Influence of Co-adsorbates on CO2 induced phase transition in functionalized pillared-layered metal-organic frameworks

Title Influence of Co-adsorbates on CO2 induced phase transition in functionalized pillared-layered metal-organic frameworks Author(s) Schneemann, Andreas; Takahashi, Yukiko; Rudolf, Robin; Noro, Shin-ichiro; Fischer, Roland A. Citation Journal of Materials Chemistry A, 4(33): 12963-12972 Issue Date 2016-09-07 Doc URL http://hdl.handle.net/2115/63199 Rights(URL) http://creativecommons.org/licenses/by/3.0/ Type article File Information C6TA03266D.pdf


Introduction
Among the porous materials, metal-organic frameworks are a class that has gathered a lot of attention in recent years, owing to their outstanding properties. 1 They are built up by the combination of metal clusters and multitopic organic linkers i.e. aromatic carboxylates or N-donor ligands. 2 The reasons for the high interest in this class of materials arise from their ultrahigh surface areas, their high degree of tunability and their structural exibility. By combining appropriate linker molecules and metal clusters, record holding materials have been prepared with unprecedented pore volumes and surface areas. 3 The aspect of tunability of material properties leads to new tailor-made materials and ranges from simple linker functionalization to more complex functionalization and tuning principles such as defect engineering, 4 postsynthetic exchange/ incorporation of metals/linkers, 5 the precise assembly of different building blocks per pore, 6 or postsynthetic metalation of metal sites by atomic layer deposition. 7 Structural exibility describes the possibility of certain frameworks to undergo a reversible phase transition between at least two distinct states. 8 The transition between these two states can be initiated by different stimuli, including guest ad/desorption, 9 temperature, 10 light, 11 or mechanical pressure. 12 Frameworks that can undergo this kind of transition are denoted as exible MOFs, 3 rd generation MOFs or so porous crystals in the literature. 8b, 13 The ability to switch depending on an external stimulus between different states makes exible MOFs amenable for a handful of potential applications, including chemical sensing, 14 as dampers, 15 drug release 16 or gas separation and storage. 17 In our study we are focusing on several issues regarding gas separation problems in exible MOFs, using as an example functionalized derivatives of the well-known compound [Zn 2 (bdc) 2 (dabco)] n (bdc ¼ 1,4-benzenedicarboxylate, dabco ¼ 1,4-diazabicyclo[2.2.2]octane). 18 These so-called fu-MOFs feature alkoxy sidechains (fu) pinned to the bdc linker to yield fu-bdc (2,5-functionalized-1,4-benzenedicarboxylate). These fu-MOFs are built up from Zn 2 paddle-wheels which are fourfold coordinated by fu-bdc linkers, thus generating a [Zn 2 (fubdc) 2 ] n 2D sheet (Fig. 1). The paddle wheels are coordinated in the axial position by dabco pillars, connecting in this fashion the adjacent layers and forming a 3D structure. In the parent material with bdc as the linker only small structural changes are observed (i.e. bending of the linker molecules) when the assynthesized DMF containing material is desolvated. In contrast, for the functionalized derivatives a large pore contraction is found aer desolvation. 19 This contraction is mostly initiated by interactions of the solvent-like side chains. The solvated state of the material is also referred to as the large pore (lp) form and the contracted state aer solvent removal is termed as the narrow pore (np) form. Interestingly, strong sorption selectivities are initiated by the right choice of fu e.g. for C 2 H 2 over C 2 H 4 when the material is present in the np form. 20 Sorption selectivity can also be observed in related materials bearing the same fu-bdc linkers but with a different topology. 21 Notably, during CO 2 adsorption a transition from the np to the lp phase is detected, for the functionalized pillared-layered MOFs. 22 Fig. 1(e) shows the shape of a CO 2 adsorption isotherm typically observed for these kinds of fu-MOFs (pink dotted line). At low pressures the material takes up a small portion of CO 2 . When a certain threshold pressure is surpassed, the pore opens which is indicated by a drastic increase of the CO 2 uptake in the material. The threshold pressure of the phase transition can be modulated by the choice of the fu-bdc linker and can be further tuned by the preparation of mixed linker MOFs. 23 Only very little N 2 uptake is monitored for these materials and the common N 2 isotherm shape is also depicted in Fig. 1(e) (green dotted line). For many exible and guest responsive MOFs this kind of situation is observed for single component adsorption: pore opening at threshold pressure p for gas A and no pore opening for gas B. Aer the pore opening by A, a steep increase of the adsorption of A is observed, while the uptake of B remains low over the whole pressure range of the adsorption isotherm. From the data obtained from such single component adsorption experiments, it is usually deduced, that the material can separate a gas mixture consisting of the two components A and B. However, the entire pore surface of the exible MOF is changed aer gas A induced the np / lp phase transition in the material and hence two different scenarios are imaginable ( Fig. 1(f)): (a) Gas A opens the pore and is adsorbed inside the framework and gas B remains outside. (b) Gas A opens the pore and is adsorbed inside the framework and gas B is co-adsorbed with gas A in the lp.
The rst scenario would be necessary for a good separation of two compounds. The other scenario, i.e. the transport of species B into the framework opened by gas A, is less desirable and would reduce the sorption selectivity.
With our study presented herein, we wish to address these issues mentioned above. In particular, we want to pursue the question of what happens if the material is exposed to two different gases that both initiate a phase transition (i.e. np / lp). [Zn 2 (bdc) 2 (dabco)] n (1) as the non-functionalized and essentially rigid reference material and the two functionalized and responsive fu-MOFs [Zn 2 (BME-bdc) 2 (dabco)] n (H 2 BME-bdc ¼ 2,5-bis(2-methoxyethoxy)-1,4-benzenedicarboxylic acid) (2) and [Zn 2 (DiP-bdc) 2 (dabco)] n (H 2 DiP-bdc ¼ 2,5-diisopropoxy-1,4benzenedicarboxylic acid) (3) were employed (Fig. 2). A series of co-adsorption experiments was conducted using CO 2 (gas A) and a selection of gases B such as N 2 , CH 4 , C 2 H 6 and C 3 H 8 which feature different pore opening properties. The results gathered here should be transferable to other exible MOF materials showing promising adsorption isotherms for gas separation applications.

Experimental section
All materials were bought from commercial suppliers, such as Sigma Aldrich, TCI and Alfa Aesar and used without further purication unless otherwise noted.

Linker synthesis
The functionalized linkers H 2 BME-bdc and H 2 DiP-bdc used for the preparation of fu-MOFs [Zn 2 (fu-bdc) 2 (dabco)] n were prepared via Williamson ether synthesis from dimethyl 2,5dihydroxy-1,4-benzenedicarboxylate. A detailed description of the synthetic route to obtain the organic precursors and the linker molecules can be found elsewhere. 19b,20

MOF synthesis
Compounds 1-3 were prepared under solvothermal reaction conditions only slightly modied from already reported synthesis methods. 18,19b [Zn 2 (bdc) 2 (dabco)] n (1), [Zn 2 (BME-bdc) 2 (dabco)] n (2), [Zn 2 (DiP-bdc) 2 (dabco)] n (3) Zn(NO 3 ) 2 $6H 2 O (1.33 g; 4.5 mmol), H 2 bdc/H 2 fu-bdc (4.5 mmol) and dabco (270 mg; 2.4 mmol) were put in a beaker and 100 ml DMF were added. The mixture was sonicated until the reagents were dissolved. Aerwards, the synthesis solution was divided into three equal portions and the portions were put into three individual 50 ml screw top jar. The jars lled with the solutions were then placed in a preheated oven and le for 24 hours at 120 C. The oven temperature was then reduced to room temperature. A white, microcrystalline precipitate formed in all three cases (off-white in the case of materials 2 and 3). The solvent from the synthesis was replaced by fresh DMF and a stirring bar was added to each vessel. The mixture was vigorously stirred for 30 minutes, in order to assure a good solvent exchange and was aerwards le to settle for 24 hours. The solvent was removed by decantation and exchanged with CHCl 3 and again stirred for 30 minutes and le to settle for 24 hours. This process was repeated a second time, also with CHCl 3 . The three identical batches were combined. Now, the material was ltered and washed three times with 15-20 ml CHCl 3 , transferred to a Schlenk tube, and dried in vacuo (oil pump) for 24 hours at elevated temperature (120 C). The purity of the isolated compounds was controlled by measuring 1 H-NMR spectra (aer sample digestion) and PXRD patterns (see ESI †). Yield (based on H 2 bdc/H 2 fu-bdc): 1 -79% (1.03 g, 1.8 mmol), 2 -56% (1.09 g, 1.26 mmol), and 3 -63% (1.15 g, 1.42 mmol).

Adsorption measurements
The single component adsorption isotherms and co-adsorption measurements were performed using a Belsorp VC apparatus (MicrotracBEL Corp.) coupled with an Agilent 490 Micro gas chromatographic system equipped with a thermal conductivity detector. For each measurement approximately 1 g of sample was used. The MOF was placed in the sample cell, which was sealed with a metal O-ring and connected to the instrument. The sample cell is located within the isothermal part of the Belsorp VC and the temperature within this area can be precisely controlled. The lowest stable temperature was at 278 K and hence all measurements were performed at this temperature. Prior to each measurement, the samples were evacuated at elevated temperatures (120 C) overnight (rst measurement on sample) or at least for three hours (each additional measurement). The single component adsorption isotherms were measured volumetrically. For the co-adsorption measurements, a pre-dened gas mixture with distinct partial pressures for each component was introduced into the manifold of the instrument. In the manifold the two gases are mixed for 2200 seconds and dosed onto the sample. The gas phase was then allowed to equilibrate until the pressure variation was minimized. Aerwards, the total uptake was determined volumetrically. Subsequently a small amount of the gas phase remaining in the manifold was removed and the composition was analyzed by GC (gas chromatography). From the composition of the gas phase not adsorbed on the sample, the exact adsorbed portions of each component of the mixture on the sample were calculated. The process of sampling the gas above the sample followed by equilibration (until pressure variation is minimal) was repeated ve times, thus measuring ve data points with relatively similar total gas pressure and uptake.

Calculation of selectivity coefficients
The co-adsorption measurements provided experimental values for the molar fraction of both components remaining in the gas phase (y i ) above the sample and adsorbed on the sample (x i ) and hence the selectivity coefficients were calculated by the equation: Illustration of the organic linker molecules used for the synthesis of the MOF materials 1-3.

General methods
Additional data including PXRDs of the as-synthesized and activated MOFs, IR spectra, TG and NMR spectra of compounds 1-3 are given in the ESI. †

Results and discussion
In the following paragraph, all co-adsorption experiments, measurements and data are represented in a similar manner. Three different data sets were acquired in each experiment. The total uptake was determined volumetrically and is always depicted as green diamonds in the gures. The CO 2 uptake (determined by GC as described in the experimental part) is indicated by red diamonds. The co-adsorbate is depicted by a differently colored diamond (N 2 , CH 4 , C 2 H 6 and C 3 H 8 are shown in blue, black, turquoise and orange, respectively). Five measurements were undertaken during each co-adsorption experiment with the selected gas mixtures. The black arrows point from the rst to the h measurement point. In addition, the relevant single component adsorption isotherms are plotted in the same diagrams for better understanding and comparison of the effect of the simultaneous presence of the two components.
[Zn 2 (bdc) 2 (dabco)] n (1) The rst material under study, in order to test the experimental set-up and measurement conditions, was the parent MOF [Zn 2 (bdc) 2 (dabco)] n and the respective single component adsorption isotherms for N 2 , CO 2 , CH 4 , C 2 H 6 C 3 H 8 at 278 K are shown in Fig. 3(a). The measurements were undertaken in the range from 0 to 3200 kPa (C 2 H 6 and C 3 H 8 adsorptions were measured in a smaller range, due to their lower condensation pressures). It is directly visible that the adsorbed amount of CO 2 on 1 is much higher than for the other gases. The maximum uptakes are for CO 2 , 330 cm 3 (STP) per g (1624 kPa); for N 2 , 107 cm 3 (STP) per g (3100 kPa); for CH 4 , 141 cm 3 (STP) per g (3050 kPa); for C 2 H 6 , 127 cm 3 (STP) per g (1047 kPa) and for C 3 H 8 , 95 cm 3 (STP) per g (189 kPa). The shapes of the isotherms do not indicate gate opening behaviour, since for all adsorbed gases no steps are visible in the adsorption isotherm. In fact, this is in good accordance with the expectations, since the parent material 1 only undergoes small structural changes upon guest loading. Fig. 1 highlights the slight structural exibility upon ad/desorption of the solvent DMF inside the pores of 1.
Material 1 was chosen as a reference for comparison and will be also used to explain and evaluate the co-adsorption data. Fig. 3(b) shows the co-adsorption of a mixture of N 2 and CO 2 . The applied mixture has a total gas pressure of 925 kPa. The partial pressures of the two components amount to 513 and 421 kPa of N 2 and CO 2 , respectively. At the applied pressures barely any N 2 from the mixture is adsorbed in the pore (below detection limit). Moreover, 274 cm 3 (STP) per g CO 2 are taken up, which coincides nicely with the single component adsorption isotherm at the same pressure. During the CO 2 /CH 4 measurement, the measured pressure points are not overlaying with the single component gas adsorption isotherms as nicely as for the CO 2 /N 2 mixture. During the measurement a total gas pressure of 1324 kPa was applied to the sample, containing 640 kPa CO 2 and 713 kPa CH 4 . A total amount of 169.9 cm 3 (STP) per g of CO 2 is adsorbed which is distinctively less than for the single component adsorption isotherm at the same pressure. From the mixture a total of 27.7 cm 3 (STP) per g CH 4 are adsorbed.
[Zn 2 (BME-bdc) 2 (dabco)] n (2) Among the exible fu-MOF materials that have been introduced by Fischer and co-workers over the recent years, [Zn 2 (BME-bdc) 2 (dabco)] n has been the most studied material. 10c,19,22 The material undergoes a np / lp phase transition when the desolvated np phase is exposed to suitable guest molecules (Fig. 1). In Fig. 4 the single component gas adsorption isotherms of this material measured at 278 K for CO 2 and Fig. 3 Depiction of the excess single component and co-adsorption measurements of material 1. N 2 , CO 2 , CH 4 , C 2 H 6 and C 3 H 8 single component isotherms are blue, red, black, turquoise and orange, respectively. CO 2 , N 2 and CH 4 fractions of the co-adsorption experiments are shown in red, blue and black. Total adsorption is shown in green.
N 2 are displayed. From the isotherms it is directly observable, that the material does not take up any N 2 at 278 K. However, the CO 2 adsorption isotherm shows the anticipated phase transition known for this material featuring a considerably wide hysteresis loop. 22 In the low pressure region of the isotherm, before the np / lp transition occurs, already a decent amount of CO 2 is adsorbed (10 cm 3 (STP) per g). Aer surpassing the threshold pressure at 1700 kPa, a marked increase of the adsorbed volume of CO 2 can be observed. The overall uptake reaches a value of 83 cm 3 (STP) per g. The desorption of CO 2 starts at pressures below 1367 kPa and the initial plateau observed during the adsorption is reached again.
In order to characterize this gate opening and how it is affected by the presence of N 2 , co-adsorption experiments were performed using CO 2 /N 2 mixtures containing ratios of 986 : 1051 kPa ( Fig. 4(a)) and 1469 : 1068 kPa ( Fig. 3(b)). The total pressure (1975 and 2468 kPa, respectively) is higher than the characteristic single-component CO 2 phase transition pressure of material 2. However, the partial CO 2 pressure is below the phase transition pressure. A total CO 2 uptake of 15 cm 3 (STP) per g is measured at the rst equilibration step (Fig. 4(a)) (986 kPa partial CO 2 pressure), slightly more than that of the CO 2 single component adsorption isotherm. It should be noted, that the N 2 uptake during co-adsorption is below the detection limit and thus might have an inuence on the correct determination of the CO 2 uptake. During the 5 equilibration steps, the CO 2 uptake slightly increases, most likely due to the longer exposure of the material to CO 2 (higher equilibration time). For the second experiment a substantially higher CO 2 content is present in the CO 2 /N 2 mixture, however the partial CO 2 pressure is still below the threshold pressure inducing the phase transition. The total CO 2 uptake is only slightly higher compared to the rst experiment (18 cm 3 (STP) per g). As for the rst experiment, the CO 2 uptake increases over the course of the ve set equilibrium points. The N 2 uptake again is below the detection limit. Unfortunately, it was not possible to increase the pressure of CO 2 to surpass the threshold pressure of the phase transition due to limitations of the applicable pressures of the co-adsorption instrument. Nonetheless some valuable conclusions can be drawn. Material 2 will not undergo a np / lp phase transition for CO 2 /N 2 mixtures unless the characteristic CO 2 threshold is surpassed, i.e. the pore cannot be pushed open by a great excess of N 2 in the presence of CO 2 , even when the applied partial CO 2 pressure in the mixture is only slightly lower than the threshold pressure.
[Zn 2 (DiP-bdc) 2 (dabco)] n (3) Since material 2 showed some restrictions, due to the relatively high phase transition pressure under the used Fig. 4 Depiction of the excess single component and co-adsorption measurements of material 2. N 2 and CO 2 single component isotherms are blue and red, respectively. CO 2 and N 2 , fractions of the co-adsorption experiments are shown in red and blue, respectively. Total adsorption is shown in green. p(CO 2 ) and p(C 2 H 6 ) < p pt p(CO 2 ) < p pt ; p(C 2 H 6 ) > p pt p(CO 2 ) > p pt ; p(C 2 H 6 ) < p pt CO 2 /C 2 H 6 Pore opening before reaching p pt ; cooperative induction of phase transition (enhanced uptake of both component) Pore opening; C 2 H 6 transports CO 2 into pore (enhanced CO 2 uptake) Pore opening; CO 2 transports C 2 H 6 into pore (enhanced C 2 H 6 uptake) a p pt is the phase transition pressure of the respective single component adsorption isotherm.
conditions, we chose the differently functionalized material [Zn 2 (DiP-bdc) 2 (dabco)] n (3) because it reveals the lowest np / lp CO 2 phase transition pressure among our library of fu-MOFs, 19b making this material much more amenable for examining the behaviour when exposed to gas mixtures, in particular when using the specic set up of this study. Fig. 5(a) and (b) shows the single component adsorption isotherms for CO 2 , N 2 , CH 4 , C 2 H 6 and C 3 H 8 recorded at 278 K. Notably, a stepped isotherm with a hysteresis loop, indicating the np / lp phase transition of the activated material is observed for all probe gases except N 2 and CH 4 . The uptake of these two gases is very low. Before the np / lp transition occurs, CO 2 , C 2 H 6 , and C 3 H 8 uptakes of 6.5, 3.1 and 2.6 cm 3 (STP) per g were recorded. Aer exceeding the threshold pressures of the np / lp transition at 306 kPa (CO 2 ), 115 kPa (C 2 H 6 ) and 11 kPa (C 3 H 8 ) the uptakes were drastically increased, peaking at 92.1 (CO 2 ), 53.2 (C 2 H 6 ) and 47.2 (C 3 H 8 ) cm 3 (STP) per g. For N 2 and CH 4 only comparatively low values of 2.4 and 4.6 cm 3 (STP) per g were found. Fig. 5(c) and (d) shows the co-adsorption experiments on 3 for CO 2 /N 2 mixtures. In one case the partial CO 2 pressure is kept below the threshold pressure at 290 kPa (Fig. 5(c)) and in the other case the partial CO 2 pressure is above the np / lp transition pressure at 621 kPa (Fig. 5(d)). The total pressure of the gas mixture was above the threshold pressure in both cases. The rst experiment shows that an excess of N 2 (903 kPa) cannot help to induce the np / lp transition (Fig. 5(c)), as was also observed for material 2. When this mixture is applied the framework takes up Fig. 5 Depiction of the excess single component and co-adsorption measurements of material 3. N 2 , CO 2 , CH 4 , C 2 H 6 and C 3 H 8 single component isotherms are blue, red, black, turquoise and orange, respectively. CO 2 , N 2 , CH 4 and C 2 H 6 fractions of the co-adsorption experiments are shown in red, blue, black and turquoise, respectively. The total adsorption is shown in green.
8.5 cm 3 (STP) per g CO 2 and barely any N 2 (amount below the detection limit). Both values are in line with the values found for the single component isotherms. For the second experiment the threshold pressure of the np / lp transition was surpassed by the applied partial CO 2 pressure (621 kPa) (Fig. 5(d)). In comparison to the previous experiment, the CO 2 uptake is signicantly increased and completely in line with the uptake of 3, determined from the single component adsorption isotherm. Interestingly, the N 2 uptake is not increased (compared to the single component isotherm), attesting that the material does not take up any N 2 , even when being present in the lp. Thus, this material might be feasible for CO 2 /N 2 separations. Aer these experiments we moved on to a different gas mixture, to assess the ability of 3 to separate CO 2 from a more polarizable co-adsorbate as compared to N 2 and chose CH 4 . From the single component adsorption we know that CH 4 does not induce a phase transition and the overall adsorbed amount is only slightly higher than for N 2 ( Fig. 5(a)).
Again two experiments were performed. In the rst run, the partial CO 2 pressure is slightly below the np / lp transition pressure and during the second experiment the partial CO 2 pressure was above the np / lp transition pressure. The results are depicted in Fig. 5(e) and (f). From the rst experiment (Fig. 5(e)) it can be seen, that the adsorbed amount of CO 2 during the co-adsorption experiments is much higher than for the single component adsorption at the same pressure (23.47 cm 3 (STP) per g compared to 6.5 cm 3 (STP) per g), while the CH 4 uptake is only slightly increased (6.5 cm 3 (STP) per g compared to 4.6 cm 3 (STP) per g). We propose a synergy of the CH 4 and CO 2 co-adsorption that allows for a lower partial CO 2 pressure to trigger the np / lp transition. In the second experiment ( Fig. 5(f)) it can be nicely seen that the CO 2 uptake coincides with the CO 2 uptake of the single component experiment at this pressure aer the np / lp pore transition. Most interestingly, also the amount of CH 4 adsorbed in the framework is now drastically increased when compared to the rst CO 2 /CH 4 co-adsorption experiment. We assume that aer the CO 2 induced opening CH 4 is able to co-adsorb in the lp phase. This behaviour is substantially different from the ndings obtained from the CO 2 /N 2 co-adsorption experiments and attributed to the much higher polarizability of the CH 4 molecule compared to N 2 (see ESI † for information on properties of used adsorptives).
In order to see how material 3 behaves when two guests are adsorbed that both initiate a np / lp transition CO 2 /C 2 H 6 and CO 2 /C 3 H 8 co-adsorption experiments were conducted (Fig. 5(g)-(i) (C 2 H 6 ) and ESI † (C 3 H 8 )). Three different measurement conditions were chosen for CO 2 /C 2 H 6 mixtures. In the rst case, the partial pressures of both gases were below the respective threshold pressures of the np / lp transition (306 and 115 kPa for CO 2 and C 2 H 6 , respectively). In the second and third case the partial pressure of one of the components was below the corresponding phase transition pressure and the other one was set above. The applied mixtures contained CO 2 /C 2 H 6 ratios of 250 : 89 kPa (Fig. 5(g)), 249 : 271 kPa (Fig. 5(h)) and 546 : 91 kPa (Fig. 5(i)), respectively. In the rst measurement (Fig. 5(g)) a marked increase of the adsorption was found for both components when compared to the respective single component isotherms. Values of 24.4 cm 3 (STP) per g adsorbed CO 2 and 11.5 cm 3 (STP) per g of C 2 H 6 are measured in contrast to 6.1 cm 3 (STP) per g and 2.8 cm 3 (STP) per g (single component adsorption). This suggests that strong cooperative effects between the two gases allow the gate opening transition to occur at lower partial pressures of each component. In the experiments depicted in Fig. 5(h) and (i) in both cases very similar behavior is observed. In Fig. 5(h), p(CO 2 ) and p(C 2 H 6 ) amount to 249 kPa and 271 kPa, respectively. The measurement reveals, that the CO 2 uptake (28.7 cm 3 (STP) per g) is signicantly increased in comparison to the single component adsorption (6.1 cm 3 (STP) per g). We conclude that the CO 2 is co-adsorbing with the C 2 H 6 that opened the pore. A similar explanation can be drawn for the experiment in Fig. 5(i), where p(CO 2 ) is above (545.7 kPa) and p(C 2 H 6 ) is below (90.7 kPa) the np / lp transition pressure. The CO 2 adsorption at this point nicely coincides with the single component isotherm, while the C 2 H 6 adsorption at 90.7 kPa is notably increased from 2.9 cm 3 (STP) per g to 10.4 cm 3 (STP) per g, which is in line with the uptake found in the rst co-adsorption experiment of these two species when both partial pressures are below the np / lp transition pressure ( Fig. 5(g)).
For the CO 2 /C 3 H 8 mixtures similar results were found as for CO 2 /C 2 H 6 and a detailed description can be found in the ESI (Fig. S9 †). A summary of the results found for material 3 is shown in Table 1.

Calculation of selectivities
For all co-adsorption measurements the selectivity coefficients were also calculated based on the molar fractions adsorbed on the material and le in the gas phase during each co-adsorption measurement. The calculated values for the 1 st equilibration point for material 3 are presented in Table 2 (values for the other materials and other equilibration points are listed in the ESI †). First of all, it needs to be mentioned, that the CO 2 /N 2 measurements conducted on all three materials look very promising and a good separation of the two components is directly visible from the co-adsorption experiments shown before. However, the values determined during the N 2 adsorption are below the detection limit and a meaningful evaluation was not possible. However, in particular for material 3 some interesting conclusions can be drawn. In the case of the CO 2 / CH 4 co-adsorption experiments a decent selectivity towards CO 2 is calculated, for the case that the CO 2 pressure is below the threshold pressure of the phase transition (5.85). In this case CH 4 helps inducing a phase transition at lower partial pressures as expected. When the CO 2 pressure is increased and the material is present in the lp, CH 4 is co-adsorbing and the selectivity is drastically increased to 11.57 (the increase in CO 2 adsorption is higher than the increase in CH 4 adsorption induced by the observed cooperative effects). From the CO 2 / C 2 H 6 co-adsorption measurements only very low selectivities are determined. The values towards CO 2 are in the range of 0.75 to 1.02the material seems to slightly favour the adsorption of C 2 H 6 and a separation of these two gases over 3 does not seem feasible. The CO 2 /C 3 H 8 co-adsorption measurements show a clear trend towards a favoured adsorption of C 3 H 8 in the presence of CO 2 (see ESI Table S1 † for data). In particular at low overall pressures, when the C 3 H 8 partial pressure is below the threshold pressure of the phase transition determined by the single component isotherm, a good selectivity of C 3 H 8 over CO 2 (7.32 and 3.24) results. This result seems a bit odd, as the adsorbed amount of C 3 H 8 is very low compared to the adsorbed CO 2 . However, the molar ratios of C 3 H 8 /CO 2 in the gas phase are much lower than for the adsorbed state. Thus C 3 H 8 is relatively enriched over CO 2 by adsorption.

Conclusions
The results leave a lot of room for discussion and further experiments and theoretical modelling. Below we summarize our ndings (I-IV) regarding the adsorption of binary gas mixtures on fu-MOFs of the type [Zn 2 (fu-bdc) 2 (dabco)] n and suggest some conclusions that might hold true for other exible MOFs.
(I) The presence of N 2 is not affecting the CO 2 uptake in fu-MOFs. The results found for 2 and 3 show that the phase transition cannot be shied to lower pressures and N 2 does not assist in the pore expansion process.
(II) N 2 is not adsorbed in the pore aer the CO 2 threshold pressure is surpassed and the np / lp phase transition occurred.
(III) More polar guests, with stronger adsorption enthalpies at the adsorption sites and more favourable interactions with the pendant side chains can help inducing a preliminary CO 2 induced phase transition. Additionally, CH 4 can travel into the pore with the adsorbent and the CH 4 uptake is strikingly higher than during single component adsorption when only the np is present.
(IV) Adsorbates that show stepped isotherms (gate opening behaviour) can induce strong cooperative effects with CO 2 . In the presence of C 2 H 6 and C 3 H 8 the CO 2 adsorption is strongly increased below the CO 2 phase transition pressure.
We conclude that a separation of two gases that possess a two-step isotherm over a exible MOF is highly unlikely. The gas with the lower threshold pressure will open the pore and considerable amounts of the second gas will be adsorbed in the pore as well. Nevertheless, if the concentrations/pressures of both components are wisely chosen, our data suggest that a separation can be achieved under some specic conditions.
Finally it needs to be highlighted that the np / lp phase transition is a phenomenon governed by thermodynamics. The energy needed for the phase transition must be compensated by the adsorption enthalpy of the respective gases at lp. From our recent molecular dynamics simulations and NMR studies we know that the favoured adsorption sites for CO 2 within this class of fu-MOFs are located in the vicinity of the carboxylate groups connecting to the metal-ion. 23a Thus, the interactions of the guest with the side chains need to be taken into account in order to pass through and reach the adsorption site. Furthermore, the attractive side-chain/side-chain and side-chain/linker interactions favour the pore contraction and need to be compensated to initiate the phase transition. In case of coadsorption, both components (guests) contribute to the over-all energetics of the np / lp gate-opening phenomenon. It is thus quite conceivable that a cooperative gate opening is possible (as we observed) in case of setting the partial pressures of the two components below the individual phase transition pressures (CO 2 /C 2 H 6 and CO 2 /C 3 H 8 ). If a guest is chosen (N 2 ) exhibiting low uptake and adsorption enthalpies for the np (experimental evidence) and also (presumably) in the lp state of the material, it will neither assist in the pore opening nor will it signicantly coadsorb. In the case of CO 2 /CH 4 experiments, we assume that the adsorption enthalpy of CH 4 is increased in the lp as compared to the np state (increased accessibility of favourable adsorption sites, less hindrance by the side chains to reach adsorption sites). We are aiming to understand the effects reported in this study more quantitatively by focusing on a comprehensive thermodynamic analysis of the co-adsorption process in combination with molecular dynamics simulations in the future. Nevertheless, we suggest some exciting potential functions for exible MOFs from the current results. For example, it is imaginable that the one gas A of a mixture can work as a switching gas. The second gas B can undergo a reaction or cause a signal inside the framework, only if a certain partial pressure of gas A is set. The obvious necessity of ne-tuning the chosen exible framework could be done by choosing a suitable linker-functionalization as it is offered by our library of fu-MOFs including mixed-component (or multivariate) fu-MOFs (with different metal ions, linkers and functionalized side chains). Such systems may be tailored to operate at different temperature and pressure regimes and feature guest-specic and adjustable gate-opening properties.