Hydrogen bonding effects on infrared and Raman spectra of drug molecules

Spectrochimica Acta Part A 66 (2007) 213–224

Hydrogen bonding effects on infrared and Raman spectra of drug molecules a ˚ Laban Bondesson a,b,c , Kurt V. Mikkelsen c , Yi Luo a,∗ , Per Garberg b , Hans Agren a

Theoretical Chemistry, Royal Institute of Technology, SCFAB, SE-106 91 Stockholm, Sweden b Biovitrum AB, SE-112 76 Stockholm, Sweden c Department of Chemistry, University of Copenhagen, DK-2100 Copenhagen Ø, Denmark

Received 6 September 2005; received in revised form 27 January 2006; accepted 23 February 2006

Abstract Infrared and Raman spectra of three drug molecules, aspirin, caffeine and ibuprofen, in gas phase and in aqueous solution have been simulated using hybrid density functional theory. The long range solvent effect is modelled by the polarizable continuum model, while the short range hydrogen bonding effects are taken care of by the super-molecular approach with explicit inclusion of water molecules. The calculated spectra are found to compare well with available experimental results. The agreement obtained make grounds for proposing theoretical modeling as a tool for characterizing changes in the bonding environments of drug molecules in terms of particular variations in their IR and Raman spectra. © 2006 Elsevier B.V. All rights reserved. Keywords: Drug molecules; Infrared; Raman; Solvent effect; Hydrogen bonding; DFT

1. Introduction Infrared (IR) and Raman spectroscopies are important and well-established tools for identifying drug molecules [1–4] and for quantitative analysis of drugs [2,1]. Despite the general utility of these techniques, it is often an arduous and time-consuming task to interpret the spectroscopic data without having appropriate microscopic information at hand. Tools that accurately can predict the spectra and interpret them in terms of molecular structure or in terms of so-called structure–property relationships would greatly promote the usefulness of these characterization spectroscopies. For instance, in order to interpret specific frequency shifts one needs to understand which vibrational modes correspond to the particular measured experimental frequencies. Theoretical modeling of the spectra based on specified molecular structures could constitute such tools and lead to the basic understanding provided the modeling is sufficiently accurate and reliable from case to case. Modern theoretical methodologies possess indeed a large inherent potential for simulating spectra of different kinds and show often high accuracy for molecules in the gas phase. How-

∗

Corresponding author. Tel.: +46 8 55378414; fax: +46 8 55378590. E-mail address: [email protected] (Y. Luo).

1386-1425/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.saa.2006.02.045

ever, most of the experiments can only be conducted when the sample molecules are in crystal form or dissolved in a liquid. The spectra obtained in these aggregation forms show in general distinct changes compared to corresponding gas phase spectra. These changes could involve general broadening and distortion of the spectral bands, as well as relative shifts in band energy and intensity, depending on which type of interaction prevails with the surrounding medium. In case of strong bonding completely new bands can appear. Therefore, in order to maintain the utility of theoretical modeling in relating spectral properties with structure, it is essential to extend its applicability to the solvent or solid cases. In this paper we will address precisely that, and investigate how the calculated spectra changes if molecules are dissolved in a medium. This could be a polar medium represented by a dielectric continuum, or a medium with hydrogen bonds to the solute which is the case for many crystals and liquids. The latter case is modelled by the semicontinuum approach in which shells of solvent molecules are accounted for explicitly. We believe that a successful modelling of the intermolecular or medium perturbations of the IR and Raman spectra can have significant ramifications concerning the important problem of drug solubility. In this first investigation we have chosen to study three drugs, namely aspirin, caffeine and ibuprofen as the representative model systems. These three molecules possess typical functional groups that can be found

214

L. Bondesson et al. / Spectrochimica Acta Part A 66 (2007) 213–224

in most drugs. Furthermore, for the selected molecules, one can find few experimental results to compare with. 2. Computational details Four different types of calculations have been performed for each of the investigated compounds. The first two started with a gas phase geometry optimization. The optimized geometry was then used for an IR and Raman frequency calculation in gas phase (GS) and in a polarized continuum model calculation with water as solvent (PCM). The last two types of calculations were also performed in two steps. The first step is a geometry optimization with explicit water molecules at positions where the molecules are assumed to form hydrogen bonds. The second step is an IR and Raman frequency calculation for the optimized structure with (HB-PCM) and without (HB) a polarized dielectric continuum. For all calculations Gaussian 03 [5] ab initio programs have been used. In this study we have used the B3LYP functional as it has frequently been proved to produce results at the explicitly correlated MP2 level of theory for hydrogen bonding calculations [6]. The 6−31+ G(d+, p) basis set is employed for both geometry optimization and frequency calculations. 3. Results and discussion 3.1. Structure information The structures of the investigated molecules when they are hydrogen bonded to water are shown in Fig. 1. The bond length for the atom pairs that have one or two hydrogen bonded atoms are assumed to change most and will therefore have an noticeable impact on the spectra. In comparison with the gas phase result, for aspirin molecule the hydrogen bonds have increased the car˚ the C O distance in the carboxylic bonyl C O bond by 0.007 A, ˚ the O H bond by 0.041 A ˚ and the C O acid group by 0.013 A, ˚ For the other non hydrogen bonded atoms in bond by 0.026 A. the aspirin molecule the bond length changes only slightly. For ˚ the caffeine molecule the C O1 bond is increased by 0.005 A, but the bond length of the second carbonyl group C O2 actually decreases by the same amount. The last hydrogen bonded atom, N3 , has two C N bonds, the one that is bonded to the carbon in the imidazole ring hardly change its bond length irrespective of how the molecule is optimized. The bond length of another N C ˚ Overall, for the caffeine molecule bond is increased by 0.003 A. one can see that the bond length does not drastically change with the inclusion of the hydrogen bonded. For the ibuprofen molecule there are three atoms in the carboxyl group that can form hydrogen bonds. All of them have shorter distance with the hydrogen bonds. The distances for the O H bond, the C O bond and the C O bond are decreased by 0.024, 0.016 and 0.016 ˚ respectively. A, 3.2. Aspirin The aspirin molecule has been investigated in numerous studies with different objectives, and also in several different stud-

Fig. 1. Structure of aspirin, caffeine, and ibuprofen with hydrogen bonded water molecules.

ies addressing infrared and Raman spectroscopy (see Ref. [2] and references therein). Only a few of the latter though deals with theoretical studies of the spectroscopy[2,3,7]. The study by Binev et al. [2] was both experimental and theoretical, the former utilized FT-IR spectroscopy applied on aspirin dissolved in hexadeutero-dimethyl sulfoxide (DMSO-d6 ). The presented spectra ranged between 1100 and 2000 cm−1 . Their theoretical work consisted of a HF/3-21G frequency calculation in gas phase. The study by Boczar et al. [3] also utilized FTIR experiment. However, in this case the aspirin sample was ground with KBr and pressed, the investigated spectra ranged

L. Bondesson et al. / Spectrochimica Acta Part A 66 (2007) 213–224

215

Table 1 High vibrational frequencies, IR and Raman intensities (in parenthesis) of aspirin calculated from different approaches Description ␯ (OH) ␯ (CHϕ ) ␯ (CHϕ ) ␯ (CHϕ ) ␯ (CHϕ ) ␯ (CH3a ) ␯ (CH3a ) ␯ (CH3s ) ␯ (OH)

Gas phase (103,125)a

3763 3232 (2,111) 3213 (3,150) 3203 (10,123) 3187 (2,67) 3170 (6,67) 3130 (3,49) 3061 (2,163) NA

PCM

HB + PCM

HB

3589 (481,237) 3203 (6,361) 3179 (5,597) 3172 (1,297) 3158 (0,203) 3167 (5,130) 3127 (1,122) 3057 (0,360) NA

(653,102)w

3394 3228 (4,164) 3222 (0,93) 3207 (8,139) 3193 (2,63) 3172 (6,63) 3129 (2,50) 3061 (1,165) 2956 (2019,252)

3373 (931,227)w 3224 (3,293) 3188 (8,462) 3175 (1,450) 3163 (0,204) 3169 (5,131) 3126 (1,118) 3058 (0,363) 2906 (2897,401)

w: Interacting modes with water molecules; ␦: in-plan bending; ␯: stretching; ␥: out-of-plan bending; ␶: torsion. a

Frequency (IR intensity, Raman intensity).

between 70 and 4000 cm−1 . They also presented a Raman spectrum for the aspirin molecule. Their theoretical study was performed in a different manner with semi-empirical fitting, see reference [3] for details, the geometry used for their calculation is achieved by a DFT/6−31++G(d, p) gas phase geometry optimization. 3.2.1. High frequencies For the high frequencies in aspirin (see Table 1) there are only one significant peak for the calculated IR spectra in gas phase (see Fig. 2) at frequency 3763 cm−1 corresponds to O H

Fig. 2. IR and Raman spectra of aspirin in high frequency region. Solid line: calculations. Dotted line: experiment (see text for reference).

stretching. The corresponding frequency for the other computational methods are considerably lower and are decreased by 174 cm−1 for PCM, 370 cm−1 for HB and 390 cm−1 for HBPCM. When looking at the absorbance of the peak one has to bear in mind that pure water is interacting with the two in the latter methods. The experimental data obtained in the study of Binev et al. [2] shows only a broad peak at this frequency, while in the study of Boczar et al. [3] this peak is missing. The HB calculation also reveals another peak at the frequency 2956 cm−1 , which is shifted down to 2906 cm−1 when both HB and PCM are included. This peak is associated with vibrations that are connecting both water and aspirin. In the experimental study of Binev et al. [2] with DMSO-d6 as solvent an absorbance of this magnitude was not seen at this frequency, while in the study by Boczar et al. [3], investigating the aspirin dimer, a broad peak was observed at the frequency of 2870 cm−1 . In that study, like the results of HB and HB + PCM calculations, the effects of hydrogen bonding are distinguished. For the rest of the frequencies in this region the absorbances are small and the frequencies are similar for the different computational methods. The stretching for these frequencies relates to the C H bonds in the benzene ring and the CH3 groups. The calculated frequencies are generally higher than the experimental results [2,3]. However, the agreement can be improved if a scaling factor is used. One can also see that the frequency is lowered with a few wave numbers if PCM is added to the gas phase calculation. The Raman intensities at the high frequency region of aspirin are quite high. In Fig. 2 one can see how the O H stretch changes dependent on which kind of calculation method that is used. One can also see how small the changes of the frequencies are for the C H stretching in the region 3000–3300 cm−1 . The peaks with frequency higher than 3400 cm−1 obtained from the HB and HB + PCM calculations relate only to pure water vibrations. Compared to the experimental data of Boczar et al. [3] our calculated values seem to be slightly overestimated from all four calculations. 3.2.2. Middle frequencies The middle region follows the same pattern as the high frequency region, that is the unpolar groups show only small differences in frequency between various methods employed, and the major difference in frequency are present at the C O and O H groups. As one can see in Fig. 3 and Table 2 there are

216

L. Bondesson et al. / Spectrochimica Acta Part A 66 (2007) 213–224

five motions that have a detectable IR absorbance and Raman intensity with a noticeable difference in frequency. These are:

Fig. 3. IR and Raman spectra of aspirin in middle frequency region. Solid line: calculations. Dotted line: experiment (see text for reference).

(i) C O stretching with a frequency of 1828 cm−1 in the gas phase. This frequency is lowered by 15 cm−1 with PCM, by 32 cm−1 with HB, and by 47 cm−1 with HB + PCM. The experimental frequency by Binev et al. [2] where DMSO-d6 is used as solvent is around 1767/1753 cm−1 , which is slightly smaller than the calculated PCM value of 1813 cm−1 . In the experiment of Boczar et al. [3] for the aspirin dimer the frequency becomes 1754 cm−1 , which should be comparable with the HB calculated result of 1781 cm−1 . (ii) C O stretching combined with O H bending, for which the gas phase calculation gives a frequency of 1783 cm−1 and is lowered by 20, 51 and 77 cm−1 , respectively, for PCM, HB and HB + PCM calculations. The experimental frequency by Binev et al. [2] for aspirin in DMSO-d6 is 1706 cm−1 which should be compared to the value of 1763 cm−1 for the PCM calculation without HB. The experimental data by Boczar et al. [3] when HB are present in the dimer form of aspirin the frequency becomes 1691 cm−1 which is compared well to the calculated value of 1606 cm−1 from HB + PCM. (iii) O H bending at frequency 1367 cm−1 in gas phase which is only decreased by 3 cm−1 from the PCM calculation. However, for the two calculations including HB the frequency is increased by 116 and 106 cm−1 , respectively, for HB and HB + PCM. One can also note that both the IR absorbance and Raman intensity are decreased for the calculations with HB which are not the case for most of the

Table 2 Middle vibrational frequencies, IR and Raman intensities (in parenthesis) of aspirin calculated from different approaches Description

Gas phase

PCM

HB

HB + PCM

␯ (C O) ␯ (C O)+ ␥ (OH) ␯ (CCϕ )+ ␦ (CHϕ ) ␯ (CCϕ )+ ␦ (CHϕ ) ␦ (CHϕ ) ␦ (CHϕ ) ␦ (CH3a ) ␦ (CH3a ) ␦sCH3 ␥ (OH) ␦ (HO)+ ␯ (CCϕ ) ␦ (CHϕ ) ␯ (COC)+ ␦ (CHϕ ) ␥ (OH) ␦ (CHϕ )+ ␥ (OH)+ ␦CH3 ␦ (CHϕ )+ ␥ (OH) ␦ (CHϕ )+ ␥ (OH) ␯ (CCϕ )+ ␦ (CHϕ ) ␶ (CH3 ) ␦ (CHϕ )+ ␶ (CH3 ) ␥ (CH3 ) ␥ (CHϕ ) ␥ (OH) ␥ (CHϕ ) ␥ (CHϕ )

1828 (215,6) 1783 (410,76) 1649 (45,53) 1617 (16,14) 1517 (39,5) 1480 (8,3) 1476 (15,4) 1471 (14,6) 1395 (64,1) 1367 (93,12) 1344 (14,5) 1296 (8,2) 1239 (4,21) 1210 (487,11) 1198 (168,25) 1180 (84,11) 1150 (46,1) 1074 (165,5) 1063 (10,25) 1059(11,4) 1018 (62,2) 993 (0,0)

1813 (482,23) 1763 (840,354) 1642 (111,267) 1615 (29,54) 1511 (75,18) 1473 (31,7) 1465 (14,10) 1460 (24,16) 1388 (71,4) 1364 (165,59) 1344 (40,20) 1290 (14,5) 1235 (11,100) 1034 (490,33) 1190 (728,63) 1173 (47,32) 1158 (48,45) 1120 (103,13) 1054 (7,5) 1061 (4,135) 1013 (127,5) 991 (0,0)

1796 (276,9) 1732 (404,93) 1648 (25,54) 1618 (13,11) 1522 (41,6) 1480 (8,2) 1474 (13,5) 1470 (16,7) 1399 (70,2) 1483 (5,1) 1353 (95,9) 1301 (16,0) 1245 (348,3) 1314 (451,32) 1226 (171,39) 1187 (14,7) 1165 (14,3) 1101 (105,2) 1061 (9,5) 1065 (1,26) 1023 (45,3) 1001 (0,0)

1781 (515,24) 1706 (775,307) 1643 (92,255) 1617 (12,26)w 1515 (74,15) 1472 (22,6) 1463 (13,9) 1459 (29,17) 1391 (81,4) 1473 (3,4) 1349 (67,24) 1295 (433,85) 1232 (234,15) 1286 (386,48) 1215 (547,128) 1169 (25,13) 1153 (28,7) 1094 (202,7) 1055 (16,4) 1062 (1,106) 1019 (82,10) 995 (2,1) 992 (9,0) 972 (9,1) 933 (44,7)

974 (3,0) 926 (40,3)

968 (6,1) 918 (89,5)

978 (4,0) 938 (20,3)

L. Bondesson et al. / Spectrochimica Acta Part A 66 (2007) 213–224

other compared frequencies. The experimental results by Binev et al. [2] who used DMSO-d6 as solvent a peak at frequency 1339 cm−1 was presented which compares well to the PCM calculated value of 1364 cm−1 . For the aspirin dimer[3] a peak at 1370 cm−1 was detected which is in the between of the results from HB and HB + PCM calculations, which give the frequencies of 1473 and 1349, respectively. (iv) O H bending at the frequency 1210 cm−1 in the gas phase, which is decreased by 176 cm−1 when performing a PCM calculation. It is increased by 104 and 76 cm−1 , respectively, from HB and HB + PCM calculations. The calculated results are found to be in good agreement with the corresponding experimental results [2,3]. (v) Benzene C H bending accompanied by O H bending at the frequency of 1198 cm−1 , decreased by 9 cm−1 with the PCM. The inclusion of HB has small effect on the frequency, which is slightly decreased by 28 and 17 cm−1 for HB and HB + PCM calculations, respectively. This peak is one of the strongest in the IR spectra. The calculated frequencies fit well with the experimental results of Binev et al.[2] and Boczar et al. [3]. As mentioned above the rest of the calculated frequencies does not change in frequency as much as the the ones described

217

above. For the comparison of the different methods and with experimental data of Binev et al. [2] we refer to the Table 2 and the Fig. 3. 3.2.3. Low frequencies For the low frequency region of aspirin as one can see in Fig. 4 and Table 3 neither IR or Raman spectra have any peaks with significant intensity. This region has only a few bending modes for HB groups. The rest of the frequencies concerns C H bonding in CH3 groups and in the benzene ring. It should be noted that for the lowest region (e.g frequencies under 250 cm−1 ) it is difficult to compare the peaks for the different computational methods and the comparisons should therefore be examined with precautions. 3.3. Caffeine There are several studies dealing with experimental IR spectra of caffeine. Some of them deals with the determination of the amount of caffeine in beverages [8–11]. To the best of our knowledge, the theoretical simulations of IR or Raman spectra for caffeine are not available in the literature. We have compared our calculated results to experimental data of Ohnsmann et al. [8], who have performed a FT-IR study of caffeine in tea samples. Both the tea samples and the standard caffeine were dissolved in CHCl3 and gave similar spectra.

Table 3 Low vibrational frequencies, IR and Raman intensities (in parenthesis) of aspirin calculated from different approaches

Fig. 4. IR and Raman spectra of aspirin in low frequency region. Solid line: calculations. Dotted line: experiment (see text for reference).

Description

Gas phase

PCM

HB

HB + PCM

␥ (CHϕ ) ␥ (CHϕ ) ␥ (CHϕ ) ␥ (CHϕ ) ␥ (CHϕ ) ␥ (OH)+ ␥ (CHϕ ) ␥ (CH3 )+ ␥ (CHϕ ) ␦ (OH)+ ␦ (CCϕ ) ␥ (OH) ␥ (CH3 ) ␥ (CH3 )+ ␦ (CCϕ ) ␥ (CH3 )+ ␦ (CCϕ ) ␥ (CH3 )+ ␦ (CCϕ ) ␥ (CCϕ ) ␥ (CH3 ) ␥ (CH3 ) ␥ (CH3 ) ␥ (ϕ) ␥ (OH) ␦ (O COH) ␥ (ϕ) ␶ (CH3 ) ␥ (ϕ) ␶ (CH3 ) ␥ (CH3 ) ␥ (CH3 ) ␥ (CH3 )

889 (10,1) 831 (20,2) 812 (1,1) 773 (21,1) 742 (35,16) 713 (48,5) 664 (9,5) 638 (32,0) 591 (68,2) 581 (18,1) 556 (4,9) 526 (13,1) 515 (4,1) 435 (7,0) 423 (4,4) 357 (1,2) 311 (2,2) 272 (1,1)

883 (19,4) 824 (41,10) 806 (1,6) 761 (29,4) 728 (61,45) 708 (65,20) 661 (12,13) 623 (46,0)

893 (5,1) 835 (16,1) 813 (6,1) 801 (23,16) 763 (37,9) 713 (18,2) 682 (37,2) 666 (40,4)

888 (10,4) 830 (24,4) 809 (9,5) 795 (35,51) 752 (63,21) 709 (44,4) 675 (30,1) 659 (66,13)

577 (54,1) 552 (7,25) 515 (21,2) 494 (47,2) 451 (5,1) 423 (11,13) 329 (7,7) 311 (5,4) 291 (34,5) 199 (39,7) 142 (10,2) 100 (11,8) 91 (9,7) NA 75 (3,2) 50 (6,3) NA

598 (9,0) 569 (32,2) 544 (13,4) NA 453 (6,1) 426 (7,3) 323 (63,1) 223 (56,3) 254 (13,1)

594 (11,1) 562 (12,9) 545 (19,15) NA 452 (21,4) 422 (6,7) 376 (97,4) 322 (5,5) 285 (63,3)

149 (2,0) 154 (2,1) 99 (4,2) NA 92 (0,1) 84 (3,1) 66 (1,1) 48 (0,1) 29 (1,0)

151 (1,2) NA 95 (2,4) NA 74 (10,7) 67 (0,4) 59 (0,1) 51 (1,1) 26 (3,1)

229 (1,0) 140 (2,1) 109 (1,0) 102 (1,3) 97 (1,3) 85 (4,1) 78 (3,2) 30 (1,1)

218

L. Bondesson et al. / Spectrochimica Acta Part A 66 (2007) 213–224

Table 4 High vibrational frequencies, IR and Raman intensities (in parenthesis) of caffeine calculated from different approaches Description

Gas phase

PCM

HB

HB + PCM

␯s (CH) ␯a (Me1 ) ␯a (Me2 ) ␯a (Me3 ) ␯a (Me3 ) ␯a (Me1 ) ␯a (Me2 ) ␯s (Me1 ) ␯s (Me2,3 )

3258 (2,104) 3188 (1,29) 3179 (0,31) 3154 (6,60) 3144 (5,54) 3135 (13,79) 3132 (11,78) 3066 (15,261) 3066 (43,88)

3218 (44,286) 3185 (1,89) 3178 (0,96) 3153 (4,149) 3139 (6,146) 3131 (14,183) 3128 (12,178) 3065 (30,434) 3062 (17,490) 3062 (33,374)

3264 (5,88) 3199 (2,28) 3190 (6,53) 3161 (6,62) 3155 (7,51) 3151 (5,59) 3140 (8,70) 3080 (11,180) 3071 (16,197) 3070 (17,166)

3220 (57,280) 3196 (1,88) 3190 (5,121) 3159 (2,143) 3151 (7,139) 3145 (9,151) 3135 (10,176) 3076 (19,423) 3068 (22,470) 3067 (15,404)

3.3.1. High frequencies For the high frequency region of caffeine shown in Table 4 and Fig. 5 there are only small IR absorbance in the lower frequency regions (e.g 3250 cm−1 and lower) except a few peaks from the HB and HB + PCM calculations due to the presence of the OH bond in water. For the Raman spectrum there are several non-water related peaks in the high frequency region. The highest frequency related to the C H stretching has frequency of 3258 cm−1 in gas phase. This

value is lowered by 40 cm−1 with inclusion of the PCM. The HB can slightly increase the frequency by 6 cm−1 . The combination of HB and PCM has resulted in the increase of the frequency by as much as 38 cm−1 , which is slightly smaller than that caused by the pure PCM. Such a small HB effect is expected since the C H is not hydrogen bonded. The intensive spectral peaks are concentrated in the high frequency region between 3062 and 3080 cm−1 and are related to the C H stretching in the methyl groups. It is thus ex-

Fig. 5. IR and Raman spectra of caffeine in high frequency region. Solid line: calculations. Dotted line: experiment (see text for reference).

Fig. 6. IR and Raman spectra of caffeine in middle frequency region. Solid line: calculations. Dotted line: experiment (see text for reference).

L. Bondesson et al. / Spectrochimica Acta Part A 66 (2007) 213–224

219

Table 5 Middle vibrational frequencies, IR and Raman intensities (in parenthesis) of caffeine calculated from different approaches Description

Gas phase

PCM

HB

HB + PCM

␯s (C O1,2 )+ ␦ (Py) ␯a (C O1,2 )+ ␦ (Py) ␯ (C C)+?+ ␯a (C O1,2 ) ␦ (C C)+ ␦ (Im) ␥ (Me3,2 ) ␶ (Me3,2 )+ ␶ (Me1 ) ␶ (Me1,2 ) ␶ (Me1 )+ ␶ (Me3,2 ) ␶ (Me1 ) ␶ (Me1,3,2 ) ␥ (Me1 ) ␥ (Me1 )+ ␶ (Me3,2 )+ ␯s (CH) ␥ (Me1 )+ ␶ (Me3,2 )+ ␯s (CH) ␥ (Me1 )+ ␶ (Me3,2 )+ ␯s (CH) ␯ (Im)+ ␯ (Py) ␦ (Im) ␦ (Im) ␦ (Im) ␦ (N1 )+ ␶ (Me1 )+ ␦s (CH) ␦ (Im)+ ␦ (Py)+ ␦ (CH) ␦ (Im)+ ␦ (Py) ␦ (CH)+ ␦ (Me2 ) ␥ (Me1,3 ) ␥ (Me1,2 ) ␥ (Me3 ) ␦ (CH)+ ␦ (Me3 )+ ␦ (Me2 ) ␦ (Me3,2,1 ) ␦ (Me1 )+ ␦ (CH) ␯ (N Me1,2 ) ␦ (C O1 )

1752 (416,57) 1713 (857,32) 1630 (73,46) 1579 (149,3) 1527 (39,8) 1509 (26,15) 1504 (16,10) 1492 (9,7) 1488 (9,6) 1489 (147,8) 1476 (10,8) 1464 (16,2) 1451 (27,10) 1442 (29,7) 1417 (16,2) 1387 (57,69) 1363 (49,29) 1306 (33,28) 1273 (23,10) 1260 (63,8) 1241 (13,0) 1208 (13,2) 1156 (0,0) 1155 (1,0) 1151 (0,0) 1090 (2,7) 1086 (1,3) 1042 (72,2) 991 (32,1) 942 (2,3)

1735 (924,221) 1678 (1806,134) 1628 (171,197) 1569 (314,19) 1522 (82,42) 1502 (58,50) 1497 (27,29) 1483 (12,14) 1478 (13,13) 1480 (226,20) 1469 (14,14) 1459 (12,10) 1446 (59,38) 1437 (65,33) 1414 (26,4) 1386 (121,278) 1357 (75,76) 1302 (76,82) 1267 (34,48) 1257 (130,22) 1238 (37,1) 1205 (31,7) 1150 (0,0) 1149 (1,0) 1146 (0,0) 1087 (5,26) 1084 (7,6) 1033 (139,18) 989 (73,2) 939 (7,9)

1738 (531,69) 1697 (975,31)w 1626 (38,44) 1582 (154,3) 1536 (53,11) 1515 (75,17) 1503 (22,11) 1498 (58,4) 1492 (6,6) 1495 (37,10) 1483 (32,8) 1472 (7,4) 1456 (17,9) 1448 (31,8) 1416 (11,3) 1392 (62,66) 1363 (24,37) 1307 (41,32) 1280 (12,10) 1260 (69,7) 1237 (18,0) 1206 (18,2) 1157 (0,0) 1147 (1,0) 1149 (0,0) 1099 (5,7) 1091 (3,2) 1049 (73,3) 989 (16,0) 941 (5,3)

1720 (990,211) 1667 (1739,110) 1624 (123,179) 1569 (202,19) 1530 (106,45) 1507 (138,51) 1496 (53,27) 1488 (101,12) 1486 (19,23) 1485 (36,5) 1476 (68,19) 1466 (8,19) 1450 (41,27) 1440 (62,27) 1413 (25,15) 1391 (140,246) 1359 (35,99) 1303 (87,91) 1275 (20,45) 1056 (138,21) 1235 (40,1) 1203 (40,7) 1150 (0,0) 1143 (1,0) 1148 (0,0) 1095 (15,21) 1089 (4,11) 1040 (142,17) 988 (35,1) 938 (9,9)

Table 6 Low vibrational frequencies, IR and Raman intensities (in parenthesis) of caffeine calculated from different approaches Description

Gas phase

PCM

HB

HB + PCM

␥ (C H) ␦ (CH3 (1))+ ␦ (Me1 ) ␥ (C1,2,3 ) ␥ (C4 ) ␯ (N Me3 ) ␥ (C1,3 ) ␦ (Py) ␥ (N3,4 ) ␯ (Py)+ ␯ (Im) ␦ (Py) ␦ (Py) ␦ (Py)+ ␦ (Im) ␦ (C O2 )+ ␦ (N Me1,2 ) ␥ (N2 ) ␦ (N Me1,3 ) ␦ (N Me2,1 ) ␥ (N Me1 ) ␥ (N Me3 ) ␦ (N Me3 ) ␶ (N Me2,3,1 ) ␶ (N Me1,3 )

843 (15,1) 810 (1,3) 793 (11,0) 764 (10,0) 751 (14,7) 713 (0,0) 646 (1,5) 621 (14,0) 555 (1,27) 485 (25,3) 445 (15,3) 407 (13,1) 393 (15,1) 365 (0,1) 355 (9,1) 298 (2,1) 287 (0,0) 223 (0,0) 208 (9,0) 165 (5,0) 131 (8,0)

846 (21,6) 808 (4,14) 793 (30,2) 759 (24,0) 749 (21,24) 710 (1,2) 646 (4,21) 614 (32,1) 553 (0,85) 485 (52,8) 445 (32,6) 403 (17,2) 389 (29,2) 370 (1,2) 358 (23,2) 301 (6,0) 290 (1,0) 227 (1,1) 210 (19,1) 171 (11,1) 132 (4,0)

854 (19,5) 813 (3,13) 797 (36,2) 759 (21,0) 750 (28,25) 713 (2,2) 646 (45,24) 616 (32,1) 558 (21,76)w 488 (46,6) 451 (33,6) 414 (35,1)w 402 (15,3) 378 (4,2) 366 (20,2) 316 (2,1) 296 (2,0) 227 (135,2)w 224 (65,0)w 179 (26,1) 149 (2,1) 146 (3,1)

␶ (N Me2,3,1 ) ␶ (N Me2 )+ ␶ (N Me3,1 ) ␶ (N Me2,3,1 )

119 (1,0) 95 (0,0) 90 (1,1) 81 (2,0) 73 (0,0)

123 (7,0) 106 (1,0) 98 (5,1) 85 (5,3) 79 (0,0)

857 (13,1) 815 (1,4) 799 (14,0) 765 (9,0) 753 (22,8) 718 (3,1)w 648 (18,8) 625 (15,0) 556 (4,27) 489 (26,3) 452 (18,3) 420 (15,0)w 408 (10,1)w 365 (78,0)w 363 (11,1) 312 (18,1) 295 (1,0) 241 (9,0) 227 (4,1) 191 (13,0) 168 (16,0) 152 (2,1) 144 (10,0) 120 (27,0) 108 (13,1) 95 (16,0) 92 (2,0)

133 (16,0) 117 (24,1) 110 (6,2) 93 (6,0) 80 (11,1)

220

L. Bondesson et al. / Spectrochimica Acta Part A 66 (2007) 213–224

pected that they are insensitive to the inclusion of PCM and HB. 3.3.2. Middle frequencies In the middle frequency region shown in Table 5 and Fig. 6 one can find most of the observable peaks for the IR spectra. For the Raman spectra the spectral intensities are smaller than those in the high frequency region. The highest frequency in this region which corresponds to C O stretching and pyrimidine ring bending are 1752 cm−1 for the gas phase calculation. When using PCM the frequency is decreased by 17 cm−1 . For the calculation with HB the frequency becomes decreased by 14 and 32 cm−1 compared to gas phase, respectively, for without and with PCM. These values can be compared to experimental data by Ohnsmann et al.[8] where caffeine was dissolved in CHCl3 . The second highest peak in the experiment has a frequency of 1705 cm−1 which fits the best with the calculated value from HB + PCM calculation that gives a result of 1720 cm−1 . The frequencies with the highest absorbance in the caffeine molecule also corresponds to C O stretching and pyrimidine ring bending. For the gas phase calculation the frequency is 1713 cm−1 and is decreased by 34 cm−1 when PCM is added. The hydrogen bonded calculation without and with PCM decreases the energies by 16 and 46 cm−1 , respectively, compared to gas phase calculation. The highest ab-

Fig. 7. IR and Raman spectra of caffeine in low frequency region. Solid line: calculations. Dotted line: experiment (see text for reference).

sorbance peak in the experimental data by Ohnsmann et al. [8] carried out in CHCl3 has a frequency of 1659 cm−1 which also is in good agreement with the PCM + HB value of 1667 cm−1 . Next peak with high absorbance corresponds to C C stretching and imidazole ring bending. The wave number for the gas phase calculation are 1579 cm−1 and is decreased by 10 cm−1 for the PCM calculation. Also for this peak there exist an experimental value by Ohnsmann et al. [8] presenting a peak at frequency 1554 cm−1 . This experimental value agrees the best with the PCM value of 1569 cm−1 . The last peak with relatively high absorbance corresponds to rotations of the three methyl groups. The frequency for the gas phase calculation is 1489 cm−1 and the PCM calculated frequency is decreased by 9 cm−1 . The inclusion of hydrogen bonds further slightly reduces the frequency. For the rest of the calculated frequencies in the middle region the frequency changes among different methods are small. It is seen that the stretching of double bonds gives the largest absorbance in this region. 3.3.3. Low frequencies The result of the calculations of the low frequency region is shown in Fig. 7 and in Table 6. As one can see in the IR spectra of this region there are no strong spectral peaks except for those related solely to the water. Also in the Raman spectra there is

Fig. 8. Calculated IR and Raman spectra of ibuprofen in high frequency region.

L. Bondesson et al. / Spectrochimica Acta Part A 66 (2007) 213–224

221

Table 7 High vibrational frequencies, IR and Raman intensities (in parenthesis) of ibuprofen calculated from different approaches Description

Gas phase

PCM

HB

HB + PCM

␯ (O H) ␯ (ϕCH6,5 )S ␯ (ϕCH2,3 )S ␯ (ϕCH6,5 )A ␯ (ϕCH2,3 )A ␯ (CHMe1 )A ␯ (CHMe1 )A ␯ (CHMe2 )A ␯ (CHMe3 )A ␯ (CHMe2,3 )A ␯ (CHMe2,3 )A ␯ (CH1 ) ␯ (CH2) ␯ (CHMe1 )S ␯ (CHMe2,3 )S + ␯ (CH2) ␯ (CHMe2,3 )S ␯ (CH2) ␯ (CH2 )

3743 (72,147) 3190 (3,108) 3179 (16,123) 3168 (11,54) 3161 (12,49) 3137 (13,28) 3116 (28,104) 3108 (36,53) 3095 (41,114) 3090 (58,107) 3083 (9,20) 3070 (7,61) 3058 (17,30) 3047 (24,197) 3026 (45,356) 3020 (44,31) 3016 (15,18) 3004 (5,120)

3568 (370,238) 3163 (0,476) 3151 (2,555) 3142 (3,262) 3137 (3,179) 3131 (19,99) 3113 (37,244) 3104 (53,147) 3092 (57,246) 3085 (92,290) 3080 (13,52) 3045 (3,188) 3052 (28,149) 3043 (27,543) 3022 (51,858) 3016 (54,64) 3010 (31,156) 2986 (5,503)

3255 (1028,214) 3194 (2,126) 3187 (2,116) 3171 (12,53) 3165 (13,63) 3140 (5,27) 3115 (26,88) 3107 (36,53) 3095 (41,113) 3089 (61,113) 3082 (8,20) 3082 (1,70) 3057 (17,29) 3046 (27,202) 3026 (49,366) 3019 (39,22) 3016 (18,20) 3004 (4,123)

3217 (1456,361) 3191 (0,313) 3181 (1,352) 3146 (2,311) 3141 (3,338) 3138 (11,80) 3113 (32,226) 3103 (53,147) 3092 (55,239) 3085 (96,299) 3080 (10,47) 3081 (2,220) 3052 (28,147) 3044 (30,467) 3022 (53,881) 3016 (49,38) 3010 (32,153) 2987 (5,507)

only low intensity peaks in this region. The frequency is almost the same for all four computational models. Like for the aspirin molecule at the lower frequencies (e.g. below 160 cm−1 ), it is difficult to compare the same groups from different calculations.

Fig. 9. IR and Raman spectra of ibuprofen in middle frequency region. Solid line: calculations. Dotted line: experiment (see text for reference).

3.4. Ibuprofen Ibuprofen is also a common painkiller, however, it is quit new compared to the above described drugs. The spectroscopical studies are also less than for the above described drugs. We

Fig. 10. Calculated IR and Raman spectra of ibuprofen in low frequency region.

222

L. Bondesson et al. / Spectrochimica Acta Part A 66 (2007) 213–224

Table 8 Middle vibrational frequencies, IR and Raman intensities (in parenthesis) of ibuprofen calculated from different approaches Description

Gas phase

PCM

HB

HB + PCM

␯ (C O)+ ␥ (OH) ␦ (ϕ) ␦ (ϕ) ␦ (CHϕ ) ␥ (CHMe2,3 ) ␥ (CHMe2,3 ) ␥ (CHMe1 ) ␥ (CHMe2,3 )+ ␥ (CH2) ␥ (CHMe1 ) ␥ (CHMe2,3 )+ ␥ (CH2) ␥ (CHMe2,3 )+ ␥ (CH2) ␦ (ϕ) 1452 ␥ (CHMe2,3 ) ␥ (CHMe1 ) ␥ (CHMe2,3 ) ␥ (OH)+ ␯ (CH1 ) ␦ (CHϕ )+ ␥ (CH2 )+ ␥ (CH2) ␥ (CH2 )+ ␥ (CH2) ␦ (CHϕ )+ ␥ (CH1,2 ) ␦ (CHϕ )+ ␥ (CH2 ) ␥ (CH2 )+ ␥ (CH2) ␥ (OH) ␦ (CHϕ )+ ␥ (CH1 )+ ␥ (OH) ␥ (CH2) ␦ (CHϕ )

1803 (270,15) 1657 (1,98) 1612 (0,3) 1545 (25,1) 1509 (12,1) 1503 (4,9) 1497 (8,5) 1494 (3,1) 1490 (5,5) 1487 (1,13) 1479 (1,4) (9,0) 1418 (5,1) 1409 (10,1) 1398 (4,0) 1389 (49,4) 1369 (3,19) 1365 (0,3) 1360 (0,2) 1347 (2,6) 1310 (2,15) 1303 (6,3) 1276 (1,4) 1244 (1,5) 1226 (1,27)

1777 (624,56) 1653 (2,240) 1611 (0,20) 1539 (48,1) 1500 (15,2) 1494 (4,19) 1491 (10,14) 1487 (4,1) 1482 (6,11) 1480 (1,25) 1473 (3,17) 1447 (17,1) 1410 (4,5) 1401 (14,4) 1393 (4,0) 1377 (82,9) 1363 (1,39) 1361 (1,12) 1356 (4,3) 1341 (3,17) 1308 (3,35) 1290 (31,9) 1257 (23,11) 1238 (2,12) 1224 (3,97)

1721 (566,70) 1652 (5,239) 1611 (2,18) 1541 (74,6)w 1499 (13,1) 1493 (4,20) 1499 (12,12) 1486 (4,2) 1488 (8,13) 1481 (1,24) 1475 (3,16) 1450 (16,2) 1410 (5,6) 1409 (7,1) 1393 (4,0) 1370 (46,8) 1366 (1,38) 1362 (2,9) 1358 (22,4) 1345 (4,23) 1310 (3,32) 1435 (55,9) 1289 (45,8) 1240 (2,13) 1227 (56,13)

␦ (CHϕ )+ ␥ (CH1 ) ␦ (CHϕ )+ ␥ (CH1 ) ␥ (OH)+ ␦ (CHϕ )+ ␥ (CH1 ) ␥ (CHMe2,3 )+ ␥ (CH2) ␥ (OH)+ ␦ (CHϕ )+ ␥ (CH1 ) ␦ (CHϕ ) ␥ (Me2,3 ) ␥ (CH2)+ ␥ (Me1,2,3 ) ␥ (Me1 )+ ␥ (CH1 ) ␥ (Me1 )+ ␥ (CH1 ) ␥ (OH) ␦ (CHϕ ) ␥ (Me1 )+ ␥ (CH1 ) ␥ (CHϕ ) ␥ (CHϕ )+ ␥ (Me2,3 )

1214 (11,14) 1206 (8,41)

1206 (2,90) 1195 (1,60)

1746 (313,26) 1656 (7,102)w 1612 (0,4) 1548 (28,0) 1508 (11,1) 1502 (4,10) 1507 (7,5) 1494 (3,2) 1496 (5,6) 1487 (1,12) 1480 (1,5) 1454 (4,3) 1418 (6,1) 1416 (6,0) 1399 (5,0) 1385 (65,4) 1373 (0,15) 1366 (1,3) 1364 (8,9) 1350 (1,9) 1312 (3,14) 1459 (22,4) 1295 (38,3) 1245 (1,5) 1241 (170,8) 1231 (9,22) 1215 (7,23) 1200 (36,34)

1187 (8,2) 1159 (180,1) 1144 (34,1) 1132 (2,16) 1101 (29,4) 1091 (15,2) 1079 (50,2)

1183 (11,8) NA 1138 (20,4) 1130 (13,34) 1117 (77,7) 1091 (47,6) 1086 (23,5) 1052 (337,7) 1020 (3,2) 1001 (6,23) 975 (0,0) 962 (0,5) 959 (0,3)

␥ (OH) ␥ (Me2,3 ) ␥ (Me2,3 )+ ␥ (CH2 )

1025 (4,0) 1005 (4,9) 982 (0,0) 964 (0,2) 963 (0,1) 952 (1,5) 926 (2,1)

have found one article that contains an experimental IR spectra although this article only discuss the wave number of one of the peaks. The article that we refer to is written by Garrigues et al. [1] which have performed a FT-IR spectra of ibuprofen dissolved in tetrachloride. 3.4.1. High frequencies The data for ibuprofen at high frequencies are shown in Table 7 and plotted in Fig. 8. For the IR spectra the highest wave number is also the peak with the highest absorbance and corresponds to O H stretching. This peak is present at the frequency 3743 cm−1 for the gas phase calculation. If the PCM is applied the frequency is decreased by 175 cm−1 . The frequency for the same stretching when the group are hydrogen bonded the frequency is 489 cm−1 lower than for the gas phase calculation and

948 (0,13) 923 (2,3)

1186 (11,2) NA 1153 (5,1) 1132 (0,18) 1110 (9,4) 1094 (12,1) 1088 (13,3)

1223 (185,114) 1202 (10,60) 1192 (161,52) 1182 (16,7) NA 1144 (13,2) 1130 (0,35) 1106 (25,9) 1088 (19,4) 1084 (27,6)

1029 (2,3) 1015 (5,10) 993 (0,0) 977 (0,1) 965 (0,3) 954 (110,1) 952 (0,6) 927 (2,1)

1024 (2,7) 1010 (7,25) 986 (0,1) 969 (0,2) 962 (0,5) 890 (166,5) 949 (0,12) 923 (2,3)

when adding PCM to the HB calculation the frequency is decreased with 526 cm−1 compared to gas phase. There are also a number of high absorbance peaks in the IR spectra that relates to pure water that appears in the frequency region 3500– 3900 cm−1 for the calculations including HB. When looking at the Raman spectra there are several peaks that have high or medium high intensity. The peak described above for the IR spectra is the one that changes most in frequency although it is not the one with largest intensity. Two C H stretching groups have a high/medium intensity. The first one has the frequency of 3190 cm−1 in the gas phase and is decreased by 27 cm−1 when PCM is applied to the calculation. It is increased by 4 and 1 cm−1 , respectively, for HB and HB + PCM calculations. The second peak is located at 3179 cm−1 in the gas phase and the PCM calculation decreases the wave number by 28 cm−1 . With-

L. Bondesson et al. / Spectrochimica Acta Part A 66 (2007) 213–224

223

Table 9 Low vibrational frequencies, IR and Raman intensities (in parenthesis) of ibuprofen calculated from different approaches Description

Gas phase

PCM

HB

HB + PCM

␥ (CH2)+ ␥ (Me2,3 ) ␥ (CHϕ ) ␥ (CHϕ ) ␥ (CHϕ )+ ␥ (Me1 ) ␥ (Me1 ) ␯ (Me2,3 ). ␥ (CHϕ ) ␥ (ϕ) ␦ (ϕ) ␦ (ϕ)+ ␥ (OH) ␦ (ϕ)+ ␥ (OH)

889 (3,3) 867 (15,2) 852 (1,1) 848 (3,5) 834 (1,20) 819 (8,8) 791 (17,6) 751 (5,12) 707 (36,1) 657 (9,4) 636 (21,3) 616 (21,3) 590 (47,1) 508 (1,0) 470 (10,1) 417 (1,0) 415 (0,0) 403 (3,3) 377 (1,1) 350 (1,1)

885 (7,6) 862 (33,10) 841 (1,9) 847 (6,23) 831 (1,63) 817 (13,16) 787 (30,16) 749 (8,37) 694 (20,5) 650 (10,10) 626 (24,8) 587 (84,2)

886 (25,5) 869 (42,3) 857 (9,) 851 (0,3) 840 (0,82) 821 (1,20) 786 (35,16) 756 (8,34) 702 (20,3) 669 (43,3) 640 (19,14) 597 (47,5)

513 (28,2) 472 (16,3) 414 (2,1) 411 (0,0) 403 (16,7) 372 (3,2) 347 (2,5)

304 (1,1) 296 (1,1) 249 (0,0)

294 (5,3) 271 (57,3) 242 (3,1)

230 (0,2) 223 (0,0) 220 (0,1) 213 (1,0) 161 (0,0)

227 (1,5) 220 (1,1) 214 (0,1) 299 (12,1) 151 (4,2)

890 (6,3) 874 (18,1) 863 (1,0) 859 (3,2) 841 (0,27) 822 (1,9) 790 (16,5) 758 (9,12) 706 (27,2) 674 (51,2)w 645 (86,6)w 638 (156,3)w 599 (15,1)w 530 (23,1) 484 (16,2) 420 (4,0) 418 (1,0) 409 (6,3) 385 (32,1)w 352 (2,1) 328 (77,2) 320 (5,1) 304 (65,1)w 259 (13,1) 240 (0,1) 232 (4,2) 225 (6,0) 217 (6,0)

184 (6,0) 178 (22,1)

102 (0,2) NA 59 (0,2) 55 (1,2) 44 (0,1) 35 (1,1) 22 (0,0)

85 (1,3) 55 (5,3) 51 (1,2) 35 (1,11) NA

191 (9,0) 181 (4,0) 158 (10,1) 70 (2,2) 90 (4,1) 58 (6,2) 54 (2,0) 51 (1,0) 42 (4,1) 37 (2,1)

␥ (CHϕ)+ ␥ (OH) ␯ (C C) ␶ (CH2)+ ␦ (ϕ) ␦ (CHϕ)+ ␶ (CH2) ␶ (CH2 ) ␶ (ϕ) ␥ (C C) ␥ (OH) ␥ (C C) ␶ (Me1,2,3 ) ␶ (Me1,2,3 ) ␶ (Me1,2,3 ) ␥ (Me1,2 ) ␯ (C C) ␥ (C O) ␶ (ϕ) ␥ (Me2,3 ) ␶ (Me2,3 )

out or with the PCM, the HB calculations increase the gas phase value by 8 and 2 cm−1 , respectively. There are also five frequencies corresponding to the C H stretching in the methyl groups that show high/medium intensities in the Raman spectra. These five modes are not sensitive to the choice of the computational methods. One more frequency with high/medium intensity is associated with the C H stretching in the CH2 group. This stretching has a gas phase value of 3001 cm−1 and is decreased by 17 cm−1 when the PCM is included. The inclusion of HB alone does not change the frequency while the combination of HB and PCM decreases the gas phase value by 17 cm−1 . 3.4.2. Middle frequencies For the middle region of ibuprofen, the vibrational frequencies are shown in Table 8 and the IR and Raman spectra are given in Fig. 9. One can notice that one stretching and four bending modes possess high IR intensities and are affected greatly by computational techniques employed. The stretching mode is related to C O stretching mixed with O H bending. The cal-

517 (128,1)w 476 (12,5) 416 (12,2) 415 (0,0) 407 (32,7) 377 (12,2) 349 (5,5) 313 (18,3) 299 (15,4) 252 (30,4) 234 (0,1) 228 (11,4) 219 (17,0) 211 (0,0)

104 (1,1) 77 (5,2) 67 (5,2) 59 (3,4) 48 (3,5) 48 (5,8) 38 (15,1)

culated gas phase frequency is 1803 cm−1 and it is decreased by 26, 57 and 82 cm−1 for PCM, HB and HB + PCM calculations, respectively. The experimental frequency obtained by Garrigues et al. [1] for this vibration is 1710 cm−1 which compares well with the value of 1721 cm−1 from HB + PCM calculation. The four bending modes are mostly associated with the bending of O H group and are sensitive to the different solvent environments. The first vibration is located at 1389 cm−1 in the gas phase. Its frequency decreases slightly with the inclusion of PCM and HB. The second bending mode at 1303 cm−1 in the gas phase is very sensitive to the hydrogen bonding which can reduce the frequency by 156 and 132 cm−1 , respectively, for HB and HB + PCM models. The third peak with visible intensity only appears in the case of HB + PCM calculation and it is located at 1192 cm−1 . The last bending mode has a frequency of 1159 cm−1 in the gas phase. Its frequency decreases by 107, 205 and 269 cm−1 when PCM, HB and HB + PCM are included, respectively. It should be noted that there are several similar vibrations that can appear around the last bending mode depending on the computational approaches used. This makes it

224

L. Bondesson et al. / Spectrochimica Acta Part A 66 (2007) 213–224

slightly difficult to trace the same vibration in different solvation models. There are also several C C and C H bending modes presented in this region. However, their frequencies do not show strong dependence on the computational methods and most of them have low IR absorbance and Raman intensity. 3.4.3. Low frequencies In the low frequency region of ibuprofen the IR absorbance and the Raman intensities are low as shown in Fig. 10. The frequencies are presented in Table 9. Most of the frequencies are found to remain relatively intact, independent which method is used. The largest difference is, once again, found for the frequencies for vibrational modes involving O H bending. There are two frequency regions that are of main interest. The first concerns both benzene ring and O H bond bending where the gas phase spectrum has four frequencies ranging from 657 to 590 cm−1 , in which calculations with inclusion of hydrogen bonding produce only three peaks. The second group concerns only O H bending and is only found in PCM calculation where the frequency is 328 cm−1 . The vibrations with even lower frequencies are a mixture of different forms of bending, rotations and stretching, depending on the computational methods employed. 4. Conclusion Interactions between drug molecules or between such molecules and surrounding media constitute key research problems for drug research, for instance for the important research area of drug solubility. Such interactions can sometimes be difficult to master directly by theoretical means owing to the smaller free energies involved, the complexity of the kinetics and the potential energy landscapes involved. The present paper represents a different route to obtain information of drug interaction, namely one which makes use of theoretical modeling to predict and interpret characterizing spectroscopies. The goal of such modeling is to assign salient features in the spectra to molecular structure, i.e. to find so-called structure–property relationships, and to trace specific changes of the spectral features, that follow upon the binding or solvation of the sample molecule, to the particular nature of the intermolecular interaction. Thus the notion is that the combined use of modelling and spectral characterization can give input to important problems like drug solubility. A prerequisite to such an effort is evidently that modeling can reproduce the spectra and their characteristic changes with sufficient accuracy to be reliable for studies of unknown situations. The outcome is evidently different considering different wavelength regions and spectroscopies, but as general trend one can state that modern, quantum chemical methods, n.b. density functional theory, is capable to fulfill these demands for free species for most wavelength regions and spectroscopies. The situation is different, and much unexplored, concerning solvated or condensed species, which indeed is the common form of nature. Thus the theoretical predictability of characterizing spectroscopies of solvated species is a study goal. We have presented results of accurate ab initio calculations of IR and Raman frequencies for three common drugs. The inves-

tigation covers the solvation effect of both hydrogen bonded and non hydrogen bonded systems. It is shown that solvation effects have an impact on the spectra, especially for vibrational modes involving the motion of oxygen atoms. For the atoms that are responsible for the hydrogen bonding, one can see a large impact on the spectra of their characteristic motion by the presence of a solvent. It is also seen that some other atom pairs motions changes in frequency when PCM is used. The polarizable continuum model is found to provide good agreement to experimental results in the case of unpolar solvents. For the crystal experiments the hydrogen bonding model used here is not sufficient for the whole molecule although one can see that the frequencies are better correlated for the hydrogen bonded atoms. Acknowledgments This work was funded by Biovitrum AB and by the European Research and Training Network “Molecular Properties and Molecular Materials” (MOLPROP, contract HPRN-CT-200000013). LB wants to thank his colleagues at the Copenhagen University and at Biovitrum AB and KTH in Stockholm for helping him with computational problems and for discussing the nature of hydrogen bonding. K V M thanks the Danish Natural Science Research Council, the Danish Technical Research Council, Danish Center for Scientific Computing and the EUnetwork MOLPROP for support. References [1] S. Garrigues, M. Gallignani, M. de la Guardia, Talanta 40 (1993) 89. [2] I.G. Binev, B.A. Stamboliyska, Y.I. Binev, J. Mol. Struct. 378 (1996) 189. [3] M. Boczar, M. W´ojcik, K. Szczeponek, D. Jamr´oz, A. Zie¸ba, B. Kawałek, Chem. Phys. 286 (2003) 63. [4] S. Gunasekaran, G. Sankari, S. Ponnusamy, Spectrochim. Acta Part A 61 (2005) 117. [5] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, J.A. Montgomery, Jr., T. Vreven, K.N. Kudin, J.C. Burant, J.M. Millam, S.S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G.A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J.E. Knox, H.P. Hratchian, J.B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W. Ochterski, P.Y. Ayala, K. Morokuma, G.A. Voth, P. Salvador, J.J. Dannenberg, V.G. Zakrzewski, S. Dapprich, A.D. Daniels, M.C. Strain, O. Farkas, D.K. Malick, A.D. Rabuck, K. Raghavachari, J.B. Foresman, J.V. Ortiz, Q. Cui, A.G. Baboul, S. Clifford, J. Cioslowski, B.B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R.L. Martin, D. J. Fox, T. Keith, M.A. Al-Laham, C.Y. Peng, A. Nanayakkara, M. Challacombe, P.M.W. Gill, B. Johnson, W. Chen, M.W. Wong, C. Gonzalez, J.A. Pople. Gaussian 03, Revision C.02, Gaussian Inc., Wallingford, CT, 2004. [6] P.R. Rablen, J.W. Lockman, W.L. Jorgensen, J. Phys. Chem. A 102 (1998) 3782. [7] M. W´ojcik, Chem. Phys. Lett. 83 (1981) 503. [8] J. Ohnsmann, G. Quint´as, S. Garrigues, M. Guardia, Anal. Bioanal. Chem. 374 (2002) 561. [9] Y. Daghbouche, S. Garrigues, M.T. Vidal, M. de la Guardia, Anal. Bioanal. Chem. 69 (1997) 1086. [10] Z. Bouhsain, J.M. Garrigues, S. Garrigues, M. de la Guardia, Vib. Spectrosc. 21 (1999) 143. [11] J.M. Garrigues, Z. Bouhsain, S. Garrigues, M. de la Guardia, Fresen. J. Anal. Chem. 366 (2000) 319.