ARTICLE pubs.acs.org/JPCA
Conformations of Trimethyl Phosphite: A Matrix Isolation Infrared and ab Initio Study N. Ramanathan, K. Sundararajan, Bishnu Prasad Kar, and K. S. Viswanathan* Chemistry Group, Indira Gandhi Centre for Atomic Research, Kalpakkam 603 102, India ABSTRACT: The conformations of trimethyl phosphite (TMPhite) were studied using matrix isolation infrared spectroscopy. TMPhite was trapped in a nitrogen matrix using an effusive source maintained at two different temperatures (298 and 410 K) and a supersonic jet source. The experimental studies were supported by ab initio computations performed at the B3LYP/6-31++G** level. Computations identified four minima for TMPhite, corresponding to conformers with C1(TG(G(), Cs(TG+G ), C1(G(TT), and C3(G(G(G() structures, given in order of increasing energy. Computations of the transition state structures connecting the Cs(TG+G ) and C1(G(TT) conformers to the global minimum C1(TG(G() structure were also carried out. The barriers for the interconversion of Cs(TG+G ) and C1(G(TT) to the ground state C1(TG(G() conformer were 0.2 and 0.6 kcal/mol, respectively. Comparison of conformational preferences of TMPhite with the related carbon compound, trimethoxymethane, and the organic phosphate, trimethyl phosphate, was also made using natural bond orbital analysis.
1. INTRODUCTION Hyperconjugative interactions have received considerable attention because of their importance in determining structure and reactivity in organic compounds. Importantly, the conformations of molecules with O P O and O C O linkages decide the overall structures of biologically important molecules such as phosphates and carbohydrates.1,2 In particular, study of the reactions of phosphorus esters evoked interest as these reactions serve as model systems for understanding many biological processes. For example, hyperconjugative interactions decide the reactivity in the hydrolysis of ortho esters, the reactivity being largely determined by the lone pair alignment with respect to the leaving group.3,4 It was also suggested that the hyperconjugative effect impacts phosphotransfer potential by stabilizing groundstate geometries that are closer in energy and geometry to the hydrolyzed products.5 The conformations of trimethyl phosphate (TMP), the first member of the series of organic phosphate, has been reasonably well studied.6 10 In an effort to understand the reasons for the conformational preferences in the organic phosphates, we also studied the conformations of a number of acetals and ketals,11 15 which have O C O segments, similar to the O P O segments in the phosphates. Some related methoxysilanes were also studied.16 19 It was concluded from these studies that the hyperconjugative interactions are important in determining the conformations of such molecules. In addition to the O P O segments, the organic phosphates have an additional PdO group, that can play an important role in conformational preferences. To address this issue, it was thought r 2011 American Chemical Society
interesting to study trimethyl phosphite (TMPhite), which lacks a PdO group. A comparison of the conformations of TMP and TMPhite would highlight the role of the PdO group in conformational preferences. In addition, the motivation for the study of TMPhite is manifold. TMPhite itself is industrially important; it is used for the synthesis of phosphorus doped diamond films, which has important application in electronic industries.20 It is also used as an electrolyte additive for improving the performance of electrolytic systems.21 One of the earliest works on the spectroscopy of TMPhite was that of Nyquist, who, using infrared and Raman spectroscopy, suggested that the symmetry of the molecule was less than C3v or C3.22 However, this work did not specifically discuss the structure of the different conformers of TMPhite. Recently, Belyakov et al.,23 using electron diffraction and ab initio calculations, studied the conformations of TMPhite, and concluded that the ground state conformer has a C1(TG(G() structure. However, to the best of our knowledge, no detailed study on the conformations of TMPhite has been reported in the literature. We therefore studied the conformers of TMPhite using matrix isolation infrared spectroscopy. We performed ab initio computations on the structures, energies and vibrational frequencies for the various conformers of TMPhite, at the B3LYP/6-31++G** level, to corroborate our experimental results. We also performed Received: June 8, 2011 Revised: July 22, 2011 Published: July 25, 2011 10059
dx.doi.org/10.1021/jp205368k | J. Phys. Chem. A 2011, 115, 10059–10068
The Journal of Physical Chemistry A
ARTICLE
Table 1. Selected Structural Parametersa of the Conformers of TMPhite Calculated at the B3LYP/6-31++G** Level C1(TG(G() Cs(TG+G ) C1(TTG() C3(G(G(G()
parameter O2 P1
1.66
1.66
1.64
1.65
O3 P1 O4 P1
1.68 1.64
1.66 1.65
1.68 1.65
1.65 1.65
O2 C5
1.43
1.43
1.44
1.43
O3 C6
1.43
1.43
1.43
1.43
O4 C7
1.44
1.44
1.44
— O2P1O3
1.43
95.9
92.4
96.8
97.9
— O2P1O4
98.1
102.8
105.4
97.9
— O3P1O4
103.7
102.8
101.6
97.9
— C5O2P1 — C6O3P1
118.7 118.0
117.5 117.5
127.9 117.6
118.6 118.6
— C7O4P1
118.6
123.3
122.7
125.3
tor — C5O2P1O3
72.3
169.9
20.9
83.1
tor — C6O3P1O2
168.9
169.2
170.9
177.9
tor — C7O4P1O2
45.5
47.8
49.6
83.1
tor — C7O4P1O3
52.7
47.7
50.9
177.7
a
Bond distances in Å; bond angles and torsion angles in degrees. Torsion angles of the fragment ABCD denote the angle between ABC and BCD planes.
NBO analysis for understanding the reasons for conformational preferences in TMPhite. A comparison of the conformations of TMPhite with trimethoxymethane (TMM) and TMP was also made.
were carried out by maintaining the effusive nozzle at two different temperatures, viz. 298 and 410 K. In the hot nozzle experiments, the nozzle was heated to 410 K, over a length of 35 mm, just prior to the exit of the gas mixture. The hot-nozzle experiments were performed to increase the population of the higher energy conformers. We also performed a few experiments using a pulsed supersonic jet source (Parker Hannifin Corporation), with a pulse width of 25 ms and a period of 30 s between pulses, to deposit the matrix. A stagnant pressure of ∼1.3 atm was used in these experiments. At this deposition rate, the temperature of the cryotip did not rise. The supersonic jet deposition was designed to suppress the population of the higher energy conformers.
2. EXPERIMENTAL SECTION Matrix isolation experiments were carried out using a Leybold AG helium-compressor-cooled closed cycle cryostat. The details of the vacuum system and experimental setup are described elsewhere.6 9 TMPhite (Merck, >98%) was used without further purification. The sample was, however, subjected to several freeze pump thaw cycles before performing the experiments. Nitrogen (INOX; purity: 99.9995%) and argon (INOX; purity: 99.9995%) were used as the matrix gases. The sample and the matrix gas were mixed in appropriate proportions using manometric techniques; this gas mixture was then allowed to stream out of a single jet nozzle and deposited onto a cold KBr substrate maintained at ∼12 K. Typical sample to matrix ratios in all our experiments was 1:1000. Deposition was carried out at the rate of ∼3 mmol/h and a typical deposition lasted for ∼1 h. The spectra were recorded using a BOMEM MB 100 FTIR spectrometer with a spectral resolution of 1 cm 1. After a spectrum was recorded, the matrix was warmed to ∼32 K, kept at this temperature for 15 min, and recooled to ∼12 K. The spectra of the matrix thus annealed were again recorded. Experiments
3. COMPUTATIONAL DETAILS Computations were carried out for the different conformations of TMPhite using the Gaussian G94W suite of programs on a Pentium IV machine,24 using a B3LYP/6-31++G** level of theory. All of the structures were optimized without imposing any symmetry constraints during the optimization process. Harmonic vibrational frequency calculations were performed using analytical gradients, first, to ensure that the computed structures did correspond to minima on the potential surface and also to assign the vibrational features observed in the experiments. When using the computed frequencies to assign the experimental infrared features, the computed frequencies were scaled on a mode-by-mode basis. For example, to arrive at the scaling factor, in the C O stretching region near 1030 cm 1, the experimentally observed strongest feature at 1025.1 cm 1 was correlated with the strongest computed feature for the ground state conformer in this region. A scaling factor of 0.9880 was required to bring the two values in agreement. This scaling factor was used to scale other features of the ground and higher energy conformers, in this region. Likewise, a scaling factor of 1.032 was calculated for the P O stretching region. A mode-by-mode
Figure 1. Matrix isolation infrared spectra of TMPhite trapped in a N2 matrix; spanning the region 1100 1000 cm 1 (grid A) and 800 700 cm 1 (grid B). Spectra of TMPhite deposited using (a) a room temperature effusive source, (b) an effusive source at 410 K, and (c) a supersonic nozzle source. The spectra shown here are those recorded in preannealed matrixes at 12 K. The typical sample to matrix ratio is 1:1000. The feature marked with the label M is due to methanol.
10060
dx.doi.org/10.1021/jp205368k |J. Phys. Chem. A 2011, 115, 10059–10068
The Journal of Physical Chemistry A
ARTICLE
Figure 2. Structure of the conformers of TMPhite. Relative energies of the conformers (kcal/mol) with respect to the ground state conformer are given against each structure.
Table 2. Total Energies, Relative Energies, and Dipole Moments of Different Conformers of TMPhite Computed at the B3LYP/631++G** Level conformer
degeneracy
total energiesa (hartrees)
relative energya (kcal/mol)
dipole moment (debye)
C1(TG(G()
6
686.681697
0.00
1.64
Cs(TG+G ) C1(TTG()
3 6
686.679999 686.679440
1.06 1.42
0.46 2.81
C3(G(G(G()
2
686.677955
2.35
1.60
TSb
686.679708
1.25
0.21
TSc
686.678431
2.05
2.72
Energies corrected for ZPE. b Transition state connecting Cs(TG+G ) and C1(TG(G() structures. c Transition state connecting C1(TTG() and C1(TG(G() strutures. a
scaling is necessary, in our opinion, to account for the varying influence that the matrix has, on the vibrational frequencies in the different regions of the infrared spectrum.25,26 The computed
frequencies and intensities were used to simulate a vibrational spectrum using the SYNSPEC program, 27 by assuming a Lorentzian line profile with a full-width half-maximum of 1 cm 1. 10061
dx.doi.org/10.1021/jp205368k |J. Phys. Chem. A 2011, 115, 10059–10068
The Journal of Physical Chemistry A
ARTICLE
Table 3. Scaled Computed (B3LYP/6-31++G**) and Experimental Vibrational Frequencies (in a N2 Matrix) of the C1(TG(G(), Cs(TG+G ), and C1(TTG() Conformers of TMPhitea frequency computational mode assignments b
P O stretch
O C stretchc
CH3 rockd CH3 scissord
CH3 symmetric stretchd
CH3 asymmetric stretchd
a
C1(TG(G()
Cs(TG+G )
C1(TTG()
720.0 (179)
729.6 (134)
711.0 (173)
760.1 (150)
742.4 (152)
727.5 (135)
773.2 (79) 1025.1 (326)
784.5 (111) 1017.1 (257)
745.0 (88) 1020.0 (195)
1040.8 (308)
1037.0 (462)
1046.9 (388)
1030.8, 1035.0
1068.7 (88)
1082.3 (4)
1062.3 (128)
1066.8
1195.6 (21)
1195.5 (40)
1186.4 (6)
1174.5, 1176.3, 1183.7, 1187.7
1199.2 (16)
1201.3 (6)
1187.6 (6)
1494.9 (3)
1493.7 (8)
1494.8 (0.2)
1496.1 (9)
1498.6 (4)
1498.1 (8)
1498.3 (11) 1507.7 (6)
1498.8 (18) 1506.3 (6)
1500.9 (10) 1506.9 (9) 1509.6 (5)
experimental (nitrogen matrix) 731.4, 736.7, 742.2 760.1 775.8, 780.6 1022.2, 1025.1
1455.6, 1459.0, 1468.0, 1471.2, 1472.4, 1474.1
1510.0 (10)
1507.8 (8)
1510.2 (7)
1511.1 (3)
1511.6 (10)
3019.8 (79)
3020.8 (118)
3022.0 (71)
2842.0, 2900.7,
3021.5 (59)
3021.8 (18)
3035.4 (29)
2924.2
3053.4 (47)
3048.2 (49)
3046.7 (71)
3085.8 (36)
3085.6 (30)
3087.6 (33)
3095.9 (30) 3131.5 (26)
3085.7 (40) 3123.7 (28)
3110.4 (11) 3123.7 (45)
3134.4 (19)
3138.2 (17)
3133.8 (24)
3139.3 (21)
3138.2 (26)
3134.5 (14)
3149.8 (19)
3146.4 (19)
3148.6 (15)
2944.3, 2958.3, 2975.1, 3001.3, 3011.9, 3018.7
The intensities, in km/mol, are given in paranthesis. b Scaling factor was 1.032. c Scaling factor was 0.9880. d No scaling factors were used.
The structure and energy of the transition states connecting the ground state and higher energy conformers were also computed. To understand the nature of the orbital interactions in determining conformational preferences, natural bond orbital analysis (NBO, version 3.1) was performed, invoked through Gaussian G94W.28 The use of NBO has been shown to provide a detailed understanding of conformational preferences in several systems.29 34 To ascertain the role of orbital interactions, we also performed calculations, where specific interactions were deleted from the function space, using the NBODEL option.
4. RESULTS AND DISCUSSION Experimental Details. Figure 1 shows the matrix isolation infrared spectra of TMPhite trapped in nitrogen. The spectra span the region 1100 1000 cm 1 (grid A) and 800 700 cm 1 (grid B), which correspond to C O stretching and P O stretching vibrations, respectively, of TMPhite. Figure 1, trace a, shows the infrared spectra of matrix isolated TMPhite deposited using an effusive source at room temperature. Trace b, in the same figure, shows the spectra deposited using a hot nozzle effusive source, where the nozzle temperature was maintained at 410 K, whereas trace c shows the spectra deposited using a supersonic source. The main spectral features of TMPhite occur at 1066.8, 1035.0, 1030.8, 1025.1, 1022.2, 780.6, 775.8, 760.1, 742.2, 736.7, and 731.4 cm 1. Spectra recorded with a hot nozzle and
supersonic jet source essentially appeared similar to that obtained using a room temperature effusive source. When the matrix was annealed at 32 K, no perceptible changes in the features were observed. In some of the spectra, particularly those recorded with a hot effusive nozzle, features due to methanol were observed, due to the reaction of TMPhite with water; water being an inevitable impurity in any matrix isolation experiment. Where they appear, the features due to methanol are marked with the label M. Computations. We performed geometry optimizations at the B3LYP/6-31++G** level of theory. At this level of theory, we obtained four minima for TMPhite, corresponding to conformations with C1(TG(G(), Cs(TG+G ), C1(TTG() and C3(G(G(G() symmetries, given in order of increasing energy.35 Selected structural parameters of the four conformers are given in Table 1. The corresponding structures are shown in Figure 2, together with the energies of each conformer, relative to the ground state conformer. The energies of the various conformers of TMPhite, corrected for ZPE, are also given in Table 2. We computed the transition state structures connecting the ground state to the two higher energy conformers, Cs(TG+G ) and C1(TTG(). Vibrational frequency calculations confirmed that the computed transition state structures were indeed firstorder saddle points, with the wavenumber of the single imaginary mode being ∼60 cm 1, for each of the saddle point structures. Intrinsic reaction coordinate calculations were performed to confirm that the transition structures did indeed connect the 10062
dx.doi.org/10.1021/jp205368k |J. Phys. Chem. A 2011, 115, 10059–10068
The Journal of Physical Chemistry A minima in question. The energies of these transition state structures relative to the ground state conformer are given in Table 2. The barrier for interconversion from the Cs (TG+G ) and C1(TTG() to the ground state C1(TG(G() conformer were calculated to be 0.2 and 0.6 kcal/mol, respectively, after correcting for ZPE. Vibrational Assignments. The spectral features at 1066.8, 1030.8, 1025.1, 780.6, 760.1, and 736.7 cm 1 marked in Figure 1a are essentially those due to the C O and P O stretch of the ground state conformer, which is indicated by our computations to have a C1(TG(G() structure. The computed and scaled frequencies for these modes of the conformer, agree well with the spectral features obtained with a room temperature effusive source, as can be seen from Table 3. Some weak features were also observed at 1035.0, 1022.2, 775.8, 742.2, and 731.4 cm 1, which can possibly be assigned to the higher energy conformers of TMPhite, a discussion of which is presented later. In addition to the frequencies of the P O and C O stretch, Table 3 also presents the unscaled frequencies of the computed vibrational features for the CH3 bending and stretching vibrations. Since these vibrational modes for the different conformations are not resolved, we have made no attempt to assign these features for the different conformations. It can be seen from Figure 2 and Table 2 that the conformers with structures Cs(TG+G ), C1(TTG(), and C3(G(G(G() occur at 1.06, 1.42, and 2.35 kcal/mol, respectively, above the ground state C1(TG(G() conformer. At room temperature, the ratio of the population of the four conformers, C1(TG(G(), Cs(TG+G ), C1(TTG(), and C3(G(G(G(), would be 84.7:7.0:7.8:0.5. The C3(G(G(G() conformer has negligible population (∼0.5%) to be of any experimental significance. The combined population of the other two higher energy conformers, Cs(TG+G ) and C1(TTG(), are about equal and add up to ∼15% at room temperature; hence, the contributions of these higher energy conformers to the room temperature effusive spectra cannot be ruled out. To clearly discern the features of the higher energy conformers, we carried out experiments using a hot nozzle source, where the nozzle temperatures were maintained at 410 K. At 410 K, the population of the four conformers was calculated to be 75.2:10.2:13.2:1.4. The combined population of the higher energy conformers thus amounts to ∼25%, which is a definite change from the population distribution at room temperature. If the weak features at 1035.0, 1022.2, 775.8, 742.2, and 731.4 obtained in the room temperature effusive spectra are due to higher energy conformers, the intensities of these features should have increased in the hot nozzle experiments. However, no discernible change in intensities was observed for these features when the hot nozzle was used. Furthermore, these features did not decrease in intensity in the experiments where a supersonic jet source was used for depositing the matrix. It is therefore unlikely that the weak features in question are due to higher energy conformers and are likely sitesplit features of the ground state conformer, C1(TG(G(). To confirm if the splitting is really due to site effects, a few experiments were also conducted using Ar as a matrix. It was found that similar to the N2 matrix splitting was also observed in Ar, though the structure of the splitting was slightly different in Ar compared with that in N2. To avoid a repetitive display of the spectra, the spectra recorded using Ar as matrix is not shown. Even though, the overall population of the higher energy conformers is rather significant (∼15%) at room temperature, features of the higher energy conformers could not be discerned in
ARTICLE
Figure 3. Comparison of the experimental and computed spectra of TMPhite. The spectra span the region 1100 1000 cm 1 (grid A) and 800 700 cm 1 (grid B). (a) Computed spectra of C1(TG(G() conformer at 298 K. (b) Matrix isolation infrared spectra of TMPhite recorded using room-temperature effusive source at 12 K. (c) Same as in b but recooled to 12 K after annealing at 32 K. The typical sample to matrix ratio is 1:1000.
the room temperature or in hot nozzle effusive experiments. It may be recalled that the barriers for conformer interconversion were computed to be
