Reductive Degradation of CCl₄ by Sulfidized Fe and Pd-Fe Nanoparticles: Kinetics, Longevity, and Morphology Aspects

Abstract

In this study a systematic comparison in morphology, long-term degradation, regeneration and reuse were conducted between palladized and sulfidized nanoscale zero-valent iron (Pd-Fe and S-Fe). Pd-Fe and S-Fe were prepared, after the synthesis of precursor Fe⁰ nanoparticles (spherical, ~35 nm radius) for carbon tetrachloride (CTC) treatment. With HAADF-TEM-EDS characterization, dispersive Pd islets were found on the Fe core of Pd-Fe. However, the Fe core was covered by the FeS_x shell of S-Fe (FeS/FeS₂ = 0.47). With an excessive Pd dose (10 mol%), the Pd-Fe were dramatically deformed to dendritic structures which significantly decreased reactivity. For CTC degradation, Pd-Fe (0.3 atomic% Pd) increased the degradation rate by 20-fold (k_sa= 0.580 Lm⁻²min⁻¹) while S-Fe presented a greater life time. The major intermediate chloroform (CF) was further degraded and less than 5% CF was observed after 24 h using Pd-Fe or S-Fe while above 50% CF remained using Fe. During aging, the Fe core was converted to FeOOH and Fe₃O₄/γ-Fe₂O₃. The restoration of Fe⁰ was achieved using NaBH₄, which regenerated Fe and Pd-Fe. However, the formed FeS_x shell on S-Fe was disappeared. The results suggest that S-Fe extends longevity of Fe, but the loss of FeS_x after aging makes S-Fe eventually perform like Fe in terms of CTC degradation.

Graphical Abstract

1. Introduction

Zero-valent iron (ZVI) has been widely studied for decades, especially in the treatment of water contaminants, such as chlorinated organic compounds [1–5], dyes [6–9] and heavy metals [10–13]. In industrial applications, as filling materials in permeable reactive barrier (PBR), more than 200 ZVI-PBRs have been installed worldwide by 2014 [14]. However, the long-term performance of ZVI has been reported to be limited by water corrosion and subsequently oxidation/surface passivation in groundwater [15–17]. Field tests using subsurface injection methods have also revealed some obstacles including particle agglomeration and lack of mobility [18, 19].

In the area of nanoscale ZVI (nZVI) treatment methods, both physical and chemical countermeasures were studied including various stabilizers/porous medias [20–29], bimetallic (Ni, Pd) [30–33] and sulfidized nZVI-based nanoparticles [16, 34–36]. The stabilizers, such as carboxymethyl cellulose (CMC), could well disperse the particle and even control the size of formed particles [35, 37]. In addition, incorporating nanoparticles within membrane filters could prevent the leaching of Fe and allow for the regeneration of Fe based particles [38]. More important, the versatile formation of Fe-based nanoparticles, such as palladized (Pd-Fe) or sulfidized Fe (S-Fe), grants new functionalities: (1) incorporating Pd as a catalyst, the Pd-Fe nanoparticles activate the produced H₂ (water corrosion of Fe) to atomic hydrogen for a hydrodechlorination reaction [31, 39] which could effectively dechlorinate recalcitrant chlorinated aromatic hydrocarbons (such as polychlorinated biphenyl and 2,4-dichlorophenol [40, 41]). (2) the formation of a FeS_x (mackinawite, greigite and pyrite) shell on Fe blocks the H₂ evolution from water corrosion of Fe⁰ [34], which extends the longevity of Fe⁰ and allows for a higher usage of Fe in dechlorination (better electron selectivity) [42, 43] as well as an increased reaction rate [44, 45].

Both Pd-Fe and S-Fe can be post-prepared using a precursor of nZVI. The difference in the standard reduction potential (Pd²⁺/Pd > Fe²⁺/Fe > S/S^2-) leads to distinct mechanisms of particle formation: the preparation of Pd-Fe involves redox reactions, i.e., Fe⁰ reduces Pd²⁺ to metallic Pd [39] while S-Fe post synthesis requires the water corrosion of Fe⁰ to produce Fe²⁺ for precipitation with S^2- [46]. Besides post preparation of S-Fe, one-pot methods (adding reduction agent and dithionite simultaneously) [47, 48] and ball-milling methods (mechanically mixing element sulfur and bulk Fe particles) [49] have also been reported.

Since Pd-Fe and S-Fe have different formation mechanisms, it is interested to compare the morphology and eventually to uncover the correlation between the morphology/composition of particles and their degradation rate in water treatment. Ling et al. developed an advanced scanning transmission electron microscopy (STEM) characterization method for visualizing the formation of metallic silver on nZVI as well as bimetallic Ni-nZVI [50, 51]. Integrating STEM with X-ray energy dispersive spectroscopy (EDS), the morphology and elemental composition of Fe based nanoparticles could be clearly observed. In addition, long-term degradation studies (7 −60 days) were achieved for both Pd-Fe and S-Fe nanoparticles [16, 18, 43, 52, 53], but the morphology characterization during the aging process was limited as well as the investigations of the potential for regeneration.

We agree extensive work has been published, in the preparation of Fe based nanoparticles and the mechanisms of reductive/oxidative degradation, including excellent review papers [33, 36]. However, a systematic comparison of Pd-Fe and S-Fe particle properties and long-term dechlorination is missing. Especially changes of morphology and composition during the aging period (31 days) and after the regeneration process. Therefore, Pd-Fe and S-Fe nanoparticles were post-prepared (under various synthesis conditions) in order to study the correlation between particle morphology/composition and reactivity during the long-term degradation of carbon tetrachloride (CTC). The objective of this study includes: (1) visualizing the formation of Fe, S-Fe and Pd-Fe under various preparation conditions using HAADF-TEM-EDS; (2) correlating particle morphology/elemental composition with the degradation rate of CTC at different preparation conditions; (3) evaluation of particle longevity and the potential of particle regeneration; (4) understanding the changes of particle morphology and composition during the long-term studies, especially to reveal the reason why S-Fe performed like Fe in terms of CTC degradation after the aging and regeneration processes.

2. Materials and methods

2.1. Materials

The chemicals were used as received and are listed in SI section 1. For instance, ferric chloride hexahydrate (>97%) was obtained from Fisher Scientific. Sodium borohydride (99.99%), sodium sulfide nonahydrate (≥98%), ferrous chloride (98%), carbon tetrachloride (99.9%) and potassium tetrachloropalladate (II) (98%) were purchased from Sigma-Aldrich. The analytical standard of carbon tetrachloride (100 ppm in methanol, part # HC-040–1) was obtained from Ultra Scientific. To create an anoxic environment, deoxygenated water was prepared by purging DI water with N₂ for 60 min. The dissolved oxygen was tested as 0.3 ppm.

2.2. Synthesis of nanoparticles in aqueous phase

As a precursor, Fe⁰ nanoparticles were synthesized in aqueous ethanol (30 vol%) using a borohydride reduction method (reported as [54]). Briefly, FeCl₃ (0.035 M) was dissolved into 50 ml aqueous ethanol (30 ml ethanol) in a 3-neck round bottle flask with a mechanical stirrer at a speed of 100 rpm. 50 ml NaBH₄ (0.140 M) solution was then drop-wise delivered to the flask (5 ml/min) and to convert the dissolved Fe³⁺ to Fe⁰ nanoparticles according to equation 1. Post preparation of S-Fe and Pd-Fe was conducted based on literature [2, 46]. For S-Fe synthesis, prepared Fe⁰ nanoparticles were added in Na₂S solution and the solution was placed on an orbit shaker (300 rpm). For Pd-Fe synthesis, Fe⁰ nanoparticles were mixed with a K₂PdCl₄ solution on an orbit shaker at 300 rpm. A 10 min sonication was conducted in both cases for a better dispersion. The FeS particles (mackinawite) were synthesized by directly mixing FeCl₂ solution and Na₂S solution, which was modified based on literature [16]. Details of particle synthesis and kinetic studies are illustrated in SI section 2.

Reductive formation of Fe⁰ was accomplished as follows:

$$4F{e}^{3+}+3B{H}_4^{-}+9H_{2}O\to 4F{e}^0\downarrow +3H_{2}B{O}_3^{-}+12H^{+}+6H_2\uparrow$$ (1)

2.3. Characterizations

Particle morphology was investigated using a transmission electron microscope (TEM, FEI Talos F200X). The elemental composition of the particles was analyzed using an energy-dispersive X-ray spectroscopy (EDS). Particle structure was characterized using X-ray diffraction (XRD, Siemens D500) analysis with Cu Kα (1.5418 Å) radiation. The composition of S-Fe nanoparticles was analyzed using X-ray photoelectron spectroscopy (XPS, Thermo Scientific K-Alpha), which uses an aluminum mono-chromatic X-ray source (1486.69 eV) and an electron flood gun for charge neutralization. Wide scans were performed at a pass energy of 160 eV and high-resolution scans were performed at a pass energy of 20 eV. Details of characterization processes are summarized in SI section 3.

2.4. Analytical methods

Dissolved oxygen (DO) was tested using a DO meter (edge, HANNA). The probe was polarized at 800 mV leading to the continuous consumption of DO when it passes through the probe’s gas permeable membrane. Calibration was achieved using water-saturated air and standard zero oxygen solution (HANNA).

Volatile compounds, such as carbon tetrachloride (CTC) and its dechlorination intermediates (chloroform (CF) and dichloromethane (DCM)), were analyzed according to EPA method 624. This method includes gas chromatography mass spectrometry (GC-MS, Varian 3900 GC and Saturn 2100T MS) and a purge and trap auto-sampler (Tekmar 3100). A 0.3 ml sample was added into a 40 ml vial with 0.1 ml nitric acid to dissolve metal particles. The vial was then filled up with DI water with no head space left. The auto-sampler transferred water samples to a gas-tight water column, the purge gas (helium) brought volatile species to an adsorbent. Trapped species were then desorbed at 225°C and went to the GC column. A calibration curve from 1 to 1000 ppb was made (R² > 0.99) with an internal standard pentafluorobenzene. Information on the operation protocol and sample preparation are recorded in SI section 4.

2.5. Long-term degradation studies and regeneration

The Fe, S-Fe and Pd-Fe nanoparticles were evaluated in long-term degradation studies under anoxic condition (6 cycles during a total 31-days of aging). All of the degradations were conducted in 8 ml vials with septa. No head space left during the degradation experiment (300 rpm shaking at 22 °C). A stock solution of CTC (0.25 mM) was made by two steps of dilution: (1) 100-fold dilution in methanol and then (2) 400-fold dilution in deoxygenated DI water. The CTC stock solution was replaced among different cycles, which corresponds to an aging-time of 0, 3, 5, 10, 17 and 31 days, respectively. This replacement involves magnetic separation of nanoparticles, drainage of the solution and refilling with stock solution. Each degradation cycle was operated for 24 h and all the vials were then placed on a magnet for 15 min to separate Fe based nanoparticles ([Fe]_initial = 250 ppm). Only deoxygenated water was replaced daily between the degradation cycles for the purpose of aging. After the first 31-days of aging, the collected Fe based nanoparticles were regenerated using NaBH₄ (400 mol% of [Fe]_initial). The nanoparticles (30 min regeneration) were then washed three times before the degradation study. Duplicates were tested for the long-term study. Particle morphology and composition at different stages during the long-term studies were characterized using TEM, EDS, XRD and XPS. For each degradation cycle, the nanoparticles were sonicated for 10 min for a well dispersion.

3. Results and discussion

3.1. Visualizing the formation of Fe, S-Fe and Pd-Fe nanoparticles

After the synthesis of precursor Fe, various dosage (S or Pd) and preparation time (Table 1) in the preparation of S-Fe and Pd-Fe were evaluated with particle morphology and reactivity. The discrepancy of unit that was used in dosage (mole percent) and particle composition (atomic percent) is because that the mole ratio of S/Fe can not show the total S/Fe atomic ratio due to the difference in stoichiometry of the presence of FeS (S/Fe = 1) and FeS₂ (S/Fe = 2). The Fe batches show a chain-like structure which was formed by multiple individual near spherical Fe nanoparticles (70 ± 20 nm) (Fig. 1a). A core-shell structure of individual particles was also observed as a 2.5–5 nm thin oxidized shell outside the Fe core. The atomic ratio of O/Fe is 1.4 in shell area (area 2) and 0.105 in core area (area 1), respectively (Fig. 1d). Besides EDS mapping, the oxygen shell was also observed as a darker ring out of the bright Fe core in the high-angle annular dark-field (HAADF) mode. The lighter oxygen atoms scatter electrons less intensely than the heavier iron atoms, leading to a darker region in the HAADF mode.

Table 1.

Reactivity of post-prepared S-Fe and Pd-Fe at various synthesis conditions. After synthesized, [Pd/Fe] and [S/Fe] of the particles were measured using EDS. [Fe]_initial was same for all the batches (50 ppm). [CTC]_initial = 0.25 mM. T = 22°C.

Batches	Synthesis conditions		Composition*	CTC degradation, residual%
	Pd or S dose	t, min	Pd/Fe or S/Fe	after 5 min	after 120 min
Precursor Fe	N/A	N/A	N/A	0.94	0.14
Pd-Fe (A)	1 mol%	30	0.3 atomic%	0.31	< 0.01
Pd-Fe (B)	10 mol%	3	1.4 atomic%	< 0.01	< 0.01
Pd-Fe (C)	10 mol%	30	12.1 atomic%	0.56	0.53
S-Fe (A)	20 mol%	30	1.7 atomic%	0.92	0.06
S-Fe (B)	200 mol%	30	3.0 atomic%	0.93	0.10
S-Fe (C)	200 mol%	300	13.9 atomic%	0.98	0.29

^*The mole ratio of S/Fe can not show the total S/Fe atomic ratio due to the difference in stoichiometry of the presence of both FeS (S/Fe = 1) and FeS₂ (S/Fe = 2).

Fig. 1.

TEM characterization of initial Fe⁰ nanoparticles: (a) bright field (b) HAADF mode. EDS mapping of particles (iron in red and oxygen in blue): (c) Fe core (d) Fe core and O shell. The chain-like structure was formed by individual near-spherical Fe⁰ nanoparticles (core-shell).

The post preparation of Pd-Fe or S-Fe changed the morphology of the precursor Fe nanoparticles. As shown in Fig. 2a to 2f, the scattered formation of Pd islets was found on the surface of initial Fe nanoparticles. The formation mechanism of bimetallic Fe particles (Ni, Ag) was reported to be dependent on the standard reduction potential E⁰ [50]. With a E⁰ close to Fe, the metallic Ni was formed via sorption and reduction which tends to form hollow nanoparticles. While dendritic structures were formed for Ag-Fe bimetallic particles through rapid reduction (Ag⁺/Ag has more positive E⁰ than that of Fe). Since [PdCl₄]^2- has more positive E⁰ than Fe (equation 2, 3) [55], Pd islets were rapidly formed on Fe particles via reduction reaction (equation 4). However, with an equal preparation time of 30 min, the higher Pd dose (10 mol% compared to 1 mol%) dramatically deformed part of the near-spherical nanoparticles to dendritic structures (Fig. 2g to 2i). This structure is made by a shell of oxidized Fe and a Pd-rich core (SI section 5). The increased Pd/Fe atomic ratio in the core area (2.32 compared to 0.077 in the edge) indicates the reduction of Pd prefers Fe⁰-rich area at high Pd dosage, which leads to a rapid replacement of Fe in the core area (Kirkendall effect [50, 56]). Details of elemental composition of all conditions are summarized in Fig. S1 and Table S1.

Fig. 2.

Bright field TEM imaging (image a, d, g), HAADF imaging (image b, e, h) and EDS mappings (image c, f, i) for three types of synthesized Pd-Fe nanoparticles (Table 1). In EDS mapping: iron in red, oxygen in blue, and palladium in yellow.

The post preparation of Pd-Fe was accomplished through the following equations (2–4):

$${\left[PdC{l}_4\right]}^{2-}+2e^{-}\rightleftharpoons P{d}^0+4C{l}^{-}{E}^0=+0.61V$$ (2)

$$F{e}^{2+}+2e^{-}\rightleftharpoons F{e}^0{E}^0=-0.44V$$ (3)

$$F{e}^0+{\left[PdC{l}_4\right]}^{2-}\to Pd+F{e}^{2+}+4C{l}^{-}$$ (4)

While a significant deformation of Pd-Fe was observed with a higher Pd dose, the formation of S-Fe was observed to be less correlated to the S dose but more to the preparation time. Using the same 30 min preparation time, a similar S shell was formed on the surface of precursor Fe nanoparticles at batches with 20 mol% S dose ([S/Fe]_shell = 6.3 atomic%) and with 200 mol% S dose ([S/Fe]_shell = 9.1 atomic%) (Fig. 3c, 3f). However, a significant thicker S shell (4.5–7 nm) was observed when the preparation time was increased 10-fold for the batches S-Fe(C) with 200 mol% S dose ([S/Fe]_shell = 56 atomic%). The details of element distribution are reported in SI section 5. The rate-limiting step in the preparation of S-Fe is the water corrosion of Fe at neutral pH (equation 5), which produces Fe²⁺ to precipitate with S^2-. This caused a 10–53 times lower S/Fe ratio detected in particles compared to that in dosage. Furthermore, less than 0.2 atomic% of Na was detected in all synthesis conditions, which indicates a negligible adsorption and residual of Na₂S on the S-Fe nanoparticles.

Fig. 3.

Bright field TEM imaging (image a, d, g), HAADF imaging (image b, e, h) and EDS mappings (image c, f, i) for three types of synthesized S-Fe nanoparticles (Table 1). In EDS mapping: iron in red, oxygen in blue, and sulfur in green.

The post preparation of S-Fe (FeS, FeS₂) was achieved through the following reaction at neutral pH [46]:

$$F{e}^0+2H_{2}O\to F{e}^{2+}+2O{H}^{-}+H_2\uparrow \text{\hspace{1em}(rate – limiting step)}$$ (5)

$$N{a}_{2}S+H_{2}O\to 2N{a}^{+}+H{S}^{-}+O{H}^{-}$$ (6)

$$F{e}^{2+}+2H{S}^{-}\to FeS\downarrow +H_{2}S\uparrow$$ (7)

3.2. Correlation between particle morphology/composition and reactivity

The differences in morphology and composition significantly affected the reactivity of Pd-Fe and S-Fe nanoparticles. As shown in Table 1, Pd-Fe(B) (10 mol% Pd dose in a 3 min synthesis time) and S-Fe(A) (20 mol% S dose in a 30 min synthesis time) show a better performance among its individual category of particles. Though Pd served a catalysis for H₂ hydrodechlorination, excess [Pd/Fe]_particle has been reported to decrease degradation rate due to the hindrance of production of H₂ (water corrosion of Fe) [38]. The formation of dendritic structures in Pd-Fe(C) demonstrated a consumption of the Fe core and a greater coverage of the Fe⁰ surface, which leads to an insufficient H₂ production and therefore causes the insignificant hydrodechlorination degradation between a 5 min and a 120 min reaction time. Compared to the ineffective degradation using Pd-Fe(C), Pd-Fe(A) (with 10% Pd dose) and Pd-Fe(B) (with 10% preparation time) show more than 99% degradation of CTC within 120 min and 5 min, respectively.

For S-Fe nanoparticles, S-Fe(A) (20 mol% S dose in a 30 min synthesis time) and S-Fe(B) (200 mol% S dose in a 30 min synthesis time) show similar degradation performance. This similarity was achieved due to the similar morphology and composition of the formed particles (i.e. the correlated [S/Fe]_particle is 1.7 atomic% and 3.0 atomic%, respectively). The composition of the formed FeS_x (20 mol% S dose) was detected as 68% FeS₂ and 32% FeS on the surface of nanoparticles using XPS (details are illustrated at section 3.3). However, the reactivity decreased when the preparation time is increased by 10-fold (S-Fe(C), [S/Fe]_particle = 13.9 atomic%). Excess FeS_x has been reported to decrease hydrophilicity of such particles [43] and to hinder the adsorption on metal surface [34, 57], which could eventually block the reactive sites for degradation. Mangayayam et al. show that the excessive formation of FeS_x layers (S/Fe > 10 atomic%) on Fe decreased the dechlorination rate of trichloroethylene even though an increase of reactivity was observed within the S/Fe ratio between 0–10 atomic% [16]. A similar trend was also observed when S-Fe nanoparticles are stabilized using carboxymethyl cellulose [58].

Regarding the degradation performance and the cost of materials, the batches Pd-Fe(A) and S-Fe(A) were selected for a comparison of reaction rate (Fig 4). CTC degradation curves can be fitted as a pseudo-first-order reaction (equation 8).

$$\frac{dC}{dt}=-k_{sa}{\rho }_m{a}_{s}C$$ (8)

Fig. 4.

CTC degradation among Fe, Pd-Fe and S-Fe nanoparticles (with same [Fe]_initial = 50 ppm). (a) Fe precursor (b) Pd-Fe(A) with [Pd/Fe]_particle = 0.3 atomic%, (c) S-Fe(A) with [S/Fe]_particle = 1.7 atomic%. (d) FeS_mackinawite particles were also tested for a comparison ([Fe]_initial = 2500 ppm). [CTC]_initial = 0.25 mM. T = 22°C.

Where k_sa is surface area normalized reaction rate (Lm⁻²min⁻¹). Since the ratio of Pd/Fe and S/Fe is minor (< 2 atomic%), only the surface area of the Fe precursor was used in this calculation. ρ_m is nanoparticle loading density, which in this case is 0.05 gL⁻¹, and a_s is surface area per unit mass, which is 10.9 m²g⁻¹. This value was calculated using average spherical nanoparticle size (~35 nm, based on the TEM characterization) and iron density (7870 gL⁻¹). In this study, only Fe surface was used for calculation.

The k_sa of CTC degradation followed the order of Pd-Fe (0.3 atomic% Pd) >> S-Fe (1.7 atomic% S) > Fe (Table 2). With the Pd catalyst, degradation using Pd-Fe involves an activation of produced H₂ (water corrosion of Fe) for hydrodechlorination [31], whereas for Fe and S-Fe the H₂ production is an undesirable side reaction of degradation through an electron transfer mechanism. This mechanistic difference significantly enhanced the k_sa of CTC degradation using Pd-Fe (13–20 fold). Around 90% conversion from CTC to chloroform (CF, a major intermediate) was observed in all types of Fe-based nanoparticles. The formed CF was further degraded and less than 5% CF remained after 24 h degradation using both Pd-Fe and S-Fe, while nearly 60% CF remained for Fe batches (Fig. 5). Since the degradation rate is largely dependent on the fraction of Pd, the k_sa of trichloroethylene is 0.152 Lm⁻²min⁻¹ using 0.3 atomic% Pd in our study (not shown) and 0.0405 Lm⁻²min⁻¹ using 0.05 atomic% Pd [59]. According the fast degradation in this study (t₅₀ was 2.2 min), it is true that the high Pd loading (0.3 atomic% of Pd) is not necessary for effective degradation. Our previous publication [38] reported the effects of the Pd fraction (Pd/Fe nanoparticles) on H₂ production as well as the degradation of 2-chlorobiphenyl (PCB 1). A lower Pd fraction (0.026 atomic% Pd) was simulated to degrade 90% PCB 1 within 100 min (simulated as pseudo first order reaction). The reason for a 11.5-fold higher Pd fraction in this present manuscript is to obtain a significant signal in the EDS detection, which is important for the study of Pd distribution on the precursor Fe nanoparticles as well as the transformation of Pd-Fe nanoparticles during aging and regeneration processes.

Table 2.

The surface area (Fe core) normalized reaction rates (ksa, Lm⁻²min⁻¹) of CTC degradation for Fe, S-Fe and Pd-Fe ([Fe]_initial = 50 ppm for all the particles, [CTC]_initial = 0.25 mM, [Fe/CTC]_initial = 3.6, 22°C).

Batches	Composition [S/Fe] or [Pd/Fe]	ksa Lm−2min−1	R2	Derived t1/2 min
Fe	N/A	0.029	0.98	43
S-Fe(A)	1.7 atomic%	0.042	0.97	30
Pd-Fe(A)	0.3 atomic%	0.580	0.98	2.2

Fig. 5.

A long-term degradation study with 31-days of aging and followed regeneration of (a) Fe, (b) S-Fe ([S/Fe]_particle= 1.7 atomic%) and (c) Pd-Fe ([Pd/Fe]_particle= 0.3 atomic%) nanoparticles. Same [Fe]_initial (250 ppm) was used for all the batches. Six degradation cycles were conducted during the aging process, corresponds to start the day 0, 3, 5, 10, 17 and 31. Every cycle lasts for 24 h and a fresh stock solution of CTC (0.25 mM) was replaced for individual cycles. [Fe]_initial/ [CTC]_initial = 18. T = 22°C. The number above bars represents the total mass balance of DCM, CF and CTC.

Compared to bare Fe, S-Fe has been reported to limit the H₂ evolution (side reaction in degradation via electron transfer) either by decreasing the corrosion tendency [43] or by restricting hydrogen recombination reactions on H adsorption sites of particles [60, 61]. With higher usage of Fe in degradation rather than the evolution of H₂, S-Fe presented a faster degradation rate of CTC compare to that of Fe and even better degradation of the intermediate, CF. Furthermore, Cheng et al. reported that the oxidation of dissolved oxygen by FeS can generate hydroxyl radicals resulting in the oxidization of As (III) in Fenton-like reaction [62, 63]. However, in the system of S-Fe nanoparticles, nanoscale Fe core could play a vital role in consuming DO compared to that of the formed FeS/FeS₂ layer (1.7 atomic% S of Fe). The DO measurement shows the decrease of DO from 2.21±0.05 to 0.12±0.01 ppm from the control sample to the Fe⁰ samples after 24 h degradation, which was even less than that of the initial deoxygenated DI water (0.30±0.01 ppm). In addition, the FeS particles (mackinawite) were also studied. Lan and Butler reported that 90% CTC was degraded after 14 days using FeS_mackinawite ([FeS/CTC]_initial = 5) [64], however, only 12% CTC degradation was observed after 24 h reaction in this study even with a [FeS/CTC]_initial = 180 (Fig. 4d). The limited degradation of CTC within 24 h could be explained by the bigger size of particle (micro scale, Fig. S2) and the slower reaction rate. Hence, the reactivity of FeS_mackinawite has a negligible contribution to the increased degradation rate using S-Fe.

3.3. Long-term degradation studies and regeneration of Fe, S-Fe and Pd-Fe nanoparticles

In addition to the reactivity comparison, a study of the degradation (over 31 days) and the regeneration processes (using NaBH₄, reducing agent) was conducted in batch mode among Fe, S-Fe and Pd-Fe nanoparticles. Within the 31-days of aging, all the particles were tested for six degradation cycles (24 h per cycle) with a magnetic separation and replacement of CTC stock solution between individual cycles. Considering the fast degradation rate reported on Table 2, a cycle time of 24 h was selected for the purpose to observe the aging effects on particle reactivity and morphology/composition. Since the sample vials were sealed and with no head spaceleft, >98% CTC remained after each degradation cycle of 24 h.

As shown in Fig. 5, more than 99% CTC was degraded using all types of particles (Fe, S-Fe, Pd-Fe) during the first three cycles (0 to 5-days of aging). CF and dichloromethane (DCM), two major intermediates of CTC degradation) were produced and then further degraded using Fe, S-Fe and Pd-Fe. S-Fe also presented a better long term degradation of produced CF compared to bare Fe. No obvious CTC degradation was observed for the Fe and Pd-Fe at the fourth cycle (10-days of aging), but more than 99% degradation was achieved for the S-Fe batches until the fifth cycle (17-days of aging). Herein, the S-Fe was proven to extend the longevity of bare Fe nanoparticles, which could attribute to the hindrance of the side reaction between Fe⁰ and water (discussed in section 3.2).

No degradation was observed at the sixth cycle (31-days of aging) for all types of particles. Regeneration process was then achieved using NaBH₄ (400 mol% of [Fe]₀). Regenerated Fe and Pd-Fe present a similar degradation pattern compared to that when using freshly made particles. However, S-Fe performed like Fe after the regeneration process: S-Fe no longer extended longevity like the original S-Fe and was less effective in the degradation of CF.

Advanced material characterizations (TEM, XPS and XRD) were conducted during the long-term study to investigate the changes of particle morphology/composition as well as to explain the loss of longevity of S-Fe after aging and regeneration processes. Compared to the chain-like structure observed at the initial stage of Fe, S-Fe and Pd-Fe (Fig. 6a, 6d, 6e), an agglomeration of particles was observed in all batches after 31-days of aging (Fig. 6b, 6e, 6h) and after regeneration (Fig. 6c, 6f, 6i). For individual particles, a transformation from near-spherical structures to thin rods, round and translucent polygons (such as square and hexagon) was also observed after aging, which could attribute to the oxidation formation of the FeOOH and magnetite (Fe₃O₄)/maghemite (γ-Fe₂O₃) [65]. As shown in Fig. 7, the oxidation phases were confirmed by observing the interplanar spacing as 0.27 nm in the area 1 (thin rod) and as 0.25 nm in the area 2 (polygons), which correspond to the (021) plane of goethite (α-FeOOH) [66] and the (311) plane of Fe₃O₄ [67], respectively. After regeneration, the near-spherical particles were recovered and the EDS analysis also show a conversion of an oxygen/iron mixed nano-rods and polygons to an oxygen shell and iron core composition (Fig. S3). This conversion of iron oxidation products to Fe⁰ leads to the restoration of reactivity of Fe, S-Fe and Pd-Fe nanoparticles.

Fig. 6.

A comparison of particle morphology at three stages: initial (first column), 31-days of aging (second column) and after regeneration (third column). Types of particles: Fe (image a, b, c), S-Fe with 1.7 atomic% S (image d, e, f) and Pd-Fe with 0.3 atomic% Pd (image g, h, i).

Fig. 7.

Characterization of the interplanar spacing of S-Fe (1.7 atomic% S) after 31-days of aging. Two typical areas of thin rod (area 1) and polygons (area 2) were zoomed in.

Specific to S-Fe, the Fe core was initially covered by a thin shell, which accounts to the formation of FeS_x [68, 69]. This thin shell disappeared after 31-days of aging (Fig. 6d, 6e). The EDS analysis (Fig. 8) also detected a negligible amount of S within the whole mapping area (<0.2 atomic%) after aging, comparing to that at the initial stage (1.7 atomic%) (Table 3). This loss of S during aging was also observed in the EDS spectra (Fig. S4). The significant decreases of O/Fe ratio after NaBH₄ treatment (from 3.17 to 0.43) demonstrated the restoration of Fe core. Therefore, the loss of FeS_x made S-Fe performed like Fe in terms of CTC degradation after the aging and NaBH₄ treatment.

Fig 8.

EDS mapping of S-Fe (20 mol% S dose in 30 min) at initial, aged (31-days of aging) and regeneration stages of the long-term test. Iron in red, oxygen in blue, and sulfur in green. The [S/Fe] of the marked areas was summarized in Table 3.

Table 3.

EDS analysis of elemental composition of S-Fe particles (20 mol% S dose in 30 min) during long-term study. The target regions are marked in Fig. 8.

S-Fe batch	ID	Region	[S/Fe] atomic%	[S/Fe], atomic% whole mapping area	[O/Fe], atomic% whole mapping area
Initial	1	Edge	6.3	1.7	31
Initial	2	Core	0.6
31-days of aging	1	Edge	1	<0.2	317
31-days of aging	2	Core	<0.1
NaBH4 treatment	1	Edge	1.4	<0.2	43
NaBH4 treatment	2	Core	0.4

The formation of FeS_x and the loss of FeS_x during aging were both confirmed using XPS. As shown in Fig 9a, the scanned S 2p spectra were fit with doublets corresponding to 2p_1/2 and 2p_3/2. No obvious sulfur peak was observed after 31-days of aging. However, different valence states of sulfur such as disulfide (S₂^2-), sulfide (S^2-) and sulfate (SO₄^2-) were observed at both initial and 5-days of aging S-Fe batches (Fig. 9b, 9c). The detected sulfur at the surface of freshly made S-Fe nanoparticles was made of 38 atomic% S₂^2-, 18 atomic% S^2- and 44% SO₄^2-. Therefore, the FeS_x layer was made by 68% FeS₂ and 32% FeS on the surface of S-Fe nanoparticles. Xu et al. also found an overall higher fraction of FeS₂ compare to that of FeS [43]. This existence of SO₄^2- in the initial stage indicates the exposure to oxygen during preparation process and this exposure could also happen in the long-term study when the CTC solution was replaced between the degradation cycles. The oxidation of Fe sulfide minerals (mackinawite and pyrite) has been widely reported and could convert FeS/FeS₂ to insoluble elemental sulfur [62, 70] and soluble sulfur species such as SO₃^2- and SO₄^2- [71–74]. The measurement of elemental sulfur can be conducted using LC-HPLC [75] and analysis of SO₃^2-/SO₄^2- can be made using wet chemistry methods or ion chromatography [76]. This oxidation conversion could lead to a decrease in sulfur content during aging which was confirmed from S/Fe atomic ratio as 49%, 25% and <0.1% corresponding to initial, 5-days of aging and 31-days of aging. Xu et al. reported the formation of 18% SO₄^2- at the initial stage of post-prepared S-Fe while no oxidized sulfur species (besides S_n^2-) was detected when S-Fe particles were prepared in a one-step method without using Fe⁰ nanoparticles as a precursor. Hence, they suggest one-step prepared S-Fe has a better stability of oxidation compared to the post-prepared S-Fe. Interestedly, they also reported post-prepared S-Fe (nano-scale, [S/Fe]_particle = 0.8 atomic%) was hydrophilic but one-step prepared S-Fe (micron-scale, [S/Fe]_particle = 5.4 atomic%) was hydrophobic. [43]. The difference in hydrophilicity might also impact the adsorption of soluble oxidized sulfur species. This lack of oxidized sulfur species in XPS analysis was also detected during the aging process of synthesized micron-scale FeS_mackinawite (hydrophobic) [64, 77]. Therefore, the post-prepared S-Fe (with a greater hydrophilic property and nano-scale particle size), could either have a greater adsorption of soluble sulfur species or is more active for oxidation conversion. Less than 5% Fe (ICP-MS test) was leached after 7-days of aging compared to freshly made particles (SI section 7). The rapid formation of iron oxide/iron hydroxide could hinder the further consumption of Fe⁰ core. However, our main objective is to study the changes on morphology/composition and long-term performance of iron-based nanoparticles. The systematic study of leaching of Fe, Pd and S can be conducted in the future.

Fig 9.

XPS analysis of S 2p spectra of S-Fe nanoparticles at initial stage (image b), 5-days of aging (image c). A disappearance of S 2p peaks were observed after 31-days of aging (image a).

XRD analysis was also conducted to identify phases changes during the aging and regeneration processes. For a better XRD signal, 2.5 atomic% Pd-Fe was prepared just for characterization. The diffraction peaks of Fe⁰ appeared at 44.5°, 64.9° and 82.2° (Fig. 10a), and correspond to the crystaline planes of (110), (200) and (211), respectively (JCPDS 06–0696) [78]. These peaks were observed at the initial Fe, S-Fe and Pd-Fe particles. The peak at 40.1° observed in Pd-Fe agrees with the major crystal plane (111) of Pd (JCPDS 05–0681) [79]. For S-Fe (Fig. 10b), the characteristic peak of FeS_x were not detected and this phenomena has also been reported [16, 80]. The explanation might be the low [S/Fe] (1.7 atomic%) and the poor crystallization when FeS_x was synthesized at room temperature. After 31-days of aging, the characteristic peaks of magnetite (Fe₃O₄)/maghemite (γ-Fe₂O₃) (30.2°, 35.6°, 57.4°, 62.8°) were detected at all the batches [81, 82]. This correlates with the polygons observed in TEM imaging (Fig. 6b, 6e, 6h). However, no obvious characteristic peak of FeOOH were observed, which might due to a poor crystallization [83]. Seha et al. reported the primary formation of the poorly crystalline akaganeite (β-FeOOH) during Fe⁰ treatment of metolachlor, while α-FeOOH and Fe₃O₄ were detected in the similar treatment containing FeSO₄ [65]. Greenlee et al. also observed that lepidocrocite (γ-FeOOH) is the primary oxidation product of Fe⁰ under oxygenated water [84]. These studies suggest the formation of Fe oxidation products is significantly related to water conditions, such as dissolved oxygen level and various inorganic salts. Furthermore, after NaBH₄ treatment, the characteristic peaks of the iron oxides had disappeared and similar diffraction patterns of the initial particles were present. This conversion demonstrates the regeneration of Fe⁰.

Fig. 10.

XRD analysis of Fe (image a), S-Fe (image b), and Pd-Fe (image c) nanoparticles at three stages: initial, 31-days of aging and after NaBH₄ regeneration. Marks of characteristic peaks: Fe⁰ (Fe), magnetite (Fe₃O₄)/maghemite (γ-Fe₂O₃) (M) and palladium (Pd).

4. Conclusion

This study evaluates the correlations between particle morphology/composition and reactivity under various preparation conditions and at the different stages during the aging and regeneration process. Though using the same Fe⁰ precursor, the distinct formation mechanisms of Pd-Fe (redox reaction) and S-Fe nanoparticles (water corrosion and precipitation) lead to the different key factors in particle morphology and reactivity. Higher Pd dose (1 to 10 mol%) dramatically deformed the original structure of near-spherical Fe particles with Pd islets to dendritic structures which significantly lowered the reactivity. However, an increase of S dose from 20 mol% to 200 mol% merely changed the particle properties. Thicker layers (4.5 −7 nm) of FeS_x were formed when the preparation time was increased from 30 min to 300 min, which resulted in a decreased reaction rate. The degradation rate of CTC (k_sa = 0.580 Lm⁻²min⁻¹) was greatly enhanced (20 fold higher than that of Fe) when the Pd catalyst (0.3 atomic% Pd/Fe) activated the produced H₂ for hydrodechlorination. In contrast, for S-Fe and Fe, H₂ evolution is an undesired side reaction of degradation via electron transfer mechanism. S-Fe has been reported to limit H₂ evolution and thereby enhance CTC degradation and particle longevity compared to that of bare Fe: only < 5% CF (major intermediate of CTC degradation) remained after 24 h degradation using S-Fe while nearly 60% CF remained when using Fe. The FeS_mackinawite was reported to produce hydroxyl radical upon oxidation. However, in the system of S-Fe nanoparticles (1.7 atomic% S of Fe), nanoscale Fe core could play a vital role in consuming DO compared to that of the formed FeS/FeS₂ layer (DO level decreased from 2.21 ± 0.05 to 0.12 ± 0.01 ppm). The limited degradation of CTC using prepared FeS_mackinawite also proven the the reactivity of FeS_mackinawite has a negligible contribution to the increased degradation rate using S-Fe.

For the long-term study, S-Fe (FeS_x layer was made of 68% FeS₂ and 32% FeS) presented better longevity compared to Pd-Fe and Fe. Significant agglomeration was observed in all types of particles after 31-days of aging. Individual near-spherical Fe core were oxidized to thin rods and polygonal structures, which were identified as FeOOH and magnetite (Fe₃O₄)/maghemite (γ-Fe₂O₃) respectively via interplanar spacing and XRD analysis. Similar morphology and degradation performance were recovered for both regenerated Fe and Pd-Fe comparable to that of freshly made particles (using reducing agent NaBH₄). However, only the Fe core was restored for S-Fe but not for the previous formed FeS_x shell. Although S-Fe was proven to extend the longevity of bare Fe, the loss of FeS_x during aging made S-Fe nanoparticles eventually have similar morphology and CTC degradation performance compared to that of bare Fe after the aging and regeneration (using NaBH₄) processes. This disappearance of S species might be due to oxidation during the aging process, which leads to a continuous conversion of insoluble FeS_x species to soluble sulfur species (SO₄^2-). Compared to the reported one-step synthesized S-Fe and synthesized FeS_mackinawite, the post-prepared S-Fe (with nanoscale near-spherical structures and hydrophilic nature) might either have a greater adsorption of soluble sulfur species (SO₄^2-) or are less stable to oxidation processes. To better apply Fe-based nanoparticles in the field, the leaching profile, recycling method as well as the transformation of oxidized sulfur species should also be considered

Highlights:

• Particle properties in nanoscale were characterized using HAADF-TEM and EDS

• Pd-Fe catalyze the CTC degradation (greater k_sa) and S-Fe shows a better longevity

• Pd-Fe and S-Fe effectively degraded chloroform (intermediate of CTC) but not for Fe

• Aged Fe-based particles were converted to FeOOH and Fe₃O₄/γ-Fe₂O₃

• Aged and regenerated S-Fe lost S species and performed like Fe in CTC degradation