CdIn₂S₄ Micro‐Pyramids for Reductive Photocatalytic Degradation of Perfluorooctanesulfonic Acid

Abstract

Per‐ and poly‐fluoroalkyl substances (PFAS) constitute a class of persistent organic pollutants that severely affect human health and the environment owing to their resistance to degradation by traditional water treatment methods. Semiconductor‐assisted photocatalysis has the potential to be a green method to achieve complete mineralization of PFAS. Cadmium indium sulfide (CdIn₂S₄) is an exciting photocatalytic material because of its high visible light harvesting capacity and high reduction potential. However, CdIn₂S₄ has not been investigated for PFAS degradation. Herein, CdIn₂S₄ micro‐pyramids are fabricated via solvothermal synthesis, and their photocatalytic activity toward the photodegradation of perfluorooctanesulfonic acid (PFOS) is investigated. The reaction conditions, such as the light source and the light power are optimized. The findings show the capability of CdIn₂S₄ to achieve almost complete degradation and defluorination of PFOS (removal% = 99 ± 7% and defluorination% = 97 ± 22%) under optimized conditions. From radical quenching experiments, it is found that the mechanism of degradation is via photoelectron reduction. CdIn₂S₄ shows an outstanding performance toward the degradation of the PFAS sample derived from a contaminated facility in South Australia. This work opens the door for investigating other CdIn₂S₄‐based photocatalysts and other metal sulfide‐based photocatalysts for the degradation of PFAS.

Cadmium indium sulfide (CdIn₂S₄) is a promising visible‐light‐active photocatalyst which has not previously been investigated for photodegradation persistent per‐ and poly‐fluoroalkyl substances (PFAS). CdIn₂S₄ micro‐pyramids achieved complete photocatalytic degradation and mineralization of perfluorooctanesulfonic acid (PFOS) in both simulated lab sample and field‐derived mixed PFAS sample.

1. Introduction

Per‐ and poly‐fluoroalkyl substances (PFAS) are a large class of synthetic organo‐fluorine compounds with per‐/poly‐fluoro moiety (C_nF_2n+1‐) and a hydrophilic head group (e.g., carboxylic acid ─COOH, sulfonic acid ─SO₃H).^{[ 1 ]} Since the 1950s, PFAS have been extensively used in manufacturing surfactants and polymers for wide applications such as non‐stick coatings (e.g., Teflon), water‐repelling coatings (e.g., Scotch Guard), anti‐fogging coatings on glass/mirrors, anti‐corrosive coatings on metals (e.g., Zonyl FSP, a fluorosurfactant phosphate), wetting agents in paints, firefighting foams, textiles, leather treatments, adhesives, inks, insecticides, and cosmetics.^{[ 2} , ^{3 ]} Owing to these numerous applications, PFAS rapidly accumulated in the soil, sediments, surface water, and underground water. PFAS are persistent organic pollutants due to the strong electronegativity of F atoms and extremely strong C─F bonds with a high bond dissociation energy of 400–550 kJ mol⁻¹.^{[ 4} , ⁵ , ⁶ , ^{7 ]} PFAS have become an emerging environmental and health threat, and they may lead to the disruption of hormones, increased risk of cancers, higher cholesterol levels, and reproductive problems.^{[ 8} , ⁹ , ¹⁰ , ¹¹ , ^{12 ]} Perfluorooctanesulfonic acid (PFOS, C₈F₁₇SO₃H) and perfluorooctanoic acid (PFOA, C₇F₁₅COOH) are among the most widely used PFAS, and hence most commonly detected PFAS in the environment. The research into complete degradation and mineralization of PFAS is essential. This is aligned with target 6.3 of the United Nations’ Sustainable Development Goals (UN SDG‐6) for improving water quality by reducing pollution, eliminating pollutants, and enhancing safe water recycling.

Currently, research is focused on removing PFAS from water bodies using non‐destructive traditional water treatment methods (e.g., adsorption, ion exchange, reverse osmosis, and nanofiltration) followed by attempts to destroy PFAS. Incineration (thermal degradation) can be applied for PFAS destruction at high temperatures (>1000 °C).^{[ 13 ]} Unfortunately, incineration results in incomplete degradation of PFAS to smaller PFAS molecules in addition to toxic byproducts (gas or ash) that are released in the environment resulting in the spread of PFAS instead of complete destruction.^{[ 13 ]} For instance, Martin et al. ^{[ 14 ]} undertook a pilot study to inspect the distribution and concentration of PFAS in soil samples surrounding a hazardous waste incinerator and found that almost all the samples have PFOS (50–8300 ng kg⁻¹), PFOA (51–1300 ng kg⁻¹), and other short‐chain length PFAS. Among destructive technologies, semiconductor‐assisted photocatalysis is a promising green technology that can achieve complete degradation of PFAS molecules.^{[ 15 ]} Under light irradiation, semiconductors produce reactive species to achieve degradation through an oxidative degradation pathway via holes ($h_{VB}^{+}$) and hydroxyl radicals ( ^∙ OH) or reductive degradation pathway via photoelectrons ($e_{CB}^{-}$), superoxide radicals ($O_2^{.-}$), or hydrated electrons ($e_{aq}^{-}$). Reductive degradation pathway using $e_{CB}^{-}$ and $e_{aq}^{-}$ have been found to be more effective in PFAS degradation compared to oxidative degradation.^{[ 16} , ¹⁷ , ^{18 ]} Some semiconductor photocatalysts with high bandgap energies (E_g) have been investigated for the photocatalytic degradation of PFAS under ultraviolet (UV) light such as titanium oxide (TiO₂, E_g = 3.2 eV/388 nm),^{[ 19 ]} bismuth oxychloride (BiOCl, E_g = 3.3 eV/376 nm),^{[ 20 ]} gallium oxide (β‐Ga₂O₃, E_g = 4.8 eV/258 nm),^{[ 15 ]} hexagonal boron nitride (h‐BN, E_g = 6 eV/207 nm),^{[ 21 ]} and zinc sulfide (ZnS, E_g = 3.4 eV/365 nm).^{[ 22 ]} Due to their large bandgap, these semiconductors are only active upon illumination with UV light.

Recently, some research has been dedicated to the synthesis of visible‐light active photocatalysts such as cadmium sulfide (CdS, E_g = 2.4 eV/517 nm) for PFAS photodegradation.^{[ 16} , ²³ , ^{24 ]} The investigation of other visible light active photocatalysts toward PFAS degradation is important for large‐scale applications where sunlight is a cheap and abundant source of photons. Indium‐based ternary metal sulfides (MIn₂S₄, M = Zn, Cd, Co) have displayed promising photocatalytic performance under visible‐light irradiation in various applications.^{[ 25} , ²⁶ , ^{27 ]} There is currently only one research article investigating this promising class of visible‐light materials for PFAS photodegradation. Liu et al. ^{[ 24 ]} found the performance of ZnIn₂S₄ toward the photodegradation of sodium p‐perfluorous nonenoxybenzenesulfonate (OBS) compared to P25‐TiO₂ under simulated sunlight. After 24 h reaction, the OBS defluorination values were 7.6% and 1.7% for ZnIn₂S₄ and P25‐TiO₂, respectively, after 24 h‐reaction and the higher activity of ZnIn₂S₄ is logical due to its bandgap resulting in very few photons being absorbed. There is much room for investigating ZnIn₂S₄ and other MIn₂S₄ toward the photodegradation of different PFAS.

CdIn₂S₄ is a promising photocatalyst with visible‐light harvesting capacity (E_g = 2.0–2.4 eV/620–517 nm)^{[ 28 ]} and has not previously been investigated for PFAS photodegradation. Herein, we report the solvothermal synthesis of CdIn₂S₄ and investigate the photocatalytic performance of CdIn₂S₄ toward the photocatalytic degradation of PFOS for the first time. The PFOS removal and complete mineralization of PFOS have been investigated. The mechanism of photoelectron reduction is determined from radical quenching experiments using specific scavengers. Finally, CdIn₂S₄ was investigated in the treatment of a field‐derived PFAS sample.

2. Results and Discussion

2.1. Physicochemical Characterizations

CdIn₂S₄ was synthesized via a simple solvothermal method in an aqueous/ethanol solution in the presence of polyvinylpyrrolidone (PVP) as a surfactant (Figure 1a). The X‐ray diffraction (XRD) pattern of CdIn₂S₄ (Figure 1b) showed the synthesis of the cubic spinel phase of CdIn₂S₄ with the space group $Fd\overline{3}m$ (No. 227) compared to the reference card recorded on the Joint Committee on Powder Diffraction Standards (JCPDS, card No. 27–0060).^{[ 25} , ^{28 ]} The corresponding unit cell (Figure 1c) shows the CdIn₂S₄ spinel structure composed of tetrahedral CdS₄ and octahedral InS₆.^{[ 28 ]} Diffuse reflectance spectra (DRS) measurements (Figure 1d) showed the relatively wide optical absorption range of CdIn₂S₄ in the UV‐visible range of 200–570 nm compared to that of TiO₂ in the UV range of 200–390 nm. The calculated Tauc plots (inset of Figure 1d) revealed the bandgap values of the as‐synthesized CdIn₂S₄ and the commercial P25‐TiO₂ are ≈2.2 eV and ≈3.2 eV, respectively. These bandgap values are consistent with the bandgap energy ranges reported in the literature for both CdIn₂S₄ (2.0–2.4 eV)^{[ 29} , ^{30 ]} and TiO₂ (3.0–3.2 eV).^{[ 31} , ^{32 ]} Scanning electron microscopy (SEM) imaging was performed to investigate the micromorphology of the as‐synthesized CdIn₂S₄. SEM image (Figure 1e) showed that the as‐prepared CdIn₂S₄ sample exhibited a micro‐pyramid / octahedral morphology shown in Figure 1c. This morphology has been observed in some previous studies.^{[ 33} , ³⁴ , ^{35 ]} Figure S1a,b (Supporting Information) demonstrates the particle size distribution of the as‐synthesized CdIn₂S₄ micro‐pyramids with a mode particle size of 868 nm. Scanning transmission electron microscopy (STEM) images demonstrate the micro‐pyramids surrounded by some nanosheets as shown in the inset of Figure 1e and Figure S1c–e (Supporting Information).

Figure 1.

a) Schematic diagram of solvothermal synthesis of CdIn₂S₄ b) the XRD pattern of as‐prepared CdIn₂S₄ matched with the reference Cubic CdIn₂S₄ (JCPDS card #27‐0060), d) the polyhedral unit cell structures of cubic CdIn₂S₄ with the $Fd\overline{3}m$ space group showing octahedral InS₆ and tetrahedral CdS₄ (Created by VESTA software), d) the UV–vis absorption spectra (based on K‐M transform) of CdIn₂S₄ and P25‐TiO₂ (inset: the corresponding Tauc plots of CdIn₂S₄ and P25‐TiO₂, and e) SEM image showing the octahedral/micro‐pyramidal structure of the as‐synthesized CdIn₂S₄ (inset: STEM image showing micro‐pyramids surrounded by nanosheets).

2.2. Kinetics of PFOS Degradation

The photocatalytic activity of CdIn₂S₄ was investigated toward the photodegradation of PFOS in an N₂‐purged aqueous solution in the presence of triethanolamine (TEOA, a hole scavenger) at different irradiation conditions. These conditions were inspired by work by Arima et al. ^{[ 16 ]} investigating PFOS degradation with CdS quantum dots. The PFOS concentration was measured using liquid chromatography‐mass spectrometry (LC‐MS). The PFOS removal% was estimated using Equation (1):

$$PFOSremoval\%=\frac{{\left[PFOS\right]}_o-{\left[PFOS\right]}_t}{{\left[PFOS\right]}_o}\times 100$$ (1)

where [PFOS] _o and [PFOS] _t are the initial PFOS concentration and after a certain time (t) of irradiation, respectively.

The conditions (light source and light intensity) were optimized as shown in Figures S2, S3 (Supporting Information) and more details are stated in the supporting information. We then investigated the kinetics of the PFOS photodegradation and determined the order of the reaction (n) and the observed rate constant (k). Time‐series experiments were performed to follow the variation of PFOS concentration over the photocatalytic reaction using CdIn₂S₄. Figure 2a,b shows the fitting of data in the [PFOS] versus time and [PFOS]/[PFOS]_o versus time plots, which follow an exponential decay behavior indicating it is not a zero‐order reaction. The linear fitting of ln[PFOS] versus time plot (Figure 2c) indicates that the photocatalytic removal of PFOS follows first‐order kinetics with an observed rate constant (k_obs ) of ≈0.2 h⁻¹. The optimized power intensity (1197 mW cm⁻² using 2 × 365 nm LEDs) showed the highest observed rate constant for PFOS removal compared to the other conditions (Figure S3, Supporting Information). Figure 2d shows the temporal increase of PFOS removal% with the progress of the photocatalytic reaction until reaching almost complete PFOS removal. The complete degradation of PFOS is accompanied by the production of fluoride ions resulting from the breaking of C─F bonds. The quantification of the produced F⁻ ions was completed using an F⁻ ion selective electrode (F‐ISE). PFOS defluorination% and mineralization% can be estimated using Equations (2) and (3):

$$Defluorination\%=\frac{{\left[F^{-}\right]}_t-{\left[F^{-}\right]}_0}{n_{C-F}\left({\left[PFOS\right]}_o-{\left[PFOS\right]}_t\right)}\times 100$$ (2)

$$Mineralization\%=\frac{{\left[F^{-}\right]}_t-{\left[F^{-}\right]}_0}{n_{C-F}{\left[PFOS\right]}_o}\times 100$$ (3)

where [ F ⁻] _t and [ F ⁻]₀ are the molar concentration of fluoride produced after a certain time (t) of irradiation and the initial molar concentration of fluoride, respectively, [PFOS] _o (M) and [PFOS] _t (M) are the initial PFOS concentration and concentration after a certain time (t) of irradiation, respectively, and n _{C − F} is the number of C─F bonds per molecule (17 for PFOS, C₇F₁₇SO₃H).

Figure 2.

The temporal change in PFOS concentration using CdIn₂S₄ in the presence of TEOA under 365 nm LED irradiation: a) ([PFOS] versus time plot (inset: photograph of the experimental setup of the quartz cell reactor and the 2 × 365 nm LEDs), b) [PFOS]/[PFOS]_o versus time plot, c) ln[PFOS] versus time plot, and d) the temporal increase of PFOS removal% with the progress of the photocatalytic reaction and the defluorination% at the end of the reaction. (Experimental conditions: photocatalyst = 5 mg CdIn₂S₄, PFOS solution volume = 1 mL, [PFOS]_o = ∼40 ppm, [TEOA] = 50 g L⁻¹, light = 2 × 365 nm LEDs, 1197 mW cm⁻²). The error bars were obtained upon applying a regression analysis with a 95% confidence in the slope of the calibration curves and the rules of propagation of errors to all calculations.

After 24 h of reaction, the concentration of produced F⁻ ions was 23 ± 5 ppm corresponding to a defluorination% of 97 ± 22%. This high defluorination% suggests the complete degradation of C─F bonds in the degraded PFOS molecules. This is consistent with the LC‐MS results (Table S1, Supporting Information) that did not show any increase in the concentration of smaller PFAS (C_nF_2n+1SO₃H or C_nF_2n+1COOH, n = 7,6,5,4). Overall, we have observed near complete defluorination and mineralization at 24 h using CdIn₂S₄. At the beginning of the reaction (usually 30–90 min), the removal of PFOS can be attributed to tandem adsorption processes and photodegradation. The defluorination at the end of the photocatalytic reaction is evidence of PFOS removal via photodegradation while no defluorination was observed in the dark control (without irradiation) or without catalyst control (irradiation only) as shown in Figure S4 (Supporting Information).

2.3. Mechanism of PFOS Photodegradation

Specific scavengers/trapping agents are used in a set of radical quenching experiments to investigate their influence on photocatalytic activity and to assess the crucial reactive species participating in the photodegradation process. TEOA is a hole scavenger while methyl viologen dichloride hydrate (MV) is an electron scavenger. Figure 3a shows the effect of using scavengers on the PFOS removal% and defluorination% using CdIn₂S₄ after 24 h reaction under 2 × 365 nm LEDs irradiation (1197 mW cm⁻²). The molar concentration of scavengers was maintained constant during this set of experiments ([TEOA] = [MV] = ≈0.33 M, corresponding to ≈50 g L⁻¹ TEOA and ≈85 g L⁻¹ MV).

Figure 3.

a) Effect of scavengers on the PFOS removal% and defluorination% using CdIn₂S₄ under 2 × 365 nm LEDs irradiation (1197 mW cm⁻²) and b) the proposed reductive photodegradation of PFOS using the photoinduced e ⁻ at the CB of CdIn₂S₄. (Experimental conditions: photocatalyst = 5 mg CdIn₂S₄, PFOS solution volume = 1 mL, [PFOS]_o = ∼40 ppm, [TEOA] = 50 g L⁻¹, [MV]= 85 g L⁻¹, light = 2 × 365 nm LEDs, 1197 mW cm⁻²). The error bars were obtained upon applying a regression analysis with a 95% confidence in the slope of the calibration curves and the rules of propagation of errors to all calculations.

The observed removal% can be attributed to the combination of adsorption processes and degradation. A high removal% with no defluorination was obtained in the dark control (without light irradiation). The adsorption equilibrium usually occurs after 60–90 min. The defluorination is attributed to photocatalytic reactions that occur using the reactive species produced by CdIn₂S₄. Arima et al. ^{[ 16 ]} found that complete PFOS removal% occurs in the first hour via the adsorption onto CdS quantum dots capped with mercaptopropionic acid ligands while the defluorination process occurs gradually with the progress of the photocatalytic reaction. The standard conditions using TEOA showed significant defluorination (105 ± 40%) within 24 h of photocatalytic reaction using CdIn₂S₄. However, in the absence of TEOA, the CdIn₂S₄ photocatalysts did not achieve any defluorination of PFOS (although there was 71 ± 14% removal via adsorption). Similarly, the use of MV (an electron scavenger) in the absence of TEOA achieves a negligible defluorination% (1.77 ± 0.74%). These results show that the oxidative species (e.g., h ⁺, ^∙ OH) did not contribute to the PFOS photodegradation over CdIn₂S₄ micro‐pyramids. The approximate valence band (VB) and conduction band (CB) edge potentials of CdIn₂S₄ are +1.44 and −0.76 eV versus standard hydrogen electrode (SHE), respectively, based on the experimentally measured E_g and the absolute electronegativity of CdIn₂S₄ (more details are shown in Section  4 ). This low oxidation power of CdIn₂S₄ (E_VB = +1.44 eV versus SHE) produces h⁺ , which are unable to oxidize C─F bonds in PFOS that require a minimum potential of +2 eV as suggested by Park et al. ^{[ 36 ]} It is unlikely that any ^∙ OH is participating in the PFOS photodegradation, and this can be attributed to their formation requiring an oxidation potential of +2.8 eV.^{[ 37 ]}

However, a significant reduction (≈9.4 fold) in the defluorination% from 105 ± 40% to 11 ± 4% was observed upon using a combination of TEOA/MV. This indicates that the photoinduced electrons (e⁻ ) at the CB of CdIn₂S₄ are crucial for the photocatalytic degradation of PFOS using CdIn₂S₄.

The presence of TEOA is essential to trap h⁺ and decrease the e⁻ ‐h⁺ recombination so electrons could be more available for the reduction of C─F bonds. Van Hoomissen and Vyas^{[ 38 ]} used density functional theory (DFT) to investigate the reductive defluorination of perfluoroalkyl substances and found that the reduction potential of the C─F bonds (near the head group) was ≈−1.1 eV and more negative (≈−1.5 eV) for bonds further down the chain. The potential of C─F bond reduction has been experimentally determined by Park et al. ^{[ 36 ]} to be ≈−1 eV, which is slightly higher than our approximated estimated CB potential (−0.76 eV). The approximated value of CdIn₂S₄ could be slightly underestimated, the reduction potential of PFOS could be overestimated, or band bending at the solution interface may alter the CdIn₂S₄ reduction potential. Despite the apparent mismatch in reduction potential, we observe high defluorination% (≈100%). Based on these experiments, we propose that the mechanism of the PFOS degradation is through reduction via the photoinduced e ⁻ using CdIn₂S₄ micro‐pyramids while TEOA consumes h⁺ as shown in Figure 3b.

2.4. CdIn₂S₄ versus TiO₂

The photocatalytic performance of CdIn₂S₄ toward PFOS removal and defluorination was compared to commercial P25‐TiO₂ under 365 nm irradiation (which is suitable for the bandgaps of both photocatalysts, Figure 1d). Figure 4a shows that TiO₂ achieved low PFOS removal% (19.51 ± 7.33%) and low defluorination% (10.45 ± 6.25% corresponding to a negligible mineralization% of 2.04 ± 0.96%) under the standard optimized conditions (TEOA, 24 h reaction, 2 × 365 nm LEDs irradiation, 1197 mW cm⁻²). CdIn₂S₄ achieved a complete PFOS removal% (≈100% ± 8) and complete defluorination%/mineralization% (105 ± 39%). Defluorination% and mineralization% are the same in the case of PFOS degradation using CdIn₂S₄ because of complete PFOS removal while they are different in the case of using TiO₂ because of the low PFOS removal%. These results reveal the outstanding photocatalytic activity of CdIn₂S₄ compared to TiO₂. This could be attributed to the higher reduction power of CdIn₂S₄ compared to TiO_2, especially since the PFOS degradation depends on a direct reductive degradation pathway in this system. The approximate VB and CB edge potentials of TiO₂ are 2.91 and −0.29 eV versus SHE based on the experimentally measured E_g and the absolute electronegativity of TiO₂. The higher reduction potential of CdIn₂S₄ (−0.76 eV versus SHE) than that of TiO₂ (−0.29 eV versus SHE) is most likely responsible for the difference in PFOS degradation (Figure 4b). The low (but non‐zero) defluorination achieved could be attributed to the high oxidation potential of TiO₂ that could oxidize C─F bonds, either directly or via producing hydroxyl radicals^{[ 39 ]} (Figure 4b).

Figure 4.

a) Comparison of the photocatalytic performance of TiO₂ and CdIn₂S₄ toward the PFOS removal%, defluorination% and mineralization% in the presence of TEOA under 2 × 365 nm LEDs irradiation (1197 mW cm⁻²) and b) the energy diagram comparing TiO₂ and CdIn₂S₄ edge potentials (versus SHE) with oxidation and reduction potential of C─F bonds. (Experimental conditions: photocatalyst = 5 mg CdIn₂S₄ or TiO₂, PFOS/TEOA solution volume = 1 mL, [PFOS]_o = ∼40 ppm, [TEOA] = 50 g L⁻¹, light = 2 × 365 nm LEDs, 1197 mW cm⁻²). The error bars were obtained upon applying a regression analysis with a 95% confidence in the slope of the calibration curves and the rules of propagation of errors to all calculations.

2.5. Photodegradation of Field‐Derived PFAS Sample

The photocatalytic activity of CdIn₂S₄ was investigated toward the photodegradation of a field‐derived PFAS sample. The PFAS sample was obtained from a groundwater treatment plant contaminated by aqueous film‐forming foam (AFFF) in South Australia. LC‐MS analysis of the as‐received AFFF‐contaminated water sample showed the presence of various PFAS with a total concentration of ≈150 ppb. Concentration of PFAS is a typical step to have a high PFAS amount in a low volume instead of treating a high volume of diluted PFAS resulting in higher operating costs.^{[ 40 ]} Figure 5a shows a schematic of the pre‐treatment procedure employed for PFAS concentration: 1) PFAS capture from groundwater via adsorption using granular activated carbon (GAC), 2) PFAS elution and concentration from GAC via solvent extraction and distillation.^{[ 40 ]} Before the photodegradation experiments, TEOA (50 g L⁻¹) was mixed with the as‐prepared concentrated PFOS sample, and the total PFAS concentration was ≈50 ppm in the final field‐derived PFAS/TEOA solution. LC‐MS analysis of the final field‐derived PFAS solution includes at least 8 different PFAS molecules as shown in Figure 5 and Table S2 (Supporting Information). The two main PFAS in this sample were perfluorohexanesulfonic acid (PFHxS, C₆F₁₃SO₃H, 25.43 ± 1.38 ppm) and PFOS (12.78 ± 1.64 ppm). There are other PFAS molecules were detected in concentration (>1 ppm) such as perfluorobutanesulfonic acid (PFBS, C₄F₉SO₃H, 1.80 ± 0.13 ppm), perfluoropentanesulfonic acid (PFPeS, C₅F₁₁SO₃H, 3.09 ± 0.19 ppm), and perfluoroheptanesulfonic acid (PFHpS, C₇F₁₅SO₃H, 1.53 ± 0.11 ppm).

Figure 5.

a) Scheme of the treatment of the photodegradation of different PFAS in a field‐derived PFAS sample collected from aqueous film‐forming foam (AFFF) contaminated water in the presence of TEOA using CdIn₂S₄ micropyramids under 365 nm LED irradiation. (Experimental conditions: photocatalyst = 5 mg CdIn₂S₄, Field‐derived PFAS/TEOA solution volume = 1 mL, total [PFAS]_o = ∼50 ppm, [TEOA] = 50 g L⁻¹, light = 365 nm LEDs, 813 mW cm⁻²). The error bars were obtained upon applying a regression analysis with a 95% confidence in the slope of the calibration curves and the rules of propagation of errors to all calculations.

The CdIn₂S₄ micro‐pyramids showed high total PFAS removal% (78 ± 7%) and complete defluorination% (109 ± 42%) of the removed PFAS. This reveals the capability of the as‐synthesized CdIn₂S₄ micro‐pyramids toward the defluorination of a field‐derived sample containing PFAS and other PFOS. Focusing on the main PFAS in the field‐derived sample (>1 ppm). Figure 5b shows that the PFAS removal depends on the carbon chain length of PFAS. As the PFAS chain length increases, the PFAS removal% increases as follows: PFOS (C₈) > PFHpS (C₇) > PFHxS (C₆) > PFPeS (C₅) >>> PFBS (C₄). This is attributed to the long‐chain PFAS being more hydrophobic and less soluble in water than the short‐chain PFAS. These hydrophobic long‐chain PFAS have a higher tendency to be adsorbed on the surface of the catalyst. This leads to PFAS degradation and defluorination due to the stronger interaction between the adsorbed PFAS molecules with the reactive species produced on the catalyst. Future work may look toward altering the surface ligand of the photocatalyst to enhance the adsorption of shorter‐chain PFAS.

2.6. Stability of CdIn₂S₄ and Future Work

The photocatalytic activity and stability of CdIn₂S₄ were investigated over 3 cycles. Initially, 10 mg of CdIn₂S₄ was mixed with 2 mL of PFOS/TEOA. After each photocatalytic cycle, the post‐reaction suspension was centrifuged and the filtrate was collected for F‐ISE, fluorine‐nuclear magnetic resonance (F‐NMR), and inductively coupled plasma mass spectrometer (ICP‐MS) measurements. The catalyst was separated, washed 3 times with MilliQ water by centrifugation, and left for drying overnight in the fume hood. The collected catalyst is re‐dispersed in fresh 2 mL PFOS/TEOA solution for the subsequent photocatalysis cycle. Figure 6a shows that the amounts of F⁻ produced after the first, second, and third cycles were about 7 ± 1, 18 ± 2, and 12 ± 2 ppm (F‐ISE measurements), respectively. A similar trend for produced F⁻ was confirmed from F‐NMR measurements (Figure S5, Supporting Information). The low yield of F⁻ after the first cycle may be because of the incomplete PFOS removal as suggested by F‐NMR measurements (Figure S5, Supporting Information) and this sample requires activation to produce a high amount of F⁻ produced after the second photocatalytic cycle. The decrease in produced F⁻ after the third cycle may be attributed to catalyst degradation. ICP‐MS analysis was used to follow the leaching of catalyst elements (Cd, In, and S). Figure 6b shows the leaching of elements over the three cycles. The total element leaching% after three photocatalytic cycles was 2.6 ± 0.7% for Cd, 2.4 ± 0.7% for In, and 10.0 ± 2.7% for S. This indicates the catalyst degradation in the reaction medium including high pH, high [F⁻], and reactive radical species. The morphology of the residual catalyst after the photocatalysis cycles was investigated by SEM. Figure 6c and Figure S6 (Supporting Information) show the deformation of some CdIn₂S₄ micro‐pyramids confirming the degradation of some catalyst particles. These results confirm that the reduction in the performance of the CdIn₂S₄ catalyst toward PFOS defluorination can be attributed to catalyst degradation and metal leaching with multiple uses.

Figure 6.

Recyclability of CdIn₂S₄. a) The produced [F⁻] over 3 cycles (estimated from F‐ISE), b) the element leaching % in the post‐reaction solutions over 3 cycles (estimated from ICP‐MS measurements), and c) SEM image of the post‐reaction catalyst. (Experimental conditions: photocatalyst = 10 mg CdIn₂S₄, PFOS/TEOA solution volume = 2 mL, [PFOS]_o = ∼40 ppm, [TEOA] = 50 g L⁻¹, light = 2 × 365 nm LEDs, 1197 mW cm⁻²). The error bars were obtained upon applying a regression analysis with a 95% confidence in the slope of the calibration curves and the rules of propagation of errors to all calculations.

Sulfur/metal leaching is commonly observed for CdIn₂S₄ ^{[ 41 ]} and numerous other metal sulfides including CdS,^{[ 42 ]} ZnIn₂S₄,^{[ 43 ]} Co₉S₈,^{[ 44 ]} and Ni₃S₂‐Co₉S₈.^{[ 45 ]} Inhibition of sulfur/metals leaching and/or post‐reaction treatment/separation (e.g., ion exchange, precipitation, adsorption, etc.) is essential for practical application to remove any residual metals in the treated water samples. As this is the first work discussing the use of CdIn₂S₄ toward the PFAS degradation, there is room to optimize the reaction conditions (e.g., varying the hole scavenger, reaction pH, etc.) to enhance the photocatalytic performance and photocatalyst stability. Generally, the enhancement of the stability of the CdIn₂S₄‐based photocatalysts could be investigated through formation heterojunctions (e.g., Z‐scheme, type II), using co‐catalysts, and support materials.^{[ 46 ]} For instance, UiO‐66/CdIn₂S₄ ^{[ 47 ]} and CQDs/TiO₂/CdIn₂S₄ ^{[ 48 ]} were used in other photocatalytic applications and the post‐reaction samples showed stability and minor metal leaching.

In addition to enhancing stability, the formation of heterojunctions to a significant enhancement in the photocatalytic activity compared to bare CdIn₂S₄.^{[ 46} , ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ^{52 ]} The activity of the photocatalysts could also be improved through changes to the morphology of the CdIn₂S₄ in order to achieve less recombination during photocatalysis, but also to increase PFAS adsorption to the surface.^{[ 33} , ³⁵ , ⁵³ , ^{54 ]} For instance, Zhang et al. ^{[ 33 ]} fabricated CdIn₂S₄ with a hybrid morphology of octahedra (or micro‐pyramids) and nanosheets, forming an S‐scheme homojunction with enhanced charge separation and photocatalytic performance compared to the bare nanosheets and bare octahedra. Coupling CdIn₂S₄ with relatively high‐surface‐area carbon materials (e.g., graphene, carbon nanotubes, metal carbides, metal‐organic‐frameworks) could enhance both the adsorption of PFAS molecules and their photodegradation via enhancing the charge separation compared to bare CdIn₂S₄.^{[ 46} , ^{55 ]} To sum up, there is much room to investigate the above‐mentioned strategies to enhance the performance/stability of CdIn₂S₄‐based photocatalysts toward PFAS degradation.

3. Conclusion

Cubic crystalline CdIn₂S₄ micro‐pyramids were synthesized using a solvothermal method. CdIn₂S₄ showed promising photocatalytic performance with almost complete removal and defluorination of PFOS in 24 h. Radical quenching experiments revealed that PFOS degradation follows a reductive degradation pathway and photoexcited e⁻ are the key reactive species. CdIn₂S₄ showed outstanding PFOS removal%/ defluorination% compared to TiO₂ due to the relatively higher reduction potential. Finally, CdIn₂S₄ showed similar outstanding performance toward the degradation of field‐derived PFAS samples containing PFOS and other smaller PFAS (PFAS removal% is proportional to the chain length). The stability of CdIn₂S₄ was investigated over three cycles and there is a significant sulfur leaching (≈10%) and minor Cd/In leaching (≈2%). The stability of CdIn₂S₄ can be enhanced by the construction of heterojunctions, anchoring co‐catalysts, and/or using supporting material. Finally, this work sheds light on the use of other indium‐based ternary metal indium sulfides (MIn₂S₄, M = Zn, Co, Mn, Fe, Mg, etc.) that may be promising candidates for PFAS photodegradation due to their high reduction potential.

4. Experimental Section

Chemicals and Materials

Ammonium acetate (CH₃COONH₄, ≥99.99%), cadmium chloride (CdCl₂, ≥99%), indium chloride (InCl₃, 98%), iron (III) acetylacetonate (Fe(acac), ≥99.9%), P25 titanium (IV) oxide (P25‐TiO₂, ≥99.5%), polyvinylpyrrolidone (PVP, (C₆H₉NO)_n, M.W. = 360 000 g mol⁻¹), thioacetamide (CH₃CSNH₂, ≥99%), triethanolamine (TEOA, N(CH₂CH₂OH)₃, 99%), trifluoracetic acid (TFA, CF₃COOH, 99%), and methyl viologen dichloride hydrate (MV, 98%) were purchased from Sigma‐Aldrich. Perfluorooctanesulfonic acid (PFOS) was received as a gift from Prof. Duan Losic (School of Chemical Engineering, the University of Adelaide). Deuterium oxide (D₂O, 99.9%) was purchased from Cambridge Isotope Laboratories. Methanol (UNICHROM, HPLC‐grade) was purchased from Ajax Finechem (part of Thermo Fischer Scientific). Ethanol (100% undenatured) and total ionic strength adjustment buffer II solution (TISAB II, pH 5.3‐5.5 at 25 °C) were purchased from ChemSupply. Native perfluorinated compound solution/mixture (PFAC‐MXC) containing 13 native perfluoroalkyl carboxylic acids (C₄‐C₁₄, C₁₆, and C₁₈) and 8 native perfluoroalkanesulfonates (C₄‐C₁₀ and C₁₂) with a chemical purity of >98% was purchased from Wellington Laboratories. Mass‐labeled perfluorinated compound injection standards solution (MPFAC‐C‐IS) containing a solution/mixture of ¹³C‐labeled perfluoroalkyl carboxylic acids (M3PFBA, M2PFOA, and MPFDA) and a ¹³C‐labeled perfluorooctanesulfonates (MPFOS) with a chemical purity of >98% and an isotopic purity of ≥99% was purchased from Wellington Laboratories. Milli‐Q water was obtained from Milli‐Q IQ 700 (Merck) and used to prepare all aqueous solutions.

Synthesis of CdIn₂S₄

The solvothermal method is the most commonly used synthesis route of CdIn₂S₄ because 1) it was a simple solution‐based synthesis method performed at relatively lower temperatures (<200 °C) compared to solid‐sate synthesis routes that require high temperatures (>800 °C) and 2) It was a bottom‐up synthesis approach that could be controlled to tune the morphology of the target nanoparticles through adjusting the reaction conditions such as solvent composition, reaction time, and surfactants (type/concentration).^{[ 46 ]} Hence, three batches of CdIn₂S₄ were prepared via a one‐pot solvothermal method^{[ 28 ]} as summarized in Figure 1a. In ≈100 mL Teflon reactor, 1.2 mmol (220 mg) of CdCl₂, 2.4 mmol (532 mg) of InCl₃, 9.6 mmol (720 mg) of thioacetamide (in excess amount), and 600 mg of PVP (a surfactant) were dissolved in 50 mL of a mixed solvent of ethanol and Milli‐Q water (50/50 V/V). After vigorous stirring for 30 min, the Teflon reactor was put in a 100 mL stainless‐steel autoclave for the solvothermal process, which was kept at 180 °C for 24 h, where the heating rate was adjusted to be 2.3 °C min⁻¹ and the cooling rate was adjusted to be 10 °C min⁻¹. The samples were collected and washed several times with Milli‐Q water and pure ethanol via centrifugation and decantation. Finally, the samples were dried overnight at 60 °C and the dried samples from the three batches were mixed and ground in an agate mortar to obtain fine powder.

Materials Characterization

The crystalline phase of CdIn₂S₄ was investigated from XRD patterns measured on a Rigaku Miniflex powder X‐ray diffractometer using Cu Kα radiation. The morphology of the as‐prepared CdIn₂S₄ was investigated using an ultra‐high‐resolution scanning electron microscope (Hitachi SU7000 UHR‐SEM). Particle size measurements were performed using ImageJ software. The morphology of CdIn₂S₄ was further investigated by scanning transmission electron microscopy (STEM) images, which were recorded on a FEI Titan Themis 80–200 TEM. DRS measurements were performed on a Cary 5000 UV–vis–NIR spectrophotometer equipped with a Praying Mantis Diffuse Reflection Accessory (Harrick, DRP‐SAP) where a PTFE disc was used as reference. The reflectance measurements were obtained in the wavelength range from 200 to 800 nm and the Kubelka‐Munk transform was applied to the data and used to calculate the bandgap of the catalyst. Approximated values of VB and CB edge potentials of a semiconductor could be estimated based on its absolute electronegativity and measured optical bandgap energy using Equations (4) and (5):^{[ 31} , ^{56 ]}

$$E_{\mathrm{VB}}={\chi }_{\mathrm{sc}}-E_e+\frac{1}{2}{E}_g^{\mathrm{opti}\mathrm{cal}}$$ (4)

$$E_{CB}=E_{VB}-E_g^{optical}$$ (5)

where E_VB and E_CB are the edge potentials of the VB and CB of a semiconductor versus the standard hydrogen electrode (SHE), respectively, χ _sc is the absolute electronegativity of the semiconductor, $E_g^{optical}$ is the optical bandgap energy in eV, and E_e is the energy of the free electron (4.5 eV versus SHE). χ _sc is the geometric mean of the electronegativity values of the isolated component atoms.^{[ 56 ]} χ _sc of CdIn₂S₄ and TiO₂ were 4.84 eV and 5.81 eV, respectively. χ _sc values were calculated using Equation (6):^{[ 57 ]}

$${\chi }_{SC}\approx {\left[{\chi }_A^a\times {\chi }_B^b\times {\chi }_C^c\right]}^{1/\left(a+b+c\right)}$$ (6)

where χ _A , χ _B , and χ _C are the absolute electronegativity values of elements (A, B, & C) constituting the semiconductor, and a, b, and c are the stoichiometric coefficients of these elements in the semiconductor molecular formula. For example, using the χ values of Cd, In, and S as 4.33, 3.10, and 6.22 eV, respectively, then ${\chi }_{CdI{n}_2{S}_4}={({4.33}^1\times {3.1}^2\times {6.22}^4)}^{1/(1+2+4)}=4.84$ eV. Similarly, ${\chi }_{Ti{O}_2}={({3.45}^1\times {7.45}^2)}^{1/(1+2)}=5.81$ eV.

Photocatalytic Experiments

A ≈3.5 mL quartz cuvette (CV10Q35EP, Thorlabs, USA) with a lid containing a PTFE septum was employed as a batch‐style reactor for photocatalytic degradation of PFOS. Typically, 5 mg of CdIn₂S₄ were dispersed in 1 mL of an aqueous solution containing ~40 ppm PFOS and 50 g L⁻¹ TEOA (a hole scavenger/electron donor). Before the irradiation, the reactor was purged with nitrogen gas for ≈15 min. Then the reactor was irradiated for 24 h. Initially, different light conditions were used: 1 × 365 nm LED, 2 × 365 nm LEDs (UVTao Yuan UV‐LED Pens with a diameter of 12 mm), or 300 W Xe light (Beijing Perfect Light 300 W Xenon, equipped with a 400 nm cutoff filter, 815 mW cm⁻²). The emission spectra of the used light sources are demonstrated in Figure S2. The 2 × 365 nm LED pens (UVET UV‐LED Light source) were operated at different power %: 25% (286 mW cm⁻²), 50% (742 mW cm⁻²), 75% (1197 mW cm⁻²), and 100% (1626 mW cm⁻²). The post‐reaction solution was centrifuged, and the liquid layer was collected for further analysis.

The quantification of PFAS concentration was determined via LC‐QQQ‐MS analysis using an Agilent 1290 Infinity II Series high‐performance LC with negative ion electrospray ionization (ESI) coupled with Agilent 6495B triple quadrupole mass spectrometer (QQQ‐MS). An Agilent ZORBAX Eclipse Plus C18 UHPLC guard column (2.1 × 5 mm × 1.8 µm) was used connected pre‐injection to reduce background PFAS, and an Agilent ZORBAX Eclipse Plus C18 Rapid Resolution (3.0 × 50 mm × 1.8 µm) was used as the analytical column. The analytical column was heated to 40 °C, and the experiments were conducted at a constant flow rate of 0.4 mL min⁻¹. The aqueous solvent (solvent A) was ammonium acetate (20 mM), and the organic solvent (solvent B) was HPLC grade MeOH. The A/B solvents ratio varied over the experiment starting from 90:10% (0–0.5 min), 45:55% (0.5–2.5 min), 30:70% (2.5–7.5 min), 1:99% (7.5–16 min) and returning to the starting ratio 90:10% at the end of the run (16 min). Multiple reaction monitoring (MRM) was used for quantification and peak areas were standardized with the use of ¹³C labeled internal standard (sodium perfluoro‐1‐1(1,2,3,4‐¹³C₄)octanesulfonate, MPFOS, [¹³C₄ ¹²C₄F₁₇SO₃]⁻→[FSO₃]⁻, MRM: 503→99). The PFOS removal% was estimated using Equation (1).

The complete defluorination and mineralization of PFOS can be estimated by quantifying the amount of fluoride ions (F⁻) produced during the photocatalytic degradation reaction. Produced F⁻ ions were quantified using a fluoride ion selective electrode (F^_ISE). F^_ISE measurements were performed using a Thermo Scientific Orion fluoride ion selective electrode (9609BNWP Combination fluoride electrode). F‐ISE measurements can be affected by changes in pH and ionic strength. In these measurements, 200 µL of the filtrates were diluted with TISAB II buffer solution (1:1 volume ratio). TISAB II solution has three functions: 1) to buffer pH around 5.3–5.5 to avoid inaccurate measurement due to the formation of hydrogen fluoride complexes (e.g., HF, HF₂ ⁻) at low pH values or the presence of hydroxide ions at high pH, 2) to buffer pH and equalize the ionic strength in samples and standards, and 3) break metal‐fluoride complexes and complexation with the metals that can bind with fluoride using the chelating agent (e.g., CDTA) in TISAB II solution.^{[ 58 ]} Based on the calibration curve constructed, the amount of F⁻ is estimated and the PFOS defluorination% and mineralization% can be estimated using Equations (2) and (3).

In some experiments, fluorine‐19 nuclear magnetic resonance (F‐NMR) was used for F⁻ quantification using a Bruker 600 MHz NMR spectrometer. A 600 µL sample of the filtrate was mixed with 60 µL of 44 mm CF₃COOH/D₂O solution (internal standard solution) and 40 µL of 71 mm Fe(acac)/methanol (a relaxation agent). The NMR spectra were collected using a ¹⁹F no decoupling experiment (size of fid = 131 072, spectral width = 233.024 ppm, frequency offset = −100 ppm, relaxation delay = 1 s, number of scans = 512, pulse = 12µs, and sample temperature = 298 K). The peak of the internal standard is ≈−75 ppm, the organic terminal ‐CF₃ peak at ≈−80.5 ppm, and the F⁻ peak at ≈−120.5 ppm (Figure S5, Supporting Information). Based on the constructed calibration curves, the [PFOS] concentration and produced [F⁻] can be determined, and PFOS removal% and defluorination% can also be estimated using Equations (2) and (3), respectively, as described above.

An inductively coupled plasma mass spectrometer (ICP‐MS, Agilent 8900x QQQ‐ICP‐MS) was used to investigate the leaching of catalyst elements (Cd, In, and S) in the post‐reaction solutions. The initial PFOS/TEOA solution and the post‐reaction solutions were acidified and diluted (×100 times) by 2%HNO₃.

Statistical Analysis

The error analysis of the obtained data was performed to estimate the uncertainty of the experimentally measured data. Based on the fitting of calibration curves constructed for LC‐MS analysis, ISE measurements, and F‐NMR analysis, a regression analysis with a 95% confidence to the slope was performed. This gives ± absolute error in the measured quantity (e.g., 50 ± 2 ppm). Propagation of errors was applied in all calculations following standard procedures.^{[ 59 ]}

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Hamza M. A., Keltie A. J., Matthews R. K., Day M. L., and Shearer C. J., “CdIn₂S₄ Micro‐Pyramids for Reductive Photocatalytic Degradation of Perfluorooctanesulfonic Acid.” Small 21, no. 36 (2025): 21, e04601. 10.1002/smll.202504601

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.