Experimental investigation into methane production from hydrate-bearing clayey sediment by CO2/N2 replacement

The replacement of gas hydrate in clayey sediment by a CO2/N2 (20:80) gas mixture injection was experimentally studied to explore the influence of clay on the gas exchange behaviours in the gas hydrate. Clay (montmorillonite) and silica sand were mixed in three different proportions (clay mass ratios of 10%, 30% and 50%) to simulate the host sediments of natural gas hydrate while pure silica sand sediment was selected for comparison. Experimental results showed that clay hindered gas diffusion during the initial replacement stage and thus reduced the methane recovery rate. In the later stage, the gas exchange between CO2/N2 and methane in the hydrate structure might be subject to thermodynamic inhibition and geometric constraints of the clay interlayer. Moreover, the CO2 sequestration ratio was lowered significantly in the sediment with a 50% clay mass ratio. Therefore, it was determined that clay has an inhibitory effect on gas hydrate replacement by CO2/N2.


Introduction
Natural gas hydrate (NGH) is a nonstoichiometric, crystalline compound formed by water and light hydrocarbons (mainly CH 4 ) under certain temperature and pressure conditions (Li et al., 2012;Sloan and Koh, 2007). NGH has great potential as a next-generation energy source due to large reserves; NGH resources are mainly stored in both clayey and sandy sediments (Boswell and Collett, 2011;Chen et al., 2019;Collett, 2002). Clayey NGH deposits are distributed worldwide and NGH resources account for more than 90% of all global hydrate resources (Johnson, 2011;Li et al., 2018b). At present, the three main methods of NGH production involve depressurization (Wang et al., 2016;Yang et al., 2018), thermal stimulation (Sun et al., 2014) and inhibitor injection (Zhao et al., 2016); all three are based on breaking down the NGH phase equilibrium through external stimulation (Chong et al., 2016;Li et al., 2016). Since NGH decomposition may cause hydrate reservoir sediment instability which can result in submarine slides or subsidence (Moridis et al., 2011), CO 2 replacement is recognized as a promising option for recovering natural gas without hydrate decomposition . It has the added benefit of permanent CO 2 sequestration in the hydrate reservoir (Koh et al., 2016;Zhao et al., 2015).
The concept of NGH production with CO 2 replacement was first proposed by Ohgaki et al. (1996), and both its thermodynamic and kinetic feasibilities have been demonstrated (Smith et al., 2001;Uchida et al., 2010). Subsequent work utilized N 2 as an additional replacement media to enhance the CH 4 -CO 2 replacement in NGH (Shin et al., 2008). Park et al. (2006) compared the replacement process by injecting a CO 2 /N 2 gas mixture (20/80 mol% to simulate flue gas from a power plant) as well as pure CO 2 . They found the total CH 4 production increased from 64% to 85% upon the addition of N 2 to CO 2 . Moreover, the direct injection of a CO 2 /N 2 gas mixture not only reduced CO 2 separation costs but also prevented CO 2 liquefaction and potential pipeline blockage problems as compared with the injection of pure CO 2 into hydrate reservoirs . The successful application of the replacement technique by using a binary CO 2 -N 2 gas mixture on the Alaska North Slope in the United States demonstrated the feasibility of this technique in the field (Boswell et al., 2016). In order to explore the NGH replacement mechanism, the phase equilibria of CO 2 /N 2 , CO 2 /CH 4 and CH 4 /CO 2 /N 2 gas mixtures and the distribution of the mixed gas molecules in hydrate structures were investigated (Herri et al., 2011;Lim et al., 2017;Sun et al., 2017). Additionally, extensive research efforts have focused on factors influencing CH 4 -CO 2 replacement that include temperature, pressure, sediment particle size, permeability, water salinity, water saturation and hydrate saturation (Lee et al., 2014;von Solms, 2016, 2018;Sun et al., 2018;Wang et al., 2018). However, previous studies were mainly focused on gas hydrates in water or silica sand and scant attention was paid to the influence of clay on CH 4 -CO 2 replacement.
The host sediments of NGH mainly consist of fine-grained silt, clay and sand (Boswell and Collett, 2011;Koh et al., 2016); the clay content of hydrate deposits varies in different regions (Kumar et al., 2015;Liu et al., 2015). Compared to silt and sand, clay has a special structure which causes NGH behaviour to differ from gas hydrates in water or silica sand (Cygan et al., 2004;Guggenheim and van Groos, 2003); therefore, CH 4 -CO 2 swapping process in NGH may be affected by clay. Moreover, clay concentration variations also affect the thermodynamic phase behaviour of NGH (Seo et al., 2009). However, only a few reports have dealt with the effects of clay on CH 4 -CO 2 replacement; even though clay is one of the main components of NGH host sediments. Yang et al. (2008) investigated CH 4 -CO 2 replacement in silica sand and a kaolinite-sand mixture using liquid CO 2 . They observed a significantly lower methane recovery rate in the kaolinite-sand mixture relative to silica sand. Koh et al. (2012) compared gas replacement in pure CH 4 hydrates, NGH sediments and CH 4 hydrates bearing clay (montmorillonite) by injecting a CO 2 and CO 2 /N 2 gas mixture. They found that the total recovered CH 4 for the all three samples were nearly identical; however, the methane recovery rate was largest in the pure CH 4 hydrate, followed by the NGH sediments and finally the CH 4 hydrate-bearing clay. However, those previous studies were mainly focused on CH 4 recovery rates that were affected by clay. The influence of clay on CH 4 , CO 2 and N 2 variations in both gas and hydrate phases as well as CO 2 sequestration, which are important for analysing the mechanism of gas hydrate replacement (Li et al., 2018a;Yang et al., 2017), have not been investigated. Thus, the influence of clay on CH 4 -CO 2 exchange behaviours may not be well understood. In addition, although the clay concentration varies in the NGH host sediments, the effect of different clay levels on CH 4 -CO 2 replacement has received very little attention. Therefore, it is significant to completely reveal the CO 2 -CH 4 replacement mechanism in hydrate-bearing clayey sediments and investigate the influence of clay content on gas exchange behaviours.
In this work, replacement of CH 4 hydrate in clayey sediments by a CO 2 /N 2 gas mixture was experimentally studied and the replacement in CH 4 hydrate-bearing sand sediment was analysed for comparison. The clayey sediments used in these experiments were simulated from the host sediments of NGH in the Shenhu area of the South China Sea. In that area, montmorillonite and illite are the primary clay mineral types and the clay content in the NGH reservoirs ranged from 0.4% to 30% (Li et al., 2018b;Liu et al., 2012). Thus, clay (montmorillonite) and silica sand were mixed in three different proportions (the three clay mass ratios were 10%, 30% and 50%). Montmorillonite was also chosen because it is generally the richest clay mineral in NGH bearing sediments worldwide (Zhou et al., 2011). The gas mixture of 20 mol% CO 2 and 80 mol% N 2 was used to simulate flue gas. During gas replacement, the evolution of CH 4 , CO 2 and N 2 in the gas phase were examined using gas chromatography (GC) to study gas exchange behaviours between the CO 2 /N 2 and CH 4 hydrate-bearing clay/sand sediments.

Materials and apparatus
Analytically pure (99.99%) CH 4 , CO 2 and N 2 gases used were supplied by Beifang Special Gas Industry Corporation (China). The silica sand was 200-400 mesh and the montmorillonite was 1250 mesh, both were supplied by Hunyuan Junhong New Materials Co., Ltd. (China). Double-distilled water was used. A schematic diagram of the experimental apparatus is shown in Figure 1. The experimental apparatus is composed of a reactor, three gas cylinders, a glycol bath, a GC, a set of data acquisition system and so on. The reactor is made of 316 L stainless steel, is 18 cm tall with an internal diameter of 6 cm for an effective volume of 509 mL. The temperature was measured by two secondary platinum resistance thermometers (type pt100), which were uniformly arranged in the vertical direction inside the reactor; a pressure sensor was used to monitor the reactor pressure. The small cylinder was used to store CH 4 and provided a sufficient gas source for hydrate formation in the reactor, thus avoiding the secondary injection of CH 4 to form a hydrate. The glycol bath was used to regulate the reactor temperature of the reactor. The uncertainties of measured pressure and temperature are AE0.01 MPa and AE0.1 K, respectively.
Procedures CH 4 hydrate formation in clayey sediment. At first, a certain amount of silica sand and montmorillonite were dried at 373.2 K for 72 h. Silica sand, montmorillonite and distilled water were thoroughly mixed at predetermined mass ratios and those mixtures were loaded into the reactor in small quantity batches. Each batch was compacted to remove voids in the reactor. The entire system was then evacuated and purged three times with CH 4 to vent the air completely. The reactor and the small cylinder were pressurized to approximately 9 MPa using CH 4 and set at room temperature for 24 h to monitor for leaks. When the reactor pressure stabilized, the reactor and small cylinder were put into the glycol bath at 275.2 K. The CH 4 hydrate began to form once the system temperature dropped to a certain value. The system pressure gradually decreased with hydrate formation. Finally, when the pressure and temperature stabilized for 48 h (Figure 2), CH 4 hydrate formation was assumed to be finished. Preparation of the silica sand sediment hydrate sample was consistent with the aforementioned procedure.
CO 2 /N 2 injection and CH 4 recovery. After preparing the CH 4 hydrate sample, the glycol bath was set to 268.2 K and the small cylinder valve was closed. When the temperature and pressure of the reactor stabilized, residual CH 4 in the reactor was released through a sixway valve and the reactor was evacuated for 20 s. Afterwards, the room temperature CO 2 / N 2 was injected into the reactor through a cold pipe (a 5 m pipeline placed in the glycol bath) until the pressure in the reactor reached the desired value and the initial gas sample was taken. This CO 2 /N 2 injection process was also controlled by the six-way valve and lasted for approximately 30 s. The reactor was sealed and the bath was set to 274.2 K for gas hydrate replacement. During replacement, a series of gas samples (100 mL, atmospheric pressure) were taken at predetermined time intervals and measured by GC. Experimental parameters for CH 4 hydrate replacement are shown in Table 1.

CH 4 hydrate formation
Figure 2(a) shows temperature and pressure variations during hydrate formation in the clayey sediment (Run 4, clay mass ratio of 50%). The temperature and pressure evolution in Run 4 was similar to Runs 2 and 3 and was selected to represent typical results of CH 4 hydrate formation. As shown in Figure 2(a), the trend of temperature 1 and temperature 2 is almost the same, which indicates that the hydrate is formed evenly in the clayey sediment. Meanwhile, both temperature and pressure decreased linearly after the reactor is placed in the bath. After approximately 2 h, the pressure dropped sharply, but the temperature increased suddenly and reached a maximum of 281.4 K, which indicated the rapid formation of hydrate. The pressure stabilized after $5 h, indicating the end of hydrate formation. Figure 2(b) shows temperature and pressure data during hydrate formation in silica sand sediment. Temperature and pressure variations over time in Run 1 agreed with those from Run 4; however, the pressure drop in Run 1 was much larger than Run 4. The pressure drop difference between the two runs was $1.75 MPa, indicating that hydrate formation in silica sand was significantly larger than clayey sediment. This is attributed to the high water absorption and expansion characteristics of clay. When water was mixed with clayey sediment, a number of water molecules were orderly adsorbed and incorporated into the interlayer space of clay sheets (Uchida et al., 2004). During hydrate formation, the ordered water structure of clay inter-layer promoted CH 4 hydrate nucleation (Seo et al., 2009;Uchida et al., 2004), but growth of the hydrate crystal was limited due to the water activity reduction caused by geometrical constraints (Uchida et al., 2002). Furthermore, the pore size in clayey sediment is smaller than silica sand sediment since clay particles swell after absorbing water; that swelling hinders hydrate formation outside the clay inter-layers. Figure 3 illustrates the mole ratio of initial CH 4 conversion to hydrate. As can be seen, the conversion ratio of CH 4 to hydrate decreased with a clay mass ratio increase; for Run 1 in silica sand sediment, a 43% conversion ratio was achieved and was the highest for all four runs. Consequently, it can be concluded that the increase in clay content will enhance the inhibition of hydrate formation.

Variation of gas compositions in the gas phase
Direct component determination by GC is helpful for quantitative analysis of the gas phase in the reactor. Based on gas-phase analysis and the gas equation of state, mole percent variations CH 4 , N 2 and CO 2 in the gas phase before sampling during replacement were obtained (Figure 4). Runs 1-4 all showed similar gas phase trends, CH 4 increased with time, while N 2 and CO 2 decreased. That indicated CH 4 was recovered from the hydrate in the reactor after injection of the CO 2 /N 2 gas mixture. In particular, for all four runs, the mole content of CH 4 initially increased rapidly but decreased with time. Using Run 2 as an example, from 0 to 4 h (Stage 1), the mole percentage of CH 4 increased from 0 to 9.65%, but over the next 274 h (Stage 2), the mole percentage increased by 21.35 mol%. The CH 4 recovery rate of Run 2 in Stage 1 was 32 times higher than that in Stage 2. This observation is likely due to the reduction of driving forces over time during hydrate replacement. Previous studies have shown that the driving force of gas hydrate replacement is the chemical potential gradient between the hydrate phase and the gas phase (Schicks et al., 2011).  With the progress of the hydrate replacement by CO 2 /N 2 gas mixture, the recovered CH 4 will gradually dilute the CO 2 /N 2 gas mixture, which reduces the gradient of the chemical potential between the hydrate phase and the gas phase. Table 2 presents the average recovery rate of CH 4 at Stages 1 and 2. For all four runs, the average recovery rate of CH 4 in Stage 1 was more than 30 times higher than that in Stage 2. However, it should be noted that in Stage 1, the average recovery rate in silica sand was higher than clayey sediment. This is attributed to the different diffusion ranges of CO 2 /N 2 in silica sand and clayey sediments. For the CH 4 hydrate in silica sand, after CO 2 /N 2 injection, the mixed gas diffuses rapidly in the void space of the sediment system and then contacts the hydrate for CH 4 recovery. But the clayey sediment has smaller pore sizes than silica sand due to expansion of the clay particles which hinders diffusion of the CO 2 /N 2 gas mixture in the void space and results in a smaller gas diffusion range.
Therefore, as the clay mass ratio increases, gas diffusion is hindered and lowers the average CH 4 recovery rate. However, as listed in Table 2, the recovery rate in Run 4 was higher than Runs 2 and 3, although the clay mass ratio in Run 4 was the largest. This abnormal observation could be associated with CH 4 hydrate saturation in Run 4. An analysis suggested that an increase in hydrate saturation reduced the void space in the porous media system and restricted diffusion of the replacement gas. As shown in Table 1, the hydrate saturation in Run 4 was 15.75%, significantly lower than the other runs. Despite the clay particle expansion that leads to a void space reduction in the porous media system of Run 4, the low-hydrate saturation partially offsets this and increases the diffusion range of the injected CO 2 /N 2 gas mixture. Figure 5 illustrates the mole percent variation CH 4 , N 2 and CO 2 in the hydrate phase before gas sampling in the replacement process. As can be seen from Figure 5, the trend of variations for all runs was similar; the mole content of CH 4 in the hydrate phase decreased with time, while N 2 and CO 2 increased. This indicated that gas exchange occurred between the CO 2 /N 2 and the CH 4 hydrate. Furthermore, this phenomenon was also observed during early research on gas hydrate replacement using a CO 2 /N 2 gas mixture or pure CO 2 (Li et al., 2018a;Sun et al., 2018). In Figure 5, for Run 1, the mole percent of CO 2 in the hydrate phase was nearly identical to the mole percent of N 2 at the end of Stage 1 (0-4 h); however, the CO 2 mole percent was higher than N 2 for Runs 2-4. In particular, the CO 2 mole percent in Run 4 maximized at the end of Stage 1 (16.03%). This observation could be related to the formation of mixed gas hydrate after CO 2 /N 2 injection. The equilibrium pressure of CO 2 /N 2 binary hydrate in all experimental runs was $7 MPa at the experimental temperature. The pressures in Runs 1-4 during replacement were higher than this phase equilibrium pressure, indicating the formation of a CO 2 /N 2 binary hydrate could occur after gas injection. For ease of analysis, the CO 2 to N 2 molar ratio in the hydrate phase is referred to as the Sequestration Ratio (C/N). Figure 6 shows C/N variations over time for all four runs. At 4 h, the C/N was 3.18 when the clay mass ratio was 50%, close to the theoretical consumption ratio of CO 2 /N 2 (3:1) (Seo et al., 2015). This means that gas exchange between the CO 2 /N 2 mixture and the CH 4 hydrate was dominant in Stage 1 compared to formation of the mixed gas hydrate. Moreover, the value C/N at 4 h decreased with the decreasing clay content in the clayey sediment and was lowest in the silica sand sediment; this indicated that the extent of mixed hydrate formation in Stage 1 was largest for silica sand sediment and decreased as the clay content increased. This analysis suggested the difference in the extent of mixed hydrate formation between the silica sand sediment and the clayey sediment was due to the difference in the amount of free water (water unabsorbed by clay) in the system. As mentioned, clay has a strong water absorption capacity and many water molecules were adsorbed into the clay inter-layer system. The amount of free water in the system decreased as the clay content increased, thus causing gas exchange to dominate in Run 4. Meanwhile, the free water content in Run 1 was higher due to the weak water absorption capacity of silica sand which favoured formation of the CO 2 /N 2 binary hydrate in the silica sand sediment relative to the clayey sediment. It should be noted that the diffusion range of the CO 2 /N 2 gas mixture in Stage 1 may also affect the extent of mixed hydrate formation in the system. In addition, C/N remains low after 40 h and is related to the distinct gas exchange behavior of CO 2 and N 2 . Figure 7 shows water consumption at 4, 40 and 280 h for all four runs during replacement. Water consumption in Run 1 was significantly higher than Runs 2-4 over 4 h, which further supports the aforementioned observation and analysis. Furthermore, water consumption at 40 h was slightly less than that at 280 h, indicating that after 40 h, the amount of mixed hydrate formation in Runs 1-4 was very small and all experimental runs were dominated by gas exchange between the replacement gas and the CH 4 hydrate. Combining results from Figures 5 and 8, after 40 h, the increase in CO 2 mole percent slowed considerably while the N 2 mole percent continued to increase; meanwhile, the ratio of the N 2 increments in the hydrate phase to the CH 4 reductions in hydrate phase from 40 to 280 h was between 0.8 and 0.9. This indicated that after 40 h, the gas exchange amount between N 2 and CH 4 molecules in the hydrate structure was significantly higher than for CO 2 molecules. These observations are attributed to the stronger diffusibility of N 2 relative to CO 2 molecules. Since N 2 is smaller than CO 2 (Liu et al., 2016), N 2 molecules migrate deeper into the hydrate layer to recover CH 4 molecules occupied in water cavities. Therefore, all experimental runs were dominated by the gas exchange between N 2 and CH 4 the hydrate from 40 to 280 h.

Variation of gas compositions in the hydrate phase
The percentage increase of N 2 in the hydrate phase for Runs 1-4 after 40 h is also illustrated in Figure 8. As can be seen, the percentage of N 2 in silica sand sediment was higher than clayey sediment and the increased percentage of N 2 decreased as the clay mass ratio increased in the clayey sediment. This indicated the replacement extent in silica sand sediment after 40 h was higher than clayey sediment since the gas exchange between the N 2 and CH 4 hydrate dominates all experimental runs. This may be due to difficulties in replacing CH 4 of the clay inter-layer with N 2 . As the replacement process proceeds in the clayey sediment, N 2 molecules penetrate the clay inter-layer to extract CH 4 molecules; however, since the thermodynamic stability of the CH 4 hydrate in the clay inter-layer exceeds the pure  hydrate (Park and Sposito, 2003;Seo et al., 2009), CH 4 extraction in the clay inter-layer might be thermodynamically inhibited. Additionally, N 2 migration in the clay inter-layer is likely to be restricted due to geometric constraints resulting in lower N 2 penetration depths in the hydrate layer of clayey sediment relative to the hydrate layer of silica sand.
CO 2 sequestration and hydrate replacement efficiency CO 2 sequestration. The CO 2 sequestration ratio is defined as the molar quantity of consumed CO 2 in the gas phase to the initial molar quantity of CO 2 after gas mixture injection. CO 2 sequestration rates and CO 2 sequestration ratios are shown in Table 3. The CO 2 sequestration rate in silica sand sediment was highest and the sequestration rate in Run 4 was higher than Runs 2 and 3 despite the elevated clay mass ratio for Run 4. Remarkably, CO 2 sequestration rates for the four runs agreed with H 4 recovery rates discussed in Variation of gas compositions in the gas phase section. As a consequence, the different CO 2 sequestration rates for all four runs was mainly due to the difference in gas diffusion ranges caused by the interaction between clay expansion and CH 4 hydrate saturation. Moreover, it should be noted that a part of the CO 2 consumption rate in Stage 1 was contributed by mixed hydrate formation. On the other hand, as listed in Table 3, 80.4%, 73.1%, 69.5% and 49.6% of the initial CO 2 in the gas phase were captured and retained in the hydrates in Runs 1-4, respectively. These results demonstrate that CO 2 sequestration capacity of silica sand sediment surpassed clayey sediment. This is understandable because the amount of free water in the clayey sediment is limited and CO 2 migration into the clay inter-layer is difficult to replace CH 4 in the hydrate structure. Accounting for the sharp slope change of CO 2 sequestration in the clayey sediment with a 50% clay mass ratio, it suggested the hydrate reservoirs with high clay content are not conducive to CO 2 capture and storage by CO 2 /N 2 injection.
Hydrate replacement efficiency. Replacement efficiency is defined as the ratio of the molar quantity of CH 4 in the gas phase to the initial molar quantity of CH 4 in the hydrate phase. Figure 9 illustrates replacement efficiencies of Runs 1-4 over time. Initially, replacement efficiencies for all four runs increased rapidly (0-4 h). This increase was attributed to the initially high fugacity between the gas phase and the hydrate phase. Figure 10 shows the replacement efficiencies after hydrate replacement and the mole ratios of injected CO 2 /N 2 to the initial CH 4 hydrate before replacement for all four runs. In Runs 2-4, the replacement efficiency order was Run 4 > Run 3 > Run 2, which agreed with the mole ratio order of injected CO 2 /N 2 to the initial CH 4 hydrate. This analysis indicated that a higher mole ratio maintained a higher partial fugacity of the CO 2 /N 2 which promoted gas exchange between the gas phase and the hydrate phase and resulted in higher replacement efficiency. This phenomenon is consistent with Sun's finding that the replacement efficiency of the hydrate increased with the CO 2 fluid mole ratio increase to the initial hydrate before replacement (Sun et al., 2018). However, it should be noted that the replacement efficiency of Run 1 was Figure 9. Variations of replacement efficiencies in Runs 1-4 over time during hydrate replacement. Figure 10. Replacement efficiencies after the hydrate replacement and the mole ratios of injected CO 2 /N 2 to initial CH 4 hydrate before the replacement reaction in Run 1-4.
higher than Runs 2 and 3, although the mole ratio in Run 1 was lower than Runs 2 and 3. This anomalous result indicated the restrictive effects of clay on replacement efficiency. This is understandable since CH 4 extraction in the clay inter-layer might be thermodynamically inhibited and N 2 migration in the clay inter-layer is likely to be restricted due to geometric constraints. In conclusion, even though higher clay content has a stronger inhibitory effect on CH 4 recovery, increasing the amount of CO 2 /N 2 injected into hydrate reservoirs is advantageous for improving hydrate replacement efficiency.

Conclusions
The replacement reaction occurring in clayey sediments containing CH 4 hydrates with a CO 2 /N 2 mixture was examined and the replacement in hydrate-bearing sand sediment was analysed for comparison. The following conclusions were drawn: 1. Increased levels of clay in sediments inhibited hydrate formation. This was due to limited hydrate crystal growth in clay inter-layer by water activity reduction caused by geometrical constraints as well as the reduced pore space outside the clay inter-layer which resulted from clay particle expansion. 2. During the initial stage of hydrate replacement, the CH 4 recovery rate in clayey sediment was lower than silica sand but minor hydrate saturation reduced the diffusion barriers of the replacement gas; this improved the recovery rate. Moreover, the clay diminishes the free water content in sediment due to its strong water absorption capacity which reduced the extent of mixed hydrate formation during replacement. 3. Additional analysis of compositional evolution in the hydrate phase indicated that the clay was not conducive to hydrate replacement in latter stages, since CH 4 extraction in the clay inter-layer might be thermodynamically inhibited and gas migration was likely restricted due to geometric constraints of the clay interlayer. 4. It was also found that the CO 2 sequestration capacity of clayey sediment is lower than silica sand and the hydrate reservoirs with high clay content are not beneficial for CO 2 capture and storage by CO 2 /N 2 injection. Furthermore, clay has an inhibitory effect on hydrate replacement efficiency but increasing the amount of CO 2 /N 2 injected into hydrate reservoirs is advantageous for improving replacement efficiency.

Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was funded by the National Natural Science Foundation of China (No. 41672361 and No. 41876218), the Scientific and Technological Development Program of Jilin Province (No. 20170414044GH) and the Provincial and School Co-construction project (No. SXGJSF2017-5).