metal@cofs: covalent organic frameworks as templates for
TRANSCRIPT
Metal@COFs: Covalent Organic Frameworks as Templates for PdNanoparticles and Hydrogen Storage Properties of
Pd@COF-102 Hybrid Material
Suresh Babu Kalidindi,[a] Hyunchul Oh,[b] Michael Hirscher,*[b] Daniel Esken,[a]
Christian Wiktor,[a, c] Stuart Turner,[c] Gustaaf Van Tendeloo,[c] and Roland A. Fischer*[a]
� 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2012, 18, 10848 – 1085610848
DOI: 10.1002/chem.201201340
Introduction
Covalent organic frameworks (COFs) are organic analoguesof metal–organic frameworks (MOFs) and are constructedsolely from organic building blocks. Instead of metal–linkercoordination bonds as in MOFs, COFs are linked by strongcovalent bonds. In their stimulating work, Yaghi and co-workers showed that self-condensation of aromatic diboron-ic acids or condensation of aromatic alcohols and aromaticdiboronic acids in a controlled way yields crystalline organicpolymers.[1] These fascinating materials exhibit high thermalstability and large surface areas on a par with MOFs. Poten-tial applications of COFs in gas storage, optoelectronics, andcatalysis have been demonstrated.[2] Unlike most of the or-ganic polymers, COFs have an ordered crystalline structure,which makes them a promising new class of templates forhosting nanoparticles. The template-directed synthesis routeis one of the widely used methods for producing various
types of metal nanoparticles, and the template plays a majorrole in determining the size and shape of the nanoparticles.[3]
MOFs, microporous zeolites, mesoporous siliceous materials,and related matrices are shown to act as templates for nano-particles. Realization of nanoparticles with narrow size dis-tribution even at higher metal loadings is the key challengein this area.[4] In this regard, 3D COFs (COF-102) with thefollowing interesting structural features are potential tem-plates to realize monodisperse particles at higher metalloadings. Some of the structural features of COF-102 are:1) the guest molecule adsorbed within the framework “sees”all the atoms used for its construction; and 2) pores do notform channel-like structures, but are highly interconnected.
On the other hand, COFs have exceptionally low densitiesbecause they are made up of light elements, and thereforecan store very high gravimetric amounts of small gas mole-cules. Hydrogen, with a chemical energy per mass of H2
equal to 142 MJ kg�1, is a widely accepted green fuel.[5]
However, safe and economical storage of hydrogen formobile applications remains one of the bottlenecks for itswidespread usage. Yaghi and co-workers studied the H2 stor-age properties of 2D and 3D COFs and found that COF-102outperforms the other COFs and stores close to 7 wt % ofhydrogen at 77 K and 35 bar.[2c,d] The next challenge lies inthe tuning of adsorption enthalpies for effective storageunder ambient conditions. Several theoretical investigationshave been reported from this point of view, which involvedecoration of the COF surface with alkali-metal ions, alkali-metal clusters, and transition-metal catalysts.[6] However, ex-perimental studies in this regard are yet to be carried out.Herein, we present the synthesis of a new organic–inorganichybrid material “metal@COF”. The objectives of this workare to establish COFs as novel templates for the stabiliza-tion of nanoparticles and to study the hydrogen storageproperties of metal-decorated COF surfaces.
Abstract: Three-dimensional covalentorganic frameworks (COFs) have beendemonstrated as a new class of tem-plates for nanoparticles. Photodecom-position of the [Pd ACHTUNGTRENNUNG(h3-C3H5) ACHTUNGTRENNUNG(h5-C5H5)]@COF-102 inclusion compound(synthesized by a gas-phase infiltrationmethod) led to the formation of thePd@COF-102 hybrid material. Ad-vanced electron microscopy techniques(including high-angle annular dark-field scanning transmission electron mi-croscopy and electron tomography)along with other conventional charac-terization techniques unambiguouslyshowed that highly monodisperse Pdnanoparticles ((2.4�0.5) nm) wereevenly distributed inside the COF-102
framework. The Pd@COF-102 hybridmaterial is a rare example of a metal-nanoparticle-loaded porous crystallinematerial with a very narrow size distri-bution without any larger agglomerateseven at high loadings (30 wt %). Twosamples with moderate Pd content (3.5and 9.5 wt %) were used to study thehydrogen storage properties of themetal-decorated COF surface. The up-takes at room temperature from thesesamples were higher than those of simi-
lar systems such as Pd@metal–organicframeworks (MOFs). The studies showthat the H2 capacities were enhancedby a factor of 2–3 through Pd impreg-nation on COF-102 at room tempera-ture and 20 bar. This remarkable en-hancement is not just due to Pd hy-dride formation and can be mainly as-cribed to hydrogenation of residual or-ganic compounds, such asbicyclopentadiene. The significantlyhigher reversible hydrogen storage ca-pacity that comes from decomposedproducts of the employed organometal-lic Pd precursor suggests that this dis-covery may be relevant to the discus-sion of the spillover phenomenon inmetal/MOFs and related systems.
Keywords: covalent organic frame-works · hydrogen storage · nanopar-ticles · organic–inorganic hybridcomposites · palladium
[a] Dr. S. B. Kalidindi,+ Dr. D. Esken, C. Wiktor, Prof. Dr. R. A. FischerInorganic Chemistry II—Organometallics & MaterialsFaculty of Chemistry and BiochemistryRuhr University Bochum44780 Bochum (Germany)Fax: (+49) 234 32-14174E-mail : [email protected]
[b] H. Oh,+ Dr. M. HirscherMax Planck Institute for Intelligent SystemsHeisenbergstrasse 3, 70569 Stuttgart (Germany)E-mail : [email protected]
[c] C. Wiktor, Dr. S. Turner, Prof. Dr. G. Van TendelooEMAT, University of AntwerpGroenenborgerlaan 171, 2020 Antwerp (Belgium)
[+] These authors contributed equally to this work.
Supporting information for this article is available on the WWWunder http://dx.doi.org/10.1002/chem.201201340.
Chem. Eur. J. 2012, 18, 10848 – 10856 � 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 10849
FULL PAPER
We chose a 3D COF, COF-102, as a representative exam-ple of this family due to its promising hydrogen storageproperties. Self-condensation of tetra(4-dihydroxyborylphe-nyl)methane at 85 8C yields the COF-102 framework, whichis based on the hypothetical C3N4 net (ctn type, seeFigure 1). On the other hand, Pd is a well-studied hydrogenstorage material that can absorb hydrogen into its lattice toform PdHx (x= 0.1 to 0.6).[7] We selectively decorated the in-terior surface of COF-102 microcrystals with Pd nanoparti-cles by using a precursor chemical infiltration technique,and the hydrogen storage properties of the resultant hybridmaterial were evaluated at 77 K and room temperature.
Results and Discussion
A range of Pd@COF-102 samples with varying Pd content(30, 20, 9.5, and 3.5 wt %) were synthesized. The characteri-zation data of 30 wt % Pd are discussed in detail, since thenovelty of the COF template lies in obtaining very monodis-perse nanoparticles at higher metal loadings. The sampleswith lower Pd content were used for the hydrogen storagestudies due to high surface areas that are close to pureCOF-102, which allows comparison of the hydrogen absorp-tion properties of the materials.
COF-102 as a template for Pd nanoparticles : Solvent-freegas-phase infiltration of volatile organometallic precursorsinside porous frameworks is demonstrated as a highly versa-tile method for the synthesis of metal@MOF-like hybrid ma-terials.[3b] In this method, the inclusion compounds[MLn]a@MOF (index “a” refers to the number of organome-tallic guest molecules per MOF formula unit) are decom-posed by heat or light to yield metal@MOF composite ma-terials. The major advantage associated with this approach isthat heavy-homogeneous metal loadings can be obtained in
a single loading step. Recently, by using this procedure,some of us have studied for the first time the organometallichost–guest chemistry of COF-102 with [FeACHTUNGTRENNUNG(h5-C5H5)2], [Co-ACHTUNGTRENNUNG(h5-C5H5)2], and [Ru ACHTUNGTRENNUNG(cod) ACHTUNGTRENNUNG(cot)] (cod= cycloocta-1,5-diene,cot= cycloocta-1,3,5,7-tetraene) as guest molecules andshowed that ferrocene assumed a structure reminiscent ofthe host through p–p interactions.[8] Also, it has been estab-lished that COF-102 keeps its crystalline nature intact inmetallocene@COF-102 inclusion compounds. Herein, weused highly volatile and light-sensitive [Pd ACHTUNGTRENNUNG(h3-C3H5) ACHTUNGTRENNUNG(h5-C5H5)] as a precursor for Pd nanoparticles (Figure 1). In atypical loading experiment, activated COF-102 and [Pd ACHTUNGTRENNUNG(h3-C3H5) ACHTUNGTRENNUNG(h5-C5H5)] were placed in two separate glass boats in aSchlenk tube under vacuum (�10�3 mbar). The moleculardimensions of [Pd ACHTUNGTRENNUNG(h3-C3H5) ACHTUNGTRENNUNG(h5-C5H5)] are �2.2, 2.3, and4.1 � in the x, y, and z directions, respectively, which aremuch smaller than the pore openings (above 1 nm) of theCOF-102 framework. Therefore, the diffusion of precursorinto the COF matrix is highly feasible. The color of theCOF-102 matrix changed immediately to light red and thento dark red, due to the infiltration of [Pd ACHTUNGTRENNUNG(h3-C3H5) ACHTUNGTRENNUNG(h5-C5H5)]into the matrix.
The Schlenk tube in which the reaction was performedwas kept at room temperature for 3 h in a dark cupboard.The unstable nature (due to light sensitivity) of the inclusioncompound (1) did not allow us to complete its characteriza-tion. The IR spectrum (Figure S1 in the Supporting Informa-tion) of compound 1, recorded immediately after isolationfrom the Schlenk tube, showed some additional peaks alongwith peaks corresponding to the COF-102 structure. TheXRD data (Figure S2 in the Supporting Information) ofcompound 1 obtained immediately after isolation alsoshowed that the COF-102 crystalline structure is intact afterloading with the [Pd ACHTUNGTRENNUNG(h3-C3H5) ACHTUNGTRENNUNG(h5-C5H5)] complex. From ele-mental analysis data, the number of [Pd ACHTUNGTRENNUNG(h3-C3H5) ACHTUNGTRENNUNG(h5-C5H5)]molecules per COF-102 formula unit was determined to beclose to two.
The inclusion compound 1 was subjected to UV-light irra-diation for 16 h at room temperature, and a change in thecolor of the sample from dark red to black indicated the for-mation of Pd nanoparticles. Subsequently, thermal treatmentunder vacuum at 400 K for 12 h was performed to removeany physisorbed organic byproducts. A range of sampleswith varying Pd contents (30, 20, 9.5, and 3.5 wt %) weresynthesized.
The IR spectrum (Figure 2) of the obtained materialPd30wt%@COF-102 (2) showed no signs of decompositionof COF-102; the IR spectrum perfectly matches that of thepure matrix. Additional bands were noted around3000 cm�1, which could be due to leftover organic com-pounds from the decomposition of the organometallic pre-cursor (details are given later). Corroborating this, the13C NMR magic-angle spinning (MAS) spectrum (Figure S3in the Supporting Information) showed four intense peaksat d= 151.4, 134.2, 128.5, and 66.0 ppm, which again perfect-ly match those for the host framework. Once more, low-in-tensity peaks corresponding to leftover organic compounds
Figure 1. Schematic representation of the synthesis of the Pd@COF-102hybrid material. Left: the ctn structure of COF-102; carbon, boron, andoxygen atoms are represented in gray, red, and green, respectively. Right:the reaction carried out inside the pores of COF-102 for the synthesis ofPd nanoparticles.
www.chemeurj.org � 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2012, 18, 10848 – 1085610850
M. Hirscher, R. A. Fischer et al.
were observed at around d= 40 ppm. From the isotherms ofthe N2 sorption measurements of the composite, an equiva-lent BET surface area of 1419 m2 g�1 was calculated. The de-crease in surface area corresponds to filling of pores of theframework with Pd nanoparticles. The powder XRD pattern(Figure 3) of 2 reveals the characteristic reflections of the
host matrix along with a very broad peak around 2q=39.78assigned to approximately 2.5 nm sized Pd nanoparticles, asestimated from the Scherrer equation. These characteriza-tion data suggest that small Pd nanoparticles are formedand are present along with COF-102 in compound 2. Exten-sive transmission electron microscopy (TEM) studies werecarried out on compound 2 to gain complete insight into thedistribution of nanoparticles in the COF-102 matrix.
Bright-field TEM (Figure S4 in the Supporting Informa-tion) and high-angle annular dark-field and scanning trans-mission electron microscopy (HAADF-STEM; Figure 4 a)images of 2 showed the presence of COF-102 crystals sys-tematically and heavily loaded with Pd nanoparticles. Theiraverage size was measured to be (2.4�0.5) nm (Figure 5).Larger agglomerates of Pd were not found. The observedparticle size is larger than the pore diameter (0.9 nm) of theCOF-102 framework, and is attributed to the highly inter-connected pores in COF-102 with openings close to 1 nm,possibly allowing the growth of the particles up to the ob-served size regime without distortion of the matrix.
The chemical nature of the nanoparticles was evidencedby local energy-dispersive X-ray (EDX) spectra (Figure S5in the Supporting Information) and proven to be cubic Pd0
(Fm3̄m, space group 225) by indexing Fourier transforms ofparticles imaged in high-resolution (HR) TEM (Figure 6)and HAADF-STEM (Figure 4 b–e) along zone-axis orienta-tions. Some imaged Pd particles exhibited the presence ofsingle or multiple twins (e.g., the fivefold-twinned nanopar-ticle in Figure 4 f). Material 2 was not stable enough underthe TEM electron beam to be able to reveal electron dif-fraction information. However, it was possible to obtain aweak diffraction pattern from empty COF-102 (Figure S6 inthe Supporting Information), thereby proving its composi-tion and crystallinity.
To determine the 3D distribution of the Pd nanoparticleswithin the COF-102 crystals, electron tomography was per-
Figure 2. IR spectrum of a) activated COF-102 and b) Pd30wt%@COF-102 hybrid material after heating at 130 8C under vacuum. The peaks inthe zoomed in region correspond to leftover byproducts, such as bicyclo-pentadiene.
Figure 3. Powder XRD patterns of a) Pd30wt%@COF-102 hybrid materi-al and b) activated COF-102; the inset shows a broad peak correspondingto Pd nanoparticles.
Figure 4. a) HAADF-STEM of Pd30wt%@COF-102; the image revealsthe even distribution of Pd particles throughout the framework crystals.b) HRHAADF-STEM image of a Pd nanoparticle imaged along its [110]zone axis at atomic resolution. c) Fourier transform of the area highlight-ed by the white frame in (b) showing evidence that the particle is cubicPd0. d) HRHAADF-STEM image of the Pd nanoparticle imaged alongits [100] zone axis. e) The indexed Fourier transform of the area in thewhite box in (d). f) HRHAADF-STEM image of a fivefold-twinned Pdnanoparticle. Its fivefold symmetry is highlighted by white arrows; thescale bars in (b), (d), and (f) are 5 nm.
Figure 5. Size distribution of the Pd nanoparticles in Pd30wt%@COF-102, with particle sizes measured from HAADF-STEM images. The aver-age particle size was calculated to be (2.4�0.5) nm.
Chem. Eur. J. 2012, 18, 10848 – 10856 � 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 10851
FULL PAPERMetal@Covalent Organic Frameworks
formed. An overview of the reconstruction of a loaded crys-tal together with a slice through the reconstructed volume isshown in Figure 7 (see the Supporting Information for a 3D
rendering (MPEG)) reveals that the particles are denselyand homogeneously distributed throughout the whole COF-102 crystal. Larger particles at or close to the surface thatare quite common in loaded MOF systems were absent (seeFigure 7 b).[4b, 9] The excellent distribution of particles insideCOF-102 crystals is unambiguously established from thesetomography data. In fact, this composite system is a veryrare, if not the only, example of metal-particle-loadedporous crystalline materials with very narrow size distribu-tion and an absence of larger agglomerates even at higherloadings (30 wt %). This suggests that COF-102 is an excel-lent template for the stabilization of and as host matrix forhighly monodisperse nanoparticles. The COF-102, which hasits entire matrix (all the atoms involved in the frameworkconstruction) available to guest molecules (nanoparticles),could provide a more efficient stabilization to nanoparticlesthrough interaction with the framework surface, thus leadingto the formation of highly monodisperse nanoparticles.
Pd@COF-102 samples with lower Pd content : Similar obser-vations were made for the samples with lower Pd content.
At loadings of 20, 3.5, and 9.5 wt% the particles still exhibit-ed the same small size and were still monodisperse. Theamount of Pd loading has no effect on the particle size inPd@COF-102 hybrid material. The majority of the crystalsexhibited a homogeneous particle distribution throughoutthe COF-102 crystals even at lower loadings of Pd (Figure 8
and Figure S7 in the Supporting Information). The particlesize remained almost the same (around 1–4 nm) irrespectiveof the amount of Pd present in the samples. Very fewloaded crystals were found with particles only present at orvery close to the surface (Figure S7a–e in the Supporting In-formation). This may be due to the gradient diffusion of themetal precursor through COF-102 crystals at lower concen-trations and preferential aggregation or clustering of precur-sor molecules at levels of precursor adsorption below thesaturation limit. Also, for all these samples, any detectablechanges corresponding to decomposition of the frameworkwere not noticed in the powder XRD and IR spectroscopydata (Figures S8 and S9 in the Supporting Information),after loading with Pd. The BET (N2) surface areas measuredwere 1419 and 3145 m2 g�1 for 30 and 3.5 wt % samples, re-spectively. The adsorption/desorption curves are shown inFigure 9. As expected, the micropore volume decreasedwith the increase in Pd loading, which suggests incorpora-tion of Pd inside COF-102 pores.
Hydrogen storage properties of Pd@COF-102 : Currentcryogenic hydrogen storage capacities for mobile applica-tions have been mainly demonstrated in the materials withhighest specific surface area to approach the U.S. Depart-ment of Energy target.[10] However, the storage capacity ofthese materials exponentially decreases at ambient tempera-ture due to the small enthalpy of H2 adsorption (typicalrange of 3.5–7 kJ mol�1[11]), which is well below the15 kJ mol�1[12] that is estimated at optimal operating condi-tions under moderate temperature and pressure. Withregard to this, the study of metal–organic hybrid materialsrepresents one promising route for improving the enthalpy
Figure 6. HRTEM of Pd30wt%@COF-102; the insets show the Fouriertransforms of the areas highlighted by the white frames. Only reflexes ofcubic Pd0 are present.
Figure 8. Bright-field TEM image of a Pd3.5wt%@COF-102 crystal. Theparticles are in the size range of 1–4 nm.
Figure 7. a) Tomographically reconstructed volume; the Pd particles arerendered in grey, the COF framework in white. b) Orthoslice through the3D reconstruction of loaded Pd30wt%@COF-102. The particles visible aswhite dots are clearly present in the whole crystal without any preferen-tial distribution in size or position.
www.chemeurj.org � 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2012, 18, 10848 – 1085610852
M. Hirscher, R. A. Fischer et al.
of H2 adsorption and consequently the storage properties aswell. Recently, Latroche and co-workers[13a] suggested thatthe Pd-decorated MOF is a good strategy to enhance the hy-drogen storage performance. Thus, in the work reportedherein, we intend to use our novel Pd@COF-102 hybrid ma-terials to investigate hydrogen storage properties.
Hydrogen adsorption isotherms at 77 K for COF-102,Pd3.5wt%@COF-102, and Pd9.5wt%@COF-102 samples arepresented in Figure 10 a. As shown in Figure 10 a, COF-102had a hydrogen storage capacity of 5.27 wt % at 77 K and20 bar. The value measured for COF-102 is in good agree-ment with the previous work from Yaghi and co-workers.[2c,d]
After decoration of Pd on COF-102, however, the hydrogenuptakes on Pd3.5wt%@COF-102 and Pd9.5wt%@COF-102were decreased to 4.60 and 4.38 wt %, respectively (Fig-ure 10 a). The loss of excess hydrogen uptake at low temper-ature can be correlated with the decrease of the specific sur-face area and pore volume after Pd loading. Hence, the con-tribution of Pd nanoparticles to the hydrogen storage ca-pacity is negligible at 77 K. In contrast, at 298 K and 20 barthe hydrogen storage capacity was enhanced by a factor of2–3 relative to the pristine material (Figure 10 b). Note thatthe adsorption isotherm was completely reversible after fullevacuation overnight, although there appeared to be a hys-teresis at room temperature. It can be seen from Figures S10and S11 in the Supporting Information that the third andfourth adsorption isotherms at room temperature and 77 Kwere in good agreement with the first and second adsorptionisotherms, thus indicating preservation of the framework.Moreover, the absolute value of the hydrogen uptake ofPd@COF-102 hybrid material is significantly higher thanthat for similar systems at a pressure of 20 bar as reportedrecently.[13] Figure 11 and Table S1 in the Supporting Infor-mation show the direct comparison of H2 adsorption iso-therms at 298 K for various Pd contents impregnated inCOFs (this work) and MOFs (taken from ref. [13]). At298 K, the hydrogen uptakes on Pd3.5wt%@COF-102 andPd9.5wt%@COF-102 were 0.38 and 0.42 wt %, respectively,
Figure 9. N2 sorption isotherms of a) COF-102 (~), b) Pd3.5wt%@COF-102 (&), and c) Pd30wt%@COF-102 (*). Filled symbols: adsorption,open symbols: desorption. The micropore volume decreased with the in-crease in Pd loading, which suggests incorporation of Pd inside COF-102crystals.
Figure 10. Excess hydrogen sorption isotherms of pristine COF-102 (&),Pd3.5wt%@COF-102 (*), and Pd9.5wt%@COF-102 (^) at a) 77 andb) 298 K. i) The amount of hydrogen adsorbed by the physisorption proc-ess, ii) the amount of hydrogen adsorbed by hydrogenation of residual or-ganic compounds, and iii) the amount of hydrogen adsorbed by Pd hy-dride; iii-a is due to the 6 wt % difference in Pd loading, �0.036 wt % ofH2; iii-b is due to the 3.5 wt % of Pd impregnated in COF-102,�0.02 wt % of H2. Filled symbols: adsorption, open symbols: desorption.
Figure 11. Comparison of H2 adsorption isotherms at 298 K for variousPd contents impregnated in COFs and MOFs: COF-102 containing3.5 wt % Pd (*), COF-102 containing 9.5 wt % Pd (^), MIL-100(Al) con-taining 10 wt % Pd (": taken from ref. [13a]), and [SNU-3]0.54 + ACHTUNGTRENNUNG(NO3
�)0.54
containing 3 wt % Pd (&: taken from ref. [13b]).
Chem. Eur. J. 2012, 18, 10848 – 10856 � 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 10853
FULL PAPERMetal@Covalent Organic Frameworks
at the moderate pressure of 20 bar. On the other hand, thehydrogen uptakes on Pd10wt%@MIL-100(Al)[13a] andPd3.0wt%@ ACHTUNGTRENNUNG[SNU-3]0.54+ ACHTUNGTRENNUNG(NO3
�)0.54[13b] were 0.35 and
0.30 wt % at higher pressures of 40 and 95 bar, respectively.The remarkable enhancement observed relative to pristineCOF-102 can be only partially explained by Pd hydride for-mation; some other mechanism must be considered for im-proving the hydrogen storage capacity of the Pd@COF-102samples.
“Hydrogen spillover” is a conventional way to describethis phenomenon, which is solely defined as the dissociativechemisorption of hydrogen molecules on metal nanoparti-cles, followed by migration of monatomic hydrogen (some-how) into the porosity of the host MOF.[14] However, thismechanism is not yet fully understood and is still open to ar-gument.[15] Therefore, we leave this point behind in our dis-cussion, and try to explain our data with a new approach. Aconsistent explanation is the following: the 6 wt % increaseof the Pd content between Pd9.5 wt %@COF-102 andPd3.5 wt %@COF-102 results in an increase in the H2
uptake of 0.036 wt % at 20 bar, as shown in Figure 10 b iii-a.Under the assumption that this increase is solely due to hy-dride formation, an upper limit for the ratio of atomic hy-drogen and Pd can be calculated, which leads to H/Pd =0.6.Considering the 20 bar pressure, the ratio of 0.6 H/Pd in Pdnanoparticles agrees well with the literature (e.g., between0.2 and 0.55 H/Pd for the different Pd sizes (2–9 nm) at1 bar H2 and 300–310 K).[7c,16] Hence, from this H/Pd ratioof 0.6, the maximum amount of hydrogen adsorbed by Pdhydride formation can be calculated for Pd3.5wt%@COF-102 (0.02 wt % of H2), as shown in Figure 10 b iii-b. Further-more, the amount of physisorbed hydrogen (0.16 wt% ofH2) coming from the pristine material can be subtracted(Figure 10b i). The further enhancement of the hydrogen ad-sorbed at 20 bar is 0.2 wt %, which has to be related to anadditional mechanism (Figure 10 b ii).
It should also be noted that a relatively strong hysteresisappeared in Pd-impregnated COF-102, especially in the low-pressure region (below 6 bar). To elucidate this point, theadsorption kinetics of Pd-impregnated COF-102 and pristinematerials were investigated (Figure S12 in the SupportingInformation). It can be seen that the equilibration of Pd/COF-102 in isothermal measurements required a muchlonger time at low pressure than at high pressure. The equi-libration time at high pressure is equivalent to the physi-sorption kinetics of pristine materials. This remarkable dif-ference in equilibration time, based upon pressure, indicatedthat a chemical interaction dominated at low pressurewhereas physical interaction prevailed at high pressure.
Pd-catalyzed hydrogenation/dehydrogenation of phenylrings pinned inside the framework is highly unlikely due tothe steric strain barrier; rather, H/D exchange of phenylrings involving C�H/D activation is possible as in the caseof the Ru@MOF-5 system.[17] However, this effect cannotaccount for the observed excess of H2 uptake. Corroboratingthe intact nature of the framework, IR, TEM (Figure S13 inthe Supporting Information), NMR, and XRD data of the
hydrogenated sample (PdH0.6@COF-102) did not indicateany modification of the framework. Considering thesepoints (extra 0.2 wt % of H2 adsorption and hysteresis at lowpressure), we tentatively presumed that residual organiccompounds coming from the decomposed products of theorganometallic Pd precursor might still exist inside COF-102, thereby leading to remarkable enhancement of H2 ad-sorption and hysteresis. A close comparison of the MAS 13CNMR spectra of H2-loaded and -unloaded Pd@COF-102supports our conclusion (Figure S14 in the Supporting Infor-mation). The hydrogenation of residual organic compoundsis consistent with a slight shift in the peak position fromaround 45 to 34 ppm. Thus, the question arises of whatkinds of residual compound exist in sample 2. The decompo-sition (UV-light/thermal) pathways for the [Pd ACHTUNGTRENNUNG(h3-C3H5) ACHTUNGTRENNUNG(h5-C5H5)] precursor have been well studied in the literature.[18]
Based upon these studies, the major decomposition pathwayfor [Pd ACHTUNGTRENNUNG(h3-C3H5) ACHTUNGTRENNUNG(h5-C5H5)] under UV light is shown inFigure 12.
Initially, the Pd precursor complex was loaded inside theframework through gas-phase infiltration, and then decom-posed under UV light. At the end, the compound washeated under a vacuum at 400 K. Therefore, most of theallyl and cyclopentadienyl radical products, such as propy-lene, propane, biallyl, benzene, and cyclopentadiene, shouldevaporate, but dicyclopentadiene, which has low volatility(b.p. �443 K), might have stuck to the surface of Pd nano-particles. NMR signals (Figure S14 in the Supporting Infor-mation) also confirmed the peaks of dicyclopentadiene at136, 132 (overlaps with framework peaks, so not observed),50, 46, 45, 41, and 34 ppm (seen as broad peaks). The exis-tence of dicyclopentadiene inside Pd@COF-102 is also cross-checked by thermal desorption spectroscopy measurements(Figure S15 in the Supporting Information). The major vola-
Figure 12. Most probable reaction pathways for decomposition of the[Pd ACHTUNGTRENNUNG(h3-C3H5)ACHTUNGTRENNUNG(h5-C5H5)] precursor in the presence of UV light. ThePd@COF-102 samples were heated at 403 K under vacuum. TB =boilingtemperature.
www.chemeurj.org � 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2012, 18, 10848 – 1085610854
M. Hirscher, R. A. Fischer et al.
tile residual organic compounds originating from the precur-sor are stripped off from the hybrid material by heating thecompound at 403 K for 12 h under vacuum. We attemptedto remove all the byproducts of [Pd ACHTUNGTRENNUNG(h3-C3H5) ACHTUNGTRENNUNG(h5-C5H5)] byheating to temperatures above 473 K, but the Pd@COF-102hybrid material lost its crystalline nature (Figures S16 andS17 in the Supporting Information). The alkene chemistry(hydrogenation/dehydrogenation) on Pd surfaces (singlecrystalline, as well as supported nanoparticles) has been wellstudied in the literature.[19] The high activity of (nanosized)Pd surfaces leads to strong adsorption of alkenes in a di-s-bond fashion even at temperatures as low as 80 K. The inter-mediate undergoes facile hydrogenation or dehydrogenationthrough b-H elimination depending on the reaction condi-tions via h1-alkyl species. Therefore, we attribute the signifi-cant excess of 0.2 wt % H2 uptake at room temperature and20 bar to hydrogenation/dehydrogenation of organic com-pounds permanently present on the Pd nanoparticles� sur-face in sample 2.
Conclusion
The suitability of 3D COFs for stabilization of Pd nanoparti-cles is demonstrated. The Pd@COF-102 hybrid material de-scribed herein is a very rare example of a metal-nanoparti-cle-loaded porous crystalline material with a very narrowsize distribution. Hydrogen storage studies showed that theH2 storage capacities were enhanced by a factor of 2–3through Pd impregnation on COF-102 at room temperatureand 20 bar. The hydrogen uptake is reversible. This remark-able enhancement was not just due to Pd hydride formation,and can be mainly ascribed to hydrogenation of residual or-ganic compounds. The significantly higher hydrogen storagecapacity that comes from decomposed products of the em-ployed organometallic Pd precursor suggests that our dis-covery may be relevant to the discussion of the spilloverphenomenon in metal/MOF and related systems.
Experimental Section
Synthesis Pd30wt%@COF-102 hybrid material : In a typical experiment,dry activated COF-102 (30 mg) and [Pd ACHTUNGTRENNUNG(allyl) ACHTUNGTRENNUNG(h5-C5H5)] (80 mg) wereplaced in two separate glass boats in a Schlenk tube. The tube was thenevacuated close to 10�3 mbar for 2 min and sealed under vacuum. The re-action tube was kept at room temperature for 3 h. The obtained inclusioncompound (PdACHTUNGTRENNUNG(allyl)Cp@COF-102, Cp=cyclopentadienyl) was exposedto UV light for 6 h under an Ar atmosphere. During this period the colorof the compound changed from dark red to black, thereby indicating theformation of Pd nanoparticles. The product was heated under vacuum at403 K overnight to remove other byproducts. Decreasing the loadingtimes to �30 and �50 min resulted in the formation of hybrid materialswith 3.5 and 9.5 wt % Pd content, respectively.
Hydrogen adsorption/desorption measurements : An automated Sievert�stype apparatus (PCTPro-2000) was used with a so-called microdoser(MD) from HyEnergy. The adsorption and desorption isotherms (0–20 bar) were measured at 298 and 77 K (liquid nitrogen) in a sample cellvolume of approximately 1.3 mL by using ultrahigh-purity hydrogen gas(99.999%). Before the hydrogen uptake measurements, the samples were
heated for several hours at 130 8C in a vacuum (4.5 � 10�9 bar) to removeany adsorbed gas from the surface.
To prevent the influence of the temperature gradient between the cooledsample cell (77 K) and gas reservoir with pressure transducer (298 K)during isotherm measurement at 77 K, nonadsorbing samples (sea sand)possessing the same volume were measured under identical conditions.By subtracting these two measurements, the influence of the temperaturegradient, as well as systematic errors, cancelled out. Thus, the excessamount of hydrogen nexc ACHTUNGTRENNUNG(p,T) adsorbed can be calculated by [Eq. (1)]:
nexcðp,TÞ ¼ nexperimentðp,TÞ�nsea sandðp,TÞ ð1Þ
The adsorbed amount is reported in wt %, which is defined as the massof hydrogen mads per mass of the system, which consists of the samplemass ms and the adsorbed hydrogen [Eq. (2)]:
Hydrogen uptake ðwt %Þ ¼ mads
ms þmadsð2Þ
TEM measurements : TEM measurements were carried out under lowdose conditions where possible.[20] The diffraction pattern of unloadedCOF-102 was acquired on a Philips CM20 microscope equipped with aLaB6 gun, operated at 200 kV at liquid-nitrogen temperature with aGatan cooling holder. Bright-field TEM and HRTEM images were ac-quired on a Phillips CM30 microscope equipped with a Schottky field-emission source, operated at 300 kV. HAADF-STEM images and theelectron tomography series were acquired on an aberration-correctedFEI Titan “cubed” microscope operated at 300 kV and with a Fischionesingle-tilt tomography holder for the tomography experiments. For HRimaging, the convergence semiangle a of the electron probe was set to21.5 mrad, and the acceptance semiangle b to 50 mrad. For the tomogra-phy experiments, the convergence semiangle a of the electron probe wasset to 8 mrad, keeping the acceptance semiangle b at 50 mrad. Imageswere acquired from +708 to �708 tilt angle by using a 18 tilt angle incre-ment. Tomographic reconstruction was performed by using a SIRT algo-rithm in the FEI Inspect 3D software package. 3D visualization was per-formed with the AMIRA 5.0 software. The size distribution of the Pdnanoparticles was determined by measurement of the projected area(nm2) in HRHAADF-STEM images, and calculation of the particle di-ameters assuming spherical particle morphology.
Acknowledgements
S.B.K. is grateful for a stipend from the Alexander von Humboldt Foun-dation. Partial funding by the German Research Foundation DFG withinthe priority program SSP 1362 is gratefully acknowledged. H.O. was sup-ported for this research through a stipend from the International MaxPlanck Research School for Advanced Materials (IMPRS-AM). We ac-knowledge support from the European Union under the Framework 7program under a contract from an Integrated Infrastructure Initiative(Reference 262348 ESMI). G.V.T. acknowledges the ERC grant COUN-TATOMS. S.T. gratefully acknowledges financial support from the Fundfor Scientific Research Flanders (FWO). The microscope used in thisstudy was partially financed by the Hercules Foundation.
[1] a) A. P. C�t�, A. I. Benin, N. W. Ockwig, M. O�Keeffe, A. J. Matzg-er, O. M. Yaghi, Science 2005, 310, 1166 – 1170; b) H. M. El-Kaderi,O. M. Yaghi, J. R. Hunt, J. L. Mendoza-Cortes, A. P. Cote, R. E.Taylor, M. O�Keeffe, Science 2007, 316, 268 –272.
[2] a) X. Ding, J. Guo, X. Feng, Y. Honsho, J. Guo, S. Seki, P. Maitarad,A. Saeki, S. Nagase, D. Jiang, Angew. Chem. 2011, 123, 1325 –1329;Angew. Chem. Int. Ed. 2011, 50, 1289 –1293; b) C. J. Doonan, D. J.Tranchemontagne, T. G. Glover, J. R. Hunt, O. M. Yaghi, Nat.Chem. 2010, 2, 235 – 238; c) S. S. Han, H. Furukawa, O. M. Yaghi,W. A. Goddard, J. Am. Chem. Soc. 2008, 130, 11580 –11581; d) H.Furukawa, O. M. Yaghi, J. Am. Chem. Soc. 2009, 131, 8875 –8883;
Chem. Eur. J. 2012, 18, 10848 – 10856 � 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 10855
FULL PAPERMetal@Covalent Organic Frameworks
e) S. Wan, J. Guo, J. Kim, H. Ihee, D. L. Jiang, Angew. Chem. 2008,120, 8958 –8962; Angew. Chem. Int. Ed. 2008, 47, 8826 – 8830; f) S.Wan, F. G�ndara, A. Asano, H. Furukawa, A. Saeki, S. K. Dey, L.Liao, M. W. Ambrogio, Y. Y. Botros, X. Duan, S. Seki, J. F. Stoddart,O. M. Yaghi, Chem. Mater. 2011, 23, 4094 – 4097; g) S. Y. Ding, J.Gao, Q. Wang, Y. Zhang, W. G. Song, C. Y. Su, W. Wang, J. Am.Chem. Soc. 2011, 133, 19816 –19822.
[3] a) L. M. Bronstein, Top. Curr. Chem. 2003, 226, 55– 89; b) M. Meili-khov, K. Yusenko, D. Esken, S. Turner, G. Van Tendeloo, R. A.Fischer, Eur. J. Inorg. Chem. 2010, 3701 – 3714; c) L. Vradman, L. Ti-telman, M. Herskowitz, Microporous Mesoporous Mater. 2006, 93,313 – 317; d) C. M. Yang, H. S. Sheu, K. J. Chao, Adv. Funct. Mater.2002, 12, 143 – 148; e) C. Yang, P. Liu, Y. Ho, C. Chiu, K. Chao,Chem. Mater. 2003, 15, 275 –280; f) M. Vaarkamp, B. L. Mojet, M. J.Kappers, J. T. Miller, D. C. Koningsbergert, J. Phys. Chem. 1995, 99,16067 – 16075.
[4] a) C. K. Tsung, W. Hong, Q. Shi, X. Kou, M. H. Yeung, J. Wang,G. D. Stucky, Adv. Funct. Mater. 2006, 16, 2225 – 2230; b) D. Esken,S. Turner, O. I. Lebedev, G. V. Tendeloo, R. A. Fischer, Chem.Mater. 2010, 22, 6393 –6401.
[5] L. Schlapbach, A. Zuttel, Nature 2001, 414, 353 – 358.[6] a) D. P. Cao, J. H. Lan, W. C. Wang, B. Smit, Angew. Chem. 2009,
121, 4824 –4827; Angew. Chem. Int. Ed. 2009, 48, 4730 –4733; b) Y. J.Choi, J. W. Lee, J. H. Choi, J. K. Kang, Appl. Phys. Lett. 2008, 92,173102; c) E. Klontzas, E. Tylianakis, G. E. Froudakis, J. Phys.Chem. C 2008, 112, 9095 – 9098; d) X. L. Zou, G. Zhou, W. H. Duan,K. Choi, J. Ihm, J. Phys. Chem. C 2010, 114, 13402 – 13407; e) J. Lan,D. Cao, W. Wang, Langmuir 2010, 26, 220 – 226.
[7] a) H. Kobayashi, M. Yarnauchi, H. Kitagawa, Y. Kubota, K. Kato,M. Takata, J. Am. Chem. Soc. 2008, 130, 1828; b) T.-H. Phan, R. E.Schaak, Chem. Commun. 2009, 3026 – 3028; c) S. Kishore, J. A.Nelson, J. H. Adair, P. C. Eklund, J. Alloys Compd. 2005, 389, 234 –242; d) R. J. Wolf, M. W. Lee, J. R. Ray, Phys. Rev. Lett. 1994, 73,557; e) J. Saha, A. Dandapat, G. De, J. Mater. Chem. 2011, 21,11482 – 11485.
[8] S. B. Kalidindi, K. Yusenko, R. A. Fischer, Chem. Commun. 2011,47, 8506 –8508.
[9] a) T. Ishida, M. Nagaoka, T. Akita, M. Haruta, Chem. Eur. J. 2008,14, 8456 – 8460; b) D. Esken, X. Zhang, O. I. Lebedev, F. Schroder,R. A. Fischer, J. Mater. Chem. 2009, 19, 1314 –1319; c) S. Turner,O. I. Lebedev, F. Schroder, D. Esken, R. A. Fischer, G. Van Tende-loo, Chem. Mater. 2008, 20, 5622 – 5627.
[10] M. Hirscher, Angew. Chem. 2011, 123, 605 –606; Angew. Chem. Int.Ed. 2011, 50, 581 –582.
[11] a) B. Schmitz, U. Mller, N. Trukhan, M. Schubert, G. F�rey, M.Hirscher, ChemPhysChem 2008, 9, 2181 – 2184; b) M. Hirscher, B.Panella, B. Schmitz, Microporous Mesoporous Mater. 2010, 129,335 – 339.
[12] C. O. Are�n, S. Chavan, C. P. Cabello, E. Garrone, G. T. Palomino,ChemPhysChem 2010, 11, 3237 – 3242.
[13] a) C. Zlotea, R. Campesi, F. Cuevas, E. Leroy, P. Dibandjo, C. Volk-ringer, T. Loiseau, G. Ferey, M. Latroche, J. Am. Chem. Soc. 2010,132, 2991 – 2997; b) Y. E. Cheon, M. P. Suh, Angew. Chem. 2009, 121,2943 – 2947; Angew. Chem. Int. Ed. 2009, 48, 2899 –2903.
[14] Y. W. Li, R. T. Yang, J. Am. Chem. Soc. 2006, 128, 8136 –8137.[15] a) R. Campesi, F. Cuevas, M. Latroche, M. Hirscher, Phys. Chem.
Chem. Phys. 2010, 12, 10457 – 10459; b) S. M. Luzan, A. V. Talyzin,Microporous Mesoporous Mater. 2010, 135, 201 – 205; c) M. Hirscher,Microporous Mesoporous Mater. 2010, 135, 209 – 210; d) Y. W. Li,F. H. Yang, R. T. Yang, J. Phys. Chem. C 2008, 112, 3155 – 3156;e) N. R. Stuckert, L. F. Wang, R. T. Yang, Langmuir 2010, 26,11963 – 11971; f) L. Wang, N. R. Stuckert, H. Chen, R. T. Yang, J.Phys. Chem. C 2011, 115, 4793 –4799.
[16] a) V. A. Vons, H. Leegwater, W. J. Legerstee, S. W. H. Eijt, A.Schmidt-Ott, Int. J. Hydrogen Energy 2010, 35, 5479 –5489; b) M.Yamauchi, H. Kobayashi, H. Kitagawa, ChemPhysChem 2009, 10,2566 – 2576; c) M. Yamauchi, R. Ikeda, H. Kitagawa, M. Takata, J.Phys. Chem. C 2008, 112, 3294 –3299; d) D. G. Narehood, S. Kishore,H. Goto, J. H. Adair, J. A. Nelson, H. R. Guti�rrez, P. C. Eklund,Int. J. Hydrogen Energy 2009, 34, 952 – 960; e) B. Ingham, M. F.Toney, S. C. Hendy, T. Cox, D. D. Fong, J. A. Eastman, P. H. Fuoss,K. J. Stevens, A. Lassesson, S. A. Brown, M. P. Ryan, Phys. Rev. B2008, 78, 245408; f) M. Yamauchi, H. Kitagawa, Synth. Met. 2005,153, 353 –356; g) J. A. Eastman, L. J. Thompson, B. J. Kestel, Phys.Rev. B 1993, 48, 84–92.
[17] F. Schrçder, D. Esken, M. Cokoja, M. W. E. van den Berg, O. I. Leb-edev, G. Van Tendeloo, B. Walaszek, G. Buntkovsky, H.-H. Limbach,B. Chaudret, R. A. Fischer, J. Am. Chem. Soc. 2008, 130, 6119 –6130.
[18] a) Y. G. Kim, S. Bialy, G. T. Stauf, R. W. Miller, J. T. Spencer, P. A.Dowben, S. Datta, J. Micromech. Microeng. 1991, 1, 42– 45; b) G. T.Stauf, P. A. Dowben, K. Emrich, S. Barfuss, W. Hirschwald, N. M.Boag, J. Phys. Chem. 1989, 93, 749 – 753; c) R. R. Thomas, J. M.Park, J. Electrochem. Soc. 1989, 136, 1661 –1666; d) Q.-H. Wu, M.Gunia, T. Strunskus, G. Witte, M. Muhler, C. Wçll, Chem. VaporDepos. 2005, 11, 355 – 361; e) J. C. Hierso, R. Feurer, P. Kalck,Chem. Mater. 2000, 12, 390 –399.
[19] a) D. Stacchiola, L. Burkholder, W. T. Tysoe, Surf. Sci. 2003, 542,129 – 141; b) A. M. Doyle, S. K. Shaikhutdinov, H. J. Freund, J. Catal.2004, 223, 444 – 453; c) T. P. Beebe, Jr., J. T. Yates, Jr., J. Am. Chem.Soc. 1986, 108, 663 – 671; d) A. M. Doyle, S. K. Shaikhutdinov, S. D.Jackson, H. J. Freund, Angew. Chem. 2003, 115, 5398 – 5401; Angew.Chem. Int. Ed. 2003, 42, 5240 – 5243; e) M. Neurock, R. A. van Sant-en, J. Phys. Chem. B 2000, 104, 11127 –11145.
[20] C. Wiktor, S. Turner, D. Zacher, R. A. Fischer, G. Van Tendeloo, Mi-croporous Mesoporous Mater. 2012, 162, 131 –135.
Received: April 19, 2012Published online: August 8, 2012
www.chemeurj.org � 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2012, 18, 10848 – 1085610856
M. Hirscher, R. A. Fischer et al.