As(III) and Cr(VI) oxyanion removal from water by advanced oxidation/reduction processes—a review

Abstract

Water pollution by human activities is a global environmental problem that requires innovative solutions. Arsenic and chromium oxyanions are toxic compounds, introduced in the environment by both natural and anthropogenic activities. In this review, the speciation diagrams of arsenic and chromium oxyanions in aqueous solutions and the analytical methods used for their detection and quantification are presented. Current and potential treatment methods for As and Cr removal, such as adsorption, coagulation/flocculation, electrochemical, ion exchange, membrane separation, phyto- and bioremediation, biosorption, biofiltration, and oxidative/reductive processes, are presented with discussion of their advantages, drawbacks, and the main recent achievements. In the last years, advanced oxidation processes (AOPs) have been acquiring high relevance for the treatment of water contaminated with organic compounds. However, these processes are also able to deal with inorganic contaminants, mainly by changing metal/metalloid oxidation state, turning these compounds less toxic or soluble. An overview of advanced oxidation/reduction processes (AO/RPs) used for As and Cr removal was carried out, focusing mainly on H₂O₂/UVC, iron-based and heterogeneous photocatalytic processes. Some aspects related to AO/RP experimental conditions, comparison criteria, redox mechanisms, catalyst immobilization, and process intensification through implementation of innovative reactors designs are also discussed. Nevertheless, further research is needed to assess the effectiveness of those processes in order to improve some existing limitations. On the other hand, the validation of those treatment methods needs to be deepened, namely with the use of real wastewaters for their future full-scale application.

General aspects of arsenic and chromium

The concern about water quality and sanitation has increased during the last century. Since safe drinking water is scarce in several regions of the planet and fresh water is continuously polluted by countless contaminants, the water quality must be monitored and protected to reduce adverse health effects (Magu et al. 2016). Unlike most organic pollutants, which can be biodegradable, inorganic chemicals, including oxyanions, are continuously accumulated in the environment (Sarwar et al. 2017), posing a threat to human health due to the potential risk of entry into the food chain. The ecosystem can be contaminated with inorganic chemicals from both natural and anthropogenic activities. Industrial processes such as electroplating, metal smelting, and chemical manufacturing are examples of anthropogenic sources of inorganic chemicals in water (Chowdhury et al. 2016). In order to protect water quality, the World Health Organization (WHO) published the first guidelines for drinking water quality in 1958 and the last update was in 2011. Additionally, other regulatory agencies have published local guidelines for drinking water quality based on the treatment performance, analytical achievability, and risks related to human health. Table 1 lists the maximum levels allowed in drinking water for some oxyanions contaminants by the European Union, US, and Brazil legislation, as well as the values suggested by WHO. The potential health effects and the common sources of each contaminant are also presented.

Table 1 Maximum level allowed in drinking water for some oxyanions by the European Union, USA and Brazil legislation, values suggested by the WHO and potential health effects and common sources of each contaminant

In water contaminated with inorganic chemicals, the options for its remediation usually require an oxidative or reductive process to achieve the less soluble or less mobile form. Arsenic and chromium are examples of oxyanions that require initially an oxidative and a reductive pre-treatment step before their removal, respectively. The most toxic and mobile arsenic species is the trivalent one, being necessary a previous oxidation to remove it from water. However, chromium in its hexavalent form is more toxic and mobile than the trivalent one, being necessary a reductive process for water remediation. Additionally, arsenic and chromium are in the top of 20 substances that pose the most significant potential threat to human health according to the Agency for Toxic Substances and Disease Registry (ATSDR 2017). Therefore, the present article briefly reviews the arsenic and chromium speciation in aqueous solutions, the analytical methods used for their detection and quantification, and the treatment processes for their removal from water. Advanced oxidation/reduction processes are highlighted.

Arsenic and chromium speciation

Arsenic is a metalloid belonging to group 15 of the periodic table. The inorganic form is mostly found in natural waters, as trivalent (As(III)) or pentavalent (As(V)) oxyanions. On the other hand, the organic arsenic can enter in the aquatic environment by both industrial pollution or biological activity (Smedley and Kinniburgh 2002). Arsenic is disposed to mobilization at a broad range of pH, including near-neutral values normally found in groundwater and in oxidizing and reducing redox potential conditions (Sarkar and Paul 2016).

The pH and redox potential (E) of the solution are the most relevant parameters as concerns arsenic speciation. The species H₂AsO₄⁻ and HAsO₄²⁻ are typically found in positive E values (oxidative conditions) and pH range from 4 to 8. At extremely lower or higher pH values, the predominant species are H₃AsO₄ and AsO₄³⁻, respectively. However, in negative E values (reducing conditions), H₃AsO₃ predominates for acid and near-neutral pH values (pH values lower than 9.2) (Sarkar and Paul 2016). In solution, arsenate is present as H₃AsO₄, a triprotic acid, and deprotonated forms are H₂AsO₄⁻ (pKa₁ = 2.24), HAsO₄²⁻ (pKa₂ = 6.96), and AsO₄³⁻ (pKa₃ = 11.50) (Skoog et al. 2004), while arsenite occurs as H₃AsO₃ and its deprotonated forms are H₂AsO₃⁻ (pKa₁ = 9.2) and HAsO₃²⁻ (pKa₂ = 12.7) (Bundschuh et al. 2012; Smedley and Kinniburgh 2002). Figure 1 shows the distribution diagram of As(III) and As(V) species as a function of pH. It is worth mentioning that As(III)/As(V) ratio in groundwater is variable and depends on the aquifers oxidizing or reducing conditions. Furthermore, in contrast to other oxyanions in general, arsenic barely precipitates at near-neutral pH and its adsorption by clays or metal oxides is commonly not efficient (Sarkar and Paul 2016).

Fig. 1

Arsenic Pourbaix diagram ([As] = 0.013 mM, 25 °C, ionic strength 0.06 mM)

Chromium is a metal belonging to group 6 in periodic table. It can present oxidation numbers from 0 to + 6; nonetheless, only trivalent (Cr(III)) and hexavalent chromium (Cr(VI)) are stable in environmental conditions. Trivalent chromium is the most stable form in natural water conditions; however, in specific conditions, Cr(VI) can naturally occur too. The Aromas Red Sands aquifer, California, USA, is an example, where manganese oxides promote the oxidation of mineral deposits of trivalent chromium, making hexavalent chromium available in the aquifer (Gonzalez et al. 2005). While Cr(III) adsorbs on soil particles, showing low mobility and bioavailability, Cr(VI), on the other hand, is a strong oxidant, with high mobility due to its clay repulsion. In aqueous media, Cr(VI) can exist in several anionic species, mainly as chromate in deprotonated forms HCrO₄⁻ (pKa₁ = 0.74) and CrO₄²⁻ (pKa₂ = 6.49), which are highly soluble in a wide pH range (Choppala et al. 2013). On the other hand, Cr(III) forms hydroxides with 1 to 4 hydroxyls. It starts to become insoluble at pH 6, as Cr(OH)₃ species is formed. However, for pH values higher than 12, it becomes soluble again. In oxygenated superficial waters, beyond pH and O₂ concentration, other parameters such as the presence of reducers, oxidants, and/or complexing agents influence the Cr(III)/Cr(VI) ratio (Kotaś and Stasicka 2000). Figure 2 shows the distribution diagram of trivalent and hexavalent chromium species as a function of pH.

Fig. 2

Chromium Pourbaix diagram ([Cr] = 0.02 mM, 25 °C, ionic strength 0.35 mM)

Analytical techniques

Instrumental analytical methods for arsenic and chromium quantification include flame atomic absorption spectroscopy (FAAS), graphite furnace atomic absorption spectrometry (GFAAS), electrothermal atomic absorption spectrometry (ETAAS), hydride generation followed by atomic absorption spectroscopy (HG-AAS), ultraviolet-visible spectroscopy (UV-vis), inductively coupled plasma–mass spectrometry (ICP-MS), neutron activation analysis (NAA), and anodic stripping voltammetry (ASV). These techniques present limits of detection (LOD) in the order of μg L⁻¹ and ng L⁻¹, which are coherent with the maximum contaminant levels allowed in drinking water. Since the chemical and toxicological properties change with the oxidation states, the speciation in environmental samples is very important. For aqueous matrices, this could be simpler than for biological fluids and other complex samples (Niedzielski and Siepak 2003). However, usually, a combination of chromatographic separation and preconcentration techniques is necessary for arsenic and chromium speciation and detection (Anawar 2012). A brief list of analytical techniques for arsenic and chromium determination is presented in Table 2.

Table 2 Analytical techniques for arsenic and chromium determination

Arsenic and chromium removal from water/wastewater

Several treatment options for water and wastewater contaminated with arsenic and chromium have been applied over the years, including chemical and electrochemical precipitation, oxidation/reduction processes, ion exchange, membrane separation, flotation, solvent extraction, evaporation, adsorption, and phytoremediation. This section contains a brief overview of treatment techniques, with special attention on advanced oxidation/reduction processes. Figure 3 summarizes some techniques for remediation of arsenic and chromium contaminated waters.

Fig. 3

Techniques for the remediation of arsenic and chromium contaminated waters. Adapted from Nidheesh and Singh (2017), Singh et al. (2010), and Mazur et al. (2017)

Adsorption

Adsorption has been accepted as one of the most suitable treatment options due to its generally low cost, easy operation, and good efficiency. It is fundamentally a mass transfer process, where chemical or physical forces on the adsorbent surfaces drives a substance from the liquid phase to the solid phase, and the remaining adhered by bond (Song and Gallegos-Garcia 2014). In addition, an adsorption apparatus can be easily operated and there is a large availability of adsorptive materials, from the highly used activated carbon to the raw and modified biopolymeric materials (Kahu et al. 2016; Sathvika et al. 2016).

As an example, bismuth activated carbon was used for the binding of arsenite and dichromate. Arsenic removal was mainly achieved by ligand exchange, while in chromium sorption, the most important role was a metal reduction in combination with electrostatic phenomenon (Zhu et al. 2016). Activated carbon doped with iron hydroxide and manganese dioxide was used for As(III) adsorption. This material has the advantage of having a large surface area due to active carbon and oxidative property due to the presence of FeOOH and MnO₂, which allows the oxidation of trivalent to pentavalent arsenic. Under optimized conditions, the adsorption capacity of trivalent arsenic was 1 mmol_As g_adsorbent⁻¹ at pH 3 (Xiong et al. 2017). Chitosan is another material commonly used in adsorption studies. Kumar and Jiang (2016) reported the chitosan functionalization by graphene oxide to improve arsenic adsorption from aqueous solutions. They pointed out several interactions such as cationic and anionic, electrostatic, and intermolecular hydrogen bonding between the adsorbent and arsenic oxyanion species. The adsorption of As(V) in feldspars is reported to follow a pseudo-second-order kinetics and it is guided by electrostatic forces between terminal aluminol groups and arsenic in an acidic medium (Yazdani et al. 2016). However, over a macroporous polymer coated with coprecipitated iron–aluminum hydroxides, while As(III) adsorption follows a pseudo-second-order model, pentavalent species adsorption, on the other hand, follows a pseudo-first-order kinetic model. The polymer presented an adsorption capacity of 1.1 mmol_As g_adsorbent⁻¹ for As(III) and it is attributed to the formation of a bidentate mononuclear complex with iron sites. For As(V), a 0.7 mmol_As g_adsorbent⁻¹ adsorption capacity was observed as a result of generation of a bidentate binuclear complex with aluminum on the adsorbent (Suresh Kumar et al. 2016).

Zhou et al. (2016) described the application of Fe₃O₄-loaded mesoporous carbon microspheres for Cr(VI) removal. The material showed an adsorption capacity of 3 mmol_Cr g_adsorbent⁻¹ and a regeneration ability of 2.4 mmol_Cr g_adsorbent⁻¹ during five adsorption–desorption cycles. In addition, the adsorbent may be easily removed with a simple magnetic process. Moreover, the utilization of cheaper adsorbents has been focused forward conventional materials (Wu et al. 2017). Bamboo charcoal grafted by Cu²⁺–N–aminopropylsilane complexes was used for Cr(VI) adsorption (Wu et al. 2017). The authors reported a maximum adsorption capacity of 0.3 mmol_Cr g_adsorbent⁻¹ driven by a pseudo-second-order kinetic model. Similarly, dead biomass of isolated Aspergillus fungal species immobilized in epichlorohydrin cross-linked cellulose showed a Cr(VI) adsorption capacity of 0.5 mmol_Cr g_adsorbent⁻¹ (Sathvika et al. 2016). In turn, low-cost adsorbent Hibiscus cannabinus kenaf showed a maximum Cr(VI) uptake of only 11 μmol_Cr g_adsorbent⁻¹ (Omidvar Borna et al. 2016). However, even though the adsorption technique has several advantages, some oxyanions have a low affinity for the adsorptive material, limiting the removal efficiency, principally when working at residual oxyanions concentrations.

Coagulation

Coagulation is defined by colloid chemistry as the aggregation of colloidal or fine particles in a medium through addition of electrolytic ions. It is commonly employed to remediation of water containing several pollutants classes (Song and Gallegos-Garcia 2014). This method, as the adsorption, presents the benefits of easy operation, cheap, and simple handling coagulants (FeCl₃, FeSO₄, Al₂(SO₄)₃) (Bora et al. 2016).

Since early 1970s, coagulation process for As(V) removal from water has been performed and several studies were driven to improve the process and to understand the cleaning mechanism. Currently, the use of iron or aluminum salts as coagulants is a common process to eliminate arsenic from water (Song and Gallegos-Garcia 2014). Through the use of NaHCO₃, KMnO₄, and FeCl₃, 1.33 mM of arsenic was reduced to concentrations below 0.03 mM along with iron removal to concentrations below 1.8 μM (Bora et al. 2016). Using iron salts in enhanced coagulation (coagulant in excess), Cr(VI) removal can achieve almost 100% (Golbaz et al. 2014). However, coagulation processes are subject to several drawbacks including low efficiency when dealing with residual pollutants content, additional consumption of chemical reagents, and a great deal of secondary pollutants (Song et al. 2017).

Electrochemical methods

At the end of the eighteenth century, electrocoagulation was first reported to treat sewage. However, its application is still restricted due to the large initial capital investment and high energy consumption (Song et al. 2017). Usually, aluminum or iron are used as metal electrodes and according to complex precipitation kinetics, a series of hydroxides/oxyhydroxides are electrochemically formed. These species (Fe(OH)₂, Fe(OH)₃, Al(OH)₃, FeOOH, AlOOH) have a large surface area that can aggregate or adsorb dissolved pollutants. Parallel, hydrogen bubbles are generated by cathodic reduction and can induce flotation of the suspended particles, leading to additional pollutant removal (Nidheesh and Singh 2017).

Arsenic treatment with electrocoagulation process was studied by Vasudevan et al. (2010), using an aluminum alloy as anode and stainless steel as cathode, with a current density of 0.2 mA dm⁻², achieving 98.4% removal efficiency ([As]₀ = 6.7 μM). Beyond that, treatment of well water contaminated with arsenic in La Comarca Lagunera, México, was studied by Parga et al. (2005), using a carbon steel electrode in a pilot plant. The authors obtained 99.7% of arsenic removal attributed to magnetite formation ([As]₀ = 6.7 × 10⁻² μM).

The recovery of an electroplating wastewater with high hexavalent chromium content by electrocoagulation was studied by Tezcan Un et al. (2017). At the best conditions (0.05 M NaCl as electrolyte, 20 mA cm⁻², and pH 2.4), the initial Cr(VI) concentration of 19 mM was almost completely removed over an energy consumption of 2.68 kWh m⁻³. Additionally, the sludge obtained was used as raw material to produce reddish brown and black inorganic pigments. Hamdan and El-Naas (2014) employed Fe–Fe electrode pair to achieve 100% (considering the detection limit of the analytical method) hexavalent chromium removal ([Cr]₀ = 3.8 μM) in 5 min over electrocoagulation process (current density of 7.94 mA cm⁻² and pH 8), with an estimated energy consumption of 0.6 kWh m⁻³. Furthermore, treatment of a metal plating wastewater containing copper, chromium, and nickel (in concentrations of 0.71, 0.86, and 6.71 mM, respectively) by electrocoagulation with iron and aluminum electrodes was investigated by Akbal and Camcı (2011). At the optimum conditions, current density of 10 mA cm⁻² and pH 3.0, a removal of 100% (considering the detection limit of the analytical method) was obtained for all three metals in 20 min, corresponding to an energy consumption of 10.07 kWh m⁻³.

Ion exchange

Ion exchange is a physical-chemical sorption process where an ion from the solid phase is exchanged by other ion from the solution (Lee et al. 2017). It is a process with reversible interchange (Song and Gallegos-Garcia 2014).

Arsenic removal by ion exchange was tested with nanocomposite based on N-methyl-D-glucamine groups. The resin showed high arsenic removal efficiency, reaching, even in presence of interfering anions, arsenic concentrations below WHO recommendation. The equilibrium binding was well described by Langmuir isotherms and the binding capacity was approximately 0.7 mmol_As g_resin⁻¹ (Urbano et al. 2012). In other study, an amine-doped acrylic ion exchange fiber showed an ion exchange capacity of 7.5 mEq g⁻¹ for As(V). In addition, the fiber exhibited a removal efficiency above 83% after nine regeneration cycles (Lee et al. 2017).

Cation and anion exchangers synthesized with long-chained cross-linking agents were reported by Kononova et al. (2015) to have an interesting selectivity and satisfactory kinetic properties, allowing 100% of chromium and manganese recovery in counter-current columns (Cr(VI) and Mn(II) initial concentrations of 0.02 and 0.09 M). Besides the complete solution purification, the valuable metal components could be returned back to the industrial process. The natural resin Pelvetia canaliculata was reported to allow a synergistic effect for the remediation of an electroplating wastewater containing Cr(VI) and Cr(III). The brown algae was able to reduce hexavalent to trivalent chromium at low pH values and bind by cation exchange the generated Cr(III). The protonated P. canaliculata showed a Cr(VI) reduction capacity of 2.3 mmol g⁻¹ and a Cr(III) uptake capacity of 1.9 mmol g⁻¹ (Hackbarth et al. 2016).

Membrane separation

In membrane separation process, the use of semipermeable membranes, selectively permeable to water and certain solutes, allows the separation of target particles from the solution. There are several membrane separation alternatives, including microfiltration, reverse osmosis, electrodialysis, ultrafiltration, and nanofiltration (Song and Gallegos-Garcia 2014).

Although reverse osmosis is reported as one of the most efficient alternatives to remove arsenic from contaminated waters, it is a very expensive process. As nanofiltration requires operating pressures lower than the ones in reverse osmosis, nanofiltration process could be easily applicable. Thus, the efficiency of a nanofiltration pilot plant to remove arsenic from groundwater with natural contamination ([As]₀ = 5.7 μM) was evaluated. The rejection over 95% HAsO₄²⁻ was achieved using a process integral evaluation at 7 bar (Saitua et al. 2011). In another study, combining coagulation by Fe(III) with microfiltration, 97% of arsenic removal was obtained at pH 7 ([As]₀ = 2.7 × 10⁻² μM). The cost associated with this technology was evaluated as 0.066 U$/m³ of treated water (Mólgora et al. 2013).

The fabrication of a membrane for Ni(II) and Cr(VI) nanofiltration was investigated by Hosseini et al. (2017). The membrane was fabricated with poly(acrylonitrile) as the main material and poly(ethylene glycol) and TiO₂ as additive. Under optimized conditions, nickel ([Ni]₀ = 0.19 mM) and chromium ([Cr]₀ = 0.17 mM) rejection were 87 and 83%, respectively. Moreover, using commercial composite polyamide membranes (PN40 and NF300), Gaikwad and Balomajumder (2017) obtained a Cr(VI) and fluorine rejection of 88% and 82% with PN40 and 97% and 92% with NF300 membranes, respectively (Cr and F initial concentrations of 0.10 and 0.26 mM).

However, the real application of membrane technologies has some disadvantages, such as the generation of considerable residual sludge amount, expensive energy use to ensure the system pressurization, and the need to regenerate the resin or to clean the membrane. Additionally, the associated residual by-products generated may lead to a consequently new font of secondary pollution (Ortega et al. 2017).

Phytoremediation, bioremediation, biosorption, and biofiltration

Phytoremediation comprises the use of living plants to treat a certain contaminant by bioaccumulation or to reduce its toxicity. Arsenic removal ([As]₀ = 2.7 × 10⁻² μM) by rhizosphere of helophytes was investigated in a lab-scale wetland. An artificial domestic wastewater contaminated with arsenic was used. A better performance was observed under (i) carbon deficiency, (ii) oxidizing conditions, and (iii) elevated sulfate concentration. Arsenic mass balance indicated that 42.2% was accumulated within the roots, 17.2% remained within the gravel bed sediments, 16.2% was found in the pore water, 15.3% was in the outflow, and 9% was considered as unaccountable (Rahman et al. 2014). Nonetheless, in a horizontal subsurface flow pilot-scale constructed wetland for chromium removal, an important substrate contribution in chromium retention (61%, from an Cr initial concentration of 0.08 mM) was found, while the accumulation in plant was relatively low (0.24% in stems and leaves and 0.26% in roots) (Papaevangelou et al. 2017).

Although arsenic has an elevated human toxicity, a wide variety of microorganisms, mainly bacteria, can use it in redox reactions for growth and anaerobic respiration (Yamamura and Amachi 2014). As an example, the bacteria Pseudomonas stutzeri TS44 contains genes for arsenite oxidation and arsenic resistance that allow the bacteria application for arsenite removal from the environment (Akhter et al. 2017). Similarly, chromium bioremediation by fungi and bacteria has been employed through biosorption, chromate reduction, and bioaccumulation. While bacteria mainly promotes Cr(VI) reduction, fungi present a good biosorption mechanism. Bacillus methylotrophicus was isolated from tannery sludge and used for chromate reduction. At optimized conditions, the bioremediation process showed 91.3% of Cr(VI) reduction in 48 h ([Cr]₀ = 0.25 mM) (Sandana Mala et al. 2015).

In an innovative biosorption process, the macroalgae Gracilaria and Oedogonium were treated with iron and transformed into Fe-biochar by slow pyrolysis. The produced Fe-biochar showed higher biosorption capacity for arsenic and molybdenum: 0.8–1.1 mmol_As g_sorbent⁻¹ and 0.7–0.8 mmol_Mo g_sorbent⁻¹, respectively (Johansson et al. 2016). In the same way, protonated Laminaria seaweed was appointed as effective good option of biosorbent for Cr(III) treatment in aqueous solutions, showing a maximum uptake capacity of 0.8 mmol_Cr g_sorbent⁻¹ at pH 4 (Dittert et al. 2012).

Biofilters are formed through a set of microorganisms fixed in a porous medium. The support media coated by a thin layer of iron or manganese oxides for metal/metalloid removal from water is a long-term established filtration method. However, the biological adsorptive filtration proved to be an innovative variation of this methodology. The technique consists in the use of a biofilter with native microorganisms capable to oxidize iron and manganese, which can be naturally coated on the supports. Thus, groundwater containing arsenic species can be treated by the combination of these biological/physico-chemicals sorption processes: the oxidation and adsorption onto the biogenic iron and manganese oxides (Sahabi et al. 2009). In another study, As(III) removal using biological/iron/manganese combined oxidation systems leads to a decrease in arsenic concentration from 2.0 to 0.13 mM (Yang et al. 2014). The genetic diversity of microorganisms along the depth of the biofilter was investigated and the results suggested that the iron-oxidizing bacteria (Gallionella and Leptothrix), manganese-oxidizing bacteria (Hyphomicrobium and Arthrobacter), and arsenic oxidizing bacteria (Alcaligenes and Pseudomonas) were dominant in the biofilter (Yang et al. 2014).

Oxidative/reductive processes

Arsenic chemical oxidation is feasible over various oxidant agents. Zhang et al. (2017) compared As(III) oxidation ([As]₀ = 1.00–2.67 μM)) by potassium permanganate, sodium hypochlorite, monochloramine, and chlorine dioxide. The oxidation reactions are described by Eqs. (1), (2), (3), (4), and (5):

$$ {\mathrm{H}}_3{\mathrm{AsO}}_3+\mathrm{NaClO}\to {\mathrm{H}}_2{\mathrm{AsO}}_4^{-}+{\mathrm{Na}}^{+}+{\mathrm{Cl}}^{-}+{\mathrm{H}}^{+} $$

(1)

$$ {\mathrm{H}}_3{\mathrm{AsO}}_3+{2\mathrm{MnO}}_4^{-}\to {3\mathrm{H}}_2{\mathrm{AsO}}_4^{-}+{2\mathrm{MnO}}_2+{\mathrm{H}}_2\mathrm{O}+{\mathrm{H}}^{+} $$

(2)

$$ {\mathrm{H}}_3{\mathrm{AsO}}_3+{2\mathrm{ClO}}_2+{\mathrm{H}}_2\mathrm{O}\to {\mathrm{H}}_2{\mathrm{AsO}}_4^{-}+{2\mathrm{ClO}}_2^{-}+{3\mathrm{H}}^{+} $$

(3)

$$ {5\mathrm{H}}_3{\mathrm{AsO}}_3+{2\mathrm{Cl}\mathrm{O}}_2+{\mathrm{H}}_2\mathrm{O}\to {5\mathrm{H}}_2{\mathrm{AsO}}_4^{-}+{2\mathrm{Cl}}^{-}+{7\mathrm{H}}^{+} $$

(4)

$$ {\mathrm{H}}_3{\mathrm{AsO}}_3+{\mathrm{NH}}_4\mathrm{Cl}+{\mathrm{H}}_2\mathrm{O}\to {\mathrm{H}\mathrm{AsO}}_4^{2-}+{\mathrm{NH}}_4^{+}+{\mathrm{Cl}}^{-}+{2\mathrm{H}}^{+} $$

(5)

Using the potassium permanganate and sodium hypochlorite, an As(III) oxidation of 80% was obtained after 1 and 5 min, respectively. However, using monochloramine and chlorine dioxide under similar conditions, only 70% and 50% of As(III) oxidation was achieved after 2 days (Zhang et al. 2017).

Fe(II) is widely used for hexavalent chromium reduction. CL:AIRE (2007) reported a real case of groundwater contamination with Cr(VI) (1.6 mM) where the application in situ of an acidic ferrous sulfate heptahydrate solution was capable of removing 99.95% of hexavalent chromium by a reductive precipitation mechanism. In this process, Cr(VI) was reduced to Cr(III) (Eq. (6)) and further precipitated as Cr(OH)₃ (Eq. (7)), at neutral pH (Hashim et al. 2011).

$$ {\mathrm{Cr}}^{6+}+{3\mathrm{Fe}}^{2+}\to {\mathrm{Cr}}^{3+}+{3\mathrm{Fe}}^{3+} $$

(6)

$$ {\mathrm{Cr}}^{3+}+{3\mathrm{OH}}^{-}\to \mathrm{Cr}{\left(\mathrm{OH}\right)}_{3\ \left(\mathrm{s}\right)} $$

(7)

The main disadvantage associated with this process for industrial wastewaters is related with the high amount of iron needed and the co-production of high amounts of sludge that require further treatment. Additionally, generally the ferrous ions are able to quickly reduce the hexavalent chromium at low pH values. However, rate constants of this reaction at near-neutral pH conditions have not been reported.

Among the chemical oxidation technologies, advanced oxidation processes are by definition processes where the hydroxyl radical (^•OH) acts as the main oxidant agent. It is a radical with a high oxidizing potential (E^o = 2.8 V) able to react with virtually all classes of organic and inorganic compounds in relatively short times (Zhang et al. 2017). However, some AOPs techniques can also generate electrons or reductive species, which can be applied to promote reduction reactions. In the next section, these systems are discussed over their fundamental characteristics and in regard to arsenic and chromium treatment.

Advanced oxidation/reduction processes

Different reactions can lead to hydroxyl radical generation, being the most popular techniques the photolysis of hydrogen peroxide using UVC, ozonation, iron-based processes (mainly Fenton reaction), and heterogeneous photocatalysis (mainly TiO₂ as catalyst). These processes can be divided into homogeneous or heterogeneous if reactants and pollutants are in the same phase (ozonation, UVC/H₂O₂, and Fenton) or not (semiconductor/UV). They can be also differentiated in photochemical (UVC/H₂O₂, semiconductor/UV, and photo-Fenton) and non-photochemical (O₃ and Fenton), when in the presence or absence of radiation, respectively. Among the photochemical AOPs, iron-based processes and heterogeneous photocatalysis are able to initiate both oxidative and reductive processes, being applicable to promote hexavalent chromium reduction and trivalent arsenic oxidation, while UVC/H₂O₂ process is applicable only to generate ^•OH radicals and, consequently, can only be used for As(III) oxidation. It should be noted that despite only the most recent references will be cited in this review work, reports on As(III) and Cr(VI) removal by the processes mentioned below have been published since many years ago.

Iron-based processes

Fenton reaction is the most used iron-based process. It is described by the reaction of ferrous ions with hydrogen peroxide (Eq. (8)). In photo-Fenton reaction, UV-vis radiation enhances ^•OH radical production by forming a catalytic cycle through the photoreduction of ferric ions to ferrous ions (Eq. (9)).

$$ {\mathrm{Fe}}^{2+}+{\mathrm{H}}_2{\mathrm{O}}_2\to {\mathrm{Fe}}^{3+}+{\mathrm{O}\mathrm{H}}^{-}+{}^{\bullet}\mathrm{O}\mathrm{H} $$

(8)

$$ {\mathrm{Fe}}^{3+}+{\mathrm{H}}_2\mathrm{O}+ hv\to {\mathrm{Fe}}^{2+}+{\mathrm{H}}^{+}+{}^{\bullet}\mathrm{O}\mathrm{H} $$

(9)

The ferric species with higher quantum yields (ϕ - moles of product formed or reagent consumed per moles of photons absorbed) are the hydroxide complexes Fe(OH)²⁺ and Fe(RCO₂)²⁺, where R is an organic ligand. These species allow the photo-induced ligand-to-metal charge-transfer (LMCT) mechanism to promote Fe(II) regeneration (Rodríguez et al. 2005; Sagawe et al. 2001). Liu et al. (2007) reported the combined treatment of hexavalent chromium and bisphenol A (BPA) by Fe(III)–OH complexes in a photocatalytic system. In this study, while Cr(VI) was reduced, BPA was simultaneously oxidized, leading to a synergic effect: both Cr(VI) photocatalytic reduction and BPA degradation rates were higher in the ternary (Fe(III)/Cr(VI)/BPA) system. However, this process has some drawbacks including a strict pH control (2.8–3.5) to minimize iron oxyhydroxide precipitation and to maximize the concentration of photoactive species. Therefore, the low efficiency of these processes at neutral pH can be avoided by adding iron complexing agents. Actually, polycarboxylates and aminopolycarboxylates are able to complex with Fe(III), which absorb light in near-UV and visible regions more efficiently than hydroxide complexes (Clarizia et al. 2017). In fact, since the beginning of the nineteenth century, the generation of reactive species by photochemical dissociation of hydroxylated Fe(III) complexes (Eq. (10)) has been reported (Nansheng et al. 1998). Depending on the organic ligand, the Fe(III) complexes exhibit different quantum yields in different wavelengths (Gernjak et al. 2006; Malato et al. 2009). Table 3 shows some Fe(III) complexes and their respective quantum yields.

$$ \left[{\mathrm{Fe}}^{3+}\mathrm{L}\right]+ hv\to {\left[{\mathrm{Fe}}^{3+}\mathrm{L}\right]}^{\ast}\to {\mathrm{Fe}}^{2+}+{\mathrm{L}}^{\bullet } $$

(10)

Table 3 Quantum yield for the photoreduction of Fe(III) by various carboxylic acids (Abrahamson et al. 1994)

Summarizing, the main advantages of using ferricarboxylate complexes can be pointed as follows: (i) quantum yields higher than ferric iron–water complexes, (ii) reaction in a higher fraction of the solar spectrum (UV + visible light), and (iii) allows working at near-neutral pH values (Clarizia et al. 2017).

Besides the type of organic ligand, a strong control of iron/ligand ratio is fundamental to ensure the process efficiency. The effect of ethylenediaminetetraacetic acid (EDTA), nitrilotriacetic acid (NTA), oxalic acid, and tartaric acid ratio was investigated by De Luca et al. (2014). The authors have found that an excess of all ligands is necessary, since the photodecarboxylation is a fast process and the hydroxyl radicals formed can attack the organic ligand. This excess avoids iron precipitation and increases the catalytic activity.

Krishna et al. (2001) reported the oxidation of As(III) solution with Fenton’s reagent, and the generated As(V) solution was passed through iron scrap and filtered through sand. As(III) oxidation (0.03 mM) was performed during 10 min and in the final combined process, arsenic concentration was less than 0.13 μM. Fe(II) oxidation in the presence of EDTA was reported by Wang et al. (2013). The authors observed that at acidic and neutral pH, the presence of excess EDTA inhibited As(III) oxidation, concluding that the rapid Fe(II) oxidation is not necessarily associated with a synergistic As(III) oxidation. Alternatively, Fe(III) in the presence of citrate proved to promote faster oxidation of trivalent arsenic than in citrate absence, showing that Fe(III)–CitOH⁻ complex photolysis produces oxidant species more efficiently than Fe(OH)²⁺ photolysis (Hug et al. 2001).

For hexavalent chromium reduction, in systems with ferricarboxylate complexes, the photoregenerated ferrous ions may reduce hexavalent chromium, and the formed reactive oxygen species are able to oxidize, simultaneously, organic molecules existent in aqueous solution. In addition, beside the reactive oxygen species, the strong reducing agent CO₂^•− could be formed by photodecarboxylation or oxidation of oxalic, citric, and tartaric acids. This radical is able to reduce Cr(VI) and enhance the reaction (Meichtry et al. 2011; Soares et al. 2015) since the pair CO₂^•−/CO₂ (− 2.20 eV) (Forouzan et al. 1996) has a more negative potential than the pair Cr(VI)/Cr(III) (+ 1.33 eV) (Dittert et al. 2014). Wei et al. (2014) reported the use of iron corrosion products combined with tartaric acid in an illuminated system for hexavalent chromium reduction. This system proved to be efficient, reducing more than 90% of the initial Cr(VI) (0.2 mM) after 30 min. Moreover, Hug et al. (1997) studied hexavalent chromium photoreduction by Fe(III)–oxalate and Fe(III)–citrate complexes. The authors reported similar behavior for both acids and no precipitation of Cr(III) hydroxides, suggesting the formation of an organic Cr(III) complex with the oxidation product of oxalate or citrate. Our research group also found very important deductions concerning the reported idea whether the kinetics of Cr(VI) reduction depends on carboxylic acids structure or on α-OH number. It was found that Cr(VI) removal rate involving different organic ligands decreases in the following order: citric acid > oxalic acid > EDTA > maleic acid, proposing that there should exist a correlation between the fraction of photoactive species formed and the α-OH number of each acid, as well as with the respective quantum yield (Marinho et al. 2016). In the same paper, Fe(III)/UVA-vis/citric acid system showed to be very promising to treat a real wastewater from a galvanization process, achieving Cr(VI) photocatalytic reduction after 30 min, a better result than the one achieved with synthetic solution (Fig. 4), due to the presence of organic matter in the effluent (DOC = 5.9 mg C L⁻¹). In fact, this system contributed to the combined effect of oxyanion reduction and organic pollutant removal.

Fig. 4

Photoreduction of Cr(VI) by a Fe(III)/UVA-vis/Citric acid system with 2.4 mM of citric acid and 0.16 mM of iron at pH 5.0, 25 °C in the lab-scale photoreactor (SUNTEST at 500 W m⁻²). (Square) synthetic Cr(VI) solution, (triangle) real wastewater

A scheme for Cr(VI) photoreduction in systems with ferricarboxylate complexes is presented in Fig. 5.

Fig. 5

Reaction scheme of an illuminated solution containing Cr(VI), carboxylic acid (as oxalic, citric or tartaric acids) and Fe(III). Adapted from Hug et al. (1997)

H₂O₂-UVC

Hydrogen peroxide photolysis by UVC radiation (Eq. 11) represents one of the easiest ways to produce ^•OH radicals (Oppenländer 2007). This process has a powerful oxidation ability, no undesired sludge generation, and easy operation. However, it involves the use of high amounts of hydrogen peroxide and stricter control of pH and temperature to prevent H₂O₂ decomposition (Eq. 12) (He et al. 2012; Rosario-Ortiz et al. 2010).

$$ {\mathrm{H}}_2{\mathrm{O}}_2\overset{hv}{\to }2\ {}^{\bullet}\mathrm{O}\mathrm{H} $$

(11)

$$ {2\mathrm{H}}_2{\mathrm{O}}_2\to {2\mathrm{H}}_2\mathrm{O}+{\mathrm{O}}_2 $$

(12)

As in this process only the ^•OH radical oxidant is produced, it is not applicable for Cr(VI) reduction. Concerning As(III) oxidation, the process is feasible and can even be driven “in the dark” (chemical oxidation by H₂O₂). However, Litter et al. (2010) reported that, in this case, a large excess of hydrogen peroxide is needed to reach the total oxidation. On the other hand, the As(III) oxidation by direct photolysis with UV radiation is not very efficient. Conversely, using the combined UVC/H₂O₂ process, the As(III) is improved and lower H₂O₂ doses are required when compared with dark system. In fact, Lescano et al. (2011), reported an initial H₂O₂/As(III) molar ratio in the range of 159–247 as the optimum range to promote As(III) oxidation in short reaction times. In another study, the photon flux was described to have a significant influence on the reaction rate, showing that the kinetic constant decreases linearly with the photon flux decrease (Lescano et al. 2012). In a recent work (Marinho et al. 2018b), our research group showed the successful As(III) oxidation by UVC/H₂O₂ using a micro-meso-structured photoreactor under different illumination schemes, in order to ensure a uniform irradiation of the entire reaction mixture.

Heterogeneous photocatalysis

In heterogeneous photocatalytic processes, electron (${e}_{cb}^{-}$) and hole (${h}_{vb}^{+}$) pairs are generated (Eq. (13)) by the absorption of photons with equal or higher energy than the photocatalyst bandgap (Wu et al. 2013). The generated holes are highly oxidizing, leading to the oxidative reaction with both organic and inorganic contaminants or to the reaction with water forming ^•OH radical (Eq. (14)). In contrast, the electrons may be driven to acceptors or inorganic species with a reduction potential more positive than the one of the semiconductor conduction band (Cappelletti et al. 2008). During the last decades, due to chemical stability, low cost, and the ability to use a small percentage of ultraviolet solar radiation, TiO₂ has been the most used semiconductor in photocatalytic applications. Other semiconductors, such as ZnO, CdS, WO₃, Fe₂O₃, and SnO₂, were less investigated but can also be used. When using TiO₂ as semiconductor, in the presence of dissolved O₂ as electron acceptor, superoxide radicals (${\mathrm{O}}_2^{\bullet -}$, ${\mathrm{HO}}_2^{\bullet }$) can be generated (Eqs. (15) and (16)) (Lee and Choi 2002).

$$ \mathrm{semiconductor}+\mathrm{hv}\to {e}_{cb}^{-}+{h}_{vb}^{+} $$

(13)

$$ {\mathrm{OH}}^{-}+{h}_{vb}^{+}\to {}^{\bullet}\mathrm{O}\mathrm{H} $$

(14)

$$ {\mathrm{O}}_2+{e}_{cb}^{-}\to {\mathrm{O}}_2^{\bullet -} $$

(15)

$$ {\mathrm{O}}_2^{\bullet -}+{\mathrm{H}}^{+}\to {\mathrm{H}\mathrm{O}}_2^{\bullet } $$

(16)

Heterogeneous photocatalysis efficiency is affected by several parameters, including the initial contaminant concentration, photon flux, presence or absence of oxygen, catalyst loading, pH, and temperature. It is a consensus that the oxidation/reduction promoted through heterogeneous TiO₂ photocatalysis follows a first-order kinetic model. However, despite this model has been useful, it is commonly recognized that both rate constants and orders are “apparent” and also called “pseudo-first order” (Minero 1999). Above a certain photon flux, reaction rate dependency on light intensity changes from one to half-order due to the excess of photogenerated species (${e}_{cb}^{-}$, ${h}_{vb}^{+}$, and ^•OH). With a further increase on the light intensity, reaction rate becomes independent. Under these conditions, the quantum yield decreases because of the high rate of ${e}_{cb}^{-}$/${h}_{vb}^{+}$ pairs recombination and the reaction rate remains constant (Malato et al. 2016). The catalyst loading influences positively the reaction rate until a certain value that depends on the photoreactor design and experimental conditions. However, above this limit, the reaction rate becomes independent of the photocatalyst load and could be negatively affected due to light attenuation caused by high catalyst quantities (Cassano and Alfano 2000). The solution pH affects significantly the TiO₂ particles charge. The pH of zero point charge (pHzpc) is defined as the pH where the particles surface is uncharged. Above this value, the catalyst is negatively charged and attracts positive molecules. At pH values below the pHzpc, the catalyst surface is positively charged attracting negative species (Fernández-Ibáñez et al. 2003). Since heterogeneous photocatalytic processes are activated by photons that reach the catalyst surface, the system does not need heating and could be conducted at room temperature. Although low temperatures favor reactants/pollutants adsorption, by-products adsorption is also enhanced, which can block the catalyst surface. In contrast, at higher temperatures, above 80 °C and near to the water boiling point, the reactants/pollutants adsorption is disfavored and may become a rate-limiting step. Additionally, the dissolved oxygen concentration in water decreases with the temperature increase. Therefore, the ideal temperature range is usually between 20 and 80 °C (Malato et al. 2016).

The reactions of a metal/metalloid with TiO₂ can be driven by three different mechanisms through successive one-electron steps: direct reduction, indirect reduction, and oxidative reaction. The direct reduction by ${e}_{cb}^{-}$ (Eq. (17)) is viable for species with redox potential more positive than the one of ${e}_{cb}^{-}$. However, the reoxidation of the reduced species can cause a short circuiting (Eqs. (18) and (19)). The addition of sacrificial agents, such as organic electron donors (RH), which can be irreversibly oxidized by ${h}_{vb}^{+}$ or ^•OH (Eqs. (20) and (21)), minimizes the short circuiting and can, also, produce a synergetic effect by indirect reduction. High energetic radicals can be generated when electron donors are present (Eq. (22)). Carboxylic acids, such as formic acid and oxalic acid, can react with ${h}_{vb}^{+}$, acting as an electron donor and generating the strong reducing ${\mathrm{CO}}_2^{\bullet -}$ radical. An oxidative reaction by ${h}_{vb}^{+}$ (Eq. (23)) takes place when the metal/metalloid has a redox potential more negative than the ${h}_{vb}^{+}$. The ^•OH radicals (Eq. (24)) and other species such as ${\mathrm{O}}_2^{\bullet -}$, ${\mathrm{HO}}_2^{\bullet }$, and H₂O₂ may also be generated and assist the oxidation (Litter 2017). Figure 6 shows a schematic diagram for the photocatalytic transformation of metal/metalloids over TiO₂ heterogeneous photocatalysis.

$$ {\mathrm{M}}^{n+}+{e}_{cb}^{-}\to {\mathrm{M}}^{\left(n-1\right)+} $$

(17)

$$ {\mathrm{M}}^{\left(n-1\right)+}+{h}_{vb}^{+}\to {\mathrm{M}}^{n+} $$

(18)

$$ {\mathrm{M}}^{\left(n-1\right)+}+{}^{\bullet}\mathrm{O}\mathrm{H}\to {\mathrm{M}}^{n+}+{\mathrm{OH}}^{-} $$

(19)

$$ \mathrm{RH}+{h}_{vb}^{+}\to {\mathrm{R}}^{\bullet }+{\mathrm{H}}^{+} $$

(20)

$$ \mathrm{RH}+{}^{\bullet}\mathrm{O}\mathrm{H}\to {\mathrm{R}}^{\bullet }+{\mathrm{H}}_2\mathrm{O} $$

(21)

$$ {\mathrm{R}}^{\bullet }+{e}^{-}+{\mathrm{M}}^{n+}\to {\mathrm{R}}_{\mathrm{ox}}+{\mathrm{M}}^{\left(n-1\right)+} $$

(22)

$$ {\mathrm{M}}^{n+}+{h}_{vb}^{+}\to {\mathrm{M}}^{\left(n+1\right)+} $$

(23)

$$ {\mathrm{M}}^{n+}+{}^{\bullet}\mathrm{O}\mathrm{H}\to {\mathrm{M}}^{\left(n+1\right)+}+{\mathrm{OH}}^{-} $$

(24)

Fig. 6

Schematic diagram for the photocatalytic transformation of metal/metalloids on TiO₂. Adapted from Litter (2017)

As(III) species can be oxidized through successive one-electron steps, by both ${h}_{vb}^{+}$ and ^•OH, ${\mathrm{O}}_2^{\bullet -}$ and ${\mathrm{HO}}_2^{\bullet }$ radicals (Dutta et al. 2005; Guan et al. 2012; Hérissan et al. 2017; Kim et al. 2015; Lee and Choi 2002). However, the main As(III) oxidant (${h}_{vb}^{+}$, ^•OH, ${\mathrm{O}}_2^{\bullet -}$ or ${\mathrm{HO}}_2^{\bullet }$) is still not clear, being the ^•OH radical defended by some authors (Dutta et al. 2005; Yoon et al. 2009) and the ${\mathrm{O}}_2^{\bullet -}$/${\mathrm{HO}}_2^{\bullet }$ species by others (Choi et al. 2010; Ferguson et al. 2005; Lee and Choi 2002). Table 4 shows some studies on As(III) oxidation by heterogeneous photocatalysis.

Table 4 Studies on the As(III) oxidation by heterogeneous photocatalysis

In the last years, reduction of hexavalent to trivalent chromium using several photocatalysts has received large attention (Baig et al. 2015; Celebi et al. 2016; Chakrabarti et al. 2009; Schrank et al. 2002). Because of its high photostability, activity and reasonably low cost, TiO₂ has been widely used, both in suspension and in a supported form. Actually, Cr(VI) reduction is commonly reported to be effectively achieved by heterogeneous TiO₂ photocatalysis (Joshi and Shrivastava 2011; Testa et al. 2004). Nonetheless, in an isolated system, the formed Cr(III) can be re-oxidized to Cr(VI) by the formed holes and hydroxyl radicals. Furthermore, the electron-hole recombination may also disfavor the reaction. To suppress these effects, organic compounds are commonly added as sacrificial agents, reacting with ^•OH radicals or ${h}_{vb}^{+}$ and hindering the electron-hole recombination, contributing to an enhancement of the reduction process. In many cases, organic compounds and Cr(VI) are simultaneously present in wastewaters as a result of several industrial processes, being important to take into account their contribution. Table 5 shows some studies on hexavalent chromium reduction by heterogeneous photocatalysis.

Table 5 Studies on the Cr(VI) reduction by heterogeneous photocatalysis

As it is possible to see from Tables 4 and 5, the majority of publications about water treatment by heterogeneous photocatalysis are relative to the use of suspended photocatalysts. Nevertheless, it is costly and difficult to separate the catalysts from the solutions (Ananpattarachai and Kajitvichyanukul 2016). The functionalization of inert supports with photocatalysts avoids the need for a further filtration step and allowing catalyst reuse as far as its stability is maintained. In the next section, the support types and methods of catalyst immobilization are discussed.

Photocatalyst immobilization

Several methods are commonly employed to produce catalyst thin films, including chemical/physical vapor deposition, sputtering, dip-coating, sol–gel, and anodic oxidation methods (Ding et al. 2001; Monteiro et al. 2015; Pecchi et al. 2001; Sonawane et al. 2003; Vera et al. 2018). Furthermore, several types of materials including glass, paper, ceramic tiles, fiberglass, pumice stone, and stainless steel were already studied as inert supports (Ávila et al. 2002; Cardona et al. 2004; Shephard et al. 2002; Subrahmanyam et al. 2008; Vella et al. 2010). Table 6 shows some studies regarding the degradation of different contaminants using immobilized photocatalyst, as well as the coating method and support used. As an example, Han et al. (2012) used an easy and low-cost spray coating technique to successfully support TiO₂ nanoparticles over a polyester fiber filter at low temperature. Impregnation of TiO₂ in chitosan/xylan hybrid film to remove Cr(VI) through adsorption and photocatalysis was studied by Ananpattarachai and Kajitvichyanukul (2016). This innovative material presented competitive reaction rates when compared with TiO₂ powder. On the other hand, the use of polycaprolactone, a biodegradable polymer, as TiO₂ support during Cr(VI) reduction was evaluated by Akkan et al. (2015). The results a showed successful attachment of catalyst nanoparticles using simple solvent-cast processes, that lead to an efficient Cr(VI) reduction. In addition, Ferguson and Hering (2006) used glass beads coated with TiO₂ for As(III) oxidation ([As]₀ = 2.64 μM) in continuous flow mode, using a synthetic groundwater matrix. The authors obtained 70% of As(III) oxidation in a residence time of 10 min.

Table 6 Heterogeneous photocatalysis studies using catalyst thin films immobilized in inert supports

Our research group reported several studies (da Costa Filho et al. 2017; Lopes et al. 2013a; Marinho et al. 2017a; Marinho et al. 2017b; Monteiro et al. 2015; Pinho et al. 2015) on TiO₂-P25 deposition on different supports by dip and spray coating methods. It was proved that transparent cellulose acetate monolithic structures coated with TiO₂-P25 by a simple dip-coating deposition method have an effective catalytic stability during 10 consecutive Cr(VI) reduction cycles using citric acid as a hole scavenger (Marinho et al. 2017a). In other works, a reasonable good catalytic stability was observed during 3 consecutive Cr(VI) reduction cycles using TiO₂-P25 deposited both on a UV transparent cellulose acetate sheet assembled between the two slabs of a micro-meso-structured photoreactor (NETmix) (Marinho et al. 2017b) and on the front glass slab of the same reactor using a spray system. Additionally, when the catalyst is deposited on the network of channels and chambers of the NETmix reactor, despite the better performance observed, there was observed a decrease in the reaction rate in the second and third reuse cycles, which was attributed to the blockage of the catalyst surface by scavenger agents and their by-products (Marinho et al. 2018a).

Despite the several advantages of heterogeneous photocatalysis, this process also has its own drawbacks: the catalyst surface accessibility to the reactants/pollutants and photons and external mass transfer limitations, due to the higher diffusional length. The development of innovative photocatalysts with high activity and visible light response and the combination of nanosorbents with catalysts can be a good approach to improve the catalytic activity.

Furthermore, the reactor configuration is, also, of utmost importance regarding the system effectiveness and a competitive full-scale photocatalytic system needs to overcome mass and photons transfer limitations. This way, process intensification of heterogeneous photocatalysis through the use of innovative reactor configurations can be a good approach to enhance mass and photons transfer (Li et al. 2011). In the next section, different reactor types used in heterogeneous photocatalytic systems towards process intensification will be discussed.

Photocatalytic reactors

The industrial implementation of heterogeneous photocatalysis is limited mostly due to scale-up problems and low reaction rates. The photocatalyst reactivity in combination with photoreactor is currently in the range of 0.05–0.1 mol m⁻³_reactor s⁻¹. However, it is suggested that this parameter should achieve 100- to 1000-fold increase before industrial implementation (Van Gerven et al. 2007). Recently, a great effort has been made towards process intensification using novel reactor designs, minimizing photons and mass transfer limitations when dealing with heterogeneous photocatalytic reactions (Matsushita et al. 2008).

The first part of heterogeneous photocatalysis is the light transport to the catalyst surface. Usually, the light has to travel through the bulk solution and also through a transparent wall before reaching the catalyst surface. The amount of light that reaches the catalyst surface is only a fraction of the emitted light due to the absorption and scattering effects in reactor wall and bulk solution (Van Gerven et al. 2007). When using thin-film catalyst systems, depending on the reactor configuration, distinct irradiation mechanisms can be found: back-side illumination (BSI) and front-side illumination (FSI). In the BSI, the incident irradiation and the catalyst film are on opposite sides of the support structure. On the other hand, in the FSI mechanism, both incident irradiation and catalyst film are on the same side of the support structure (Chen et al. 2001). An important parameter related to the illumination is the amount of illuminated surface per unit of reaction liquid volume inside the reactor (m²_ill m⁻³_reactor). This value has been used quite often, although care should be taken to not neglect the shadows in the reactor and overestimate this parameter (Van Gerven et al. 2007). Table 7 lists some reactor types and their reported catalyst-coated surface per reaction liquid volume.

Table 7 Comparison of catalyst coated surface per reaction liquid volume in different photoreactor configurations (adapted from Van Gerven et al. (2007))

Once the incident light activates the catalyst, the contact between catalyst and reagents/pollutants should be maximized. One important parameter related to mass transfer is the Reynolds number, and high Reynolds numbers are associated with a decrease in mass transfer limitations (Lin and Valsaraj 2005). Several reactors design have been developed or proposed to achieve the process intensification. While slurry reactors, annular reactors, immersion reactors, optical tube reactors, and optical fiber reactors are among the most tested, some innovative reactors such as the spinning disc reactor, the monolith reactor, and the microreactor have been also tested to promote photocatalytic reactions. Figure 7 shows the schemes of some reactors applied in heterogeneous photocatalysis.

Fig. 7

Different reactor designs: a slurry immersion reactor; (b) slurry tubular reactor coupled with compound parabolic collector; (c) spinning disc reactor; (d) fluidized bed reactor; (e) side-emitting optical fiber reactor; (f) monolith reactor; (g) microreactor—Adapted from Van Gerven et al. (2007) and Boyjoo et al. (2013)

Among the various reactor types, the monolithic photoreactor is known to provide a high catalyst surface-to-volume ratio, improving the contact between the catalyst and the contaminant(s)/reactant(s), in addition to a more efficient illumination of the photocatalyst. Our team published different works on Cr(VI) reduction by heterogeneous photocatalytic processes using different reactors and system configurations. To reduce photon transfer limitations through diffusional length reduction, a tubular CPC (compound parabolic collector) photoreactor was packed with transparent cellulose acetate monolithic structures (CAM) coated with TiO₂-P25 by a dip-coating method. This reactor allows 0.15 g of TiO₂ per liter of liquid inside the photoreactor. The support geometry ensures a high surface-area-to-volume ratio, providing an illuminated catalyst surface area per unit of volume inside the reactor of 211 m² m⁻³. This configuration improves the contact between the photocatalyst and the contaminant(s)/reactant(s) and provides a more efficient exposure of the photocatalyst to the radiation. It was observed a photocatalyst reactivity in combination with the photoreactor of 0.09 mmol_Cr(VI) m⁻³_{illuminated volume} s⁻¹, corresponding to 103 mmol_Cr(VI) m⁻³_{illuminated volume} kJ⁻¹ (the values are slightly different from those published due to correction of reactor area values effectively illuminated) (Marinho et al. 2017a). On the other hand, radiant power that goes through the monolithic structure walls depends on the photocatalyst coating thickness and on the monolithic structure geometry.

Microstructured reactors are devices that allow fast mixing and short molecular diffusion distance, laminar flow, and large surface-to-volume ratio, allowing high mass transfer and reduced photon transport limitations. Therefore, it is estimated that microreactors could enhance photoreactions, in function of their higher spatial homogeneity of irradiance and better light dissemination through the entire reactor depth (Gorges et al. 2004; Padoin and Soares 2017). In this context, a novel micro-meso-structured photoreactor based on NETmix technology developed by Lopes et al. (2013b) has been successfully used by our research team (da Costa Filho et al. 2017; Marinho et al. 2017b). This NETmix reactor has a regular network of interconnected chambers and small-sized channels, allowing short molecular diffusion distances and large specific interfacial areas, improving pollutant/reactants/catalyst contact. The chambers are modeled as perfectly mixing zones and the channels as plug flow perfect segregation zones, improving the radial mixing, and, consequently, the mass and heat transfer, over a laminar flow (Fig. 8). The intensification of Cr(VI) photocatalytic reduction was tested through TiO₂-P25 deposition on UV transparent cellulose acetate (CA) sheets placed between the two slabs of the photoreactor (Marinho et al. 2017b). The micro-meso-structured reactor flat configuration and the small size of the channels and chambers provided an efficient and uniform exposure of TiO₂-P25 CA sheets to radiation, with an illuminated surface per unit of volume inside the reactor of 333 m² m⁻³ and 0.74 g TiO₂ per liter of liquid inside the reactor. The photocatalyst reactivity in combination with the photoreactor was significantly increased in relation to the monolithic tubular reactor to 1.3 mmol_Cr(VI) m⁻³_{illuminated volume} s⁻¹ (corresponding to 2387 mmol_Cr(VI) m⁻³_{illuminated volume} kJ⁻¹) (once more, the values are slightly different from those published due to correction of reactor area values effectively illuminated). In fact, the NETmix enhanced the shift from surface reaction to a homogeneous radical reaction, decreasing the existing mass and photons transfer limitations, probably due to the higher degree of mixing, higher amount of catalyst per unit of liquid volume, and higher spatial illumination homogeneity.

Fig. 8

Scheme of the micro-meso-structured photoreactor based on the NETmix technology. Adapted from Marinho et al. (2018a)

In order to evaluate BSI and FSI irradiation mechanisms, a TiO₂-P25 thin film was also uniformly deposited on the front glass slab (BSI mechanism) or on the network of channels and chambers imprinted in the back stainless steel slab (FSI mechanism) of the NETmix reactor. The results achieved with the photocatalyst immobilized at the photoreactor front glass slab (GS) are similar with the ones reported with the CA sheets as support (Marinho et al. 2017b), since the irradiation mechanism is the same for both reactor configurations (BSI mechanism). Nonetheless, a 3-fold increase on Cr(VI) reduction rate was perceived when comparing BSI and FSI irradiation mechanisms. Actually, comparing both system configurations, a 3-fold increase in the TiO₂-P25 surface area in contact with the liquid per unit of volume inside the reactor is noticed, from 333 m² m⁻³ when the photocatalyst is deposed on frontal glass slab to 989 m² m⁻³ when it is immobilized at the back stainless steel slab. Simultaneously, the photocatalyst reactivity in combination with the photoreactor was considerably improved from 2139 to 6646 mmol_Cr(VI) m⁻³_{illuminated volume} kJ⁻¹.

In the photochemical processes, the high cost associated with energy consumption to produce UV radiation is one of the biggest problems. This drawback can be suppressed using solar radiation, with the advantages inherent in the use of a renewable, clean, and sustainable energy source. However, despite being an important source of energy, the use of solar radiation in heterogeneous photocatalysis could not be the optimal alternative for real applications due to the large area needed for the installation and operation of such reactors, the associated high cost and the dependency on the weather conditions (Braham and Harris 2009; Malato et al. 2009). As alternative, the use of a microscale illumination system based on UV-emitting diodes (UVA-LEDs) in photocatalytic reactors is known to have high illumination efficiency and to promote a larger catalyst area effectively illuminated due to their small-angle emittance (Ciambelli et al. 2009). In addition, having the advantage of having a low power input, high energy efficiency, small dimensions, and long lifetime, this irradiation source has become increasingly popular when using photochemical processes (Su et al. 2015). Thus, the intensification of heterogeneous photocatalytic processes could be achieved by combining the use of improved reactors with efficient radiation sources. As a matter of fact, a 1.2-fold increase on Cr(VI) reduction reaction rate was observed by our investigation group, when the TiO₂-P25 film immobilized in the back stainless steel slab of the NETmix reactor is irradiated by UVA-LEDs light instead of solar light. Furthermore, UVA-LED light enables the numbering up of several microreactors (compact treatment system) and significantly reduces the photon generation cost, contributing to a reasonably priced process. This system attained an almost 65-fold increase in reactivity for Cr(VI) reduction comparing with results achieved with a monolithic tubular photoreactor irradiated with simulated solar light. In Table 8 a brief comparison is made between the main results obtained by our research group for the reduction of 0.02 mM of Cr(VI) by heterogeneous photocatalysis using different reactors and system configurations. Several important parameters are reported to give a clearer view of the reactor performance and to be easily compared with other systems.

Table 8 Comparison of several results attained on Cr(VI) reduction (0.02 mM) by heterogeneous photocatalysis using different reactors and system configurations (Marinho et al. 2018a; Marinho et al. 2017a; Marinho et al. 2017b)

Conclusions

In this review paper, arsenic and chromium speciation in solution was discussed and some analytical techniques for their determination were appointed. In addition, the current technologies to reduce and oxidize Cr(VI) and As(III), respectively, are described, especially in relation to advanced oxidation processes.

Based on the information reported, the following can be concluded:

• (i). The use of advanced oxidation processes for Cr(VI) reduction and As(III) oxidation is an interesting and feasible treatment option that should be explored and enhanced.

• (ii). Concerning Cr(VI) removal by heterogeneous photocatalysis in slurry systems, several papers have been published. However, there is a lack of information regarding the use of catalyst thin films and its possible reuse for Cr(VI) reduction purposes. On the other hand, although it is known that the presence of organic compounds improves the reaction and that, in most cases, they are present in wastewaters simultaneously with Cr(VI), researches regarding the treatment of real wastewaters are still scarce.

• (iii). The use of advanced oxidation processes for As(III) removal is still in an early stage and much experimental work is needed for a potential application. It is necessary to clarify which are the major oxidant species responsible by As(III) oxidation.

• (iv). Significant advances on intensifying heterogeneous photocatalysis are reported regarding the use of novel reactor designs, such as microreactors. Although a range of photocatalytic reactions have been carried out in novel photoreactors to date, a lot of work remains in this area, namely the theoretical modeling of oxanion removal, considering the coupling of fluid flow and mass transfer as well as photon transfer, its optimization and evaluation of scale-up strategies.