Optimized synthesis of nano-scale high quality HKUST-1 under mild conditions 1 and its application in CO2 capture

19 This study was focused on the development of an optimized method for the rapid synthesis of 20 nano-scale HKUST-1 with high yield, high surface area and high CO 2 uptake capacity but under 21 mild conditions. A series of HKUST-1 were synthesized under different conditions, such as 22 preparation time, temperature, activation method, etc. It was found that the nano-scale HKUST- 23 1 MOFs (T85-3-Pm4-120) was successfully synthesized at a high yield (87%) under low 24 temperature (85 o C) using a mixture of Triethylamine(TEA), Cu 2+ and trimesic acid (TMA) with a 25 molar ratio of 6:3:2. The highest porosity was achieved via this pristine HKUST-1 being activated 26 (powder activation, drying at 120 o C) four times using methanol to remove impurities trapped in 27 the pores. The best HKUST-1 MOFs (T85-3-Pm4-120) hereby prepared was then tested in CO 2 28 adsorption and exhibited an adsorption capacity of 2.5 mmol/g. It is therefore demonstrated that 29 the new approach proposed in this study is a rapid and effective way to synthsize highly porous 30 HKUST-1 MOFs under mild conditions, which is of comparable surface area and CO 2 uptake 31 capacity with those MOFs prepared under harsh conditions. 32

In this study, the effort was made to develop and optimize a new method for the preparation of 67 6 phase and crystallinity were analyzed by using a Powder X-ray diffraction (XRD) with a scanning 105 rate of 0.02°/s (2θ) using monochromatic CuK radiation (Bruker D8 A25) at 40kV and 35mA. 106 The surface area (Brunauer-Emmett-Teller, BET) and pore size of the HKUST-1 were measured 107 using a Micromeritics Tristar 3020 following the method described elsewhere 23 . Thermal 108 decomposition behavior was studied using a thermogravimetric analyzer (TGA, NETZSCH STA49 109 F3),which involved the heating of the sample from 35 to 900 °C at a heating rate of 10 °C /min 110 under N2 atmosphere 24 . The pH value of the solution was measured by using a METTLER TOLEDO 111 pH analyzer (FiveEasy Plus-FE28). The crystallinity of each sample was studied to distinguish any 112 possible variation of the crystal pattern. The crystallinity percentage (%C) was defined as the ratio 113 of the sum of the relative intensity of the five most intense peaks and can be calculated using 114 Eq.(1) 25  In this calculation, the C300 (C300 is commercial product acquired from Sigma) was used as the 117 reference (100% crystallinity). 118  CO2 adsorption testing 119 CO2 adsorption was conducted in a TGA system (NETZSCH STA49 F3) following the procedures 120 adopted by others 26 . Initial treatment of the activated sample was carried out in N2 at 150 o C for 121 7 CO2 isotherm of the HKUST-1 MOFs at 27 o C was obtained using the Micromeritics ASAP 2020. 126 The degas of the activated sample was conducted at 150 °C for 12 h. The adsorption pressure 127 varied from 2.7 mbar to 1100 mbar, as described elsewhere 27 . 128 The CO2 and N2 adsorption-desorption cycle at 27°C was obtained using the TGA. Approximately 129 10 mg HKUST-1 MOFs was pre-treated at 150°C for 2 h in a nitrogen environment. The 130 adsorption-desorption cycle was carried out at a flowrate of 60 mL/min of CO2 (99.999%) gas 131 and 40 mL/min of N2 (99.999%), respectively. 132

Screening synthesis of HKUST-1 134
In this study, the effect of synthesis temperature, time, pretreatment temperature and activation 135 methods on the synthesis of HKSUT-1 MOFs was investigated. The first attempt was made to find 136 out an optimal reaction duration and temperature for the synthesis of MOFs. From Fig. 1, it can 137 be seen that the variation in reaction time from 3 h to 9 h showed no significant impact on the 138 specific surface area for samples in Region I. That means 3 hours provide sufficient time to allow 139 the synthesis reaction to complete, which is shorter than the time required for conventional 140 HKUST-1 synthesis (≥24h) 21 . This is due to the addition of the alkaline TEA, wich is consistent 141 with the finding that the alkaline TEA could accelerate the deprotonation of H3BTC and promote 142 the nucleation of particles into nanoscale 28 . Furthermore, in Region II, the pristine HKUST-1 had 143 gone through the pretreatment in an air dryer under two different temperatures, 120 and 180 8 It can be seen from Fig. 1 that in Region III, the two samples prepared at 50 and 85 o C showed 146 significantly different BET surface areas. It is obvious that 85 o C favors the formation of HKUST-1 147 with large BET surface area. This low temperature (85 o C) is much lower than the temperature 148 (180 o C) adopted for the synthesis of HKUST-1 MOFs by Chui et al. 7 The low temperature level 149 can inhibit the formation of the by-product (Cu2O) and therefore contribute to a high selectivity 150 and yield. 151 After synthesis, the pristine HKUST-1 has to be activated to remove the impurities (unconverted 152 reactant or byproduct) that are trapped in the pores. The activation can also activate the metal 153 sites, which are always surrounded by water and other gas molecules. As shown in Region IV, 154 pristine HKUST-1 was activated via either slurry state or powder state activation by ethanol or 155 methanol. The BET surface area of T85-3-Se1-120 and T85-3-Pe1-120 was lower than that of T85-156 3-Sm1-120 and T85-3-Pm1-120. It is reported that that methanol is a better activation agent and 157 can remove more impurities than ethanol 29 . It is clear that the BET surface area of T85-3-Pe1-120 158 and T85-3-Pm1-120 was higher than that of T85-3-Sm1-120 and T85-3-Pe1-120, which suggests 159 that powder state activation is more efficient in the removal of impurities from pores. 160 In Region V, it shows that repeated slurry activation resulted in lower BET surface area, which 161 might be due to the destruction of micropores to form larger pores and subsequently result in 162 lower BET surface area. However, repeated powder state activation shows different impacts on 163 9 The combination of slurry-powder state activation was applied to activate pristine HKUST-1 and 166 the BET surface area of individual samples is shown in Region VII in Figure 1. It is found that the 167 BET surface area of T85-3-Sm1Pm1-120, T85-3-Se1Pm1-120 and T85-3-Sm1Pe1 was higher than 168 that of their respective samples being activated once via slurry state activation. For example, T85-169 3-Sm1Pm1-120 has a BET surface area of 1264.6 m 2 /g, which is higher than that of T85-3-Sm1-170 120 (954.4 m 2 /g). In addition, it was found that the BET surface area of T85-3-Sm1Pm1-120 was 171 higher than that of T85-3-Sm1Pe1-120 (1022.7 m 2 /g), which again proves the powder state 172 activation using methanol as the activation agent is a reliable and efficient method. However, the 173 BET surface area of one more times of powder activation by methanol (T85-3-Sm1Pm2-120) was 174 reduced. Therefore, it can be concluded that slurry state activation removes the impurities in the 175 pores but does not significantly contribute to the formation of higher BET surface area. 176 Among the 17 samples prepared, the highest BET was found for T85-3-Pm4-120, which is 177 1542.4m 2 /g. There are also other five samples, i.e., T85-3-Pm1-120, T85-3-Pm2-120, T85-3-Pm3-178 120, T85-3-Pm5-120, and T85-3-Sm1Pm1-120, that have BET surface area greater than 1100 m 2 /g. 179 The highest BET surface area is attributed to the cleaning of pores, which is associated with a 180 final HKUST-1 yield of 67%. (25 o C), it shows crystal structure with blocky shape and obvious sharp edges and has the highest 206 average particle size of 93.6 nm, whereas samples prepared under higher temperatures 207 (50 ,75 ,85 o C) are more spherical and are of smaller particle size. It was obvious that the MOFs 208 prepared at 50 o C and 85 o C formed are of similar average particle size (~72 nm) as shown in the 209 12 In this study, further investigation was conducted to understand the influence of powder 212 activation and slurry activation process on the samples prepared under the optimal synthesis 213 temperature of 85 o C. It was found that when the pristine sample had undergone powder 214 activation using methanol as the activation agent twic, the average particle size was the smallest 215 (76.0 nm) with the peak of particle distribution in the range of 67.5-81.25nm. However, the 216 particle size increased when the sample had undergone more times of powder state activation, 217 which could be attributed to the crystal growth of the primary particle during the activation 218 process, i.e., the precursors being washed out of the pores to form new MOFs on the surface of 219 the primary MOF particle. 220 For comparison purpose, the slurry state activation using methanol as activation agent was also 221 conducted. The samples, T85-3-SmX-120 (X=1, 2, 3), were prepared and are shown in Fig. 2

(III). 222
These samples showed different morphologies and demonstrated that particle size can be 223 reduced followed by the additional times of slurry state activation, which is shown in the Fig. 3a  224 (III). It can be seen from Fig. 3b (III) and (IV) that slurry state activation leads to the formation of 225 samples with samller particle size a compared with powder state activation. However, although 226 slurry activation process can lead to the formation of smaller size nano particles, it cannot form 227 nano-scale HKUST-1 particle with BET surface area greater than 1100 m 2 /g (as shown in Region 228 V of Fig. 1). Despite the increase in particle size after 3 times of powder state activation, the 229 specific surface area of T85-3-PmX-120 (X=1, 2, 3, 4) increased from 1115.3 to 1542.4m 2 /g as a 230 result of multiple powder state activations. Therefore, it can be concluded that to form nanoscale 231 sample with large surface area, the sample shall be prepared under 85 o C and activated via 232 powder state activation for at least two times.

Nitrogen isotherm analyses 252
It is found that N2 adsorption rate increases at a low relative pressure (0.0< P/P0<0.1) (as shown 253 in Fig. 4), the shape of each line indicates that they are Type I isotherm according to the IUPAC 254 (International Centre for Theoretical and Applied Chemistry) classification 26, 30 . Samples with 255 Type I isotherm are of microporous (<2nm) structure 31 . The hysteresis loop at a higher relative 256 pressure (P/P0>0.4) indicates capillary condensation of mesopores for N2, which contributes to 257 the stacking combination of large particle of HKUST-1 32 or the creation of defects under such 258 synthesis method 33 . That is, the hysteresis loop for T85-3-n-120 and T85-3-Pm1-120 would be 259 due to the defects, because this loop was the cavitation phenomenon 34 that occurs when the 260 pore size is less than 6 nm. Besides, the hysteresis loop for T85-3-PmX-120 (X=2,3,4,5) at P/P0 > 261 0.8 appeared due to larger pores over 10 nm 35 . This loop is caused by the interparticle pores of 262 nanoparticle agglomerate 36 . 263 The influence of activation process on the surface properties of HKUST-1 was also investigated in 264 this study. As shown in Table 2, BET surface area after activation is in the range of 503-1542 m 2 /g, 265 which is very close to the values reported by Diring 37 and Ameloot 38 , but much higher than those 266 obtained by Chui 7 . Regarding total pore volume, the reported values are in the range of 0.21 -267 0.79 cm 3 /g 39 , whereas the effective pore volume was reported as 0.82 cm 3 /g 40 . In this study, the 268 as-synthesized samples exhibited a total pore volume of 0.31-0.65 cm 3 /g, which is comparable 269 with reported data. The relatively low N2 adsorption capacity of the raw material (T85-3-N-120) 270 is attributed to some micropores being blocked by TEA and/or its derivatives. The high N2 271 adsorption capacity indicates that the modulator is absent from the pores after activation. The 272 percentage of micropore volume shown in Table 2 suggests that these samples are comprised of 273 plenty of micropores. It can be seen from Table 2 that activation process (powder, slurry and  274 combination activation) has significant influences on pore properties (BET, Langmuir, Total pore 275 volume, Micropore volume, and Microporosity). As for slurry activation, both methanol and 276 ethanol can be used as the agent to clean the interior of pore and therefore improve N2 277 adsorption capacity. However, more times (>2) of slurry state activation using methanol as the 278 activation agen would result in smaller BET surface area. This is associated with the decrease in 279 the percentage of micropores, which means excessive times of slurry state activation might 280 damage micropores and form more mesopores. 281 As showin in Table 2, the slurry state activation followed by power state activation once led to 282 the formation of HKUST-1 with better pore properties. However, powder state activation is more 283 effective in the removal of impurities trapped in pores and multiple powder state activation could 284 result in high BET surface area (T85-3-PM4-120, 1542.4m 2 /g). 285 Meanwhile, the pH value of the solution after activation was monitored by using a pH meter, the 286 results of which are shown in Table 2. It is clear that pH value generally increased with BET specific 287 surface area, which is associated with the removal of impurities and TEA derivatives. It is 288 speculated that powder state activation retained the high BET surface area and did not damage 289 the pore structure of HKUST-1. Therefore, the powder state activation by methanol is considered 290 as an appropriate method for the treatment of the pristine HKUST-1. 291

XRD analysis 292
XRD analysis was conducted to show the crystalline phases of the porous HKUST-1. It can be seen 293 that all the diffraction peaks in Fig 5 (a) and (b) match well with the pattern of C300 and 294 simulation, indicating that these samples are pure phase of HKUST-1. From Fig 5, it is evident that 295 the XRD peak positions and relative intensities of the synthesized MOFs also agree well with those 296 of the simulated HKUST-1 (red spikes labelled by star at the bottom of Fig. 5) 41 . Due to the low 297 temperature condition adopted for the synthesis of HKUST-1, the diffraction peaks of Cu2O (36.7 298 o , PDF#04-003-6433) do not show in the XRD spectrum, which means that (Cu2O) was not formed. 299 It can be seen from Table 3 that the synthesis temperature has significant impacts on the 300 crystallinity of the samples. In the samples of TX-3-Se1-120 (X=25, 50, 75, 80, 85), the crystallinity 301 percentage of T85-3-Se1-120 is very close to that of the C300 (%Crystallinity=95.4%), which 302 demonstrates that the low synthesis temperature (85 o C) can lead to the formation of HKUST-1 303 with appropriate crystal structure. After the samples were further processed via powder state 304 activation, the crystallinity percentage of each sample did not vary significantly compared with 305 the variation in temperature. The crystallinity percentage is improved until 5 times of powder 306 state activation. However, further powder state activation does not show much influence on 307 crystallinity percentage. Therefore, to obtain the highest crystallinity percentage of HKUST-1, the 308 sample shall be prepared under 85 o C with the four times of powder activation process. 309 In addition, the hydration degree of the HKUST-1 could be determined by the I200/I220 ratio 42 . 310 where higher I200/I220 ratio indicates a smaller hydration degree. Normally, a smaller hydration 311 degree indicates that more copper coordination sites are accessible for other molecules, such as 312 CO2. The T85-3-Pm4-120 shows the highest value of I200/I220, which suggests that it can capture 313 more CO2 than any other samples. This result suggests that the proper condition (T85-3-Pm4-120) 314 lead to the formation of HKUST-1 with desired properties in CO2 adsorption.

TGA analyses 322
The HKUST-1 (Cu3(BTC)2(H2O)3 . xH2O, x≈3) was tested in TGA to show its thermal stability. As 323 shown in Fig. 6, these samples have similar TG curves from 35 to 900°C. At temperatures below 324 120°C, the weight loss is due to the desorption of physisorbed water or gases. This was followed 325 by the release of water trapped in the pores when the temperature was raised up to 180°C. The 326 HKUST-1 was then heated up to 350°C and exhibited modest loss of weight, as shown in DTG 327 curve of the Fig. 6. However, the weight loss increased significantly when temperature was raised 328 above 350°C. At higher temperatures, some metal were reduced and MOFs were decomposed 329 to form carbon. The weight loss levelled off at higher temperatures when MOFs were completely 330 transformed into CO2, CO, Cu, Cu2O, and CuO. This finding is consistent with what has been 331 reported by others, the MOFs retained its molecular formula but lost the microporous nature 332 after being heated to 350 to 427°C 43 . 333

CO2 adsorption analyses 337
It was found that the presence of divalent metals significantly increased CO2 binding strength and 338 resulted in higher selectivity in CO2 adsorption 44 . Due to its crystalline structure and the existence 339 of Cu 2+ metal ions, HKUST-1 is expected to have high affinity toward CO2. Normally, CO2 340 adsorption can be evaluated by two methods, i.e., static and dynamic adsorption. In this study, 341 the static adsorption test was carried out at 27 o C with pressure from 0-1bar. In the dynamic test, 342 a thermogravimetric study using pure CO2 was carried out at 27 o C. 343 The CO2 uptake of T85-3-PmX-120 (X=1,2,3,4,5) and T85-3-Sm1Pm1-120 is shown in Fig. 7. The CO2 344 uptake capacity of T85-3-Pm4-120 sample exhibits a steep rise in a short time, and reach a maximum 345 of 8.12% wt. (1.84 mmol/g), at 27 o C and 1 bar. It is generally believed that CO2 adsorption capacity is 346 dependent on pore volume of the adsorbent 45 . The larger the microporosity, the higher the CO2 347 adsorption capacity. Besides, the I200/I220 ratio of each sample can be used as an indicator for CO2 348 adsorption capacity. To reveal the CO2 adsorption property of T85-3-Pm4-120, CO2 adsorption isotherm was obtained 352 and is shown in Fig. 8. The CO2 recycle of T85-3-Pm4-120 was measured at 27 o C under 1 bar. The 353 results showed that the CO2 desorption process finished rapidly during N2 purging under the 354 same conditions and the CO2 adsorption capacity remains almost unchanged after ten 355 adsorption/desorption cycles, which demonstrated the good adsorption stability of T85-3-Pm4-356 120. It is clear that T85-3-Pm4-120 has a CO2 uptake of 11 wt% (2.5mmol/g) under static 357 adsorption, which is higher than reported data under similar experimental conditions as shown 358 in Table 4.  In conclusion, the optimized hydro/solvo-thermal approach developed in this research is a cheap 365 and efficient method for the synthesis of nanoscale HKUST-1 MOFs under low temperature and 366 atmospheric pressure using methanol as the activation agent. Both the slurry and powder state 367 activation methods were found to have significant influence on the specific surface area, 368 micropore volume and mesopore size, while powder state activation is more effective in the 369 removal of impurities trapped in HKUST-1. The T85-3-Pm4-120 showed a high BET surface area 370 of 1542.4m 2 /g and an average size of 87 nm. It was also found that the HKUST-1 prepared in this 371 study showed a high CO2 adsorption capacity with an uptake around 11wt% (2.5mmol/g) at 27 o C 372 and 1 bar. 373