Through-bond and through-space connectivities of amorphous aluminophosphate by two-dimensional 27Al–31P heteronuclear NMR

LETTER TO THE EDITOR

Journal of Non-Crystalline Solids 353 (2007) 4227–4231 www.elsevier.com/locate/jnoncrysol

Letter to the Editor

Through-bond and through-space connectivities of amorphous aluminophosphate by two-dimensional 27Al–31P heteronuclear NMR Koji Kanehashi a

a,*

, Takahiro Nemoto b, Koji Saito

a

Advanced Technology Research Laboratories, Nippon Steel Corporation, 20-1 Shintomi, Futtsu, Chiba 293-8511, Japan b JEOL Ltd., 3-1-2 Musashino, Akishima, Tokyo 196-8558, Japan Received 24 March 2006; received in revised form 12 May 2007

Abstract The connectivities between Al and P through chemical bond and internuclear distance have been studied for an amorphous aluminophosphate (a-AlPO4) using two-dimensional (2D) solid-state 27Al–31P correlation NMR (MAS J-HMQC and CP HETCOR). Whereas the conventional 31P MAS spectrum provides less informative results because of poor resolution caused by large distributions of the nucleus surroundings, the 2D HETCOR shows much better resolution and at least four non-equivalent P sites in the a-AlPO4. These P sites are found to be correlated with one [4]Al, two [5]Al and one [6]Al species, and have different chemical shifts. This result might indicate that the mean P–O–[n]Al (n = 4, 5, 6) bond angles are different each other, and they are estimated using the relationship with the 31P chemical shifts in crystalline AlPO4 previously reported. Ó 2007 Elsevier B.V. All rights reserved. PACS: 61.18.Fs; 61.43.Er Keywords: Nuclear magnetic (and quadrupole) resonance; Phosphates; NMR, MAS-NMR and NQR; Medium-range order; Short-range order

1. Introduction The alminophosphate (AlPO4) system is one of the useful molecular sieves, and is widely utilized as catalysts and molecular sieves in industrial processes. AlPO4s are usually synthesized by the hydrothermal reaction, and become either crystalline or amorphous depending on pH in a solution and on calcined temperature [1]. It is important to elucidate the connectivities between PO4 tetrahedrons and AlOn (n = 4, 5, 6) polyhedrons, because they greatly affect the molecular designs. Double resonance NMR is a very useful technique to obtain information about the connectivities between heteronuclei. The connectivities between Al and P via bridging oxygen have been well investigated by solid-state HETCOR (HETeronuclear CORrelation) *

Corresponding author. E-mail address: [email protected] (K. Kanehashi).

0022-3093/$ - see front matter Ó 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.jnoncrysol.2007.05.020

NMR for various crystalline AlPO4s (c-AlPO4s) such as hydrated VPI-5 [2–6], AlPO4-5 [7], AlPO4-8 [4], AlPO4-14 [8–10], AlPO4-31 [11], AlPO4-34 [11], AlPO4-40 [5] and AlPO4-41 [12]. These HETCOR NMR techniques are of two distinct types: the through-bond connectivities via J couplings (2JAlP) [6,8,10,13,14] and the through-space connectivities via 27Al–31P dipolar couplings [2–5,7–9,11,12]. On the other hand, the Al–P connectivities in amorphous AlPO4s (a-AlPO4s) have been less examined compared with those in c-AlPO4s. The 27Al–31P REDOR and the TRAPDOR experiments have been performed to estimate interactions between Al and P species for the a-AlPO4 [15]. The through-space Al–P connectivities in the intermediate gel phases of the AlPO4 have been studied by one- (1D) and two-dimensional (2D) CP/MAS [16]. We have investigated not only the O–P connectivities by 1D 31P ! 17O and 17O ! 31P CP/MAS [17], but also the Al–P correlations by 2D 31P{27Al} HETCOR combined

LETTER TO THE EDITOR

4228

K. Kanehashi et al. / Journal of Non-Crystalline Solids 353 (2007) 4227–4231

with 3QMAS (triple quantum magic angle spinning) for the purpose of averaging out the second-order quadrupolar interaction in the 27Al dimension [18]. Because the powder X-ray diffraction pattern of a-AlPO4s provide no informative results because of lacking the long-range structural order, J coupling- and dipolar-driven HETCOR NMR should be quite powerful techniques to estimate the connectivities in a-AlPO4s. In this study, we examine the through-bond and through-space connectivities in an a-AlPO4 using 2D 27Al–31P MAS J-HMQC and CP HETCOR NMR, respectively. From the observed correlation spectrum, the mean P–O–Al bond angles of each P–O–[n]Al (n = 4, 5, 6) species in the a-AlPO4 are estimated using the relationship between the angle and the 31P chemical shift for c-AlPO4s (i.e. VPI-5 and AlPO4-14). 2. Experimental The a-AlPO4 studied here was prepared from Al(NO3)3 Æ 9H2O and H3PO4 under controlled pH according to the previous reports [1,17]. All NMR spectra were acquired on a JEOL JNM-ECA700 spectrometer with a 16.4 T (1H = 700 MHz) narrow-bore magnet. The homebuilt 4-mm XY double resonance probe was resistant to high rf fields, and produced quite a short p/2 pulse for both channels: 1.8 ls at 1000 W (probe-in) for the 27Al channel for solid AlK(SO4)2 Æ 12H2O (the quadrupolar coupling constant CQ  0) and 2.2 ls at 560 W (probe-in) for the 31 P channel for solid (NH4)2HPO4. The resonant frequencies of 27Al and 31P were 182.4 and 283.4 MHz, respectively. The sample spinning rate was 18 kHz ± 10 Hz. The 27Al and 31 P chemical shifts were referenced to 1.0 mol/l AlCl3 aqueous solution at 0.1 ppm and solid (NH4)2HPO4 at 1.33 ppm, respectively. A small (solids 15–20°) tip angle was applied for single pulse 27Al and 31P MAS experiments,

π/2

π

τ

27Al

t1/2

τ

t1/2

π/2

t2

π/2

t1

31P

π/2 27Al

31P

t1

CP

CP

t2

Fig. 1. Pulse sequences of two-dimensional 27Al–31P NMR: (a) MAS J-HMQC (through-bond connectivities) and (b) CP HETCOR (throughspace connectivities).

and pulse recycle delays were long enough to allow full relaxation. The z-filter sequence was applied for 27Al MQ (M = 3, 5) MAS [19]. The MAS J-HMQC [6] and the CP HETCOR pulse scheme were used for analysis of the connectivities through chemical bond and internuclear distance, respectively (Fig. 1). In the CP process from the quadrupolar spin (e.g. 27Al) to the spin-1/2 (e.g. 31P), its efficiency strongly depends on the adiabaticity parameter a [20]: a ¼ m2rfðAlÞ =mQ mMAS ; where mrf(Al) is the rf field of 27Al under the spin-locking, mQ is the quadrupolar parameter (= 3CQ/[2I(2I  1)]), mMAS is the MAS frequency. We chose the sudden passage condition (a  1) to preserve the transverse magnetization under the spin-locking: mrf(Al) = 6 kHz, mMAS = 18 kHz. Centers of gravity for cross sections in a 2D spectrum were determined by integration. 3. Results and discussion First, 27Al MAS and MQMAS spectra and a 31P MAS spectrum were measured (not shown). For 27Al, three distinct Al species assigned to tetrahedral ([6]Al), pentacoordinated ([5]Al) and octahedral ([6]Al) Al environments were observed at 42, 20 and 5 ppm, respectively, in the MAS spectrum. The higher coordinated Al sites ([5]Al and [6]Al) were attributed to the coordination of extra water molecules or hydroxyl groups to AlO4 tetrahedrons by the fact that a 1 H ! 27Al CP/MAS spectrum enhanced the relative intensities of these Al species. Compared with the MAS spectrum, spectral resolution of MQMAS was slightly better due to the line narrowing caused by the cancellation of the second-order quadrupolar interaction. However, the overall line shapes of the MAS and MQMAS spectra were quite similar because of the relatively small quadrupolar coupling constants (CQ = 2.0–2.7 MHz) of each Al site. All cross peaks in the MQMAS spectrum were dispersive along the chemical shift line (a slope of 1) rather than the quadrupolar induced shift line (a slope of 10/17), indicating that the line broadening of the 27Al spectrum results mainly from chemical shift distributions, and quadrupolar distributions less affects the broadening. This result means that there is a variety of combinations of AlOn polyhedrons with adjacent PO4 tetrahedrons such as the P–O–Al bond angles and the Al–O bond lengths, whereas distributions of the distortion of AlOn (n = 4, 5, 6) polyhedrons are relatively small. A 31P MAS spectrum showed one broad, featureless peak at 23.1 ppm (peak top) which is assigned to Q4 tetrahedral units (all four O ions in PO4 are shared with the adjacent PO4 or AlOn units), because no non-bridging oxygen have been found to present in this sample from the previous 17O NMR result [17]. This broad peak is probably due to distributions of the nucleus surroundings mainly. Remaining 31P–27Al and 31P–1H dipole interactions might also broaden the 31P MAS spectrum [9].

LETTER TO THE EDITOR

K. Kanehashi et al. / Journal of Non-Crystalline Solids 353 (2007) 4227–4231

Fig. 2. 27Al{31P} MAS J-HMQC spectrum of amorphous AlPO4. s = 5 ms.

A 2D 27Al{31P} MAS J-HMQC spectrum of the a-AlPO4 is shown in Fig. 2. The previous reports on the J coupling-based 2D experiments including quadrupolar nuclei often suffered from the line broadening along the quadrupolar dimension caused by the second-order quadrupolar interaction [6,8]. In case of this sample, however, it was not a serious problem, because again there was no large difference in spectral resolution between the 27Al MAS and 3QMAS spectra. Hence, we here applied this pulse sequence to the analysis of the through-bond connectivities instead of the MQ-J-HETCOR pulse trains [10] so as to prevent the sensitivity loss caused by the multiple quantum excitation in the MQ block. It has been known that the optimal s duration for the HMQC spectrum depends on the strength of J couplings, and is theoretically attained for s = 1/2J. The strength of the 2J coupling between 27Al and 31P in this sample has not been understood hitherto, and it should have some distributions because of its amorphous structure. We here regarded the mean 2JAl–P coupling of the a-AlPO4 as a typical value of 25 Hz (s = 5 ms) per 27Al–31P pair for phosphate complexes [21]. Mainly three correlation peaks were observed in the through-bond 2D spectrum (Fig. 2), which proved the existence of chemical bond of P–O–[4]Al, P–O–[5]Al, P–O–[6]Al. The peak tops of the cross sections along the 31 P dimension at the [4]Al (42 ppm) and [6]Al (5 ppm) positions seemed to have the different chemical shifts (the 31 P peak position correlated with [4]Al seemed to be slightly lower frequency), implying that each Al site is coordinated with distinct P sites with the different chemical shifts. However, the considerable low intensities of the correlation peaks of P–O–[5]Al and P–O–[6]Al made demonstration of

4229

this presumption difficult. The proportion of the [4]Al, [5] Al, [6]Al species was 75:14:11 in the sum of the 2D MAS J-HMQC spectrum, whereas 57:20:23 in the 1D 27 Al MAS spectrum. Thus, the relative signal intensity more decreased with increasing the Al coordination number. This is probably due to shorter T2 relaxation times of higher coordination Al sites although we did not measured T2 values [6]. Smaller 2JAl–P couplings of P–O–Al with increasing the Al coordination number due to smaller mean P–O–Al angles might also result in suppression of the signal intensity [6]. In order to optimize the experimental condition of the through-space 2D experiment, the dependence of the peak intensity of the 31P spectrum on a contact time ct, a duration of the magnetization transfer from 27Al to 31P, was examined by 1D 27Al ! 31P CP/MAS. Fig. 3(a) shows the CP build-up curve as a function of the carrier frequency of the 27Al irradiation. The 27Al offset frequency affected the optimal ct; the maximum intensity of the 31P spectrum was attained at ct = 2.5, 3.0, 4.5 ms when the 27Al offset was set to 42 ([4]Al position), 20 ([5]Al), 5 ([6]Al) ppm, respectively. Moreover, the relative signal intensity of PO4 depended on the 27Al offset; the best intensity of PO4 was achieved for the 27Al offset of the tetrahedral position. These results come from the fact that the bond length between Al and P ions becomes longer with the higher Al coordination number. The 2D 31P{27Al} CP HETCOR spectrum correlating 31P with 27Al through dipolar couplings is shown in Fig. 3(b). The correlation pattern of this spectrum was essentially identical to that of the MQMAS/ HETCOR spectrum in our previous work [18] because of relatively small CQ of 27Al sites, whereas signal sensitivity of this experiment was about ten times as high as that of the MQMAS/HETCOR spectrum. As shown in Fig. 3(b), the CP/HETCOR spectrum provided the much higherresolved 31P spectrum compared with the 1D 31P MAS spectrum. It was clearly seen that at least four crystallographically non-equivalent P sites with distributions of chemical shifts due to its disordered structure occur in this sample. The four P sites were distinguished by the types of connections between Al and P: one P–O–[4]Al, two P– O–[5]Als and one P–O–[6]Al. The centers of gravity of cross sections along the 31P dimension were at 24 ppm for the P–O–[4]Al peak, 15 and 31 ppm for the P–O–[5]Al peaks and 20 ppm for the P–O–[6]Al peak. It should be noted that the correlation with [4]Al was indeed at lower frequency in the 31P chemical shift than that with [6]Al as expected in the MAS J-HMQC spectrum. Several factors about P structure affect the 31P chemical shift in P-containing glasses. The 31P chemical shift largely depends on the degree of polymerization of PO4 units, denoted as Qn (n = 0–4, n is the number of bridging oxygen per PO4 tetrahedron) [22,23]. However, in this case, the Qn effects can be ignored, because again only Q4 species exists in this sample [17]. Other main possible factors are considered to be the P–O bond length and the P–O–Al bond angle. Unfortunately, however, the relationship between

LETTER TO THE EDITOR

4230

K. Kanehashi et al. / Journal of Non-Crystalline Solids 353 (2007) 4227–4231

Intensity / arbitrary units

4 3 2 1 0 0

5

10

15

Contact time / ms

–30 –20 –10

[6]Al

0

shift / ppm

10 [5]Al

20

27Al

30 40 [4]Al

50 60 10

0

–10

–20

–30

–40

–50

70 80 20

10

0

–10 31P

–20

–30

–40

–50

– 60

shift / ppm

Fig. 3. 27Al ! 31P CP build-up curve (a) and 31P{27Al} CP HETCOR spectrum (b) of amorphous AlPO4. (a) 20 ppm (square), 5 ppm (circle). (b) 27Al carrier frequency: 42 ppm, ct = 2.5 ms.

Al carrier frequency: 42 ppm (triangle),

165

Mean P–O–Al angle / deg

the 31P chemical shift and the bond length has not been well investigated to our knowledge. On the other hand, several correlations of the 31P chemical shift with the bond angle have been reported for various c-AlPO4s [6,8,24–26]. Here, we assume that the 31P chemical shift for the a-AlPO4 is affected mainly by the P–O–Al bond angle as the cAlPO4s, and try to obtain further understanding of the mean P–O–Al bond angles for the a-AlPO4. It has been found that the 31P chemical shift generally decreases with increasing the mean P–O–Al angle [24]. Fig. 4 shows the plots of the 31P chemical shift and the P–O–Al angle for the several c-AlPO4s previously reported. Although these samples are well fitted by their respective straight lines (the correlation coefficient R2 = 0.997 for VPI-5 [6,25], 0.942 for AlPO4-14 [8] and 0.997 for layered AlPO4 (AP2DAO) [26]), they are not well fitted by a single line. When all these data are plotted together, the value of R2 decreases to 0.914. Hence, we omitted data on AP2DAO

27

160 155 150 145 140 135 130 125 0

-5

-10 31P

-15

-20

-25

-30

-35

chemical shift / ppm

Fig. 4. Relationship between 31P chemical shifts and mean P–O–Al bond angles for various crystalline AlPO4. Closed circles and a solid line (R2 = 0.997): VPI-5 [6,21], closed diamonds and a coarse dotted line (R2 = 0.942): AlPO4-14 [8], closed triangles and a fine dotted line (R2 = 0.997): AP2DAO [22]. The open rectangular with a bold line in the figure represents the range of the 31P chemical shift of amorphous AlPO4 in this study.

LETTER TO THE EDITOR

K. Kanehashi et al. / Journal of Non-Crystalline Solids 353 (2007) 4227–4231

for estimation of the mean P–O–Al bond angles for the aAlPO4, because the degree of polymerization of PO4 units for AP2DAO (Q3) is different from the other two samples (Q4). In fact, the data on VPI-5 and AlPO4-14 plotted together still provided a satisfactory linear relationship between the 31P chemical shift and the bond angle (R2 = 0.954). When the data on VPI-5 and AlPO4-14 are combined in a single plot, the straight line is expressed as follows: 31

aðP–O–Al bond angleÞ ¼ 0:802dð P chemical shiftÞ þ 124:

Using this relationship, the mean P–O–Al bond angles of four P sites for the a-AlPO4 were estimated. We used the values of the centers of gravity (not the peak tops) for each cross section for calculating the bond angles to obtain the ‘mean’ values. The mean bond angles were estimated to be 143 ± 0.5° for the P–O–[4]Al, 136 ± 1° and 149 ± 1° for the P–O–[5]Al, 140 ± 0.5° the P–O–[6]Al. It should be here noted that uncertainties of the mean bond angles in this amorphous sample are quite large compared with the reported c-AlPO4s because of larger line widths due to chemical shift distributions and larger errors of integration for each cross section. Especially, the P–O–[5]Al cross sections have larger uncertainties than larger than the other two connectivities, because the signal to noise ratio of this is quite low and it is more difficult to obtain precise values. This estimation indicates that PO4 and AlO5 have especially unique connectivities with the existence of two stable regions of the mean P–O–Al bond angle. These complex connectivities are possibly attributed to difference in the arrangement of water molecules and/or hydroxyl groups coordinated to [5]Al. 4. Conclusions The through-bond and through-space solid-state NMR techniques provided very useful structural information about the Al–P connectivities for the a-AlPO4 as well as the c-AlPO4s. It was demonstrated that the a-AlPO4 has all types of the Al–O–P bonds (i.e. P–O–[4]Al, P–O–[5]Al, P–O–[6]Al) by the thorough-bond J coupling experiment. In addition, at least four P local environments were found to exist in the a-AlPO4 by the through-space dipolar-driven experiment, whereas the conventional 1D 31P MAS spectrum provided poor information about the distinct P sites. Moreover, the mean P–O–Al bond angle of each P site was estimated based on the linear correlation with the 31P

4231

chemical shift. Information about the connectivities using both through-bond and through-space methods as well as direct information on each nucleus (27Al [27], 31P [15,17], 17 O [17]) should be very effective for estimating the 3D framework of amorphous structure. References [1] W.L. Kehl, US Patent 4080311, 1978. [2] C.A. Fyfe, H. Grondey, K.T. Mueller, K.C. Wong-Moon, T. Markus, J. Am. Chem. Soc. 114 (1992) 5876. [3] C.A. Fyfe, K.T. Mueller, H. Grondey, K.C. Wong-Moon, Chem. Phys. Lett. 199 (1992) 198. [4] C.A. Fyfe, K.T. Mueller, H. Grondey, K.C. Wong-Moon, J. Phys. Chem. 97 (1993) 13484. [5] C. Fernandez, C. Morais, J. Rocha, M. Pruski, Solid State Nucl. Magn. Reson. 21 (2002) 61. [6] D. Massiot, F. Fayon, B. Alonso, J. Trebosc, J.-P. Amoureux, J. Magn. Reson. 164 (2003) 160. [7] C.A. Fyfe, K.C. Wong-Moon, Y. Huang, Zeolites 16 (1996) 50. [8] C.A. Fyfe, H. Meyer zu Altenschildesche, K.C. Wong-Moon, H. Grondey, J.M. Chezeau, Solid State Nucl. Magn. Reson. 9 (1997) 97. [9] L. Delevoye, C. Fernandez, C.M. Morais, J.-P. Amoureux, V. Montouillout, J. Rocha, Solid State Nucl. Magn. Reson. 22 (2002) 501. [10] J.W. Wiench, M. Pruski, Solid State Nucl. Magn. Reson. 26 (2004) 51. [11] G. Mali, J. -P. Amoureux, V. Kaucˇicˇ, Phys. Chem. Chem. Phys. 2 (2000) 5737. [12] S. Caldarelli, A. Meden, A. Tuel, J. Phys. Chem. B 103 (1999) 5477. [13] C.A. Fyfe, K.C. Wong-Moon, Y. Huang, H. Grondey, J. Am. Chem. Soc. 117 (1995) 10397. [14] H.-M. Kao, C.P. Grey, J. Magn. Reson. 133 (1998) 313. [15] E.R.H. van Eck, A.P.M. Kentgens, H. Kraus, R. Prins, J. Phys. Chem. 99 (1995) 16080. [16] Y. Huang, D. Machado, Microporous Mesoporous Mater. 47 (2001) 195. [17] K. Kanehashi, K. Saito, Chem. Lett. 31 No.7 (2002) 668. [18] T. Iijima, K. Kanehashi, K. Saito, M. Hatakeyama, T. Nemoto, T. Shimizu, S. Ohki, Chem. Lett. 34 (10) (2005) 1380. [19] J.-P. Amoureux, C. Fernandez, S. Steuernagel, J. Magn. Reson. A 123 (1996) 116. [20] A.J. Vega, J. Magn. Reson 96 (1992) 50. [21] J.W. Akitt, Progress Nucl. Magn. Reson. Spectrosc. 21 (1988) 1. [22] K.J.D. Mackenzie, M.E. Smith, Multinuclear Solid-State NMR of Inorganic Materials (2002). [23] R.J. Kirkpatrick, R.K. Brow, Solid State Nucl. Magn. Reson 5 (1995) 9. [24] D. Mu¨ller, E. Jahn, G. Ludwig, U. Haubenreisser, Chem. Phys. Lett. 109 (1984) 332. [25] G. Cheetham, M.M. Harding, Zeolites 16 (1996) 245. [26] A. Tuel, V. Gramlich, Ch. Baerlocher, Microporous Mesoporous Mater. 47 (2001) 217. [27] H. Kraus, R. Prins, A.P.M. Kentgens, J. Phys. Chem. 100 (1996) 16336.