Suspension Processing of Microporous Metal-Organic Frameworks: A Scalable Route to High-Quality Adsorbents.

Metal-Organic Frameworks (MOFs) have been intensively studied for applications such as gas storage, gas separation, catalysis, drug delivery, and more. Typically, the development of MOFs involves a post-synthetic solvent exchange process, which usually requires a significant investment of time, energy, labor, and resources. Herein, we propose a novel post-synthetic processing methodology for commercial and laboratory-scale MOFs called "Suspension Processing." Suspension processing is a non-destructive, agitation-based technique that provides efficient solvent exchange, pore cleaning, and surface defect removal in MOFs. Suspension processing has shown the capability to significantly improve the surface area and gas uptake properties of microporous MOFs, including PCN-250, UiO-66, and HKUST-1. Suspension processing displays improved time, energy, and labor efficiency, as well as considerably enhanced product quality. These findings confirm suspension processing as a straightforward methodology with applicability as a universal technique for the production of high-quality microporous materials.

Introduction

Since their discovery in the late 1990s, Metal-Organic Frameworks (MOFs) have turned into one of the fastest growing classes of materials studied in the chemical literature. MOFs have shown promise in a wide range of applications, including gas storage, chemical separations, chemical sensing, catalysis, ion exchange, light harvesting, and even drug delivery (Kreno et al., 2012, Li et al., 2012, Ma and Zhou, 2010, Orellana-Tavra et al., 2015, Wang et al., 2016, Zhu et al., 2017). This wide range of potential applications can be attributed to the ultra-high surface area, tunable pore environments, and high crystallinity of MOFs (Cui et al., 2016, Li and Huo, 2015, Silva et al., 2015, Sun et al., 2016). Although the potential applications of MOFs have been described as endless, commercial and industrial breakthroughs utilizing MOFs have been few and far between. A major reason for the disconnect between basic MOF research and their commercial development is the large monetary and time cost for material processing and activation.

The activation and processing of MOFs has evolved over the years yielding three well-developed primary strategies: conventional activation, solvent exchange, and supercritical CO2 activation (Mondloch et al., 2013). Conventional activation is the removal of solvent and/or other guest molecules by simultaneous heat and vacuum treatment. Unfortunately, conventional activation has resulted in minimal utility in accessing the full porosity of many MOFs owing to the harsh conditions, often resulting in the collapse or degradation of frameworks. Solvent exchange was developed to help combat the collapse of MOFs during activation. Solvent exchange methods replace a high-boiling-point solvent (e.g., dimethylformamide [DMF]), which is required for synthesis, with a lower-boiling-point solvent (e.g., chloroform), which is then removed under relatively mild conditions. Typically, lower-boiling-point solvents have weaker interactions with the MOF framework. The weaker interactions result in decreased surface tension and capillary forces exerted on the framework during the solvent removal. Solvent exchange is the most commonly used technique for MOF activation, but the time and resources required to perform a successful solvent exchange are typically too high for any production beyond the gram scale (Cui et al., 2016, Li and Huo, 2015, Silva et al., 2015) (Mondloch et al., 2013) (Cavka et al., 2008).

Another common technique for laboratory-scale research is supercritical CO2 (scCO2) activation. scCO2 builds on the premise of solvent exchange, using liquid CO2 as a solvent. For example, a solvent that is miscible with scCO2 (e.g., ethanol) is exchanged within the MOFs pores for scCO2 at high pressure (i.e., >73 atm) over the course of several hours. This method further reduces the surface tension during activation. Although scCO2 has proven successful on the laboratory scale, the large capital cost associated with the development of commercial or industrial-scale equipment for scCO2 has limited its adoption into large-scale systems.

Suspension-based processing methods have been used for cell processing for many years. Suspension-based cell-growing procedures were first used in 1956 when a suspended magnetic stirrer was used to grow cells in round-bottom flasks. Further optimization of suspension cell growth methods has allowed for a quick and easy process for achieving large quantities of high-quality cell lines (Iyer et al., 1999, Ryan, 2008). Based on the success of suspension cell growth methods, as well as to combat the issues that exist with the three current MOF activation and processing methods, we have developed a method of MOF activation named suspension processing. Suspension processing provides a universal, scalable, cost-effective, and robust technique for the effective activation of MOFs.

Results and Discussion

Suspension Processing Methodology

Suspension processing utilizes an enclosed cylindrical vessel with a suspended stir rod or agitator that extends from the top of the system downward without touching the bottom as shown in Figure 1. The as-synthesized MOF, still suspended within the reaction mixture (Figure 1 left, yellow colored area), is placed within the reaction vessel. Step 1 shows the addition of the full reaction vessel contents into the suspension processing apparatus. In step 2, the low-boiling-point solvent, such as methanol (MeOH, Figure 1 center, blue colored area), is added in an amount approximately 5x the volume of the solid product. The system is then heated to the boiling point of the low-boiling-point solvent (in the case of MeOH, the system was heated to 65°C). The system is then stirred at a low rate, typically 65 rpm, for the desired time. After the stirring has been stopped, in step 3 the contents of the suspension apparatus are filtered while heated, yielding a highly crystalline and porous MOF product with a filtrate consisting of a mixture of the process and reaction solvents as well as dissolved reaction by-products. The process requires minimal participation from the operator, with no solvent changes necessary during the timescale of the procedure. In addition, the apparatus utilized for laboratory-scale suspension processing is similar in design to commercial batch reactors that typically utilize suspended mechanical stirrers, allowing for this process to act as drop-in technology for existing chemical production.

Figure 1

Suspension Processing Methodology

Step 1, addition of reaction vessel contents into suspension processing apparatus; step 2, suspension processing; step 3, filtration.

This method was first developed for PCN-250, also known as MIL-127, which is constructed from Fe3-μ3-oxo clusters and tetratopic azobenzene-based linkers (ABTC = 3, 3′, 5, 5′-azobenzenetetracarboxylate). The high gas uptake, available open metal sites, exceptional stability, and scalability of PCN-250 have made it a well-studied material for gas storage applications (Feng et al., 2014, Yuan et al., 2017). Although PCN-250 has exceptional gas storage properties, the solvent exchange process used to obtain the maximum gas uptakes is currently reported as an 8- to 10-day process with approximately 9–12 steps. These steps use three to four different solvents and require active participation from an individual to wash and exchange the solvent used in each step. The use of suspension processing in place of traditional methods not only yields a product with increased gas uptake properties, but does so with improved time, energy, and labor efficiency.

Nitrogen Gas Uptake

Analysis of the suspension processed materials was primarily conducted by powder X-ray diffraction (PXRD) and nitrogen gas uptake experiments. A sample of as-synthesized PCN-250 was subjected to suspension processing in MeOH, with samples removed after the following times: 6 hr, 1 day, 2 days, 5 days, 14 days, and 20 days. Figure 2A showcases the PXRD for the series of PCN-250 samples, showing that the crystallinity of PCN-250 improved over the course of the processing. In Figure 2B, the N2 gas uptake of the samples displayed an increase in total gas uptake with increase in treatment time. Notably, the 5-day treated sample, PCN250-5 day, shows the same N2 gas uptake as reported in the literature (Feng et al., 2014). However, compared with the three to four solvents used in the reported solvent exchange method, suspension processing did not require solvent replacement or addition, only utilizing the initial process solvent added to the reaction mixture. No active participation was needed once the process was initiated. Even without exchanging the solvent, improvements in gas uptake were still observed after 20 days of processing.

Figure 2

Suspension-Processed PCN-250 Characterization

(A) Powder X-ray diffraction pattern of suspension-processed PCN-250.

(B) N2 adsorption isotherm at 77 K suspension-processed PCN-250.

(C) BET surface area vs. time of processing for suspension-processed PCN-250. Error bars were determined from three different measurments from three different samples.

(D) High-pressure methane uptake for PCN250-20 day compared with commercial PCN-250.

The surface area was found to dramatically increase during the first four treatments (6 hr–5 days). After the initial surface area response, the increase in surface area slows down until it reaches a peak of 1,702 m2/g after the 20th day. This surface area represents a 15% improvement over the commonly published PCN-250 surface area (1,446 m2/g). To our knowledge, this surface area is the record high among published PCN-250 samples.

Lastly, high-pressure methane (CH4) uptake measurements were performed to determine the applicability of suspension processing to an MOF's end application. Compared with a commercial PCN-250, the total methane uptake at room temperature (313 K) and 95 bar increased by 11.9%, from 194 to 217 v/v as shown in Figure 2D.

Universal Applicability

Owing to the success of suspension processing for PCN-250, we sought to investigate the universal applicability of this method by applying suspension processing to two other well-known, commercially available, and highly studied microporous MOFs, UiO-66 and HKUST-1. As seen with PCN-250, both MOFs were able to obtain higher gas uptake values using suspension processing compared with traditional solvent exchange procedures. In addition, an increase in gas uptake with an increase in processing time was also observed (Figures S3 and S4).

For UiO-66, a BET surface area of 1,675 m2/g was achieved after only 2 days of processing. This outperformed the BET surface area of the traditional solvent exchange sample of 1,290 m2/g. Furthermore, suspension processing of HKUST-1 was also observed to improve the BET surface area compared with traditional solvent exchange methods (1,808 vs 1,615 m2/g, respectively). Samples of PCN-250, UiO-66, and HKUST-1 were purchased from commercial vendors and compared with the laboratory-scale samples before and after suspension processing. Typically, the commercially purchased MOF adsorbents have lower BET surface areas compared with the laboratory prepared samples as seen in Table 1. Suspension processing of PCN-250, UiO-66, and HKUST-1 led to an increase in BET surface area and gas uptake properties over their commercially available counterparts.

Table 1

Comparison of BET Surface Area for Various MOFs

MOF	Traditional Solvent Exchange Process (m2/g)	Commercial Product (m2/g)	Suspension Processing (m2/g)
PCN-250a	1,446	1,270	1,702
UiO66a	1,290	1,045	1,675
HKUST1a	1,617	1,615	1,808

Figures S1–S8 in Supplemental Information display all characterization of these samples.

Mechanism Study

Mechanistic analysis of suspension processing was studied via scanning electron microscopy and thermal gravimetric analysis (TGA). Figure 3A shows that the as-synthesized PCN-250 particles were heavily aggregated. However, in Figure 3B, after suspension processing, the particles were well dispersed. This phenomenon indicates that unreacted organic ligands or surface residues have been successfully removed after the treatment. On the other hand, seed-like small particles are observed on the surface of as-synthesized HKUST-1 samples. However, they were completely removed after 48 hr of treatment, resulting in a smooth crystal surface for HKUST-1 (Figures 3D and 3E). This suggests the complete removal of solvent residues and defects from the MOF pores and surface. A similar phenomenon was also observed for UiO-66 samples (Figures 3G and 3H).

Figure 3

Scanning Electron Micrographs and Thermal Gravimetric Analysis Curves

(A) Scanning electron micrographs of PCN250-6 hr.

(B) Scanning electron micrographs of PCN250-20 day.

(C) TGA curve for PCN250-6 hr (black) and PCN250-20 day (red).

(D) Scanning electron micrographs of HKUST1-6 hr.

(E) Scanning electron micrographs of HKUST1-48 hr.

(F) TGA curve for HKUST1-6 hr (black) and HKUST1-48 hr (red).

(G) Scanning electron micrographs of UiO66-1 hr.

(H) Scanning electron micrographs of UiO66-48 hr.

(I) TGA curve for UiO66-1 hr (black) and UiO66-48 hr (red).

Scale bars 10μm.

The discussed observations are the result of the removal of low-crystallinity phases within or on the surface of the MOF by suspension processing. These results suggest that suspension processing aids in the removal of unreacted material, minor surface defects, and low-crystallinity coordination polymers via efficient dissolution and mass transport due to increased agitation and material-solvent contact. This improvement in bulk material purity and removal of non-porous by-products allowed for an increase in gas uptake performance compared with the as-synthesized samples.

As seen in Figures 3C, 3F, and 3I, the thermal stability of PCN-250, HKUST-1, and UiO-66 all increased following longer suspension processing times. The overall stability of the 20-day processed PCN-250 increased by 3°C compared with the 6-hr processed sample. The TGA curve of PCN250-6hr displays a mass loss of 6.3% below 100°C, likely the removal of MeOH from the framework. Between 100°C and 185°C, PCN250-6hr displays a mass loss event comprising 16.5%, which should correspond to the removal of DMF from the framework. In comparison, PCN250-20 day displays a significantly different TGA curve, showing a major mass loss of 20.5% below 100°C but with no significant mass loss between 100°C and 185°C (5.2%), which suggests that most of the DMF has been removed from the framework during suspension processing. More importantly, the mass loss event in the median temperature range (185°C–397/400°C) displays major differences, likely due to the effective removal of unreacted starting material, by-products, and surface defects. The mass loss decreased significantly from 19.6% for PCN250-6 hr to 9.7% for PCN250-20 day. Similar behavior was observed for HKUST-1 and UiO-66. We attribute the stability enhancement to improvements in pore cleaning and the removal of surface defects. Furthermore, it should be noted that further tests are currently ongoing to investigate the possibility of an internal defect self-healing mechanism.

It should be noted that suspension processing can be compared with other processes such as Soxhlet extraction (Hong et al., 2009). Soxhlet extraction is more similar to traditional solvent exchange as they are both solvent exchange processes, but Soxhlet extraction uses constant and automatic replacement of solvent, whereas traditional solvent exchange involved a manual solvent replacement. Soxhlet extraction, to the best of our knowledge, has been reported to be a good procedure for efficient pore cleaning and solvent exchange but has not been reported to help remove surface defects from MOFs. Moreover, the integration of suspension processing into current large-scale reactors system would be much more practical in our minds.

Cost Analysis

To analyze the practicality of suspension processing, we performed an operational cost analysis comparing suspension processing with traditional solvent exchange methods. For this analysis, we defined operational costs as the cost of solvent used plus the cost of labor for the duration of the process. In all cases, suspension processing had a significantly lower operating cost compared with traditional solvent exchange methods. The traditional solvent exchange for PCN-250 was performed over a 7-day period and involves a total of 14 separate washing procedures, involving 3 different solvents (full process is listed in the Supplemental Information). In total as displayed in Table 2, the solvent used in this procedure costs approximately $4,800.00 per kg of MOF. Including labor, the total cost per kg of PCN-250 using traditional solvent exchange is approximately $4,901.50. For comparison, a 5-day suspension processing has a total operating cost of approximately $795.75, which represents an 84% reduction in cost. Similar improvements in operating cost were also seen with UiO-66 and HKUST-1 (Figure 4). This section, along with how the analysis was performed, given in detail in the Supplemental Information, provides additional support to the claims that suspension processing not only leads to higher-quality MOF products but also does so with a major reduction of the cost and time. It should be noted that this is a preliminary cost analysis of the technique. Further analysis could be performed, which would be beneficial to the development of the technology in the future, such as taking into account alternative solvents such as hexane (Ma et al., 2017) or CH2Cl2 (Bae et al., 2017) or even to account for possible recycling of solvent.

Table 2

Total Operational Costs in US Dollars ($)

MOF	Traditional Solvent Exchange Process ($)	Suspension Processing ($)	Reduction in Operating Cost (%)
PCN-250	4,901.50	795.75	84
UiO66	5,868.00	264.50	95
HKUST1	6,163.50	1,689.50	73

Figure 4

Operational Cost Analysis for Suspension Processing vs. Traditional Solvent Exchange

(A) Solvent cost per kg of MOF.

(B) Labor cost per kg of MOF.

Conclusion

In conclusion, we introduced suspension processing as a method for the successful post-synthetic treatment of MOFs that is a viable alternative for traditional solvent exchange methods for both laboratory and commercial MOF syntheses. Through this treatment, three commercially available MOFs, with different compositions, stabilities, and porosities, have shown promising improvements in gas adsorption capabilities. We ascribe the gas uptake and surface area improvements to efficient pore cleaning and defect removal of the MOFs after treatment. Furthermore, the treatment itself is energy saving, economically efficient, and user friendly. Overall, suspension processing is a potentially universal, economical, and efficient post-treatment method for industrial-scale porous materials.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.