Patent Application
20250161908
Exhaled breath contains trace levels of hundreds of volatile organic compounds (VOCs) that are produced by various metabolic processes in the body, such as cancer, diabetes, or pathogen infection, and make their way into the lungs where they are exhaled. See N. Drabinska et al., J. Breath Res. 15, 034001 (2021). Changes to these metabolic processes, such as infection, organ failure, or other disorders, can alter the profile of VOCs that are exhaled, creating a detectable signal of chemical changes in the body. Therefore, exhaled breath contains a vast wealth of information about the workings of the body. However, breath analysis remains a challenging problem and is not yet routinely used as a medical diagnostic tool. See P. C. Moura et al., Biomed. J. 46, 100623 (2023); I. Oakley-Girvan and S. W. Davis, Cancer Biomarkers 21, 29 (2018); Y. Saalberg and M. Wolff, Clin. Chim. Acta 459, 5 (2016); T. Issitt et al., J. Breath Res. 16, 024001 (2022); and X. Sun et al., Anal. Bioanal. Chem. 408, 2759 (2016).
One challenge is capturing and measuring the VOCs of interest for a particular disease or condition. Deconvoluting the complex signatures of dozens or hundreds of VOCs in exhaled breath to isolate a handful of biomarkers associated with different diseases, infections, or metabolic conditions is a difficult task. The VOCs of interest are generally present in the breath in concentrations at ppb or low ppm levels and are mixed in with many other molecules that are not unique to specific conditions, including competing molecules like water and CO2. See G. Konvalina and H. Haick, Acc. Chem. Res. 47, 66 (2014); J. D. Fenske and S. E. Paulson, J. Air Waste Manag. Assoc. 49, 594 (1999); and A. Sharma et al., Mol. Diagn. Ther. 27, 321 (2023). It is a significant challenge to isolate and detect only the VOCs of interest for a particular diagnostic question. This is, in essence, a gas phase chemical separation problem of trying to selectively capture and detect particular molecules of interest from a complex mélange of competing molecules. Therefore, there is a need to strategically devise methods to achieve these separations, and furthermore develop design strategies for bespoke separations for future diagnostic needs.
Most previous portable breath analysis technologies can be divided into one of two categories: electronic nose (e-nose) technology and microGC (gas chromatograph). Both of these methods have advantages but also challenges for use as a portable screening tool.
E-nose technology generally relies on differentiated responses to specific chemicals from multiple sensors. See W. Hu et al., Adv. Mater. Technol. 4, 1800488 (2019); and H. Chen et al., IEEE Trans. Biomed. Circuits Syst. 16 (2), 169 (2022). Many groups have used metal-organic frameworks (MOFs) as sensors for electronic noses. See S. Okur et al., Angew. Chem., Int. Ed. 60, 3566 (2021); S. Okur et al., Sens. Actuators B Chem. 306, 127502 (2020); P. Qin et al., J. Mater. Chem. A 10, 25347 (2022); P. Qin et al., Chem. Sci. 12, 15700 (2021); B. A. Day and C. E. Wilmer, ACS Sens. 6, 4425 (2021); J. A. Gustafson and C. E. Wilmer, J. Phys. Chem. C 121, 6033 (2017); and R. Sousa and C. M. Simon, ACS Sens. 5, 4035 (2020). However, the response of these sensors to different gases is rarely perfectly specific to each species. Since the selectivity of MOFs or sensing polymers used for e-noses is not perfect, some competing species are captured by the sorbents. Deconvoluting these mixed and non-orthogonal signals is a difficult problem, and adding more sensors has diminishing returns after about ten sensors. See N. Gantzler et al., J. Phys. Condens. Matter 33, 464003 (2021); Y. S. Kim et al., Sens. Actuators B Chem. 108, 285 (2005); M. C. Lonergan et al., Chem. Mater. 8, 2298 (1996); and E. T. Zellers et al., Anal. Chem. 67, 1092 (1995). This results in both false positives and negatives, which underscores the need for greater selectivity and specificity.
Gas chromatography (GC) is a powerful and well-understood technique widely used in analytical chemistry, often coupled with mass spectrometry (MS). However, GC/MS systems are large and expensive. Portability is essential for the type of widely distributed test that might be used for a pandemic response or in remote areas. Portable chemical analysis systems such as MicroGC or portable ion mobility spectrometry (mini-IMS) are potential solutions. MicroGC employs gas chromatographic separation of analyte molecules through differing affinities for the stationary phase (typically a polymer coating the walls of the column). MicroGC systems produce high specificity with low limits of detection (LOD). The specificity of GC, however, is typically dictated by the length of the separation column which can be a practical issue. For example, Zhou et al. performed in-vivo breath analysis with a portable GC that utilized a 10-meter-long column (coiled up into a small form factor) and experienced significant co-elution (incomplete separation) of target gases. See M. Zhou et al., Anal. Bioanal. Chem. 411, 6435 (2019). A second column was required to achieve complete separation. However, the downside of portable chemical analyzers is that sensitivity is generally lower than lab bench scale instruments. Since breath analysis is targeting very low (ppb-ppm) concentrations of analytes, this could result in inaccurate results. The sensitivity can be improved by using a preconcentrator to capture VOCs from the breath sample prior to the detector.
The present invention is directed to a method for breath analysis, comprising providing a preconcentrator comprising one or more metal-organic framework (MOF) sorbents having selectivity for one or more target volatile organic compound (VOC) in a breath, preconcentrating the one or more VOCs on the one or more MOF sorbents, heating the preconcentrator to thermally desorb the preconcentrated one or more target VOCs from the one or more MOF sorbents, and detecting the one or more target VOCs desorbed from the one or more MOF sorbents. For example, the one or more MOF sorbents can comprise a Zr-MOF, such as MOF-808, PCN-777, MOF-818, UiO-66, or NU-1000. For example, the one or more MOF sorbents can comprise a fluorinated MOF.
Grand canonical Monte Carlo (GCMC) simulations were used to select the exemplary zirconium-based MOFs (MOF-808, MOF-818, PCN-777, and UiO-66 with NU-1000) for VOC capture from humid air based on their predicted selectivity with respect to water and CO2. All five were found to be highly stable after VOC exposure and thermal regeneration. These MOFs were tested for VOC recovery in the presence of humidity, and although humidity significantly reduces the amount of VOC recovered, the MOFs were able to capture significant quantities of each molecule from dilute humid streams. Overall, UiO-66, PCN-777, and MOF-818 had the best performance for recovering VOCs in high humidity, with average recovery above 40% (across the five VOCs tested here) at 95% RH. MOF-808 had exceptional (90%) acetaldehyde recovery up to 60% RH. These results show that highly selective MOFs can capture measurable quantities of VOCs from very humid streams with only ppm levels of the target molecules.
The invention is further directed to a method for synthesizing a fluorinated MOF, comprising providing a MOF with a metal cluster having at least one terminal formate ligand, and reacting the MOF with a fluorine-containing carboxylic acid, whereby at least one of the terminal formate ligands is substituted with a fluorinated function group of the fluorine-containing carboxylic acid. For example, the MOF can comprise MOF-808. For example, the fluorine-containing carboxylic acid can be a fluorinated benzoic acid, 4-fluorobenzoic acid, 3,5-difluorobenzoic acid, 4-(trifluoromethyl)benzoic acid, pentafluorobenzoic acid, or 3,5-bis(trifluoromethyl)benzoic acid. The fluorinated groups are shown to improve the hydrophobicity of the MOF and consequently increased VOC retention from around 60% to about 90%. These hydrophobic variants are useful as sorbents for VOC capture for breath analysis, chemical sensing, and other applications.
The detailed description will refer to the following drawings, wherein like elements are referred to by like numbers.
The present invention uses highly selective sorbents made of metal-organic frameworks (MOFs) to capture targeted VOCs. MOFs are nanoporous crystalline structures made from metal (or metal oxide) clusters (or nodes) connected via organic linkers. The chemical space of possible MOFs is enormous and diverse, as these materials can be made from many combinations of clusters and linkers with varying numbers of connections. These components can be arranged in numerous different topologies (connection frameworks) to yield materials with diverse structural properties. As a result, the pore sizes and overall pore structures of MOFs can be systematically varied through the judicious choice of clusters and linkers, tuning the adsorption properties of the materials for specific applications. MOFs have been used for many applications including catalysis, chemical sensing, gas storage, and, notably, separations. See H. Daglar and S. Keskin, Coord. Chem. Rev. 422, 213470 (2020); S. Kitagawa, Chem. Soc. Rev. 43, 5415 (2014); S. K. Firooz et al., Anal. Chim. Acta 149, 340208 (2022); M. S. Denny et al., Nat. Rev. Mater. 1, 1 (2016); A. Bavykina et al., Chem. Rev. 120, 8468 (2020); Ü. Anik et al., Mikrochim. Acta 186, 1 (2019); H. Li et al., EnergyChem 1, 100006 (2019); and H. Li et al., Mater. Today 21, 108 (2018).
The present invention takes advantage of the high, tunable adsorption selectivity of MOFs by using them as collectors or preconcentrators to capture target VOC molecules out of humid air (e.g., exhaled breath), as shown in
Choosing the correct MOFs to selectively capture the target VOCs of interest is essential. As the design space of possible MOFs is vast, computational screening can be used to narrow down the number of candidate MOFs to a feasible number. There are also practical factors to consider when choosing a MOF for an application. For VOC breath analysis, the MOF will need to be stable in the presence of humidity and able to withstand multiple regeneration cycles of exposure and heating (desorption) so that it can be reused for multiple tests. As an example, the following description focusses on zirconium-based MOFs with zirconium (Zr) clusters because they are known to generally be thermally stable and stable in the presence of water. Some progress has been made in recent years on MOF stability using other metals, but Zr MOFs remain a good choice as resilient MOFs. See K. Ahmad et al., Mater. Sci. Eng. B 262, 114766 (2020); J. H. Cavka et al., J. Am. Chem. Soc. 130, 13850 (2008); C. Gomes Silva et al., Chem. Eur. J. 16, 11133 (2010); J. E. Mondloch et al., Chem. Comm. 50, 8944 (2014); G. Mouchaham et al., Angew. Chem. 127, 13495 (2015); M. Shi et al., Colloids Surf. A Physicochem. Eng. Asp. 602, 125102 (2020); P. Deria et al., J. Am. Chem. Soc. 137, 13183 (2015); O. V. Gutov et al., Chem. Eur. J. 20, 12389 (2014); A. J. Howarth et al., Nat. Rev. Mater. 1, 1 (2016); S. Yuan et al., ACS Cent. Sci. 4, 440 (2018); T. Rasheed, Chemosphere 313, 137607 (2023); and X. Zhang et al., Coord. Chem. Rev. 423, 213507 (2020). Other type of MOFs, such as the Materials of Institute Lavoisier (MIL) and zeolitic imidazolate framework (ZIF) series of MOFs and are good candidates, as they have high chemical and thermal stability, and have been shown to utility for adsorption, gas separation, and chemical sensing.
Breath testing is non-invasive, and this system can provide results rapidly using a compact and portable form factor. Another advantage of this preconcentration method is agility, in that targeting different diseases with different VOC signatures simply requires a set of MOFs tailored to adsorb those specific VOCs. The same detector assembly can then be used in a plug-and-play fashion.
Computational screening can be used to quickly identify MOFs with good selectivity for target VOCs. As an example of the invention, computational modeling was used to select a series of Zr MOFs with promising selectivity for a variety of VOCs. The five MOFs that were indicated by modeling were synthesized and tested in a prototype sensing apparatus, demonstrating the capability to detect trace levels (ppm) of VOCs in realistic humid environments similar to human breath. This demonstrates the feasibility of using tailored MOFs as preconcentrator sorbents for breath analysis.
Grand canonical Monte Carlo (GCMC) simulations were performed using the multipurpose molecular simulation code RASPA (v2). See D. Dubbeldam et al., Mol. Simul. 42, 81 (2016); and D. Dubbeldam et al., Raspa 2.0: Molecular Software Package for Adsorption and Diffusion in (Flexible) Nanoporous Materials (2021). Surface areas were computed in RASPA using a N2 probe with Transferable Potentials for Phase Equilibria (TraPPE) parameters for N2 and Universal Force Field (UFF) parameters for the MOF. Pore sizes were computed in Zeo++ and void fractions were computed in RASPA using a helium probe. See T. F. Willems et al., Microporous Mesoporous Mater. 149, 134 (2012). GCMC simulations for single-component VOC adsorption were performed at 300 K and 0.1 Pa, or approximately 10 ppm VOC concentration at atmospheric conditions. The adsorbates (except for water) were modeled using TraPPE. See B. Chen et al., J. Phys. Chem. B 105, 3093 (2001); M. G. Martin and J. I. Siepmann, J. Phys. Chem. B 102, 2569 (1998); M. G. Martin and J. I. Siepmann, J. Phys. Chem. B 103, 4508 (1999); J. J. Potoff and J. I. Siepmann, AlChE J. 47, 1676 (2001); N. Rai and J. I. Siepmann, J. Phys. Chem. B 111, 10790 (2007); and J. M. Stubbs et al., J. Phys. Chem. B 108, 17596 (2004). Some of the specific molecules are not available in TraPPE but could be constructed from parameters from similar molecules (e.g., butanal is not available in TraPPE but is easily obtained by modifying pentanal). Water was modeled using the TIP3P model. See W. L. Jorgensen et al., J. Chem. Phys. 79, 926 (1983).
For the multicomponent GCMC simulations, selectivity is computed using the binary selectivity formula
where Si is the selectivity for molecule i over molecule j, qi is the mole fraction of i in the adsorbed phase, xi is the mole fraction of i in the gas phase (reservoir), qj is the mole fraction of j (or total components that are not i) in the adsorbed phase, and xj is the mole fraction of j in the gas phase (or total components that are not i).
For breath testing applications, the target analyte must be captured from air (breath) that contains high levels of humidity (i.e., water) and CO2. Therefore, MOFs that exhibit some degree of hydrophobicity are expected to be well-suited to this task. The 96 Zr-containing MOFs from the 2019 CoRE MOF database (all solvent removed, disordered structures removed) were selected for consideration and heats of adsorption and Henry coefficients were computed for water, CO2, and several diverse VOCs at infinite dilution.
The MOF with lowest affinity for water (−2.4 KJ/mol) is MOF-808. See H. Furukawa et al., J. Am. Chem. Soc. 136, 4369 (2014). MOF-808 is based on hexanuclear oxo/hydroxo Zr4+ clusters connected through tritopic BTC (1,3,5-benzenetricarboxylate) linkers in the space group Fd-3m. Each organic linker is connected to three inorganic clusters, and each cluster is connected to six linkers in spn topology, as shown in
The MOFs were also ranked according to the ratio of heat of adsorption of butanone (a target biomarker for COVID-19, among other diseases) to that of water. The highest ranked MOF was PCN-777 (FOTNIN). See D. Feng et al., Angew. Chem. Int. Ed. 54, 149 (2015). PCN-777 has a low heat of adsorption for water (−2.9 KJ/mol), while the heat of adsorption for butanone in PCN-777 is −42.4 KJ/mol, giving a ratio of 15.5 for butanone in water. PCN-777 is isoreticular with MOF-808, sharing the same connectivity and spn topology but with a more extended TATB (4,4′,4″-s-triazine-2,4,6-triyl-tribenzoate) organic linker creating much larger pore diameters than MOF-808 (about 19 Å in MOF-808 vs 33 Å in PCN-777). PCN-777 is also based on 6-connected, hexanuclear Zr4+ clusters, but these are typically capped by terminal OH−/H2O ligands instead of formate. Only the OH−/H2O-capped version of PCN-777 was analyzed further.
Given that these two MOFs have the same clusters and topology and vary only in the choice of linker, evaluation of both MOF-808 and PCN-777 provides an opportunity to study the effects of pore size and linker functionality (benzene vs. triazine) on adsorption behavior and selectivity. Therefore, another spn Zr6 MOF, MOF-818, was also studied. See Q. Liu et al., J. Am. Chem. Soc. 141, 488 (2018). This material is less well-studied than MOF-808 and PCN-777 but is an interesting candidate because it is bimetallic, containing trinuclear [Cu3(μ3-O)]3+ clusters in addition to the same Zr6 clusters in the other materials. The Cu3 clusters are capped by 3 ditopic pyrazole-4-carboxylate (PyC) linkers, allowing the Cu3(PyC)3 subunits to act as 3-connected supramolecular building units. See C. R. Groom et al., Acta Crystallogr. B. Struct. Sci. Cryst. Eng. Mater. 72, 171 (2016). The simulations indicate the heat of adsorption of water in MOF-818 is −1.9 KJ/mol, close to MOF-808 and PCN-777. The calculated heats of adsorption for CO2 in the three spn MOFs cover a larger range, with values of −7.4 KJ/mol in MOF-808, −17.8 KJ/mol in PCN-777, and −12.5 KJ/mol in MOF-818.
28 MOFs were rejected due to pore-limiting diameters (PLD) being less than the kinetic diameter of methane (3.8 Å), as this would preclude the adsorption of any VOCs. The MOF with the smallest permissible PLD is UiO-66 (RUBTAK) with a PLD of about 3.9 Å (largest cavity diameter (LCD) of about 8.5 Å), as shown in
NU-1000 was also included in the study, as it is a widely studied Zr4+ MOF, known to adsorb VOCs, and is highly stable in water. NU-1000 is constructed from 8-connected Zr6 clusters, as shown in
The five selected MOFs, MOF-808, PCN-777, MOF-818, UIO-66, and NU-1000, were synthesized and activated following literature protocols. See G. C. Shearer et al., Chem. Mater. 26, 4068 (2014); J. Baek et al., J. Am. Chem. Soc. 140, 18208 (2018); D. Feng et al., Angew. Chem. Int. Ed. 54, 149 (2015); and Q. Liu et al., J. Am. Chem. Soc. 141, 488 (2019). The corresponding organic linkers are widely available commercially.
Textural properties for the selected MOFs were computed using Zeo++ with a radius of 1.86 Å which corresponds to a N2 probe. Results are presented in Table 1. The MOFs were thoroughly characterized using PXRD and nitrogen adsorption, as shown in
Table 1. Textural properties of selected MOFs. LCD is the largest cavity diameter, PLD is the pore limiting diameter, LFS is the largest free sphere, VF is the void fraction, and SA is the surface area.
GCMC simulations were used to predict the adsorption of a diverse set of VOCs, including alkanes, alkenes, alcohols, ketones, and aromatics, on the five selected MOFs. Single component calculations were performed in RASPA at 300 K and 10 ppm, to approximate the concentration of target VOCs in exhaled breath. See J. D. Fenske and S. E. Paulson, J. Air Waste Manag. Assoc. 49, 594 (1999). The results of this screening are shown in Table 2. GCMC simulations were also performed at 100 ppm (not shown).
The GCMC simulations predict that NU-1000 and UiO-66 should adsorb the highest quantities of VOCs at 10 ppm. This is attributed to strong van der Waals interactions between the VOC adsorbates and the confined pore environment of UiO-66 and the smaller micropores of NU-1000, leading to favorable binding energies and high uptake values. The spn MOFs (MOF-808, PCN-777, and MOF-818) generally showed lower calculated VOC adsorption but still showed good selectivity for the target VOCs over water. The five Zr-MOFs were predicted to adsorb similar quantities of water, with MOF-818 showing somewhat higher affinity than the other materials. On the other hand, the NU-1000 and UiO-66 models gave significantly higher CO2 uptake than the spn MOFs. Water and CO2 are present in high concentrations in breadth relative to VOCs and can saturate potential VOC binding sites, implying that engineering high VOC selectivity into preconcentrator materials can be just as important as maximizing their absolute VOC binding affinities. Ultimately, high selectivity for the target VOCs with respect to both water and CO2 is necessary for realistic applications. It is noted that the pores in UiO-66 are quite small and could result in transport limitations, despite the excellent selectivity.
Any application using breath testing will include significant quantities of water vapor and CO2 in the sample. It is therefore critical that the sorbent material used to capture the target VOCs have high selectivity in the presence to these competing molecules. To help understand the impact of competitive adsorption on VOC uptake, GCMC simulations were performed with the selected VOCs, water, and CO2 in ratios that approximate the composition of exhaled breath (10 ppm VOC and 50000 ppm or 5% each of water and CO2). The results of these calculations are shown in
The selectivities of each MOF for each VOC with respect to water and CO2 are given in Table 3. NU-1000 and UiO-66 have the highest selectivity from simulation, with many values being 102-106, while selectivities of MOF-808, PCN-777, and MOF-818 generally range from 1 to about 100. All five MOFs are generally more selective for the VOCs over water and CO2, suggesting that these materials can be used effectively as sorbents to capture VOCs from breath. The main exceptions are methane and ethane. This is unsurprising given the small size of these molecules and the weakness of their dispersion interactions, implying that light hydrocarbons may be less useful as biomarker molecules compared to other (larger) VOCs.
The most notable changes between the single component and multicomponent adsorption are for 1-propanol, which is reduced by more than 50% in all five MOFs in the presence of water and CO2, and for methanol, which is increased in all five MOFs when water and CO2 are present (e.g., by 6% in UiO-66 and by 29% in NU-1000). The latter is likely due to increased adsorbate-adsorbate interactions between methanol and water.
As MOF-808, PCN-777, and MOF-818 have the same Zr6 clusters and network topology, as shown in
The stronger electrostatic interactions in MOF-808 (compared to PCN-777) is clear from decomposing the MOF-adsorbate interaction energy into van der Waals (vdW) and Coulombic components, as shown in Table 4. Adsorption of nonpolar hydrocarbons (i.e., alkanes, alkenes) is entirely driven by vdW interactions in all five MOFs, as the Coulombic contributions are minimal in each case. For the other VOCs, the breakdown of vdW vs. Coulombic contributions to adsorption energy varied widely. Coulombic interactions are generally more significant in MOF-808 and MOF-818, while vdW forces dominate in PCN-777. This can be attributed to the lower degree of charge separation between the cluster and linker in PCN-777 that creates a less polar pore environment than MOF-808 and MOF-818. For example, the percentage of Coulombic adsorption energy of acetone in MOF-808 is 31.1%, while in PCN-777 it is only 8.2%. Similarly, the adsorption energy of 1-propanol in MOF-808 is 51.5% Coulombic compared to 18.2% in PCN-777, and the adsorption of butanal in MOF-808 is 34.3% Coulombic compared to 9.9% in PCN-777.
The largest Coulombic contributions to VOC adsorption are seen for MOF-818, which has the highest charge per Zr. Additionally, the incorporation of both Cu2+ and Zr4+ clusters further increases the ionic character of its pore environment. Accordingly, GCMC simulations indicate that MOF-818 shows the least affinity for nonpolar VOCs and the highest affinity for water, while falling between MOF-808 and PCN-777 for the polar VOCs.
Notably, the percentage of Coulomb contribution for both NU-1000 and UiO-66 is low, generally less than 5% for most VOCs. These MOFs have 8 and 12 linkers, respectively, connected to each cluster, reducing access to binding sites for polar VOCs to interact with the Zr4+. This results in these interactions being dominated by electrostatics. Both benzene and toluene have a small negative Coulomb contribution in UiO-66, indicating the electrostatic contribution is repulsive.
The synthesized MOFs were exposed to N2 streams containing variable humidity percentages and 20 ppm of the specified VOCs for 60 minutes. The MOFs were then heated to desorb VOC adsorbates, which were then detected and quantified using GC-FID. Importantly, no VOCs were detected by the flame ionization detector (FID) until the MOF samples were heated, indicating that the MOFs retain adsorbed VOCs completely at ambient temperature until the heat is applied.
The VOC recovery percentages were found to generally decrease as relative humidity was increased, as expected due to competition for adsorption sites between water and the VOCs. However, the effects of humidity are complex and vary between different VOCs and MOFs. One striking example is the difference in acetaldehyde uptake between MOF-808 and NU-1000 (
UiO-66 was consistently one of the best performing materials at high humidities, again agreeing with the simulation results. UiO-66 recovered 40-60% of all VOCs at 80% humidity, including 57% for acetaldehyde. MOF-818 and PCN-777 also had good performance at high humidities with the exception of acetaldehyde. These MOFs recovered above 50% of the other four VOCS at 80% humidity. NU-1000 has generally poor performance compared to the other MOFs for any VOC at 80 or 95% RH.
These results indicate that acetaldehyde is especially difficult to capture from humid streams, with several MOFs recovering less than 20% above 40% humidity. This is a particularly difficult separation, as acetaldehyde is a small polar molecule similar to water. The best performing MOFs above 80% RH are UiO-66 followed by MOF-808. Notably MOF-808 has very high recovery for acetaldehyde up to 60% RH, but then drops significantly at higher humidity.
For butanone, the best MOFs are MOF-818, PCN-777, and UiO-66, which perform similarly. This agrees with the screening based on Henry coefficients that predicted PCN-777 to have the highest butanone affinity.
NU-1000 tends to perform poorly compared to the other four MOFs. This is significant because NU-1000 was chosen as a benchmark MOF for comparison, not based on data from the computational screening. However, NU-1000 remains an interesting case as the GCMC simulations suggest it should have high selectivity for most VOCs over water. The experiments indicate that the other MOFs have better recovery for all five VOCs tested but this discrepancy is likely due to the hierarchical pore structure in NU-1000. The smaller micropore yields very high VOC selectivity, but becomes saturated with prolonged exposure, and the selectivity of the mesopore is not as high, lowering the overall performance.
Generally, the experimental trends agree with the simulations. The results suggest that UiO-66 should have overall strong performance, and this is attributed to the small pores that induce strong van der Waal interactions. MOF-808 was predicted to have stronger Coulomb interactions than PCN-777, and consequently seems to generally be more negatively affected by humidity.
Based on these results, it appears that VOC recovery suffers significantly above 60% RH, even in the best performing MOFs. This leads to a challenge for any kind of VOC-based breath testing. However, even at 95% RH the MOFs capture detectable levels of the VOCs, indicating that using MOFs to capture VOC biomarkers from humid breath is viable. The level of humidity in exhaled breath varies significantly, ranging from 42-91%, and the MOFs presented here are generally effective in that regime. See E. Mansour et al., Sens. Actuators B Chem. 304, 127371 (2020).
In order to be practically useful as VOC collectors, the MOFs are preferably able to be regenerated and reused with no loss of selectivity. To test recyclability, octane and acetaldehyde in UiO-66 was chosen as a representative system because UiO-66 has strong performance for several of the VOCs. A UiO-66 sample was subjected to three consecutive adsorption/desorption cycles at 40%, 50%, and 60% relative humidity (for a total of 9 cycles per VOC). The recovery percentages for the three individual runs showed little change at each humidity level for both VOCs, as shown in
One of the great benefits of MOFs is their ability to be customized via choices of different clusters, linkers, and also, post-synthetic addition of functional groups. Post-synthetic modification is a convenient way to improve adsorption or chemical properties of MOFs that are already known to be stable and have existing synthesis protocols. See S. Mandal et al., Adv. Funct. Mater. 31, 2006291 (2021); Z. Wang and S. M. Cohen, Chem. Soc. Rev. 38, 1315 (2009); and M. Kalaj and S. M. Cohen, ACS Cent. Sci. 6, 1046 (2020). As described above, humidity can severely hinder VOC adsorption in MOF materials. It has been known for some time that fluorine groups added to MOFs or other sorbents will increase the hydrophobicity, with perfluorinated alkanes (PFAS) being a notable example. See K. Jayaramulu et al., Adv. Mater. 31, 1900820 (2019); and S. Mukherjee et al., Trends Chem. 3, 911 (2021). Therefore, it was hypothesized that adding fluorine groups would make the MOFs more hydrophobic and improve the selectivity for target VOCs. MOF-808 was used as a platform because it is a highly stable, robust MOF with a scalable synthesis procedure, and amenable to post-synthetic modification. See and C. Ardila-Suárez et al., CrystEngComm 21, 3014 (2019); and X. Meng et al., CrystEngComm 21, 3146 (2019). As can be seen in
To synthesize MOF-808, 1,3,5-benzenetricarboxylic acid (BTC; 0.210 g, 1.00 mmol) and zirconyl chloride octahydrate (ZrOCl2·8H2O; 0.970 g, 3.01 mmol) were added to 250 mL screw-top jar with a PTFE-lined lid. The solids were dissolved through sonication in a mixture of 30 mL N,N-dimethylformamide (DMF) and 30 mL formic acid. The reaction mixture was heated in an isothermal oven to 100° C. at 1.5° C./min, held at 100° C. for 24 hours, then cooled to room temperature at 1° C./min. A white, powdery solid formed and was collected by centrifugation and washed with DMF (3×25 mL) and acetone (3×25 mL) in a 50 mL centrifuge tube. The solid was then activated in a vacuum oven at 85° C. for 12 hours prior to use.
To synthesize the fluorinated MOF-808-Fs, MOF-808 (0.150 g, 0.66 mol) and a fluorinated benzoic acid FX (FX=F1—0.4623 g, F2—0.5217 g, F3—0.6274 g, F5—0.6998 g, F6—0.8518, 3.3 mmol) were added to a 125 mL screw-top jar with a PTFE-lined lid, followed by 25 mL of DMF. The jar was sealed, and the reaction was sonicated for 30 seconds to mix but not dissolve the solid. The reaction was heated to 60° C. in an isothermal oven for 18 hours. The solid was collected by centrifugation and washed with DMF (3×25 mL) and acetone (3×25 mL), then dried at 85° C. under vacuum for 12 hours.
The Julia software package PoreMatMod was used to in silico add functional groups to the MOF-808 structure. See E. A. Henle et al., J. Chem. Inf. Model. 62, 423 (2022). The MOF-808 unit cell contains 4 Zr6 clusters (or secondary building units, SBUs) each with 6 formates for a total of 24 potential substitution sites. Fluorinated MOFs were created with 4, 8, and 12 substitutions, equaling 17%, 33%, and 50% coverage of the formate sites. The resulting library of structures then underwent geometry optimization using density functional theory (DFT) in VASP and subsequently assigned partial charges via the DDEC6 method. See G. Kresse and J. Furthmüller, Comput. Mater. Sci. 6, 15 (1996); G. Kresse and J. Furthmüller, Phys. Rev. B 54, 11169 (1996); G. Kresse et al., Phys. Rev. B 50, 13181 (1994); and J. P. Perdew et al., Phys. Rev. Lett. 77, 3865 (1996). Grand canonical Monte Carlo (GCMC) simulations were performed in RASPA v2.0. See D. Dubbeldam et al., Raspa 2.0: Molecular Software Package for Adsorption and Diffusion in (Flexible) Nanoporous Materials (2021).
GCMC simulations were performed on MOF-808 functionalized with the five fluorinated groups for adsorption of acetone, hexane, octane, butanone, and octane. Results for hexane, acetone, butanone, and octane are shown in FIGS. 13A-13D for MOF-808-F1 and MOF-808-F5 with varying density of F-group substitutions. A consistent trend was found that as the density of substitutions increases, the VOC uptake at saturation decreases. This is a clear consequence of a reduction in free volume due to adding more functional groups. However, for applications such as VOC capture or sensing where the expected VOC concentration is very low, the uptake at low pressure is more important than the saturation level. A consistent trend was also seen that adding more functional groups shifts the step of the isotherm to lower pressures, indicating stronger adsorption interaction. This trend is seen across different types of F groups and VOCs. This is corroborated by the trend in computed enthalpy of adsorption for MOF-808 and MOF-808-F5 shown in Table 5.
The computed heats of adsorption corroborated the trends found in the isotherms. Addition of the functional groups increases the enthalpy of adsorption, which agrees with the notable shift of the isotherm step to lower pressures. Also, as the density of the F groups increases (more substitutions), the enthalpy of adsorption generally becomes stronger. The enthalpy of adsorption for water likely also increases, but less than the other VOCs which suggests the selectivity should shift to favor VOC adsorption.
A simple test was preformed to assess the gravimetric adsorption of VOCs in the MOF-808-F variants at room temperature by exposing the MOFs to VOC vapor at ambient conditions and measuring the change in mass. All five variants with monofunctional substitutions as well as several mixed bifunctional and trifunctional MOFs were tested, to see if there are synergistic effects from combining F-groups.
As shown in
Several bi-functional and tri-functional MOFs with various combinations of the F-groups were also synthesized and tested. No clear synergy was observed between different mixed functional groups. Notably, the bi-functional groups containing the F5 group (MOF-808-F1F5 and MOF-808-F2F5) did show significant increases in acetone adsorption, similar to the monofunctional MOF-808-F5. However, the monofunctional MOF-808-F5 shows a larger increase, so it was concluded there is little benefit to mixing F-groups and the monofunctional MOF-808-F5 is probably the best candidates. Based on this simple test, the exceptional acetone adsorption in MOF-808-F5 suggests that this MOF will be a good candidate for further tests in the VOC recovery experiments.
To develop a molecular level understanding of VOC-MOF interactions, DFT simulations were performed of VOCs and water binding in the MOF pore and studied the influence of the F-groups. Specifically, the binding energy of water, acetone, hexane, propanal, 1-propanol, and toluene was calculated in MOF-808, MOF-808-F1, MOF-808-F3, and MOF-808-F5. The structures used for these calculations each had 33% substitution (average 2 functional groups per cluster). The binding energies were determined by
where EMOF, Eadsorbent, and EMOF+adsorbent are the energies of the optimized MOF structure, gas-phase adsorbent, and adsorbent in the MOF structure, respectively. In this notation, positive energies indicate energetically favorable binding geometries. Three specific binding sites were explored, namely the μ3-OH, fully-coordinated metal site, and the F-groups. Using the raw calculated binding energy for each MOF-adsorbent pair the maximum (most favorable) binding energy across all binding sites was determined. Binding energies for acetone, hexane, propanal, 1-propanol, toluene, and water were calculated in MOF-808, MOF-808-F1, MOF-808-F3, and MOF-808-F5.
To showcase the increase or decrease for each VOC from water, a change in binding energy was then calculated, given by
where BEiVOC and BEiWater are the calculated binding energies for a given VOC and water. The calculated changes in binding energy are shown in
To explore this effect further, the optimized geometries of acetone and water in MOF-808 and MOF-808-F5 at the μ3-OH site were examined. Note that this is the most energetically favorable binding site for these two molecules in these two MOFs. Structural images are shown in
The performance of MOF-808-F5 compared to parent MOF-808 was tested by exposing the MOFs to 20 ppm vapor in humid N2 and measuring the mass of VOC recovered. The results in
A commercial desiccant (a mixture of Dryrite and Molsieve 13×) was also used in the line as a barrier prior to the humid VOC gas reaching the MOF. This is a practical measure that can be used to remove water in the sample stream. Including the desiccant does improve the VOC recovery at high humidity, as shown in FIGS. 19A and 19B, although recovery does still decrease at high humidity suggesting the desiccant did not capture 100% of the water from the stream. Therefore, there is still sensitivity to humidity. Notably, the VOC recovery of MOF-808 with the desiccant is similar to MOF-808-F5 without the desiccant. Depending on the requirements of a given application, this could mean that the parent MOF-808 used in conjunction with a desiccant is a good choice for VOC recovery. Alternately, if a desiccant cannot be used in a given context (perhaps due to space restraints, etc.), then MOF-808-F5 provides equally good VOC recovery without using a desiccant.
The VOC recovery with two mixed MOF-808-F variants: MOF-808-F1F3 and MOF-808-F1F3F5 was also tested, as shown in
Finally, MOF-808-F5 was tested with a mixture of VOCs in a multicomponent experiment. The vapor stream contained 20 ppm each of toluene, octane, and acetaldehyde, along with varying levels of humidity. Results are shown in
The present invention has been described as tailored nanoporous metal-organic framework sorbents for rapid, noninvasive disease detection. It will be understood that the above description is merely illustrative of the applications of the principles of the present invention, the scope of which is to be determined by the claims viewed in light of the specification. Other variants and modifications of the invention will be apparent to those of skill in the art.
This application claims the benefit of U.S. Provisional Application No. 63/599,621, filed Nov. 16, 2023, which is incorporated herein by reference.
This invention was made with Government support under Contract No. DE-NA0003525 awarded by the United States Department of Energy/National Nuclear Security Administration. The Government has certain rights in the invention.
| Number | Date | Country | |
|---|---|---|---|
| 63599621 | Nov 2023 | US |
