Vinyl chloride monomer (VCM) is a major chemical intermediate for the manufacture of polyvinyl chloride (PVC), which is the third most important polymer in use today [1, 2]. At present, the VCM is generally synthesized by hydrochlorination of acetylene in regions of the world where coal is abundant, especially in China. However, production of VCM in acetylene hydrochlorination units consumes about 1400 tons mercury per annum before the low-mercury catalyst was applied, which caused big environmental problems. With the implementation of the mercury restriction, the solution of environment pollution problems involving mercury is one of the key technologies for vinyl chloride producing by calcium carbide method [3, 4]. Although the gold based non-mercury catalyst invented by Hutchings' group shows superior performance and has been commercialized as mentioned in reference [5, 6]. The wide application of the gold catalysts by the PVC enterprises still need much time due to its high cost. While other non-mercury catalyst like nitrogen doped catalysts are not active enough for application in industry [7-9]. Early in 2012, the Chinese government required the loading of HgCl₂ to be reduced from 10 wt%–12 wt% to 4 wt%–6.5 wt%, which is called the low-mercury catalysts [10]. The fulfillment of the low-mercury catalyst in China has reduced more than 50% of the mercury consumption and the pollution of mercury has been greatly reduced in 2015. The low-mercury catalyst is still the major catalysts been used in the PVC factory nowadays before the mercury-free catalysts been mature and been widely applied. The stability of low-mercury catalysts should be different from the high-mercury catalyst due to the decreasing loading of HgCl₂ to 4%–6.5% will results in a single layer dispersion of HgCl₂ on carbon support [11]. However, the research work on the thermal stability of the low-mercury catalysts was neglected and rarely reported. The previous reports on the thermal stability of carbon supported mercury catalyst are mostly based on the high-mercury catalysts [12, 13]. If the thermal stability is improved, the sublimation of HgCl₂ will be suppressed and thus the environmental pollution can be greatly reduced since the catalysts will not be consumed during reaction.

At present, the strategies for the improvement of the stability for activated carbon (AC) supported mercury chloride (HgCl₂/AC) catalysts mainly focus on two aspects: Firstly, alkali chloride was used as ligand as reported by Li et al. [14]. They observed that the thermal stability of the HgCl₂/AC catalyst can be obviously improved when CsCl is co-impregnated with HgCl₂ to AC support. Except for the addition of various promoters, there are also a few reports about the study on the carbon support. Wei et al. [15] and Xu et al. [16] studied on the coconuts based and coal based ACs including metal oxide supported low-mercury catalysts, but they didn't give a direct thermal stability test of catalysts and the mechanism study on the role of carbon surface chemistry. The identification of the role of surface functional groups on the supported metal catalysts such as Ru [17], Pt [18] and Au [19] etc. is receiving increasing interest in carbon materials research recently. It is clearly concluded that the interaction of supported active phase with the carbon supports plays an important role which will determine the dispersion and stability of active phases [20]. Several studies also indicated that defective sites are of great importance to the adsorption [21-23]. However, to our best knowledge there are no reports which have been devoted to understand the role of carbon-oxygen surface complexes or surface defects on the thermal stability of supported HgCl₂ catalyst. The present work will concentrate on thermal stability of the low-mercury catalysts, the interaction of carbon surface functional groups and surface defects with HgCl₂ molecular by experimental and theory calculations.

The texture and surface structure of carbon supports were fully characterized and discussions are given in supporting information (Figs. S1–S7, Tables S1–S4). The coal and wood based commercialized activated carbons were used for study which is denoted as AC-C and AC-W, respectively. AC-W700 was prepared by calcination of AC-W under inert gas flow to remove the acidic functional groups. Briefly, all the activated carbons have a surface area higher than 1000 m²/g (Table 1), and the ash content is in range of 3.00%–3.25%. The impurities are mainly silica and alumina oxide with very few amount of calcium oxide. The elemental analyses of activated carbons are provided in Table S2. The main difference between these activated carbons is the surface structures and oxygen contents. AC-W has abundant oxygen containing surface functional groups (SFGs), the oxygen content is as high as 33% according to the CHNS elemental analysis. While the AC-W700 and AC-C has less SFGs, the oxygen content of AC-W700 is reduced to 16% and the oxygen content of AC-C is ca. 8%. As demonstrated by X-ray diffraction (XRD), thermogravimetric (TG), and Raman characterizations, although the graphitic degree of AC-W700 and AC-C is better than that of the AC-W, the I(D)/I(G) value is higher for AC-W700 and AC-C than that of AC-W. This means that more surface defects are produced by the thermal treatment at 700 ℃. The AC-C has more surface defects than that of AC-W due to the less oxygen content. As reported by Lu et al. [21], graphene with rich surface defects can be obtained by using an oxidation step followed by a thermal decomposition method. As demonstrated by the TPD and FT-IR characterizations, AC-W contains much acidic functional groups such as carboxylic acids, phenols, and carboxylic anhydrides, while the AC-W700 and AC-C have more basic functional groups as lactones, ketone, and quinone which are more thermally stable.

The low-mercury catalysts were prepared by an incipient-wetness impregnation method. The isothermal curves and pore size distributions (Fig. S8) and XRD patterns (Fig. S9) of these catalysts are also similar to the ACs indicates a homogenous dispersion of the HgCl₂ on the surface of ACs supports and no crystallite of HgCl₂ exist in the catalyst. The texture properties of these catalysts are given in Table 1. The decreasing of the surface area and pore volume of the catalysts compared with their corresponding ACs is small which indicates that the HgCl₂ molecule is monodispersed in the pore of ACs.

TG techniques were conducted under argon flow to evaluate the thermal stability of various low-mercury catalysts supported on the above ACs. TG and DTG profiles of the HgCl₂/AC-W, HgCl₂/AC-W700 and HgCl₂/AC-C catalysts are given in Fig. 1(a). For comparison, the TG and DTG profile of various ACs are also measured under same conditions, which are given in Fig. S10. The TG profile of HgCl₂ is provided in Fig. S11, the sublimation temperature is as low as 140 ℃. From Fig. 1, it can be seen that the mass loss curves can be obviously divided into three temperature ranges. The mass loss occurs at ca. 100 ℃ is caused by the evaporation of physical adsorbed water. The amounts of water for all the samples are below 0.6% due to all the samples were dried before the TG experiments. The second mass loss occurs at 200–500 ℃, there are obvious DTG peaks for all the catalysts. The peak of DTG profiles for HgCl₂/AC-W, HgCl₂/AC-W700 and HgCl₂/AC-C catalysts are at 302, 360 and 382 ℃, respectively. To eliminate the mass loss of the activated carbon, the subtracted data of the mass loss ratio at 200–500 ℃ of the catalysts and their related support are given in Table 1. The weight loss is in the range of 6.5%–7.3%, considering the experimental error, which is in the same level of the HgCl₂ loading. This confirms the mass loss peak in this temperature range can be assigned to the sublimation of the adsorbed HgCl₂ species. Since there is only one peak appears in TG curves for all the catalysts. It is in accord with the results characterized by the N₂ sorption and XRD techniques.

It is interesting to note that the sublimation temperature of HgCl₂ are all higher than the 200 ℃, and the maximum weight loss temperature of HgCl₂ for HgCl₂/AC-W, HgCl₂/AC-W700 and HgCl₂/AC-C are 302, 360 and 382 ℃, respectively. The higher sublimation temperature of the HgCl₂ indicates a stronger interaction of HgCl₂ with carbon support. Moreover, the difference of the HgCl₂ sublimation temperature for various catalysts and HgCl₂ may indicate that the interaction of HgCl₂ molecular with activated carbons is different and there are properly strong interaction between the HgCl₂ and the carbon supports.

X-ray photoelectron spectroscopy (XPS) was used to characterize the status of HgCl₂ on various catalysts. The high-resolution Hg 4f spectra for HgCl₂/AC-W, HgCl₂/AC-W700 and HgCl₂/AC-C catalysts were given in Fig. 1(b). It can be seen that the photoelectron spectra of Hg 4f clearly shows two doublet-peaks, the spin-orbit splitting of Hg 4f was 4.1 eV and intensity ratio was about 4:3 (Hg 4f_7/2: Hg 4f_5/2) [24]. For HgCl₂/AC-W, the main peak of Hg 4f_7/2 at 101.4 eV, which is attributed to Hg²⁺ species and there are apparently no Hg⁺ and Hg⁰ species [25-27]. For HgCl₂/AC-W700 and HgCl₂/AC-C, the main peaks are shifted to 100.8 eV. The fitting of the peaks shows there are two species exist in these samples with around 65% of mercury was ascribed to Hg⁺ species (100.8 eV) and around 30% of mercury was ascribed to Hg²⁺ species (101.4 eV).

From above analysis, it is clearly indicated that HgCl₂/AC-W700 and HgCl₂/AC-C catalysts are more stable than HgCl₂/AC-W. To further understand the origin of the enhanced stability, DFT calculations are performed to study the interactions between HgCl₂ molecule and carbon supports. One of the important differences between AC-W700, AC-C and AC-W is the oxygen group and surface defects as characterized by TPD and Raman techniques. The carboxylic and phenol groups are dominant on AC-W while AC-W700 and AC-C have more quinone and ketone groups. Therefore the binding energy of the HgCl₂ molecule at the different oxygen groups are calculated including quinone, ketone, carboxyl, lactone and phenol groups (Fig. S12, Table S5). It is clearly indicated that the binding energy of HgCl₂ is decreased along quinone, ketone, carboxyl, lactone and phenol groups. The quinone group is the most strong adsorption site for HgCl₂ while the phenol group is the weakest site among the investigated oxygen groups. Not only oxygen groups on the support but also defective sites such as vacancy and edge are considered in the calculations. The binding energy of HgCl₂ on defective site is indeed bigger than the strongest oxygen site (quinone) which are 49.20, 36.95 and 47.28 kJ/mol at zigzag, armchair edges and mono-vacancy respectively. The increased binding energy at defective sites is resulted from the saturated dangling bond. However, the binging energy of HgCl₂ at either oxygen site or edge site is still small and below 50 kJ/mol. Furthermore, the Mayer bond order of Hg–C and Hg–O is 0.37 and 0.2 respectively which indicate this is a weak intermolecular interaction. Interestingly, it is found the binding energy of HgCl₂ is up to 130 kJ/mol when adsorption takes place at the edge adjacent to a carbonyl group (ketone and quinone) as shown in Table S5. And the bond order is 0.64. It is suggested that the adjacent oxygen greatly enhances the binding energy at the edge site. As shown in our previous work, the insertion of oxygen group can decrease the aromaticity of the carbon materials and increase the reactivity [28]. Based on these discussions, the DFT results are consistent with the experimental observations. The calculations indicate either carbonyl groups or the nearby carbon atom at edge are the most stable binding sites for HgCl₂ molecule. This is well explained that the HgCl₂ catalysts on the AC-C and AC-W700 have a better stability.

Stability of the catalysts is an important issue for their application in industry, especially for the supported mercuric chloride catalysts. Because the loss of volatile mercury ultimately finds its way into the environment thus cause severe pollution of the environment [29]. The enthalpy of hydrochlorination of acetylene reaction is –124.8 kJ/mol, which is strongly exothermic and it will generate a hotspot in the reactor which moves through the fixed bed reactor in the direction of the flow of the reactants [30]. The reaction temperature range in industry are controlled in a range of 80–180 ℃, the space velocity of total reaction gas in the range of 15–100 h^-1 for a better control of temperature of reaction. In order to give a fast screen of the thermal stability of the catalysts, the reaction temperature and space velocity in our experiments was set at 180 ℃ and 1000 h^-1. The time on stream catalytic performances of the above catalysts supported on various activated carbons (HgCl₂/AC-W, HgCl₂/AC-W700 and HgCl₂/AC-C) in acetylene hydrochlorination are shown in Fig. 3(a). It can be seen that the initial activity is similar for all the three catalysts. But the deactivation rate of these three catalysts was evidently differed with the order of HgCl₂/AC-W > HgCl₂/AC-W700 > HgCl₂/AC-C. This order is same with the thermal stability of these catalysts characterized above. Combining with the XPS results and catalytic performance, both Hg²⁺ and Hg⁺ could be involved as center of catalysis in this reaction. This is similar with Au/carbon catalysts, as reported by Hutchings et al. the redox couple of Au⁺/Au³⁺ has been suggested to be the active center in VCM [6].

To further test the stability of the mercury catalyst, an experiment simulated the industrialized reaction condition has been done in lab. The conversion of acetylene and selectivity to vinyl chloride with reaction time is given in Fig. 3(b). The initial conversion of acetylene is about 96%, after 10000 h running (ca. 417 d) still as high as 88.6%. This excellent catalytic performance confirmed that thermal stability is the key point to life time of mercury catalysts as demonstrated in reference by Hutchings group's early report [12, 13]. According to TG characterization results, the initial evaporation of the most stable HgCl₂/AC-C catalyst is increased to above 300 ℃. There are almost 120 ℃ improved. The sublimation of HgCl₂ will not occur under industry operation temperatures (80–180 ℃). Therefore, the leaching of HgCl₂ for mercury catalysts used in industry caused by the thermally unstable properties can be greatly solved.

The thermal stability improvement of mercury related catalysts insures a long service life of mercury based catalysts under reaction conditions. This has significant impact on the elimination of the environmental pollution problems for mercury related catalytic process in industry. For example, commercial production of VCM (ca. 20 million tons PVC per annum in China is based on the acetylene based process) in acetylene hydrochlorination units consumes about 1000 tonnes Hg per annum. The fulfilment of the low-mercury catalysts for PVC production required the loading of HgCl₂ decreased from 10 wt%–12 wt% to 4 wt%–6.5 wt% has reduced the almost half of the mercury consumption. Therefore, the release of mercury chloride for the application of new generation low-mercury catalysts can be diminished to quite low level. This thermally stable low-mercury catalyst has already been successfully commercialized and applied in PVC factories. The environmental problems caused by thermally unstable mercury catalysts are greatly reduced. The combination of fundamental and experimental research proved essential for the successful development of this robust technology.

Acknowledgments

The authors thank Dr. Liang Wang in Zhejiang University for the Raman measurement and fruitful discussions with Professor Junling Lu in University of Science and Technology of China.