Heterogeneous Epoxide Carbonylation by Cooperative Ion-Pair Catalysis in Co(CO)4--Incorporated Cr-MIL-101.

Despite the commercial desirability of epoxide carbonylation to β-lactones, the reliance of this process on homogeneous catalysts makes its industrial application challenging. Here we report the preparation and use of a Co(CO)4--incorporated Cr-MIL-101 (Co(CO)4⊂Cr-MIL-101, Cr-MIL-101 = Cr3O(BDC)3F, H2BDC = 1,4-benzenedicarboxylic acid) heterogeneous catalyst for the ring-expansion carbonylation of epoxides, whose activity, selectivity, and substrate scope are on par with those of the reported homogeneous catalysts. We ascribe the observed performance to the unique cooperativity between the postsynthetically introduced Co(CO)4- and the site-isolated Lewis acidic Cr(III) centers in the metal-organic framework (MOF). The heterogeneous nature of Co(CO)4⊂Cr-MIL-101 allows the first demonstration of gas-phase continuous-flow production of β-lactones from epoxides, attesting to the potential applicability of the heterogeneous epoxide carbonylation strategy.

β-Lactones have received considerable attention due to their prevalence as key intermediates in numerous synthetic pathways.[1,2] Their versatility stems from the inherent ring strain in the four-membered cycles, which renders β-lactones highly susceptible to a rich variety of ring-opening and ring-expanding transformations. The high commercial value of the resulting products, namely, β-hydroxy acids,[3] biodegradable polyhydroxyalkanoates,[4] and succinic anhydrides,[5] further substantiates the industrial relevance of β-lactone chemistry. Despite their obvious utility, β-lactones have traditionally found comparatively little use industrially because their synthesis is challenging.[6] One particularly attractive solution to this challenge is the ring-expansion carbonylation of epoxides, which exploits the ready availability of epoxides and CO.[7−9] This route has become viable recently through the work of Alper et al. and Coates et al., who have demonstrated efficient carbonylation of epoxides by a series of homogeneous catalysts constituted of a Lewis acid and Co(CO)4–.[10−15] This discovery has prompted an ongoing effort in the private sector to commercialize an epoxide carbonylation process, despite the homogeneous nature of the catalyst.[16] At large scale, however, heterogeneous processes are clearly desirable, yet to date there are no competent heterogeneous catalysts for this process that can compete with the homogeneous systems. The development of an effective heterogeneous catalyst would undoubtedly aid the integration of the epoxide carbonylation process to industrial practice.

Herein, we report the synthesis and use of a Co(CO)4–-incorporated Cr-MIL-101 (Co(CO)4⊂Cr-MIL-101, Cr-MIL-101 = Cr3O(BDC)3F, H2BDC = 1,4-benzenedicarboxylic acid) as the first heterogeneous catalyst for the carbonylation of epoxides to β-lactones that is competitive with the homogeneous process. The activity and selectivity profiles of Co(CO)4⊂Cr-MIL-101 compare favorably with those of the most-active homogeneous catalysts for the liquid-phase batch carbonylation of a range of epoxide substrates. Enabled by the heterogeneous nature of our catalyst, we also report the first proof-of-concept demonstration of gas-phase continuous-flow production of β-butyrolactone from propylene oxide and CO.

In designing a heterogeneous epoxide carbonylation catalyst, we focused our attention on the proposed catalytic cycle for the carbonylation of epoxides by [Lewis acid]+[Co(CO)4]−: (1) epoxide activation by [Lewis acid]+, (2) attack by Co(CO)4–, (3) migratory insertion and uptake of CO, and (4) ring closing and extrusion (Figure A).[17] In view of the irreplaceable role of Co(CO)4– in the proposed CO insertion steps, we identified [Lewis acid]+ as the modifiable component and investigated its structure in various [Lewis acid]+[Co(CO)4]− catalysts. One recurring motif found in the homogeneous catalysts was the pseudo-octahedral Cr(III) center, in which the metal ion is coordinated equatorially by the nitrogen or oxygen atoms of a tetradentate macrocyclic ligand and axially by the oxygen atoms of solvent molecules (Figures B and 1C).[14,15] We reasoned that such a coordination environment favors the activation of epoxides for carbonylation and searched for a heterogeneous analogue. The MOF Cr-MIL-101 was a promising candidate as its metal clusters contain structurally similar octahedral Cr(III) ions that are coordinated equatorially by the oxygen atoms of the bridging terephthalate ligands and axially by a μ3-oxygen atom and a solvent molecule (Figure D).[18] Crucially, Cr-MIL-101 has a cationic framework with ion-exchangeable F–.[19,20] We surmised that exchanging F– with Co(CO)4– would lead to the isolation of a heterogeneous catalyst of the general formula [Lewis acid]+[Co(CO)4]−, mimicking that of the homogeneous species. Other innate properties of Cr-MIL-101 that we deemed favorable for catalysis were its high hydrothermal and chemical stability, large surface area (4100 m2/g as measured by N2 adsorption), large windows (12 and 16 Å) and pores (29 and 34 Å) for ready diffusion of reaction species, site-isolation of the Lewis acidic Cr(III) centers for robust catalysis, and facile synthesis using inexpensive chromium and terephthalic acid precursors.[21] Similar strategies to leverage the intrinsic stability,[22,23] porosity,[24,25] and site-isolation[26−28] of MOFs have proven to be effective in their applications to heterogeneous catalysis. Therefore, postsynthetic ion exchange of Co(CO)4– into Cr-MIL-101 was sought for the formation of a heterogeneous [Lewis acid]+[Co(CO)4]− system.

Figure 1

(A) Proposed catalytic cycle for the ring-expansion carbonylation of epoxides by [Lewis acid]+[Co(CO)4]−.[17] (B) Illustration of the structure of [(OEP)Cr(THF)2]+[Co(CO)4]− (OEP = 2,3,7,8,12,13,17,18-octaethylporphyrinato, THF = tetrahydrofuran).[14] (C) Illustration of the structure of [(salph)Cr(THF)2]+[Co(CO)4]− (salph = N,N′-o-phenylenebis(3,5-di-tert-butylsalicylideneimine)).[15] (D) Illustration of the metal cluster structure of Co(CO)4⊂Cr-MIL-101 with coordinated THF molecules.

The charge-balancing F– anions in the as-synthesized Cr-MIL-101 are directly coordinated to the Cr(III) sites of the framework.[18] To replace these framework-bound anions with uncoordinated Co(CO)4–, anion exchange was performed in two discrete steps: (1) exchange of the bound F– with mobile Cl– using AlCl3 and (2) exchange of the mobile Cl– with Co(CO)4– using Na[Co(CO)4]. In the initial anion exchange, Al3+ shows greater affinity to F– than the framework Cr(III) sites, resulting in the abstraction of F– from the MOF.[19] The consequent charge imbalance is compensated by the inclusion of Cl– into the framework. This sequence of events was tracked by energy dispersive X-ray spectroscopy (EDX) analysis of the Cr-MIL-101 sample before (Cr-MIL-101-F) and after (Cr-MIL-101-Cl) soaking in a solution of AlCl3 (Figure A). In the EDX spectra, the F Kα peak observed at 0.68 keV for Cr-MIL-101-F was replaced by the Cl Kα peak at 2.62 keV for Cr-MIL-101-Cl upon AlCl3 treatment and extensive washing. The absence of the F Kα and Al Kα peaks in the spectrum of Cr-MIL-101-Cl implies complete exchange of F– by Cl– and negligible retention of [AlCl3F]− or any potentially unreacted AlCl3. The structure of the MOF remained intact after the ion exchange as evidenced by the retention of crystallinity in the powder X-ray diffraction (PXRD) analysis of Cr-MIL-101-Cl (Figure S1).

Figure 2

(A) EDX spectra of Cr-MIL-101-F, Cr-MIL-101-Cl, and Co(CO)4⊂Cr-MIL-101. Au peaks from the preanalysis Au coating of samples. (B) ATR-IR absorption spectra of Cr-MIL-101-Cl and Co(CO)4⊂Cr-MIL-101.

Subsequent anion exchange between Cr-MIL-101-Cl and Na[Co(CO)4] was analyzed by EDX and attenuated total reflectance infrared spectroscopy (ATR-IR), both of which confirmed the elimination of Cl– and inclusion of Co(CO)4– into the framework. Upon soaking Cr-MIL-101-Cl in a Na[Co(CO)4] solution, the Cl Kα peak in the EDX spectrum of the former was replaced by Co Kα and Lα peaks at 6.92 and 0.78 keV, respectively (Figure A). The Co peaks persisted even after repeatedly washing Co(CO)4⊂Cr-MIL-101 in tetrahydrofuran (THF), which readily solubilizes Na[Co(CO)4], suggesting that Co(CO)4– is immobilized electrostatically in the MOF through ion-pairing with the Cr(III) Lewis acid sites.[29] The EDX spectrum also evidences the near-absence of the Cl Kα peak at 2.62 keV and the Na Kα peak at 1.04 keV, which are prominent in the spectra of Cr-MIL-101-Cl and Na[Co(CO)4], respectively (Figure S2). These data suggest that the observed Co signal for Co(CO)4⊂Cr-MIL-101 does not stem from residual Na[Co(CO)4] adsorbed on the surface of the MOF, but rather from substitution of Cl– by Co(CO)4–. The ion exchange and inclusion of Co(CO)4– in Cr-MIL-101 are further corroborated by the ATR-IR spectrum of Co(CO)4⊂Cr-MIL-101, which clearly shows the emergence of a single peak at 1888 cm–1 after the exchange procedure (Figure B). This band is in line with the characteristic carbonyl stretching mode of the tetrahedral Co(CO)4– ion in various metal complexes, including those reported for the homogeneous [Lewis acid]+[Co(CO)4]− epoxide carbonylation catalysts.[12−15,30,31] Finally, the structure and porosity of the MOF were retained after this final ion exchange step as evident in the unchanged PXRD pattern for Co(CO)4⊂Cr-MIL-101 (Figure S1).

The catalytic activity demonstrated by Co(CO)4⊂Cr-MIL-101 for the ring-expanding carbonylation of epoxides is competitive with that of the homogeneous catalysts. When using neat 1,2-epoxyhexane as a substrate, Co(CO)4⊂Cr-MIL-101 loaded with 0.5 mol % of cobalt produced the corresponding β-lactone in 86% yield after 5 h under 60 bar of CO at 60 °C (Table , entry 6). This corresponds to a calculated site time yield (STY) of 34 h–1, which is comparable to the values reported for a series of homogeneous catalysts under similar reaction conditions (Table , entries 1–5).

Table 1

Catalysts for the Ring-Expansion Carbonylation of Epoxides

entry	catalyst	R	solvent	PCO (bar)	T (°C)	t (h)	[epoxide]/[Co]a	yield (%)	STY (h–1)b	ref
1	BF3·Et2O + [PPN]+[Co(CO)4]−c	n-Bu	DMEh	62	80	24	50	66	1.4	(10)
2	[(Cp)2Ti(THF)2]+[Co(CO)4]−d	(CH2)2HC=CH2	DMEh	62	60	4	20	90	4.5	(11)
3	[(salph)Al(THF)2]+[Co(CO)4]−e	n-Bu	neat	62	60	6	350	40	23	(13)
4	[(TPP)Cr(THF)2]+[Co(CO)4]−f	n-Bu	neat	62	60	6	350	>99	58	(13)
5	[(OEP)Cr(THF)2]+[Co(CO)4]−g	n-Bu	neat	62	60	6	4500	>99	740	(14)
6	Co(CO)4⊂Cr-MIL-101	n-Bu	neat	60	60	5	200i	86j	34	this work
7	Co(CO)4⊂Cr-MIL-101	n-Bu	DMEh	60	60	1	200i	88j	180	this work
8	Co(CO)4⊂Cr-MIL-101	(CH2)2HC=CH2	DMEh	60	60	1.5	200i	93j	120	this work
9	Co(CO)4⊂Cr-MIL-101	CH2OEt	DMEh	60	60	4	200i	92j	46	this work
10	Co(CO)4⊂Cr-MIL-101	CH2Cl	DMEh	60	60	4	200i	56j,k	28	this work

[Epoxide]/[Co] = moles of epoxide per mole of cobalt in catalyst.

Site time yield = moles of β-lactone produced per mole of cobalt in catalyst per hour throughout overall reaction time t.

PPN = bis(triphenylphosphine)iminium.

Cp = cyclopentadienyl.

salph = N,N′-o-phenylenebis(3,5-di-tert-butylsalicylideneimine).

TPP = 5,10,15,20-tetraphenylporphyrinato.

OEP = 2,3,7,8,12,13,17,18-octaethylporphyrinato.

DME = 1,2-dimethoxyethane.

As determined from an inductively coupled plasma mass spectrometry (ICP-MS) derived cobalt content of the catalyst.

As determined by 1H NMR analysis with mesitylene as an internal standard.

32% of the substrate epoxide remained unreacted.

The solvent dependence of the carbonylation activity in Co(CO)4⊂Cr-MIL-101 also mimicked that of the reported homogeneous systems. When a range of solvents was screened for optimizing activity with Co(CO)4⊂Cr-MIL-101, reactions in weakly coordinating ethers such as 1,2-dimethoxyethane (DME) showed the highest activity. In other solvents, especially more strongly coordinating solvents such as THF, the reactions proceeded at a much slower rate. Identical solvent dependence has been reported for the Cr(III)-based [(salph)Cr(THF)2]+[Co(CO)4]− (salph = N,N′-o-phenylenebis(3,5-di-tert-butylsalicylideneimine)), a homogeneous catalyst that most closely resembles Co(CO)4⊂Cr-MIL-101 structurally (Figure C).[15]

Importantly, Co(CO)4⊂Cr-MIL-101 displays the broad functional group tolerance of the homogeneous catalysts as evidenced by its activity toward an array of aliphatic epoxides as well as glycidyl ether and epichlorohydrin (Table , entries 7–10). It is also noteworthy that the observed catalytic activity of Co(CO)4⊂Cr-MIL-101 is orders of magnitude higher than that of an immobilized homogeneous catalyst under similar reaction conditions.[32]

To validate the heterogeneous nature of the observed catalytic activity, Co(CO)4⊂Cr-MIL-101 was isolated from a portion of the epoxide carbonylation reaction mixture by filtration at ∼15% conversion (Figure S3). When the acquired filtrate and the unfiltered mixture were both subjected to the standard reaction conditions again, the filtrate did not show any increase in β-lactone yield whereas the unfiltered portion resumed its epoxide carbonylation activity. The structural integrity of Co(CO)4⊂Cr-MIL-101 was also retained after all epoxide carbonylation reactions as confirmed by PXRD analysis of spent Co(CO)4⊂Cr-MIL-101 (Figure S1).

To verify the catalytic cooperativity between the Cr(III) sites and Co(CO)4– in Co(CO)4⊂Cr-MIL-101, Cr-MIL-101-F, Cr-MIL-101-Cl, and Na[Co(CO)4] were tested for epoxide carbonylation activity (Table S1). All systems displayed negligible product formation when subjected to a 5 h reaction with neat 1,2-epoxyhexane at 0.5 mol % loading, 60 bar CO, and 60 °C. In addition, no substantial catalytic activity was observed when HKUST-1 (Cu3BTC2, H3BTC = benzene-1,3,5-tricarboxylic acid) (Figure S4), a representative Lewis acidic MOF with Cu(II) sites,[33] was subjected to the same reaction conditions along with an equimolar amount of Na[Co(CO)4] (Table S1). These results show that the epoxide carbonylation activity is specific to Co(CO)4⊂Cr-MIL-101, where the unique combination of Co(CO)4– and the strong Lewis acidic Cr(III) sites of the Cr-MIL-101 framework is required for cooperative catalysis. We note that an equimolar mixture of the as-synthesized Cr-MIL-101 and Na[Co(CO)4] also showed β-lactone formation under reaction conditions, presumably due to partial formation of the Cr/Co sites in situ (Table S1). The observed catalytic activity, however, was lower than that of Co(CO)4⊂Cr-MIL-101, which exhibits preassembled Cr/CO sites installed within the framework prior to reaction.

Encouraged by the observed heterogeneous epoxide carbonylation activity in liquid-phase batch reactions, we subjected Co(CO)4⊂Cr-MIL-101 to a gas-phase continuous-flow reaction using propylene oxide and CO. Uniquely among all known catalysts for this reaction, Co(CO)4⊂Cr-MIL-101 showed catalytic epoxide carbonylation activity to form β-butyrolactone under unoptimized reaction conditions: holding Co(CO)4⊂Cr-MIL-101 at 70 °C and subjecting it to 20 bar of 0.02 mol % propylene oxide and balance CO flowing at 125 mL/min (at standard temperature and pressure, STP) resulted in STYs of approximately 6 molβ-butyrolactone·molCo–1·h–1 and a turnover number of 60 molβ-butyrolactone·molCo–1 after 24 h on stream (Figures S5 and S6). Notably, catalytic activity ensued only after in situ removal of the pore-filling THF solvent left over from catalyst synthesis, which presumably blocks substrate access to the Cr/Co active sites. We expect optimization of the reaction conditions to further enhance catalytic performance for a more efficient continuous production of β-lactones. Nonetheless, the ability to perform reactions in the gas phase is a unique feature of our heterogeneous catalyst that is inaccessible to the homogeneous catalysts developed to date.

In summary, we have demonstrated the successful preparation of Co(CO)4⊂Cr-MIL-101 and its catalytic activity in the heterogeneous ring-expansion carbonylation of epoxides. We attribute this activity to the cooperativity between the Lewis acidic Cr(III) sites and the postsynthetically incorporated Co(CO)4–. The favorable catalytic performance exemplified by this heterogeneous system and its proof-of-concept application in the first gas-phase continuous production of β-lactone highlight the effectiveness of the heterogeneous epoxide carbonylation pathway and warrant further evaluation of its industrial applicability.