike MgO nanostructures also displayed excellent adsorption of Pb 2+ and Cd 2+ . Figure 2 shows the TEM images and adsorption isotherms of the flowerlike MgO nanostructures. With a BET surface area of 72 m 2 g −1 , MgO was able to remove Pb 2+ and Cd 2+ ions from 100 mg l −1 metal ions solution at adsorption capacities of 1980 and 1500 mg g −1 , respectively [92]. The adsorption mechanism exhibited by MgO is ion exchange, leaving some level of Mg 2+ in solution. However, Mg 2+ is not toxic and the recommended level in drinking water is 450 mg l −1 according to WHO [92]. We note that there appear to be very few reports on the use of CuO, ZnO, SnO and MgO nanoparticles for heavy metal sequestration. Though they possess impressive heavy metals adsorption capacities, their removal from the solution will still be a challenge. To impart magnetization to these materials, in order to ensure simplicity of removal from solution using an royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 8: 201485 external magnet, doping with Fe and Mn could potentially be a strategy to impart these magnetic properties to the host material.
MSðsÞ þ A 2þ ðaqÞ ! ASðsÞ þ M 2þ ðaqÞ, ð2:1Þ MSðsÞ þ xA 2þ ðaqÞ ! ½M ð1ÀxÞ ,A x Sð sÞ þ xM 2þ ðaqÞ, ðx , 1Þ ð 2:2Þ and MSðsÞ þ A 2þ ðaqÞ ! MS À À À À À A 2þ : ð2:3Þ FeS naturally scavenges for Hg 2+ and forms stable β-HgS (metacinnabar, with solubility product, K sp , 4 × 10 − 54 ) even in the presence of organic matter [86]. FeS also removes divalent metal ions such as Mn 2+ , Ca 2+ , Mg 2+ , Ni 2+ and Cd 2+ . Additionally, FeS is a reductant and acts as an electron donor, making it possible to reduce and detoxify chlorinated organic compounds and inorganic oxyanions of Cr(VI), Se(VI) and As(VI) [129]. Remarkably, carboxymethyl cellulose (CMC)-stabilized FeS offered an Hg 2+ sorption capacity of 3449 mg g −1 , and this was attributed to the dual-mode sorption mechanism where CMC-FeS adsorbs Hg 2+ through concurrent precipitation [86]. Similarly, a large Pb 2+ remediation capacity of 2950 mg g −1 has been demonstrated by Pala & Brock [128] with zinc sulfide (ZnS) gel, where the mechanism is mainly more of ion exchange than adsorption. ZnS has been successfully used to sequentially remove Hg 2+ , Cu 2+ , Pb 2+ and Cd 2+ from simulated contaminated water containing 5-678 mg l −1 of heavy metal ions at removal efficiencies of 99.9%, 99.9%, 90.8% and 66.3%, respectively [77]. The ZnS showed higher selectivity towards Hg 2+ and Cu 2+ than Pb 2+ and Cd 2+ , and this Fang et al. [77] attributed to the differences in the solubility product (K sp ) of the metal sulfides. The lower the K sp , the higher the stability of the sulfide, and hence, the higher the precipitation and adsorption. The selectivity of the sulfides may also be attributed to the hard-soft acids and bases (HSAB) theory pioneered by Pearson (vide infra), where the sulfur which is a soft base will have a higher affinity for Hg 2+ which is a soft acid. However, adequate research will be required to clearly establish the mechanism of metal sulfides selective adsorption.
Core-shell nanostructured materials, such as Fe 3 O 4 -ZnS combines the magnetic properties of the iron oxide and the better affinity of ZnS, making sure that adsorbent material is easily separable and also selective. The magnetic Fe 3 O 4 -ZnS nanoparticles gave an adsorption capacity of 129.9 mg g −1 for removal of Hg 2+ from water which is better compared with non-magnetic adsorbents [130].
The different nanomaterials discussed above for heavy metals sequestration have unique advantages and disadvantages (table 2). Thus, a search for new inorganic nanoadsorbents with efficient and robust properties or qualities is desired.

Qualities of nanoadsorbents
The qualities of nanoparticles that make them suitable as nanoadsorbents for sequestration of heavy metals include (i) simple and cheap synthesis pathway, (ii) selectivity and affinity towards the target pollutant, (iii) thermal and chemical stability in solution, (iv) ease of removal from solution after adsorption, (v) it should be easily regenerated and re-used severally, and (vi) should have the ability to perform other functions apart from adsorption.

Synthesis pathway
The synthetic pathways generally reported for nanoadsorbent synthesis usually include (i) reduction, (ii) co-precipitation, (iii) sol-gel, and (iv) hot-injection methods. Figure 3 shows the different synthesis pathways for nanoadsorbent synthesis.

Reduction
The reduction route is often employed for the synthesis of metal nanoparticles such as Au, Ag and n-ZVI. Ojea-Jiménez et al. [104] synthesized Au nanoparticles (with size 8.9 ± 1.6 nm) by reducing Au 3+ with sodium citrate (Na 3 C 6 H 5 O 7 ) and was able to achieve 100% of Hg removal from water; however, the initial concentration of Hg was only 0.16 ppm. Similarly, Lisha and Pradeep [132] synthesized Au nanoparticles by using Na 3 C 6 H 5 O 7 to reduce Au 3+ to Au 0 nanoparticles and further supported it with alumina, resulting in Hg adsorption capacity of 4.065 g g −1 . Sodium borohydride (NaBH 4 ) has also been used to reduce Ag + to Ag 0 nanoparticles for adsorption of Hg [105]. Using NaBH 4 , Yan et al. [7] royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 8: 201485 synthesized n-ZVI nanoparticles, which were able to adsorb 98% of Hg from 40 mg l −1 Hg solution within 2 min. Sodium citrate and NaBH 4 are efficient reducing agents commonly used in nanoparticles synthesis. While sodium citrate is food grade and biologically compatible, NaBH 4 introduces boron into the water which could raise public health concerns.

Co-precipitation
In co-precipitation synthesis, the solutes are usually soluble at a particular condition but precipitate out of solution when supersaturation is reached. Co-precipitation is simple and rapid, provides easy control of the particle size and composition, provides the possibility of nanoparticle surface modification, is energyefficient, and does not require organic solvents. However, co-precipitation could yield impure nanoparticles with irregular and less homogenized nanoparticle sizes; it is time-consuming and has poor reproducibility [133]. Fe 3 O 4 has been synthesized widely using the co-precipitation technique, largely due to its simplicity and easiness of modification of the nanoparticle surface. Mehdinia et al. [2] synthesized Fe 3 O 4 nanoparticles using co-precipitation, supported it with silica and functionalized the surface with dithiocarbamate, which resulted in approximately 99% adsorption of Hg from water.

Sol-gel synthesis
The sol-gel method employs solvents and chemical reagents that after hydrolysis and condensation, the mixture of solvents turns into a gel. Arshadi et al. [71] synthesized SiO 2 -Al 2 O 3 -mixed oxide nanoparticles using the sol-gel method. Typically, the chemical reagents: aluminium tri-sec-butylate and tetraethyl orthosilicate were dissolved in n-butanol and the resultant solution warmed at 60°C. Acetylacetone (H-acac) was added slowly to the solvent mixture after cooling to room temperature to obtain a clear solution. This clear solution was hydrolysed by adding an alkoxide solution and left overnight, resulting in the formation of a transparent gel. The gel was dried to remove the solvents and finally calcined at 500°C to remove the other organic components to obtain the nanoparticles. This nanomaterial was reported to have removed approximately 99.9% of Hg from water with a very high initial Hg concentration of 500 mg l −1 . Fakhri [85] also using the sol-gel technique, synthesized CuO nanoparticles which had an Hg adsorption capacity of approximately 825.21 mg g −1 . The nanomaterials obtained from the sol-gel method are usually of high purity and porous. Though the sol-gel method gives control over porosity, particle size, chemical composition and dopant incorporation, compared with the other methods, it usually requires longer reaction time and involves organic solvents and higher temperatures to remove the organic components from the gel [133].

Hot-Injection
The hot-injection method involves the injection of a solution of reagents into hot solvents, which gives control over the nucleation and growth of the nanoparticles. It is usually employed for the synthesis of metal chalcogenide nanocrystals. ZnS supported on Al 2 O 3 was synthesized using the hot-injection method by Fang et al. [77] for the sequential removal of Hg 2+ , Cu 2+ , Pb 2+ , Cd 2+ and Zn 2+ in water. The ZnS-Al 2 O 3 was prepared by mixing α-Al 2 O 3 and ZnCl 2 in ethylene glycol in a three-necked flask fitted with a condenser, and maintaining the temperature at 180°C, 1-butylamine solution of thiourea was injected into it continuously, and the resulting mixture aged for 3 h. Although the hot-injection method is efficient and ensures that good quality monodispersed nanoparticles are obtained, it requires a complex set-up and organic solvents, and the resulting nanoparticles' surfaces are already occupied by capping groups or solvents, which makes it unattractive as nanoadsorbents. Table 3 shows the advantages and disadvantages of the different synthetic methods that have been used for nanoadsorbents syntheses. It must be noted that very few synthetic methods have been explored for the synthesis of nanoadsorbents, and admittedly, this research gap makes it challenging in deciding on suitable synthetic approaches for nanoadsorbents for heavy metals remediation.

Selectivity and affinity
Activated charcoal is a widely known efficient adsorbent. This is because it is cheap, available and has a high capacity for the removal of a wide range of pollutants. However, activated charcoal is a nonselective adsorbent and has no special affinity towards any specific pollutant. Affinity facilitates the rate of adsorption [9]. This makes nanoadsorbents superior to activated charcoal and other nonspecific adsorbents because they could be tailored to possess special affinity and selectivity towards a particular pollutant. Figure 4 shows the chemical structures of compounds used for the functionalization of the nanoparticle surface for improving selectivity towards heavy metals in solution. To improve the preferential adsorption of Hg 2+ onto SiO 2 -Al 2 O 3 -mixed oxide nanoparticles, Arshadi et al. [71] covalently immobilized L-cysteine methyl ester dendrimer onto the surface of the nanoparticle and the selectivity was evaluated in the presence of Pd 2+ , Pb 2+ , Cd 2+ , Zn 2+ , Cu 2+ , Co 2+ and Mn 2+ . At an initial concentration of 250 mg l −1 of metal ions, the functionalized nanoparticles exhibited high selectivity for adsorption of Hg 2+ of about 96.2%, and low adsorption for the interfering cations at adsorption capacities of 10, 3.7, 1.13, 0.76, 5.81, 1.84 and 3.02%, respectively. However, when the metal cations concentration was increased to 1000 mg l −1 , the adsorption of Hg 2+ decreased to 90%. The selective adsorption was attributed to the -SH and -NH functional groups of the L-cysteine methyl ester which preferentially donated electrons to the Hg 2+ resulting in complexation [71].
royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 8: 201485 Zhang et al. [121] demonstrated that thiol-functionalized Fe 3 O 4 @SiO 2 could selectively remove Hg 2+ at an adsorption capacity of 110 mg g −1 even in the presence of K + , Na + and Ca 2+ cations which naturally abound in water at high concentrations. While the hydroxyl groups on the surface of the Fe 3 O 4 @SiO 2 nanoparticle are hard Lewis bases and, therefore, preferred bonding with hard Lewis acids such as alkaline metals and alkaline earth metals, the thiol group being soft Lewis base exhibited a strong preference for bonding with Hg 2+ which is a soft Lewis acid. Mehdinia et al. [2] compared thiol and amine-functionalized magnetic mesoporous silica nanoparticles and found that the thiol functional group was more selective towards Hg 2+ removal than the amine. Mercury adsorption capacity of 538.9 mg g −1 was achieved with the thiol-functionalized nanoparticles, and this was attributed to the thiol functional group. In the presence of competitive cations, the thiol-functionalized nanoparticles were selective to the cations in the order Hg 2+ > Pb 2+ > Cd 2+ > Zn 2+ . This trend was also confirmed by Košak et al. [73] when they found mercaptopropyl-functionalized SiO 2 nanoparticles to adsorb 99.9% Hg 2+ , 55.9% Pb 2+ , 50.2% Cd 2+ and 4% Zn 2+ . Unfortunately, the selectivity of the amine-functionalized nanoparticles was not tested in the presence of the competing cations.
The binding interactions between the carboxylic acid functional groups in free molecules of 2-mercato-4-methyl-5-thiazoleacetic acid (MCT), monomercaptosuccinic acid (MMSA), o-thiosalicylic acid (o-TSA) and p-thiosalicylic acid ( p-TSA), and Hg 2+ , Pb 2+ and Cd 2+ cations were explored by Hamid et al. [74]. These molecules also contain thiol functional group (figure 4). They realized that the COOH in MCT binds completely (100%) with the Pb 2+ and Cd 2+ , and only 20% with Hg 2+ ; the COOH in MMSA binds 97% with Pb 2+ , 91% with Cd 2+ and had no interaction with Hg 2+ ; the COOH  2+ . It appeared that the strength of the interaction of COOH in any given molecule follows the order Pb > Cd > Hg. However, when silica nanoparticles were functionalized with these molecules through amide linkage using the COOH groups, the selectivity for Hg adsorption was improved. This was because the COOH group which is selective towards Pb 2+ and Cd 2+ was blocked or used in the amidation reaction. The projected SH groups were said to be responsible for Hg 2+ selective adsorption. Unfortunately, the adsorption of the Hg 2+ was found to be affected by the steric hindrance of the capping molecules attached to the nanoparticle surface [74]. The fundamental principle that may explain the affinity and selectivity of functionalized nanoadsorbents is Pearson's theory of hard and soft acids and bases-HSAB [73,121]. HSAB theory states that a soft acid would prefer to coordinate and form stronger bonds and more stable complexes royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 8: 201485 with soft bases, whereas a hard acid would prefer to coordinate and form stronger bonds and more stable complexes with hard bases [135]. Hard acids are relatively small metal ions that are highly electronegative and have lower polarizabilities, whereas soft acids are relatively large metal ions that are less electronegative and have high polarizabilities [135]. Hg 2+ and Cd 2+ are classified as soft acids and usually form stable complexes with soft bases such as SH [135]. COOH and RNH 2 are hard bases and strongly bind with hard acids such as Fe 3+ , Sn 2+ , Cr 3+ and As 3+ [135]. Pb 2+ , Ni 2+ , Fe 2+ , Cu 2+ and Zn 2+ are classified as borderline acids and may prefer to bind with hard or soft bases [135]. Figure 5 shows the binding preferences of COOH and SH groups towards metal ions based on the HSAB theory. The selective adsorptivity of nanoadsorbents towards Hg 2+ can be enhanced by functionalizing the surface of the nanoparticle with thiol-rich molecules, whereas attaching carboxylic containing molecules will improve adsorption of Pb 2+ . However, bulky molecules may pose steric restrictions to the approaching cation. The process of nanoparticle surface functionalization can be complicated. Surface functionalization may lead to the reduction of the active surface area of the nanoadsorbents (electronic supplementary material, table S1).
Mehdinia et al. [2] observed the decrease of silica nanoparticle surface area by approximately 31% when it was functionalized with thiol groups. However, HA did not reduce the surface area of Fe 3 O 4 [83]. It must be stated that the HSAB classification was made based on the equilibrium constants of an acid-base adduct or complex molecule in solution [135]. Fang et al. [77] confirms HSAB by attributing the removal efficiency of heavy metals from solution to the solubility products (K sp ) of their sulfides, that is, the lower the K sp of the sulfide of the heavy metal, the higher its removal efficiency. However, the effect of K sp on selective adsorption has not been widely studied.

Regeneration, reusability and stability
The pollutants chemisorb on the surface of the nanoadsorbent, which implies that chemical bonds are formed between the nanoadsorbent and the pollutant. A suitable nanoadsorbent should not have a complicated desorption process. The adsorbed pollutants should be easily desorbed into the solution and the adsorbent re-used. Adsorbents with high recycling or regeneration potential are of greatest interest. Acids, bases, thermal treatment and amalgamation have been used in heavy metals desorption processes. Hg 2+ was desorbed from a dithio-functionalized magnetic silica nanoparticle by treating with a mixture of 1 M HNO 3 and 2% thiourea, and the recovery efficiency of the Hg 2+ was greater than 98% after three concurrent uses. The thiourea acted as a chelating agent which complexed with the Hg 2+ to detach it from the silica nanoparticle surface [2]. Iodine ions also possess royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 8: 201485 a strong affinity for Hg 2+ ions, making potassium iodide a suitable recovery agent for detaching Hg 2+ from adsorbents [134]. The ability of a nanoadsorbent to withstand the desorption processes during regeneration or its resilience in different solvent media determines its stability. Hg 2+ adsorbed L-cysteine methyl ester dendrimer-capped mixed oxide nanoadsorbent was regenerated using sequential solutions of EDTA, 0.15 M HCl and water, dried at 80°C and re-used more than eight times to achieve adsorption capacity of greater than 97.5% with only a loss of 7% of the dendrimer [71]. Fe 3 O 4 nanoparticles are susceptible to air oxidation and tend to aggregate in aqueous solutions [83,111,121]. Coating bare The rate of adsorption of heavy metal ions is dependent on the pH of the solution. At very low pH values, the high concentration of H + ions compete with the metal ions for available active sites of the adsorbent material, whereas at high pH values, the OH − ions tend to precipitate the metal ions, preventing their adsorption. The maximum adsorption of Hg 2+ has been achieved within pH values of 5-7.5 ( figure 6) [8,75,105,134]. Chitosan-bound Fe 3 O 4 nanoparticles were able to remove 92.4% of Hg 2+ ions even at a low pH of 3 [72]. Similarly, 2-mercaptobenzothiazole-coated Fe 3 O 4 nanoparticles Hg 2+ removal efficiency was not significantly affected within a wide pH range of 2.5-11 [78]. The chitosan and 2-mercaptobenzothiazole coatings provided adequate stability to the Fe 3 O 4 nanoparticles, thus, making them robust nanoadsorbents. Coating of nanoadsorbents seems a better option for improving the stability in solution and during desorption processes; however, the synthetic process of functionalization of nanoparticle surface remains a major challenge.
In addition, the information on the point of zero charge (pH pzc ) of nanoadsorbents is relevant in determining the pH at which the net surface charge is zero, which also ensures the robustness of the adsorbents in different media. The pH pzc indicates the mechanism of adsorption (whether cationic or anionic) and helps in selecting the appropriate pH at which the adsorption process should be carried out in order to minimize precipitation. However, there is little information on pH pzc   royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 8: 201485 L-cysteine had pH pzc values of 4.3-6.1 and 5.7, respectively [89,97]. Capping SiO 2 nanoparticles with APTES ((3-aminopropyl) triethoxysilane) significantly increased the pH pzc from 2.5 to 9.15, indicating improved robustness of the nanoadsorbent in a highly basic environment [88]. However, some material such as ZnO is preferred as nanoadsorbent because it possesses a wide pH pzc value of 8-9 [126].

Removal of adsorbent from solution
The current trend of nanoadsorbents possesses some level of magnetization which allows them to be easily removed from the solution by applying an external magnet [72]. This makes magnetic nanoadsorbents superior to other non-magnetic adsorbents such as activated charcoal. Fe 3 O 4 nanoparticles maintained magnetization at saturation of 79.6 emu g −1 even after being coated with HA and only require a low magnetic field gradient for separation from the solution [83]. Nonmagnetic inorganic nanoadsorbents can be doped with Fe, Mn and Co ions to impact some level of magnetization; however, very few works have been done using doped nanoparticles as adsorbents.

The ability of adsorbent to perform other functions
Apart from the adsorption process, some nanoadsorbents have the potential to perform other functions such as amalgamation. Gold forms amalgam with mercury; however, it requires that mercury is in the reduced form Hg 0 before the amalgamation could occur. Not all nanoadsorbents form stable amalgams with the target pollutant. In most instances, Hg 2+ ions are the predominant form of mercury pollution in water. Ojea-Jiménez et al. [104] developed an Au nanoparticle coated with citrate ions which together with its adsorption capabilities was able to reduce Hg 2+ to Hg 0 which resulted in the formation of amalgam with the Au nanoparticle (Au 3 Hg). Thus, the use of very strong and toxicreducing agents such as NaBH 4 was avoided. This observation makes nanoparticles of metals such as silver, copper, tin and aluminium, which form a stable amalgam with Hg attractive and promising nanoadsorbents for sequestration of mercury pollution [6,104]. Iron is a very useful magnetic adsorbent material but does not form a stable amalgam with mercury. However, a more stable amalgam can be obtained by doping the iron nanoparticles such as nZVI with metals that amalgamate with mercury, and this will ensure the effective formation of a more efficient and durable nanoadsorbent for Hg 2+ removal from water [7]. In addition, magnetite coated with Lysinibacillus sp. JLT12 was reported to efficiently reduce Cr(VI) to Cr(III) which is a less toxic and less soluble form of Cr [134]. The kinetics and spectroscopic data showed that the Lysinibacillus sp. JLT12 coating was able to remove the passive lepidocrocite and goethite layers of the magnetite to efficiently facilitate the Cr(VI) reduction process [134]. Also, Zhong et al. [80] through kinetics and spectroscopic studies attributed the formation of U-O and Mn-O-U bonds to the oxygencontaining groups in δ-MnO 2 @COF nanocomposite, which resulted in ultra-fast removal of UO 2þ 2 radionuclide from water. These extra abilities of inorganic nanoadsorbents demonstrate that efficient adsorbents can be designed and hold promise for efficient environmental remediation processes.

Environmental application
Hofacker et al. [135] in soil microcosm experiments demonstrated the potential application of inorganic nanoparticles in environmental remediation processes. The researchers set out to study mercury mobilization in flooded soil by incorporation into metallic copper and metal sulfide nanoparticles. The soil studied were topsoil samples obtained from the contaminated floodplain of the River Mulde in Germany, which were polluted with 0.91-1.26 mg kg −1 Hg and 160-279 mg kg −1 Cu. This level of pollution was 10 times the average content of European floodplain soils. Soil flooding experiments were carried out with air-dried soil (amended with lactate, which promotes activities of soil microorganisms including Fe and sulfate reducers) that was flooded with synthetic river water (containing 0.6 mM NaCl, 0.6 mM CaSO 4 and 0.3 mM Mg(NO 3 ) 2 ) and incubated at different periods and temperatures. Soil pore water samples were withdrawn, acidified, filtered and analysed for Hg, Cu, Fe, Mn and metal sulfides.
The study showed that the formation of Hg-Cu amalgam nanoparticles may be a common process induced by soil flooding under limited sulfate availability. However, during sulfate reduction, Hg was effectively incorporated into nanoparticulate metal sulfides. Figure 7 shows STEM-HAADF images royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 8: 201485 and EDX spectra of metal sulfide nanoparticles formed after 12 days and 33 days of flooding at 14°C and 5°C, respectively. The tendency for different metals to precipitate with a limited amount of sulfide during soil flooding depends on the thermodynamic stability of the respective metal sulfides (Hg > Ag > Cu > Cd ∼ Pb > Zn > Fe). Also, coprecipitated Fe or Zn may be exchanged for Hg, Ag, Cu, Cd or Pb due to the lower stability of Fe and Zn monosulfides. Thus, based on the extremely high thermodynamic stability of HgS, higher soil Hg/Cu ratios are expected to favour the formation of pure HgS over mixed metal sulfide formation. The study showed that Hg incorporation into metallic Cu and metal sulfide nanoparticles can cause substantial colloidal Hg mobilization into the pore water. When exposed to oxygen, metal sulfide nanoparticles were found to be stable against oxidation in oxygenated water for up to several weeks. The formation of HgS was suggested to probably affect the extent of Hg methylation and volatilization which are the most toxic and dangerous aspects of Hg pollution.
Although Hofacker et al.'s [135] work demonstrates how soil floods induce the formation of inorganic nanoparticles depending on the soil's mineral composition, it does not show in real time how they could be applied to environmental remediation. Their work, however, suggests potential removal of Hg from aqueous solution using metal Cu and metal (Fe and Zn) sulfide nanoparticles through the formation of Hg-Cu amalgam and highly thermodynamic stable HgS nanoparticles precipitates, which can be filtered out of solution.

Conclusion
It has been established that inorganic nanoadsorbent materials address the problem of environmental pollution. Heavy metal contamination levels of over 100 mg l −1 have been removed, attaining a very high adsorption capacity of less than or equal to 3449 mg g −1 with inorganic nanoadsorbents with less than 2 g l −1 adsorbent dosage. Although inorganic nanoadsorbents hold much promise in environmental remediation, there are very important bottlenecks that have to be addressed to harness their full potential. Some of the bottlenecks are: (i) Metal nanoparticles such as Ag and Au are generally expensive for adsorption purposes. Thus, cheaper materials or earth-abundant metals would be more useful. (ii) n-ZVI nanoparticles are very efficient but they are extremely reactive and may be unstable. More research efforts would be required to develop the core-shell counterparts to improve the stability.  royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 8: 201485 (iii) Iron oxides nanoparticles have been widely used because they are magnetic and facilitate easy separation from solution; however, their heavy metal adsorption capacities are lower compared with CuO, ZnO, SnO and MgO nanoparticles which are not magnetic. Thus, doping nanoparticles with magnetic materials becomes an attractive strategy to improve the separability property of nanoadsorbents. (iv) Metal sulfides nanoparticles have a higher affinity for heavy metals compared with metal oxides; however, they are less studied for remediation processes. (v) It is worth noting that the control of selectivity in heavy metal adsorption has been associated with Pearson's theory of hard and soft acids and bases, but unfortunately, there is very little information to clearly explain the selectivity concept. The key responsible principle has been associated with the complexation of the heavy metals by functional groups on the inorganic nanoadsorbents surfaces. Although ligands attached to surfaces of nanoparticles improve the selectivity of nanoadsorbents towards target heavy metal, the process of nanoparticle surface functionalization can be very challenging and requires considerable research effort to improve on the qualities of inorganic nanoadsorbents.
In the design of good quality inorganic adsorbent, the nanoparticle synthesis method, selectivity, stability, regeneration and reusability, and adsorbent separation from solution are critical concerns. It is intended in the future that stable nanoadsorbents that are selective to specific pollutants, easy to synthesize (in an environmentally friendly manner) and separated from the solution, and also can perform other important functions are developed.
Ethics. This statement is not relevant to this work. Data accessibility. Additional tables showing the BET surface area and point of zero charge of some inorganic nanoadsorbents have been provided in the electronic supplementary material.