Determination of the Apparent Crystal Structure of a Highly Faulted

Article pubs.acs.org/JPCC

Determination of the Apparent Crystal Structure of a Highly Faulted UPRM‑5 Type Flexible Porous Titanium Silicate via a Polymorph Based Superposition Model, a Rietveld Refinement and a Pair Distribution Function José N. Primera-Pedrozo,† Sneha Dugar,‡ María M. Martínez-Iñesta,† Riqiang Fu,‡ and Arturo J. Hernández-Maldonado†,* †

Department of Chemical Engineering University of Puerto RicoMayagüez Campus Mayagüez, Puerto Rico 00681-9000, United States ‡ National High Magnetic Field Laboratory Florida State University, Tallahassee, Florida 32310, United States S Supporting Information *

ABSTRACT: The crystal structure of a UPRM-5 titanium silicate prepared using tetraethylammonium (TEA+) has been approximated using a superposition model product of a polymorph stacking and a dual-phase Rietveld refinement method. Na+-UPRM-5 polymorphs were employed to elucidate the level of polymorphism or faulting in the crystal structure. DIFFaX simulations revealed that it was impossible to match the experimental diffraction data based solely on “pure” polymorphs. Instead, an intergrowth of combinations of polymorphs in the a and c directions resulted in the best faulting simulation scenario. The most suitable model combined two (2) orthorhombic polymorphs with faulting of 90 and 10% in the a and c directions, respectively. A refinement using this model did not yield a reliable structure, but an approximation was possible after employing a combination of orthorhombic and faulted triclinic phases. The superposition model, however, was not able to predict the final configuration of the TiO5 plausibly due to the unprecedented level of faulting in the structure. Upon convergence (χ2 = 13.68), the triclinic phase accounted for ca. 14% (molar basis) of the overall phase, being this further evidence of the level of faulting present in UPRM-5. The refined structure also revealed Si−O and Ti−O distances and angles that contrast with those reported for a titanium silicate known as ETS-4, and related to structural distortion. These changes are plausibly attributed to the presence of the TEA+ cations and the strong interaction of the framework oxygen with sodium cations, which were also exposed to the 8MR pore channel as described by a pair distribution function (PDF) refinement. In general, the UPRM-5 structural features appear to commensurate well with the gas adsorption and thermal stability properties previously reported, which differ considerably from those exhibited by Zorite type titanium silicates prepared in the absence of a quaternary ammonium cation.

■

INTRODUCTION

(SDAs), producing different levels of faulting depending on the size and alkyl chain length of cation employed.15,16 This serendipitous finding is apparently linked to the UPRM-5 exceptional adsorption working capacities upon detemplation, inclusion of an alkaline earth extraframework cation, and upon activation at moderate temperatures. According to standard X-ray diffraction (XRD) measurements made for UPRM-5,6,8,15,16 the material should possess a structure substantially different from that of ETS-4,1,3 but still related to the Zorite mineral.17,18 Comparisons made between Zorite and ETS-4 have led to several reports spanning nearly a decade now.11,19−21 Although UPRM-5 and ETS-4 are probably related to Zorite in many similar ways, UPRM-5

Porous titanium silicates exhibiting flexible frameworks (i.e., contraction) have been developed with the intention of achieving high selectivity during separation of gases via sizeexclusion principles.1−9 Although many of the flexible titanium silicates that are commercially available showcase exceptional adsorption selectivity for the removal of, for example, CO2 or CH4 from a mixture of light gases,3,10−14 their effective working capacities could be compromised as a result of the framework contraction process. This could in turn result in additional operation cycles during applications like pressure swing adsorption. In an attempt to provide a solution to this quandary, Hernández-Maldonado and co-workers developed a soft titanium silicate named UPRM-5, which is prepared via templated hydrothermal reactions using quaternary ammonium (NR4+) cations.6,8,15,16 Apparently these cations act not only as templates, but also as molecular structure-directing agents © 2014 American Chemical Society

Received: July 10, 2013 Revised: April 7, 2014 Published: April 8, 2014 8859

dx.doi.org/10.1021/jp502391e | J. Phys. Chem. C 2014, 118, 8859−8867

The Journal of Physical Chemistry C

Article

using the PDFgetX2 program27 and the refinement carried out using PDFfit2 and PDFgui programs.28 Solid-state 23Na magic angle spinning nuclear magnetic resonance (MAS NMR) measurements were carried out on an ultrawide bore 21 T magnet with a Bruker Avance II NMR console, where the 23Na Larmor frequency is 237.88 MHz, using a 3.2 mm homemade double resonance MAS NMR probe. The chemical shifts were referenced to a NaCl sample at 0 ppm. The NMR experimental parameters were: pulse length of 2 μs (∼π/7 small angle pulse), td of 1024, dwell time of 2 μs, and 128 scans with a recycle delay of 5 s.

exhibits surface areas nearly an order of magnitude larger than that of ETS-4 when thermally dehydrated at the same conditions. Remarkably, both titanium silicates share similar local chemical environments as corroborated by 29Si magic angle spinning nuclear magnetic resonance (MAS NMR) spectroscopy measurements. These correspond to Si(2Si,2Tiocta) and Si(3Si,1Tisemiocta) silicon environments, respectively.11,15 Although the existence of five-coordinated titanium in the framework is concomitant with the flexible structure and adjustable pores dimensions under thermalvacuum treatment,2,15,16 Hernandez-Maldonado and co-workers used in situ high temperature XRD and 29Si MAS NMR data to show that the Sr2+-UPRM-5 framework contraction process is related to the appearance of additional titanium coordination levels.15 In the case of Sr2+-ETS-4, Tsapatsis and co-workers found that five-coordinated titanium also contributes to the faults in the structure along a and c directions in the unit cell.21 On the basis of the similarities between the frameworks of ETS-4 and UPRM-5 and aiming at further understanding the role of a quaternary ammonium cation (tetraethylammonium, TEA+) in the synthesis of UPRM-5, here we report an extensive structural study based on XRD pattern simulations with DIFFaX,22,23 a superposition model, a Rietveld refinement and a pair distribution function (PDF) analysis. The DIFFaX simulations were based on “pure” and combinations of polymorphs also related to ETS-4 and Zorite.21 Since the supercell model that resulted in the theoretical XRD pattern that best matched the Na+-UPRM-5 experimental data pointed to a heavily faulted structure, the Rietveld refinement method was carried out using a combination of orthorhombic and triclinic phases, while the PDF refinement was used to provide additional insight on the characteristics of the refined structure. At the end, the triclinic phase appeared to be representative of the faults that are characteristic of the UPRM-5 material.

■

RESULTS AND DISCUSSION Simulations Based on 1D Faulting. As suggested in the Introduction, the structure of (TEA+, Na+)-UPRM-5 appears to be one that could be described by an intergrowth of polymorphs. Such a scenario could also help to elucidate the role of TEA+ cations in the synthesis, particularly its liaison to the level of faulting within the structure. Although the unit cell composition of the as-synthesized UPRM-5 differs from the one typical of ETS-4,6,29 it is still possible to use polymorph models already developed for ETS-421 in which the Na/Ti and Si/Ti ratios are set by modifying the atoms occupancies in all polymorphs to match those representative of UPRM-5. Simulations of faulting XRD patterns using DIFFaX22,23 were carried out on hypothetical “pure” polymorphs with a ETS-4like atomic arrangement to simulate the patterns in the a and c directions and in order to find the faulting probabilities on one direction and that best describe the XRD pattern of assynthesized Na+-UPRM-5. Figure 1 shows the four faulted polymorphs models that were employed and details about these are reported elsewhere.21 The DIFFaX simulations were performed according to the faulting map also shown in Figure 1. The results of the simulated faulted XRD data are shown in Figure 2 along with the experimental pattern that was gathered for (TEA+, Na+)UPRM-5. In most cases, the faulted XRD patterns matched the experimental results only at diffraction angles greater than 24° 2θ, which contrasts with the results reported for Sr2+-ETS-4 where the simulated and experimental diffraction data coincided at angles greater than 16° 2θ.21 Although the simulations made for UPRM-5 covered faulting probabilities near the 100% mark, none of the peaks in the 20−24° 2θ range were resolved since the faulting scenario represented by the stacking transitions employed in both a and c directions do not exhibit peaks within that range. Furthermore, upon further consideration of the crystal systems of the polymorphs, we developed a new hypothetical “pure” polymorph that contained a triclinic intergrowth to account for the missing diffraction peaks or that at least contained a contribution from a secondary phase based on results from indexing routines previously reported for UPRM-5.6,15 It is important to note at this stage of the discussion that the reflections corresponding to angles less than 20° 2θ may commensurate with a (TEA+, Na+)-UPRM-5 material that has a highly faulted crystal structure in the a and c directions, this probably induced by the TEA+ cations. In general, none of the simulations based on faulted UPRM-5 single “pure” polymorphs and along a single direction were capable of matching the complete experimental XRD pattern in a suitable, qualitative fashion. This was only possible using a simulation that accounted for the simultaneous faulting in the a and c directions and will be presented next.

■

EXPERIMENTAL SECTION Synthesis. (TEA+, Na+)-UPRM-5 was synthesized via hydrothermal synthesis with a gel mixture exhibiting the following composition: 3.4(TEA) 2 O/7.3Na 2 O/1.2K 2 O/ 1.3TiO2/10SiO2/200H2O, and using the procedures reported elsewhere.6 The final product was vacuum filtered and washed with copious amounts of distilled/deionized water. The resulting powder-like phase was dried in a forced convection oven overnight at 60 °C. Characterization. High-resolution standard powder X-ray diffraction (XRD) measurements for as-synthesized (TEA+, Na+)-UPRM-5 were gathered using a Rigaku Ultima III θ−θ goniometer unit equipped with a Cu−Kα target (λ = 1.5418 Å) and calibrated for focusing-type optics with a monochromator. The anode was operated using a voltage and current of 40 kV and 44 mA, respectively. A scanning speed of 0.04°/min and a step size of 0.01° were used to collect the high-resolution diffraction pattern in the 5−120° 2θ range. The instrument parameter file required for refinement using the General Structural Analysis System (GSAS)24 was carefully generated using diffraction data gathered for a silicon metal standard sample at scanning and speed conditions similar to those presented above. The XRD data was also normalized for flux, and corrected for X-ray polarization and Laue diffuse scattering, followed by a conversion to the total scattering structure factor S(Q).25,26 The maximum Q (Qmax = 4π sin θ/λ) collected was 7 Å−1. The atomic pair distribution function data were generated 8860

dx.doi.org/10.1021/jp502391e | J. Phys. Chem. C 2014, 118, 8859−8867

The Journal of Physical Chemistry C

Article

Figure 2. Simulated powder X-ray diffraction patterns obtained with DIFFaX between (A) polymorphs ABAB-AA and -ACAC, (B) polymorphs ABAB- and ABCD-ACAC, (C) polymorphs ABCDACAC and -AA, and (D) polymorphs ABCD- and ABAB-AA. The experimental (TEA+, Na+)-UPRM-5 XRD pattern is compared with the simulated data corresponding to the sequence numbers indicated in Figure 1.

combinations, the one that produced the best qualitative results corresponded to a combination of ABAB-AA and −ACAC orthorhombic polymorphs stacked in c-direction and performing simulations in a-direction. Figure 3 shows the simulation

Figure 1. Faulting map used for the DIFFaX simulations of a Na+UPRM-5 framework. Figures 2 and 3 show the results of the simulated XRD pattern for each number. The shaded regions indicate the range of faulting probabilities found in Sr2+-ETS-421 and Na+-UPRM5. Pure polymorphs adapted from a previous report by Tsapatsis and coworkers.21

Simulations Based on 2D Faulting. Simulations of faulted XRD patterns of Na+-UPRM-5 were performed in the a and c directions via stacking of the polymorphs that are shown in Figure 1. Supercells were constructed via random stacking of the polymorphs and simulating a faulting probability in one direction; the faulting probabilities in the other direction were then simulated using DIFFaX. Among the possible

Figure 3. Simulated powder X-ray diffraction patterns obtained with DIFFaX and a superposition model. The percentage data shown in the figure represent faulting probabilities in a-direction. 8861

dx.doi.org/10.1021/jp502391e | J. Phys. Chem. C 2014, 118, 8859−8867

The Journal of Physical Chemistry C

Article

process. Given that the level of faulting observed in Na+UPRM-5 will greatly influence the position of some if not most of these cations (i.e., due to the titanium coordination environment), it was essential to identify the possibility of having multiple sites first. Figure 4 shows 23Na MAS NMR data

results gathered with ca. 10% of faulting in c-direction and simulations of faulting probabilities from 10−90% in adirection. It can be seen that a diffraction pattern simulated with 90% of faulting in a-direction matched qualitatively very well with the Na+-UPRM-5 XRD experimental data. The reflection observed at ca. 14° 2θ is a contribution of the stacked ABAB-AA polymorph to the superposition model (see Figure 2A) and not due to the simulations. Meanwhile, the reflection or peak shoulder shown at ca. 8.5° 2θ was a product of the increment of the faulting developed in a-direction. The peaks corresponding to the 20−24° 2θ range in the experimental data were absent in the simulated data, as expected (see Discussion above). In addition, two of the characteristic reflections of the Na+-UPRM-5 XRD patterns were found at 6.8° and 11.4° 2θ (see Figure 3) and associated with the presence of fivecoordinated titanium centers along a faulted structure. These five-coordinated atoms have been observed in other titaniumbased materials.20 Still, the position of the said simulated reflections differed from the experimental value by as much as 0.855 Å d-spacing. Atoms related to the 6.8° and 11.5° 2θ reflections were carefully identified in the supercells, and their position coordinates were modified while observing correct bond distances and angles. As expected, the results of these changes (not shown here) produced further displacement of peak positions plus unwanted new reflections, clearly demonstrating the influence of the sensitivity of the supercell overall geometry on the simulations. Taking these observations into account, it is rather evident that the level of faulting in Na+-UPRM-5 is abundant and that precise simulation using DIFFaX would require development of a new set of “pure” polymorphs encompassing multidimensional faulting scenarios or even polymorphs that differ from what is typically found in Zoritelike titanium silicates (i.e., Figure 1). Nevertheless, the results gathered from DIFFaX also provided evidence that many of the Na+-UPRM-5 framework features, including the titanium coordination environment, are closely related to those of Zorite and, just like in the case of ETS-4, may explain the underpinnings of the flexibility or contraction process of the UPRM-5 framework. Dual-Phase Rietveld Refinement. Given the high level of faulting exhibited by Na+-UPRM-5, elucidating the structural features via a refinement based solely on a Zorite structure appears to be out of the question. Hence, we employed a supercell created by combining ABAB-AA and -ACAC orthorhombic polymorphs along the c-direction and the ABCD-ACAC monoclinic polymorph along a-direction. Although the results gathered from the simulations assumed a random stacking sequence in a-direction, for the sake of simplicity, we considered only a random sequence that produced a reduced supercell. The latter was assumed as a polymorph describing most of the faulted structure of Na+UPRM-5 and with atomic positions that served as preliminary values during the refinement. This approach provided also with a platform to incorporate a certain level of faulting at the beginning of the refinement routine. It should be noted that this approach did not account for faulting scenarios that could not be neglected during the refinement and, therefore, a dualphase refinement approach30−33 was eventually used in an attempt to address this. Before continuing with the refinement discussion, it is important to highlight that the positions of the extraframework sodium cations are a critical component of the refinement

Figure 4. 23Na MAS NMR spectrum for as-synthesized (TEA+, Na+)UPRM-5 recorded at 900 MHz. Fitted profiles and deconvoluted peaks are shown in red and blue lines, respectively.

gathered for UPRM-5 exhibiting two different sodium chemical environments in the resonances ca. σ = −5.152 and −11.864 ppm, respectively. Since the spectrum does not exhibit a second-order quadrupolar linebroadening, a deconvolution analysis was used to estimate the ratio between the different sodium atoms, which is Na2/Na1 = 2.59. As a first approach to elucidate the Na+-UPRM-5 structure, the supercell described at the beginning of this section was refined using GSAS in combination with a EXPGUI graphical interface.24,34 The preliminary cell parameters corresponded to a triclinic system with a = 23.021 Å, b = 7.238 Å, c = 13.930 Å, α = 100.50°, β = 92.07°, and γ = 90.05°. It should be noted that a Kα radiation with 0.5 ratio was also considered in view of the monochromatic source used (POLA = 0.97, IPOLA = 1). In addition, only the diffraction data gathered up to 80° 2θ was considered during the refinement and the background was refined up to 35 terms using shifted Chebyshev polynomials. The profile was also refined based on a pseudovoigt model while incorporating asymmetric and sample transparency parameters. Throughout the refinement process, it was sometimes necessary to release all or some of the Gaussian or Lorentzian profile parameters depending on whether or not χ2 increased between runs and, therefore, avoid divergence. In the reduced cell, each atom was refined as a unique entity due to the triclinic nature of the guess system. Flagging or freezing of a family of atoms was only employed during certain steps of the refinement to avoid divergence and no restraints/ constrains were included. Upon refinement of the atomic positions, structure factors and after accounting for some preferential orientation coefficients, the final converged value of χ2 reached 18.66. However, many of the resulting silicon tetrahedra atomic bonds and angles deviated considerable from the expected values. These also resulted in titanium centers with heavily distorted coordination environments. Upon inspection of the refined XRD pattern, in light of these incorrect environments, it became clear that the peaks located at angles below 16° 2θ were influenced the most. Interestingly, this low angle region has been related previously to faulting scenarios in Sr2+-ETS-4.21 8862

dx.doi.org/10.1021/jp502391e | J. Phys. Chem. C 2014, 118, 8859−8867

The Journal of Physical Chemistry C

Article

A refinement strategy excluding the 5−16° 2θ region was also employed, but it generated results similar to the ones described for the first approach. An inspection of the refined pattern and a comparison with the results previously reported for ETS-4 materials11,19−21 revealed other regions apparently related to additional faulting. To overcome this quandary, a third and final refinement strategy based on a dual-phase system30−33 was used to account for specific peak regions while allowing the atoms to be refined in an apparent single-phase environment (see Figure 5). One phase was employed to

Figure 5. Observed and calculated (dual-phase Rietveld) diffraction patterns for Na+-UPRM-5. The observed data are shown with markers points with the position of the Bragg reflections and difference between calculated and observed intensities shown below.

describe the diffraction peaks common to both UPRM-5 and ETS-4, while the second one was employed to describe the remaining peaks characteristics of UPRM-5 and that are consequence from faulting (excluding the 5−16° 2θ region). The first phase consisted of an orthorhombic system with a Cmmm space group and unit cell parameters a = 23.1962 Å, b = 7.2381 Å, and c = 13.9852 Å, respectively.21,29 The second phase (triclinic) corresponded to the one described above. For the first phase (orthorhombic), the unit cell and occupancies of the atoms comprising the bridging units with the stoichiometric ratio of 4Si2/1Ti2/2O5/4O6/1O7 (see Figure 6A) were constrained during the refinement. Soft constraints were also applied to the atoms distances between Si−O = 1.624 ± 0.044 and Ti−O = 1.97 ± 0.27 and none of the atoms of the five-coordinated titanium were refined since these are linked to the faulting observed through out what could be the UPRM-5 framework. The refinement of this orthorhombic phase included a background with up to 12 terms using the shifted Chebyshev polynomials, the profile and unit cell parameters and atoms positions (except TiO5), and converged to χ2 = 99.3. Upon inclusion of the triclinic phase, the refinement also proceeded with certain restrictions. For instance, Si−O and Ti−O distances, atomic occupancies of the bridging units and coordinates were constrained. The preliminary refinement also included a background (20 terms), profile, and unit cell parameters for the said phase (see Table 1), and converged to χ2 = 42.71. It should be noted that upon a sequential refinement of the atomic coordinates of the TiO5, although χ2 decreased, the five-coordination was lost showing Ti2 atoms

Figure 6. (A) Representation of the refined orthorhombic phase of the structure of Na+-UPRM-5. (B) Apparent supercell of Na+-UPRM-5 with missing five-coordinated titanium bridging. The TEA+ molecule shown occupies the location of the missing titanium unit as a plausible explanation to the framework faulting origin.

away from the O6 (bond distances >2.7 A). In addition, the O7 apical oxygen atoms appeared positioned near the center of the unit cell, missing any connection with the Ti2. Consequently, the coordinates of the atoms encompassing the fivecoordinated titanium centers were left fixed through out the rest of the refinement stages. The remaining atomic coordinates were refined, but the overall refinement also excluded a peak at ca. 17.75° 2θ that was not described by any of the two phases previously described, probably due to the high level of faulting in Na+-UPRM-5. Upon refinement of the structure factors and occupancies, and after releasing the weight factors to 0, the coordinate constraints, and introducing some spherical harmonic preferential orientation terms to both phases, the overall refinement converged to χ2 = 13.68. It is of utmost importance to mention that, upon convergence, the triclinic phase contributed to more 8863

dx.doi.org/10.1021/jp502391e | J. Phys. Chem. C 2014, 118, 8859−8867

The Journal of Physical Chemistry C

Article

refinement parameters and atomic parameters for both phases, respectively. The observed and calculated (dual-phase Rietveld) diffraction patterns are shown in Figure 5. Na+-UPRM-5 Apparent Structure. One may argue that the refinement described above yield perhaps an unreliable UPRM-5 faulted structure since it is a product of two phases used as an intergrowth of crystals (i.e., too many degrees of freedom). However, the resulting atomic positions and environments could still offer valuable insights since the refinement was executed following a logical pathway. Furthermore, although both the orthorhombic and triclinic phases were key in the determination of the refined parameters, the analysis of the atomic positions and resulting bond lengths were focused on the former since it prevailed at the end, stressing on the similitude between UPRM-5 and Zorite. When compared to Zorite, however, the refined hypothetical structure of Na+-UPRM-5 shows changes in the atoms distances and angles that could be a product of the presence of the TEA+ cations. Although Figure 6A shows bridging unit Ti−O7 bond distances that were not refined, the stoichiometric data showed that there is ca. 1Ti2/1O7 in the unit cell, confirming that the UPRM-5 structure has a TiO5 configuration. However, there was a concern about the distance between Ti2−O7 and the corresponding type of bonding that these two atoms would showcase. Figure 6A also shows that the Na+-UPRM-5 structure would exhibit two Na+ cations environments similar to the case of as-prepared ETS-4. These are represented by two sites: in the 6MR for Na1 and 7MR for Na2, respectively, and the finding correlates well with the MAS NMR data shown in Figure 4. In fact, a population analysis performed on the dual-phase refined structure gave a Na2/Na1 ratio of 2.82, which compares very well with the value obtained from the NMR deconvolution analysis. The environment of Na1 is also influenced by an octahedral coordination of the cation with four O1 and two Ow1 (see Table 3).20,21,29 This coordination exhibits a strong hydrogen bonding interaction between Na1···Ow1, with a coordination distance of 1.6116 Å. The Na2 cation site, meanwhile, appears to be coordinated to two O3, one O4 and two O6 framework oxygen atoms and one Ow1 water oxygen (see Figure 6A). When compared to the case of ETS-4, the distorted octahedral configuration of the Ti1

Table 1. Data Collection, Chemical Composition and Refinement Parameters Used for the Polymorph Superposition Model of Na+-UPRM-5 data collection X-ray source wavelength (Å) step (deg) 2θ range (deg) unit cells space group a (Å) b (Å) c (Å)

CuKα 1.5418 0.01 5−120

Na6.59Si10.38Ti4.61O36.87

refinement

Cmmm 23.23020 (6) 7.18715 (18) 6.96047 (17) P1

chemical composition (avg.)

excluded regions (deg)

5−16

no. of variables no. of observations no. reflections

17−18 79.5−120 143 6540 5876

space group a (Å)

20.8820(18)

b (Å) c (Å) α (deg) β (deg)

6.8207(7) 13.6621(15) 99.870(8) 91.057(9)

profile coefficients (Cmmm) GU, GV, GW, TRNS

γ (deg)

88.762(7)

LX, LY, ASYM

20

no. of background coefficients (shifted Chebyshev polynomial)

profile coefficients (P 1) GU, GV, GW, TRNS

0.000, −144.934, 19.512, 9.354 3.764, 0.000, 6.253 7240.810, −4038.600, 568.220, 212.672 0.000, 0.000, 19.713 4 5.3 1.5 13.68

LX, LY, ASYM Rp (%) wRp (%) R exp (%) χ2

than 14% (mole) to the overall phase. Such result is a remarkable one given that the 5−16° 2θ peak region of the asprepared material was excluded from the refinement. Tables 1, 2 and Supporting Information, SI Table S1 summarize the

Table 2. Atomic Parameters for the Superposition Model and Based on the Dominant Orthorhombic Phase atom

x

y

z

occupancy

multiplicity

Uiso

Ti1 Ti2 Si1 Si2 Na1 Na2 O1 O2 O3 O4 O5 O6 O7 Ow1 Ow2 Ow3 Ow4

0.2500 0.0000 0.1644(32) 0.0639(4) 0.2500 0.3692(7) 0.1527(9) 0.0936(5) 0.1943(4) 0.2162(9) 0.0000 0.0605(9) 0.0000 0.2854(10) 0.0745(33) 0.0140(5) 0.0224(21)

0.2500 0.5000 0.0000 0.1001(19) 0.2500 0.0000 0.0000 0.0000 0.2002(14) 0.5000 0.0000 0.315(4) 0.5000 0.057(6) 0.378(11) 0.1490(12) 0.0000

0.0000 0.0614 0.2669(11) 0.0000 0.5000 0.2107(27) 0.5000 0.1937(18) 0.2054(14) 0.0000 0.0000 0.0000 0.2823 0.5000 0.5000 0.5000 0.5000

1.000 0.342(7) 1.000 0.539(9) 0.460(14) 0.714(11) 1.000 1.000 1.000 1.000 1.000 0.637(16) 0.280(26) 0.454(10) 0.215(15) 0.213(14) 0.521(25)

4 4 8 8 4 8 4 8 16 4 2 8 4 8 8 8 4

0.0103(19) 0.0440(6) 0.0188(22) 0.0071(30) 0.0620(14) 0.0790(5) 0.0221(19) 0.0221(19) 0.0221(19) 0.0221(19) 0.0221(19) 0.0221(19) 0.0221(19) 0.0630(11) 0.0030(18) 0.0490(25) 0.1840(19)

8864

dx.doi.org/10.1021/jp502391e | J. Phys. Chem. C 2014, 118, 8859−8867

The Journal of Physical Chemistry C

Article

A real-space refinement was also performed on the already refined orthorhombic phase in order to analyze the Na+UPRM-5 structure within a short-range order. The parameters refined using PDFfit2 and PDFgui were the scale factors, width of the peaks due to the correlated motion (delta), lattice parameters, atoms coordinates, and the structure factors. Upon completion of the refinement procedure, the goodness of fit was Rw = 32.27%. Figure 7 shows both the experimental PDF

Table 3. Bond Distances and Angles Obtained from the Rietveld Refinement of Na+-UPRM-5a bond

distance (Å)

Ti1−O3 [ × 4] Ti1−O4 [ × 2] mean

1.9603 (1) 1.9605 (1) 1.9604

Ti2−O6 [ × 4] Ti2−O7 [ × 1] mean

1.9803 (1) 1.5382 1.7593

Si1−O1 [ × 1] Si1−O2 [ × 1] Si1−O3 [ × 2] mean

1.6457 (1) 1.7231 (1) 1.6541 (1) 1.6743

Si2−O2 [ × 2] Si2−O5 [ × 1] Si2−O6 [ × 1] mean

1.6769 (1) 1.6493 (1) 1.5476 (1) 1.6246

cation coordination environments Na1−O1 [ × 2] 2.8877 (1) Na1−O3 [ × 4] 2.4505 (1) Na1−Ow1 [ × 2] 1.6116 (1) Na2−O3 [ × 2] 2.6122 (1) Na2−O4 [ × 1] 2.4681 (1) Na2−O6 [ × 2] 2.5658 (1) Na2−Ow1 [ × 1] 2.8304 (1) Na2−Ow2 [ × 1] 2.5548 (1) a

atoms O3−Ti1−O3 O3−Ti1−O3 O3−Ti1−O4 O3−Ti1−O4 mean

angle (deg) [ [ [ [

× × × ×

2] 2] 4] 4]

O6−Ti2−O6 [ × 2] O6−Ti2−O6 [ × 2] O6−Ti2−O7 [ × 4] mean

93.6394 86.3606 84.4621 95.5379 90.0000

(19) (19) (9) (9)

90.3848 (19) 84.2751 (19) 102.4555 (4) 92.3718

O1−Si1−O2 O1−Si1−O3 O2−Si1−O3 O3−Si1−O3 mean

[ [ [ [

× × × ×

1] 2] 2] 1]

97.6180 108.9582 108.9454 120.9048 109.1066

(9) (5) (8) (15)

O2−Si2−O2 O2−Si2−O5 O2−Si2−O6 O5−Si2−O6 mean

[ [ [ [

× × × ×

1] 2] 2] 1]

107.0602 100.5534 116.7079 112.9296 109.3128

(17) (10) (7) (9)

Figure 7. Experimental and calculated PDFs of Na+-UPRM-5 based on the orthorhombic phase (Rw = 32.27%).

The cations environments are also included.

obtained from the XRD data and the calculated PDF based on the refined orthorhombic phase. A quick inspection of both profiles reveals that these are not in agreement, probably due the level of polymorphism or high degree of faulting present in the UPRM-5 structure. In terms of the distribution of the distances between atomic pairs in the Na+-UPRM-5 structure, the first peak shown in Figure 7 corresponds to the average of the Si−O and Ti−O distances (ca. 1.5−2.1 Å) and correlates with the results obtained from the Rietveld refinement. Meanwhile, the second and third peaks are related to Na−O (ca. 2.5−3.8 Å), O−O (ca. 2.4−4.5 Å), and Si−Ti (ca. 3.2−5.7 Å) atoms distances, respectively. Perhaps the more significant result from the short-range order analysis is related to new cations environments. Figure 8 shows the characteristic 8MR also found in Zorite type materials3,20,21,29 exhibiting positions of Na2 cations that are exposed to a pore channel. In addition, the water molecules coordinated to Na2 cations also show new coordination environments whose atomic distances are summarized in Table 4. On the basis of the Na2···Ow1 and Na2···Ow2 distances (2.124 and 1.577 Å), it appears that the cation-water coordination is quite strong. The new Na2 cation positions were missing in the Rietveld refinement results perhaps due to the long-range nature of the analysis. However, those results did show Na2 cation positions within the vicinity of the 7MR or the 12MR channels when TiO5 is absent (See Figure 7).3 When the PDF and Rietveld refinement analyses are combined (i.e., short- and long-range structures), it is possible to understand why the ion exchange and adsorption processes take place with relative ease in UPRM-5. Although the presence of the TEA+ cations cannot be accounted for during the present refinement, their presence is

chain in UPRM-5 is a product of the strong interaction of the Na2 cation and the influence of the hydrogen bonding of the Ow1 (Ow1···O3 distance of 2.7320 Å) with the O3 atoms of the Ti1 octahedron. In addition, this cation site exhibits a strong interaction with the O6 atom from the Ti2 fivecoordinated, with a coordination distance of 2.5658 Å (see Table 3). This perhaps explains the high level of flexibility exhibited by the Sr2+-UPRM-5 framework even at mild temperatures.6,15 It should be noted that Na2 has a higher occupancy than the five-coordinated bridging unit, which means that the cation could even replace the bridging unit if abscent.29 Since it is evident that the water molecules play an important role on the stability of this type of materials, it is important to briefly describe the environment coordination based on the relative refined atoms positions in the orthorhombic phase. As seen in Figure 6A, Na+-UPRM-5 probably contains four (4) different water molecules in its structure. The Ow1 coordination environment was already described, so let’s focus on the remaining ones. In the case of the Ow2, it appears to coordinate to a O7 from the bridging unit at a distance of 2.461 Å and appears to be part of a chain-like arrangement with Ow3 and Ow4. The latter could be due to hydrogen bonding as suggested by the occupancies and coordination distances. In the case of the Ow4, an occupancy of 0.521 translates also to hydrogen bonding with O2, with a coordination distance of 2.697 Å. This is perhaps responsible for the distortion of Si1 and Si2 tetrahedra, which are interconnected by an O2 atom. 8865

dx.doi.org/10.1021/jp502391e | J. Phys. Chem. C 2014, 118, 8859−8867

The Journal of Physical Chemistry C

Article

attempt to refine the Na+-UPRM-5 structure was made employing two intergrowth phases, one of which accounted for the presence of diffraction peaks related also to faulting. Furthermore, the level of faulting is such that refinement of the titanium-bridging unit atoms was not possible. The refinement also corroborated that the distortion of the Si tetrahedral and the Ti octahedral are a result of the interactions between extra framework sodium and water molecules with framework oxygen. This is linked to the contraction phenomenon observed in these materials even at mild temperatures.

■

ASSOCIATED CONTENT

* Supporting Information S

List of atomic parameters for the superposition model and based on the triclinic phase. This material is available free of charge via the Internet at http://pubs.acs.org.

Figure 8. Apparent sodium and water coordination environments on the Na+-UPRM-5 orthorhombic phase. Structure generated from the PDF refinement results.

■

Table 4. Bond Distances and Angles Calculated after a PDF Refinement of the Na+-UPRM-5 Orthorhombic Phase bond

distance (Å)

Ti1−O3 [ × 4] Ti1−O4 [ × 2] mean

1.933 1.998 1.966

Ti2−O6 [ × 4] Ti2−O7 [ × 1] mean

1.954 1.692 1.823

[ × 1] [ × 1] [ × 2]

1.705 1.746 1.544 1.665 1.585 1.604 1.745 1.645

Si1−O1 Si1−O2 Si1−O3 mean Si2−O2 Si2−O5 Si2−O6 mean

[ × 2] [ × 1] [ × 1]

coordination

AUTHOR INFORMATION

Corresponding Author

*Phone: 787-832-4040 x3748; fax: 787-834-3655; e-mail: [email protected].

distance (Å)

cation coordination environments Na1−O1 [ × 2] 3.127 Na1−O3 [ × 4] 2.475 Na1−Ow1 [ × 2] 2.196 Na2−O3 [ × 2] 3.234 Na2−O4 [ × 1] 3.236 Na2−O6 [ × 2] 3.084 Na2−Ow1 [ × 1] 2.124 Na2−Ow2 [ × 1] 1.577 Ow1−O3 [ × 1] 3.095

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors wish to acknowledge Prof. Michael Tsapatsis and Dr. Sandeep Kumar from the Department of Chemical Engineering and Materials Science at the University of Minnesota, Minneapolis for their input and valuable discussions on the polymorph stacking and refinement experiments. This material is based upon work supported by the National Science Foundation (NSF) under Award No. HRD-0833112 (CREST Program). We also wish to acknowledge support from the Puerto Rico Institute for Functional Materials Graduate Fellowships Program under the NSF Award No. EPS1002410. The NMR measurements were performed at the National High Magnetic Field Laboratory (NHMFL) supported by NSF Cooperative Agreement No. DMR-0654118, the State of Florida, and the U.S. Department of Energy.

certainly evident by its effect on the final positions of the sodium cations and the water molecules, helping to create faulting and distortion in the structure. When considering the results gathered by the DIFFaX simulations and those of the dual-phase refinement, a snapshot of the Na+-UPRM-5 supercell could be gathered. Figure 6B shows an approximation of the supercell structure viewed along the ac plane, and it includes the apparent (not calculated) location of the TEA+. The absence of a five-coordination bridge is related to faulting and the resulting void dimensions are commensurate with the overall TEA+ cation dimensions, and this explains why, upon removal of the TEA+, the material exhibits larger adsorption capacities when compared to ETS-4 at similar conditions.

■

REFERENCES

(1) Kuznicki, S. M. Preparation of Small-Pored Cristalline Titanium Molecular Sieve Zeolites, 1990; U.S. Patent 4,938,939 A. (2) Kuznicki, S. M.; Bell, V. A.; Petrovic, I.; Blosser, P. W.Separation of Nitrogen from Mixtures thereof with Methane Utilizing Barium Exchanged ETS-4, 1999; U.S. Patent 5,989,316 A. (3) Kuznicki, S. M.; Bell, V. A.; Nair, S.; Hillhouse, H. W.; Jacubinas, R. M.; Braunbarth, C. M.; Toby, B. H.; Tsapatsis, M. A Titanosilicate Molecular Sieve with Adjustable Pores for Size-Selective Adsorption of Molecules. Nature 2001, 412, 720−724. (4) Anson, A.; Lin, C. C. H.; Kuznicki, S. M.; Sawada, J. A. Adsorption of Carbon Dioxide, Ethane, and Methane on Titanosilicate Type Molecular Sieves. Chem. Eng. Sci. 2009, 64, 3683−3687. (5) Lin, C. C. H.; Sawada, J. A.; Wu, L.; Haastrup, T.; Kuznicki, S. M. Anion-Controlled Pore Size of Titanium Silicate Molecular Sieves. J. Am. Chem. Soc. 2008, 131, 609−614. (6) Primera-Pedrozo, J. N.; Torres-Cosme, B. D.; Clardy, M. E.; Rivera-Ramos, M. E.; Hernández-Maldonado, A. J. Titanium Silicate Porous Materials for Carbon Dioxide Adsorption: Synthesis Using a Structure Directing Agent, Detemplation and Inclusion of Alkaline Earth Metal Cations. Ind. Eng. Chem. Res. 2010, 49, 7515−7523. (7) Zhang, Y.-Q.; Zhou, W.; Liu, S.; Navrotsky, A. Controllable Morphology of Engelhard Titanium Silicates ETS-4: Synthetic,

■

CONCLUSIONS The presence of a molecular structure-directing agent (i.e., TEA+ cations) during the synthesis of the flexible Na+-UPRM-5 titanium silicate material results in a highly faulted crystal structure in a and c directions as described by a combination of orthorhombic polymorphs and DIFFaX simulations. Upon removal of the TEA+ cations, the faults created in the structure and the associated voids are probably responsible for the superior gas adsorption capacities previously documented in the literature for alkaline earth metal exchanged UPRM-5. An 8866

dx.doi.org/10.1021/jp502391e | J. Phys. Chem. C 2014, 118, 8859−8867

The Journal of Physical Chemistry C

Article

Photocatalytic, and Calorimetric Studies. Chem. Mater. 2011, 23, 1166−1173. (8) Hernandez-Maldonado, A. J.; Primera-Pedrozo, J. N. Method for Synthesizing a Novel Adsorbent Titanosilicate Material (UPRM-5), 2013; U.S. Patent 8,440,166 B1. (9) Song, L. Y.; Liu, X. J.; Zhan, Z. C.; He, H.; Zi, X. H.; Qiu, W. G. Improvement of Adsorption Performance Based on Metal Ion Modified ETS-4 Molecular Sieves. Adv. Mater. Res. 2013, 815, 437− 442. (10) Marathe, R. P.; Mantri, K.; Srinivasan, M. P.; Farooq, S. Effect of Ion Exchange and Dehydration Temperature on the Adsorption and Diffusion of Gases in ETS-4. Ind. Eng. Chem. Res. 2004, 43, 5281− 5290. (11) Philippou, A.; Anderson, M. W. Structural Investigation of ETS4. Zeolites 1996, 16, 98−107. (12) Pillai, R. S.; Peter, S. A.; Jasra, R. V. Adsorption of Carbon Dioxide, Methane, Nitrogen, Oxygen and Argon in NaETS-4. Microporous Mesoporous Mater. 2008, 113, 268−276. (13) Jayaraman, A.; Hernandez-Maldonado, A. J.; Yang, R. T.; Chinn, D.; Munson, C. L.; Mohr, D. H. Clinoptilolites for Nitrogen/Methane Separation. Chem. Eng. Sci. 2004, 59, 2407−2417. (14) Mitariten, M. New Technology Improves Nitrogen-Removal Economics. Oil Gas J. 2001, 99, 42−44. (15) Primera-Pedrozo, J. N.; Guerrero-Medina, K. J.; Fu, R.; Hernández-Maldonado, A. J. Sr(ii)-UPRM-5 Titanium Silicate Framework Thermally Induced Contraction: in Situ High Temperature XRD and 29Si MAS NMR. Dalton Trans. 2011, 40, 3547−3552. (16) Marcano-González, M. E.; Primera-Pedrozo, J. N.; JiménezLaureano, Z.; Fu, R.; Hernández-Maldonado, A. J. UPRM-5 Titanium Silicates Prepared using Tetrapropylammonium and Tetrabutylammonium Cations: Framework Stability, Textural Properties, and Carbon Dioxide Adsorption. J. Mater. Sci. 2013, 48, 2053−2066. (17) Sandomirskii, P. A.; Belov, N. V. The OD Structure of Zorite. Sov. Phys. Crystallogr. 1979, 24, 686−693. (18) Belokoneva, E. L. Symmetry and Topological Analysis of the OD Nenadkevichite−Labuntsovite−Zorite Family. Crystallogr. Rep. 2005, 50, 13−19. (19) Valtchev, V.; Mintova, S.; Mihailova, B.; Konstantinov, L. Comparison of Physicochemical Properties of Zorite and ETS-4. Mater. Res. Bull. 1996, 31, 163−169. (20) Cruciani, G.; De Luca, P.; Nastro, A.; Pattison, P. Rietveld Refinement of the Zorite Structure of ETS-4 Molecular Sieves. Microporous Mesoporous Mater. 1998, 21, 143−153. (21) Braunbarth, C.; Hillhouse, H. W.; Tsapatsis, M.; Burton, A.; Lobo, R. F.; Jacubinas, R. M.; Kuznicki, S. M. Structure of Strontium Ion-Exchanged ETS-4 Microporous Molecular Sieves. Chem. Mater. 2000, 12, 1857−1865. (22) Treacy, M. M. J.; Newsam, J. M.; Deem, M. W. A General Recursion Method for Calculating Diffracted Intensities from Crystals Containing Planar Faults. Proc. R. Soc. London Ser. A-Math. Phys. Eng. Sci. 1991, 433, 499−520. (23) Treacy, M. M. J.; Vaughan, D. E. W.; Strohmaier, K. G.; Newsam, J. M. Intergrowth Segregation in FAU-EMT Zeolite Materials. Proc. R. Soc. London Ser. A-Math. Phys. Eng. Sci. 1996, 452, 813−840. (24) Larson, A. C.; Von Dreele, R. B.General Structure Analysis System (GSAS); Los Alamos National Laboratory, 2004. (25) Warren, B. E. X-ray Diffraction; Dover Publications: Mineola, NY, 1990. (26) Klug, H. P.; Alexander, L. E. Diffraction Studies of Noncrystalline Materials; John Wiley & Sons: New York, 1954. (27) Qiu, X.; Thompson, J. W.; Billinge, S. J. L. PDFgetX2: A GUIDriven Drogram to Obtain the Pair Distribution Function from X-ray Powder Diffraction Data. J. Appl. Crystallogr. 2004, 37, 678. (28) Farrow, C. L.; Juhas, P.; Liu, J. W.; Bryndin, D.; Božin, E. S.; Bloch, J.; Proffen, T.; Billinge, S. J. L. PDFfit2 and PDFgui: Computer Programs for Studying Nanostructure in Crystals. J. Phys.-Condens. Matter 2007, 19, 335219.

(29) Nair, S.; Jeong, H. K.; Chandrasekaran, A.; Braunbarth, C. M.; Tsapatsis, M.; Kuznicki, S. M. Synthesis and Structure Determination of ETS-4 Single Crystals. Chem. Mater. 2001, 13, 4247−4254. (30) Lievens, J. L.; Verduijn, J. P.; Bons, A.-J.; Motier, W. J. Cation Site Energies in Dehydrated Hexagonal Faujaste (EMT). Zeolites 1992, 12, 698−705. (31) Feijen, E. J. P.; De Vadder, K.; Bosschaerts, M. H.; Lievens, J. L.; Martens, J. A.; Grobet, P. J.; Jacobs, P. A. Role of 18-Crown-6 and 15Crown-5 Ethers in the Crystallization of Polytype Faujasite Zeolites. J. Am. Chem. Soc. 1994, 116, 2950−2957. (32) Feijen, E. J. P.; Reynders, R. A.; Geerts, H.; Martens, J. A.; Grobet, P. J.; Jacobs, P. A. Use of 13C CP NMR Spectroscopy for the Quantification of FAU/EMT Zeolite Intergrowths Synthesized with 15-Crown-5 and 18-Crown-6 Ethers. J. Phys. Chem. 1995, 99, 1837− 1839. (33) Kaduk, J. A. A Rietveld TutorialMullite. Powder Diffr. 2009, 24, 351−361. (34) Toby, B. H. EXPGUI, a Graphical User Interface for GSAS. J. Appl. Crystallogr. 2001, 34, 210−213.

8867

dx.doi.org/10.1021/jp502391e | J. Phys. Chem. C 2014, 118, 8859−8867