Modelling mercury sorption of a polysulfide coating made from sulfur and limonene 1 2

Modelling mercury sorption of a polysulfide coating made from sulfur and limonene 1 2 Max J. H. Worthington,a,b Ismi Yusrina Muhti,b Maximilian Mann,a,b Zhongfan Jia,a,b Anthony 3 D. Millerb,*, and Justin M. Chalkera,b* 4 5 a) Institute for Nanoscale Science and Technology, College of Science and Engineering, 6 Flinders University, Bedford Park, South Australia 5042, Australia 7 8 b) College of Science and Engineering, Flinders University, Bedford Park, South Australia 9 5042, Australia 10 11 To whom correspondence should be addressed: 12 tony.miller@flinders.edu.au 13 justin.chalker@flinders.edu.au 14 15 16

pH = 3 to pH = 11. At neutral pH, the sorbent (500 mg with a 10:1 ratio of silica to polymer) 23 could remove 90% of mercury within one minute from a 100 mL solution 5 ppm in HgCl2 and 24 99% over 5 minutes. It was found that sodium chloride, at concentrations comparable to 25 seawater, dramatically reduced mercury uptake rates and capacity. It was also found that the 26 spent sorbent was stable in acidic and neutral media, but degraded at pH 11 which led to 27 mercury leaching. These results help define the conditions under which the sorbent could be 28 used, which is an important advance for using this material in remediation processes.

30
Statement of novelty: Previous studies of the featured mercury sorbent did not detail the 31 scope and limitations at varying pH and salt concentrations, which are critical to know in 32 remediation projects. This is the first study of the effects of pH and sodium chloride on the rate 33 of mercury uptake by a polymer made from sulfur and limonene. Additionally, the first kinetic 34 model of mercury uptake was established for this material. Finally, leaching experiments under 114 10-fold with 5% HCl to stabilise mercury species for subsequent analysis. The experiment was 115 repeated in triplicate. The experiment was also repeated with uncoated silica gel as a control.

116
Mercury concentrations were then determined by cold vapour atomic absorption spectroscopic 117 analysis (CVAAS).

120
Silica gel coated with poly(S-r-limonene) (500 mg, 10:1 silica:poly(S-r-limonene)) was added to 121 250 mL beaker and mixed with magnetic stirring. Next, an aqueous solution of 5 ppm HgCl2 122 (100 mL) was added. This solution was prepared at various pH values and sodium chloride 123 concentrations, as described below (1.2.5 and 1.2.6). The solution was sampled over time by 124 drawing 1.00 mL of the solution into a 3 mL syringe equipped with a syringe filter (nylon, 0.45 125 µm). In this way, the solution was separated from the sorbent during sampling. Samples (1.00 126 mL) were taken every 10 seconds for the first minute and then at 90, 120, 180, and 300 seconds 127 of total sorption time. The experiment was completed in triplicate. A control experiment was 128 done in which 450 mg of uncoated silica was used as the sorbent. All samples were diluted 10-129 fold with 5% HCl and then mercury concentrations were determined by CVAAS.

132
Aqueous solutions of 5 ppm HgCl2 were prepared at pH values of 3, 5, 9, and 11. The solution at 133 pH =3 contained 1 mM HCl, the solution at pH = 5 contained 10 µm HCl, the solution at pH = 134 9 contained 10 µm NaOH, and the solution at pH = 11 contained 1 mM NaOH. A 5 ppm solution 135 of HgCl2 was also used without adjusting the pH; this sample is referred to as the neutral sample 136 (pH = 6.99). The mercury solutions were added to the silica gel coated with poly(S-r-limonene) 137 and sampled and analyzed by CVAAS according to the general protocol described in 1.2.4. In a 500 mL beaker, a 20 ppm solution of HgCl2 (200 mL) was added along with silica gel coated 148 with poly(S-r-limonene) (4.00 g). The sorbent was prepared as described above and contains a 149 10:1 mass ratio of silica to the poly(S-r-limonene) coating. The mixture was stirred for 10 150 minutes and then the sorbent was isolated by filtration and dried under vacuum before splitting 151 up into 500 mg portions. The sorbent samples, bound to mercury, were then added to 50 mL 152 centrifuge tubes, followed by 50 mL of solutions of varying pH or sodium chloride concentrations 153 (pH of 3, 5, 7, 9, 11, or aqueous solutions of 6.85 mM or 599 mM NaCl, prepared as described in 4 the sorption experiments). The mixtures were rotated on an end-over-end mixer and sampled 155 at 10 minutes, 3 hours, and then 1, 2, 8, 14, 21, and 28 days. All samples were diluted 10-fold 156 with 5% HCl and then mercury concentrations were determined by CVAAS.

179
This rate equation is motivated by the following physical reasoning. The product term is a 180 measure of the likelihood of an entity of X and an entity of S coming into sufficiently close contact 181 that the reaction (1) is possible, while ! is a proportionality factor that also expresses the 182 likelihood that a close contact will actually result in the formation of the adsorbed complex XS.

183
It is to be expected that ! will depend on a number of factors, in particular, the local chemical 184 environment and temperature. The presence of other species, although they may not have any 185 obvious direct involvement in the reaction, may give rise to various forms of interactions, for 186 example crowding, shielding, attraction or repulsion, which can influence the likelihood of the 187 reaction (1) occurring and so will affect ! . Likewise, the presence of other nearby entities of X 188 and S may also give rise to interactions and so influence whether a close contact leads to a 189 reaction. So, and may affect the overall reaction rate not just through the product term , Here the reverse rate coefficients !$ and "$ have units of s -1 or equivalent. Just as for ! and 226 " , they will depend upon the chemical environment but are assumed approximately constant 227 in any given situation.

250
The latter feature means the product-referred to here as poly(S-r-limonene)-is not as 251 malodorous as in the original synthesis. Hasell and co-workers also showed that poly(S-r-252 limonene), as a soluble oligomer, can be used to coat silica gel for mercury sorption. 29 The focus 253 of this report, however, was primarily on the use of accelerators and catalysts in the 6 copolymerisation of sulfur with alkene-containing monomers, and not mercury sorption.

255
Therefore, there is a need to characterise this sorbent in greater detail to understand its scope 256 and limitations in mercury sorption. Accordingly, in this study we evaluated its use in mercury 257 sorption and desorption at varying pH and salt concentrations, and we developed a model to 258 account for the observed kinetics. Together, these results will help guide the deployment of this 259 sorbent in mercury remediation.

260
The copolymerisation was run according to Hasell's protocol, reacting

371
In most cases, varying ! from its best fitted value by ±15% or more produced a noticeably 372 poorer fit. So ±15% can be thought of as an informal confidence interval for the parameter 373 estimates. A summary of the results is given in the

384
2. For the neutral case (pH = 6.99) and the acidic cases (pH = 3 and pH = 5) the single 385 reaction model was adequate to fit the experimental data well (Fig. S31). All these cases

386
show a continuing decrease in Hg concentration with time over the sampling period.

387
Indeed, the log-log plots asymptote to a slope of -1, indicating a 1/ asymptotic behaviour 388 (in non-logarithmic units). The physical interpretation of these results is that ultimately 389 all X will be adsorbed and all adsorption sites S will be utilised.

391
3. For the basic cases (pH = 9 and pH = 11) the two-reaction model was needed to fit the 392 apparent steady state that was approached. Here the competing species W was taken to 393 be NaOH. The physical interpretation of these results is that ultimately all adsorption 394 sites S will be utilised, some by X, and the remainder by W. The ratio of the amounts of 395 X and W adsorbed depends on the ratio ! / " as well as the starting stoichiometric ratios 396 % and & . Further details can be found in the Supplementary Material (Fig. S32).

398
4. For the tap and seawater cases, the two-reaction model was again needed to fit the 399 apparent steady state that was approached (Fig. S33). Here the competing species W was 400 taken to be NaCl. The physical interpretation is similar to that for the base cases but 401 with NaCl as the competing species.

423
It is instructive to plot 1/ ! against pH or the log of the chloride ion concentration (as 424 appropriate for the case). This is shown in Figure 4. As all cases were prepared from HgCl2,

425
there was always a background Clconcentration arising from the 5 ppm HgCl2. For this reason,

426
the pH = 7 case has been treated as a (low level) reference for the acid, base and chloride cases.  As seen in Table 1 and Figure 4, the poly(S-r-limonene) sorbent is fastest at neutral or near-443 neutral pH, with reduced rates of mercury uptake below or above pH 7. For example, within 5 444 minutes, 500 mg sorbent captured 95, 96, 94 and 90% of mercury from solutions at pH 3, 5, 9 445 and 11, respectively. In contrast, the sorbent removed 99% of the mercury in the same 5 minute 446 period in the neutral sample under otherwise identical conditions. The pH might alter the 447 mercury speciation, which could account for these differences in rates of mercury uptake. Note 448 that for neutral and low pH the single reaction model fits the experimental data well, suggesting 449 that the availability of sorption sites is not significantly affected by these pH values. This is also 450 shown by Figure 3 where there is no apparent flattening of the curves for cases A, B and C. This

465
The origin of the inhibition can be multi-faceted, but the models in Figure 3F and 3G

477
To test if a higher concentration of sorbent can achieve more complete mercury binding in the 478 presence of sodium chloride, 9 times the mass of sorbent was used in an identical sorption 479 experiment. The kinetic model was used to predict that this amount of sorbent would provide 480 sufficient binding sites to overcome the competing processes with NaCl. The qualitative concept 481 here was that solely by a mass action effect, adding more sorbent would speed up both the 482 desired mercury sorption and the undesired competing reaction. However, due to the relative 483 rates of these reactions, ! / " , the sorption reaction is now able to proceed further before a 484 steady state is achieved. Accordingly, 4.5 g poly(S-r-limonene) coated silica was mixed with 100 485 mL 5 ppm Hg 2+ containing 599 mM NaCl for 5 minutes with regular sampling. Within the first 486 10 seconds the sorbent removed 83% of the mercury present-a greatly improved initial uptake 487 than observed in the original experiment with 9-times less sorbent. Over 5 minutes, 91% of the 488 mercury was removed (Fig S36).

505
All other samples had mercury levels at or below the limits of detection by CVAAS, indicating 506 leaching was not significant for pH = 2, 5, 7, or 9. Leaching was also not significant for the 507 samples with added sodium chloride. SEM analysis of the sorbent did reveal that the high pH 508 degrades the polymer coating (Figs 5B and S37). This could lead to the release of mercury, 509 perhaps bound to suspended polymer particles (or the products of polymer degradation). Further 510 study is required to determine the mechanism and speciation of the mercury leaching at 511 elevated pH. However, minimal leaching was observed for all other samples suggesting that the 512 mercury remains strongly bound to the sorbent even in highly acidic media or brine.

525
Poly(S-r-limonene)-coated silica was evaluated as a mercury sorbent over a range of pH values 526 and also in the presence of sodium chloride. The sorbent rapidly removed HgCl2 from water at 527 or near neutral pH. Slightly reduced rates of uptake were observed at both low and high pH, 528 but the sorbent was still effective across this wide pH range. One-reaction or two-reaction 529 kinetic models were fitted to the experimental sorption results. These modes suggest that under 530 low (acid) and neutral pH conditions, mercury sorption is a single reaction process which will 531 ultimately proceed to completion until one or both of the mercury or the sorbent is consumed.

532
In contrast, for basic pH and in the presence of NaCl, the models suggest that there are 533 significant competing reactions whereby some of the sorbent becomes unavailable for mercury 534 binding. It was also discovered that sodium chloride severely inhibits mercury binding, which 535 could limit the use of the sorbent in salt water systems. The spent sorbent was found to be stable 14 and did not leach significant mercury from pH = 3 to pH = 9. Aqueous sodium chloride also did 537 not lead to leaching. However, at pH = 11 the coating degraded and mercury was released into 538 the solution, possibly bound to suspended polymer particles or other polymer degradation 539 products. Together, these results suggest that the sorbent is most effective at low and neutral 540 pH and that elevated pH can lead to polymer degradation. This assessment of the scope and 541 limitations of this sorbent will help define the conditions for which it is most effective in the 542 field. While this study focussed on inorganic mercury, future studies will be carried out that 543 evaluate the sorbent on a broader range of mercury species and field samples.