Results and Discussion

3.3.1Characteristics of Sediment and Adsorbent

The characteristics of sediment (TY03) are listed in Table S3-1. TY03 had a pH value of 5.23 and soil organic content (SOC) of 3.66 wt%. Organic content in sediment is an important parameter for adsorption. As shown in previous research, SOCs of sediments have been observed as low as 1−2%, while can be as high as 15−20% [65, 105, 106].

Therefore, TY03 may be considered acidic sediment with relatively low organic content.

THg of TY03 was averaged 0.6232 mg/kg. The texture of sediment can be categorized as sandy loam with sand, silt, and clay content of 73.6, 9.6, and 16.8 wt%, respectively.

The characteristics of adsorbent AC and SAC including BET surface area, total pore volume, median pore width, and element content (C, H, O, N, S) are shown in Table 3-1, indicating that the characteristic of ACs was different after sulfur treatment. Raw AC had a high specific area of 818.3 m²/g. After sulfurization, the BET specific area decreased to 728 m²/g for SAC. Similar results have been observed in several studies[12, 14, 41, 74].

Asasian and Kaghazchi [74] reported that sulfurization with elemental sulfur to coal-based AC resulted in a decrease in surface area by 58% (from 828 to 596 m²/g). Wang et

al. [12] also observed a marked decrease of surface area by 47% (from 1884 to approximately 1000 m²/g) as sulfurizing the coconut AC with elemental sulfur. In this study, only 10.9% of surface area were lost, the reason may be due to the large granular size (i.e., 10−18 mesh) of AC leading to less block and destruction of the internal porous structure of AC. Shifting of the average pore width toward a larger value and a decrease in total pore volumes were also observed after sulfurization of AC. Comparing the results from surface area analysis and SEM images (Figure S3-1), it may suggest that the reduction of surface area was mainly due to the collapse or block of micropores by heating and sulfur attachment. QSDFT micropore structure analysis (Figure 3-2) supported the reduction of micropores within a pore width of 0.5−1.0 nm after sulfurization.

Additionally, the morphology of AC and SAC obtained by SEM images showed no significant difference in the surface structure after sulfurization.

Tables 3-1. Physical and chemical properties of precursor AC and resulting SAC.

After sulfurization, the sulfur content increased from 0.47 to 4.09 wt%. This content was smaller compared to some SAC with sulfur content as large as 10−22 wt% reported in preview research [12, 74] because we intentionally kept the sulfur content less than 5 wt%. By doing so, most of the surface area and pore volume can be maintained because the pores were not markedly blocked by the impregnated sulfur. The sulfur content within

BET surface area Total pore volume Mean pore width C H O N S

(m²/g) (cm³/g) (nm) (wt%)

AC 818.4 0.462 0.9454 78.88 1.41 4.52 0.73 0.47

SAC 728.1 0.416 0.9809 75.42 1.22 6.32 0.73 4.10

4−5 wt% should be sufficient for Hg capture with various sulfur functional groups in the SAC based on XPS results (Figure 3-3). The deconvoluted XPS S2p signals at binding energies of 163.90, 164.75, 165.24, 166.36, 167.88, 168.67, and 169.41 eV corresponded to functional groups such as sulfide (C-S-C) + thiophene (C-S-C) + thioethers (C-S-C) + mercaptans (C-SH), disulfide (C-S-S-C), sulfinyl group (C2S=O), sulfinyl group (C2S=O), sulfone (C2S(=O2)), sulfone (C2S(=O2)), and sulfonate (C-SO3) reported in previous studies[12, 107-109]. Hsi et al. [110] suggested that impregnated elemental sulfur would remain mainly as elemental sulfur groups on the surface of carbon in the heating condition of 250°C; as temperature increased to 650°C, increasing organic sulfur groups was observed.

Figure 3-2. Micropores size distribution of raw AC and SAC based on QSDFT model.

Figure 3-3. Deconvoluted S2p peak for AC and SAC.

3.3.2 Aqueous Adsorption Experiment

Kinetic adsorption tests were carried out with Hg²⁺ adsorbed to AC and SAC (Figure 3-4). The experimental results suggested that about 81% of THg adsorption occurred within the first 6 h, with removal efficiency of Hg²⁺ at approximately 56.3 and 79.1% by AC and SAC adsorption, respectively. The equilibrium was reached at a time between 16 and 24 h. The adsorption rate appeared to be slower due to the larger size of granular AC (i.e., 10−18 mesh), as compared to previous research. For example, Li et al. [13] observed 87% of total Hg adsorption in the first 3 h, achieving equilibrium at around 10 h.

The experimental data were further fitted by pseudo-first and pseudo-second order kinetic models, for which the equations were described as below:

Pseudo-first order rate equations:

𝑑𝑞_𝑡

𝑑𝑡 = 𝑘_𝑆1(𝑞_𝑒− 𝑞_𝑡) (eq. 1)

log(𝑞_𝑒− 𝑞_𝑡) = 𝑙𝑜𝑔(𝑞_𝑒) − ^𝑘𝑆1 Whereas qt = sorption of sorbate concentration on the sorbent at given time t; qe = sorption of sorbate concentration on the sorbent at equilibrium state; kS1 =rate constant for pseudo-first order model; kS2 =rate constant for pseudo-second order model.

Pseudo-first and second order models can, therefore, being employed by eq. 2 and 4 to describe the adsorption data. The kinetic parameters were calculated and shown in Table 2-2. Both the adsorption data of AC and SAC had a better fitting to the pseudo-first order model (R²=0.9803 and 0.9827, respectively) than to the pseudo-second order model (R²= 0.9318 and 0.8636, respectively). The pseudo-first order kinetic constant for AC and SAC is 2.994×10^-3 and 3.224×10^-3 min^-1, respectively, indicating that SAC had a faster rate of reaching adsorption equilibrium than AC. This adsorption rate appears to be lower than previous research [13], but it is in prediction since the AC used in this research is granular (> 300 μm in diameter) rather than in powder (< 300 μm in diameter). Coarser AC has been long regarded slower and less effective in adsorption reaction as compared to finer AC [5, 9, 47]. Zimmerman et al. [47] discovered that GAC (400−1700 μm in diameter) was less effective in reducing hydrophobic organic contaminant bioaccumulation in one-month tests. Cornelissen et al. [9] compared different field tests and found that finer AC (<45 μm in diameter) showed higher adsorption effectiveness than coarser AC (75−300 μm in diameter) but less capping stability. Although slower

adsorption rate of GAC was observed in previous studies and in this work, it is still reasonable to use coarser GAC because it is more stable to serve as the thin layer capping material [9].

Figure 3-4. Hg²⁺ adsorption by AC and SAC as a function of time. For here [Hg²⁺] = 1 mg/L, Hg solution = 50 mL, adsorbent dosage = 50 mg, and pH = 7.0±0.1.

Table 3-2. Fitting parameters of AC and SAC adsorption by pseudo-first and second

Temperature Hg concentration

(mg Hg/L)

Adsorbent

Pseudo-first order Pseudo-second order

q_e k_s1 R₁² q_e k_s2 R₂²

(°C) (mg/g) (min^-1) (mg/g)

(g/mg-min)

30 1.0

AC 0.6232 0.002994 0.9803 0.5237 0.0249 0.9318 SAC 0.8040 0.003224 0.9874 1.1130 0.0327 0.8636

order reaction models.

Distribution coefficient (KD) is a useful tool to describe the affinity of Hg to AC [64, 65] in a linear relationship, defined as the ratio of the concentration of sorbate sorbed to the sorbent divided by its concentration in solution. This concept includes an assumption that the solute concentration is very low that the sorption of sorbent only controls by concentration of solution rather than remaining sorption sites on the sorbent. It was recently discovered that the distribution of adsorbate on black carbon should be considered with two separated mechanisms (adsorption and partition) [111], and by definition, Hg uptake by black carbon should be mainly governed by adsorption mechanism because Hg uptake capacity does not increase after reaching maximum capacity, which was observed in previous research [13, 73, 74]. However, it may still be plausible to introduce KD concept since the Hg concentration used in this study was very low and far from reaching the maximum adsorption capacity of AC.

Adsorption isotherms of AC and SAC are shown in Figure 3-5. The results showed that KD of SAC (9.426×10⁴, R²=0.996) to Hg²⁺ was more than two-fold larger than that for AC (i.e., 3.694×10⁴, R²=0.958). Notably, the affinity of ACs to MeHg was almost 10-times larger than that to Hg²⁺, as shown in our results, that KD for AC and SAC to MeHg was 2.254×10⁵ (R²=0.983) and 7.661×10⁵ (R²=0.834), respectively. The reason for SAC having greater adsorption affinity to Hg²⁺ than AC is expected. As aforementioned, Hg can be immobilized by sulfide to form highly stable HgS (i.e., log K ≈ 52.7‒53.3) [23].

SAC had larger sulfur content (4.09 wt%) than that of AC (0.47 wt%). Additionally, based on the XPS results, SAC possessed various sulfur functional groups; all of which lead to the higher affinity to Hg²⁺ for SAC than for AC [12, 13, 74].

On the contrary, results of AC and SAC having higher affinity to MeHg than Hg²⁺ in this work is not completely consistent with those from previous studies. Gomez-Eyles et al. [64] compared KD of various black carbons to MeHg and Hg and found that six materials with Hg KD to MeHg KD ratios significantly greater than one, two materials with the ratios close to one, and six materials with the ratios significantly smaller than one. These results may suggest that adsorption of Hg²⁺ and MeHg involves different mechanisms.

Figure 3-5. Adsorption isotherm of AC and SAC adsorbing (a) Hg²⁺ and (b) MeHg.

Adsorption time = 24 h. Hg solution = 50 mL, adsorbent dosage = 50 mg, pH = 7.0±0.1

3.3.3 Sediment Competition Adsorption Experiment

Three spiked Hg-containing sediments with Hg content of 14.23, 106.28, and 235.78 mg/kg were prepared by 117-d incubation, and the sediment/adsorbent competition adsorption tests were carried out by measuring Hg concentrations in the sediment and porewater for 1, 3, and 6 wt% activated carbon addition, which is shown in Figure 3-6.

Under the condition of 1 wt% adsorbent dosage (Figure 3-6a), porewater THg was reduced by 73.4−96.2% by AC and 83.25−95.21% by SAC in the sediment with Hg range of 14.23−106.28 mg/kg. The decrease of effectiveness at 235.78 mg/kg may be due to reaching the equilibrium adsorption capacity of AC and SAC. At 3 wt% adsorbent amendment (Figure 3-6b), AC reduced 91.9−99.9% porewater THg in the Hg sediment range of 106.28−235.78 mg/kg, while SAC reduced 86.4−90.5% porewater THg. Notably, at the adsorbent dosage 3−6 wt%, the enhancement in porewater THg reduction was less obvious as compared to the dosage from 1−3 wt%.

Surface water Hg concentration of 1.3 μg/L was set by Preliminary Remediation Goal (PRG) for an ecological endpoint [112], and 3 wt% dosage of AC can cause porewater THg concentration beneath this level throughout the tested sediment Hg range in this study. It indicates that 3 wt% of AC or SAC could be an optimum dosage for thin layer capping remediation of Hg sediment with wide Hg concentration range up to 235.78 mg/kg, in terms of amount and cost as compared to amending 6 wt% adsorbent.

Surprisingly, amendment of SAC led to lesser porewater THg reduction as compared to that of AC, especially at the dosages of 3 and 6 wt%, which is inconsistent to the results obtained from our batch-scale aqueous adsorption tests aforementioned. Notably, for the sediment with Hg content of 14.23 mg/kg, 1 wt% of SAC addition significantly reduced porewater Hg to 83.3%; but with the amendment of 3 and 6 wt%, the reduction ratio was

only −31.3 and 6.4%, respectively. The reason of the inconsistency is not yet thoroughly understood, but may be attributed to the elongation of adsorption time (from 24 to 96 h) and the influence of inherent components in sediment leading to the instability of HgS complexes on the SAC surface. However, this possibility can be ruled out by a follow-up test (Figure S3-2) showing that no obvious Hg was re-dissolved in the aqueous phase for up to 72 h, and no aggregation was observed by measuring the aqueous samples filtered by different pore-sized membranes.

Another reason causing the lesser porewater THg reduction by SAC may be due to the re-dissolution of HgS(s) nanoparticles after adsorption on SAC, triggered by sediment.

This mechanism has not yet reported but is reasonable in explaining the inconsistency of aqueous adsorption and sediment competition adsorption experiments in this study. Hg is well-known to have high affinity to sulfur. However, the formed fine particles of HgS(s), via interaction of Hg with impregnated elemental sulfur present in SAC surface, may dissolve again into Hg²⁺ and SO42- in an oxic environment [113, 114]. In addition, as sulfur exists in an aqueous environment, Hg²⁺ and S^2- could precipitate and form β-HgS(s)

nanoparticles at sulfur concentration even as low as 1 nM [23]. Without interference, β-HgS(s) nanoparticles would gradually aggregate forming larger particles with a diameter of 200−500 nm [114]. However, when the aqueous environment exists dissolved organic matter (DOM), the aggregation of β-HgS(s) nanoparticles may be inhibited and even re-dissolved to the aqueous phase [115].

Figure 3-6. Hg concentrations in porewater versus in sediment at (a) 1 wt%; (b) 3 wt%; (c) 6 wt% adsorbent addition based on sediment competition adsorption test. DI water is 50 mL and Hg sediment is 5.0 g.

As aforementioned, SAC had higher Hg²⁺ and MeHg affinity compared to AC based on aqueous adsorption tests and the reason was attributed to the stable bonding of Hg-S.

As reported by Liu et al. [116], Hg has high tendency to bind to carbon-sulfur functionality such as C-S-S-C, C-S-C, and C-SH, which are abundant in SAC based on the XPS examination. However, in the sediment adsorption experiments, as the adsorption time increased from 24 to 96 h, with the possible release of DOM from sediment, sulfur functional groups may be destructed and dissolved, thus causing the Hg to form Hg-S-DOM cluster, as proposed in Figure 3-7. Consequently, The porewater THg reduction by SAC amendment was less pronounced as compared to that by AC amendment.

Figure 3-7. Proposed adsorption/desorption mechanisms of Hg adsorption on SAC in the sediment environment.

Although a similar phenomenon was observed previously, this is the first study to report the potential sulfur release from sulfur-impregnated activated carbon during sediment competition adsorption tests causing an increase in porewater Hg concentration.

Liu et al. [116] also reported the SO42- released from biochar causing Hg adsorption ability to decrease in a 48 h adsorptions tests. Similarly, Liu et al. [117] conducted

microcosms experiments and discovered that the release of Hg may have a high correlation to aqueous parameters such as the concentrations of Fe ions, SO42-, and DOC.

The influence of these parameters on Hg release from AC and SAC should be further comprehended.

3.3.4 Microcosm Experiment

During the 86-d microcosms operation, pH, ORP, temperature, and flow rate were recorded (Figs. S3-3−S3-6). In brief, the temperature of microcosms was maintained at around 25°C, pH value was typically in the range of 6.6−7.0, with few exceptions at the range of 6.2−7.4. Original ORP was around -200 to -300 mV after the operation started, ORP was dramatically increased and maintained at around 0‒100 mV due to the supplied water had greater ORP.

Overlying water THg and MeHg concentrations of microcosms during operation are shown in Figure 3-8. As soon as the active caps deployed, a dramatic decrease of overlying water THg concentration by 91.2−95.9% (not shown in the graph) and MeHg (77.8−99.8%) in capped microcosms was observed at day 2. Throughout the operation period, the influence of caps to overlying water THg concentration was insignificant between days 17 and 47. While after day 47, THg started to leach and caused high amount of Hg in overlying water, with the highest THg concentration at day 63 recorded at 0.791, 0.281, 0.654, and 0.417 μg/L for those from column A (control), column B (SAC+bentonite), column C (SAC+TY03), and column D (AC+bentonite), respectively.

These leaching processes were also observed in Liu et al. [117] as leaching of Hg occurred at around day 80, with high correlation with the rising of Fe and SO42- concentrations.

Notably, previous studies have shown that re-dissolution of Hg-S complexes into Hg²⁺

and SO42- may occur in an oxic environment [113, 116]. Therefore, with the increase of

ORP in our microcosms, the adsorbed Hg may be re-dissolved into the aqueous phase, which may partly lead to the increase in THg concentration after 47 d of operation time, especially for Columns A (control) and D (AC + bentonite) (Figure 3-8).

Figure 3-8. Overlying water (a) THg and (b) MeHg concentrations of vertical up-flow microcosms during operation.

Compared to THg, different inhibition patterns of MeHg into overlying water were observed in microcosms. As in the first 40 d, the sediment without active cap had high overlying water MeHg, with the highest concentration recorded as 34.7 ng/L. This value was 13 times higher than the Preliminary Remediation Goal set for the ecological endpoint (i.e., 2.6 ng/L) [112]. Microcosm results also showed that the presence of active caps had great leaching inhibition ability to MeHg, with MeHg reduction efficiency of higher than 79.1%. After day 60, all microcosms had relatively low MeHg, it may be due to increasing ORP in the system, causing microbiome shift, thus reducing biomethylation.

The microcosm results further demonstrated that using SAC or AC as thin layer capping can reduce the overlying water concentration of MeHg to a level far below ecology threat consideration, thus is applicable in reducing the risk from Hg-contaminated sediment since MeHg is the most toxic species of Hg.