Ethylene glycol: Infrared spectra, AB initio calculations, vibrational analysis and conformations of 5 matrix isolated isotopic modifications

Chemical Physjcs 2f.(1977) 271-298 0 North-Holland Publishing Company

iTHiLENE

GLiCCi:

INFRARED SPECkA,

AND CQtiORMATIOiS iI. FE&I, Taeiyb

AB INITIO CALCULATIONS, VIBRATIONAL ANALYSIS

OF-5 MATRIX ISOLATED ISOTOPIC MODIFICATIONS

HA, R. MEtiR

and Hs.H. GiiNTHARD

taboratory ofPhysical_Chemistry, Swiss Federal Institute of Technology, ._CH-8006 Ztirich. Switzerland Received 29 March 1977

Anextension of earlier ab initio calculations is reported which consists in a more detailed structure optimization of conformations investiated earlier, using the previously described as well as a more flexible basis set. The measurement and analysis of infrared spectra of five isotopic modifications, isolated in AI and Xe matrices, are described. The analysis of the spectra, carried out within the framework of two H-bonded conformations r,, and r,, predicted by ab initio calculations to be the most stable ones, is complicated by the occurrence of unusually large matrix splittings and irreversible changes of the spectra under irradiation by infrared tight. The spectra are interpreted in terms of essentially two slightly different conformers of type rer , deformed by the matrix field. The 180 isotope shifts prove to be important for correct assignment and the least squares fit. A valence force field is given, whose dependence on the conformation of the OH groups and especially on H-bonding is discussed. According to normal coordinate analysis, the effect of the internal H-bond is most pronounced in the COH bend and OH torsion region. The differences of the potential constants of free and bonded OH groups are responsible for a zero point energy difference of about 20 cm -l of two isomers of the modification CHz0DCH20H. From ex-

perimental evidence for at least a partial equilibration of these two isomers at liquid helium temperature, internal rotation of the OH groups in an Ar matrix may be deduced. Thermodynamic functions of the ret conformation of glycol in the rigid rotor-harmonic oscillator approximation are reported.

Lhtroduction

The glycol molecule, roughly represented by

H-O

H\

‘CHpLj

0

is one of the simplest nonrigid molecules with three internal degrees of freedom. From a point of view of classification of semirigid molecules it has been classified in earlier papers as a D,(r)F(C,T)2 .system [ 1,2]. Matrix spectra and ab initio calculations have shown that the internal rotational degrees of freedom are frozen to a considerable extent by the formation of an interoal hydrogen bond [l] _According to published ab in~tio work [l] there exist at least two re structures differing mainly in the conformation of the OH groups. The corresponding local minima were found to be rather shallow as shown by fig. 1 which shows a

part of the Born-Oppenheimer surface and the associated corformations. The Freezing of the internal degrees of freedom offers a possibility to study vibrational spectra without too much complications by the internal rotation modes, i.e. in the sense of quasi rigid nuclear configurations. Measurement of spectra of the free molecule are made difficult by the low vapor pressure at room temperature [3]. Gas spectra at higher temperature will almost certainly be too complex For an analysis owing to excited state spectra and possibly tbe existence of several conformers [4] _On the other hand solution, liquid and solid state vibrational spectra will dominantly be related to highly associated species [5,6] and barely may be considered as relevant for the Free molecule. Information about the latter may be obtainable from matrix isolated molecules, though it should be expected that the matrix field might exert noticeable eFfects on the conformation of free glycol. In this paper results of new ab initio calculations

272

H. Frei et oi./Ethylene glycol

a

Fig. I. (a) Section of the Born-Oppenheimer energy surface and models of conformers rel and rez. (b) Newman projections for the deftition of internal rotational degrees of freedom. Key: EIectronic energy diagram from the ab initio SCF calculation of E(UO,IJ~) for 27 = 60” (basis set 1). The energy levels are separated by 5 X 104 au (110 cm-‘).

and the measurement and analysis of infrared spectra of the isotopic modifications CH20HCH20H

CHzODCHBD

do

CD20HCD20H

Odz

CH;*OHCH;‘OH

‘8o2

timization of the conformations investigated earlier [II, using the same basis set and including optimization of the CC-torsional angle. Furthermore, a more flexible basis set was applied for energy optimization of essentially the same set of~conformations. Infrared spectra of the isotopic modititiations mentioned aforehead, isolated in Ar and Xe matrices are reported (section 4). Analysis of the matrix infrared spectra is carried out within the framework of the ab initio results, i.e. in terms of two conformatidns rel and re. predicted by ab initio calculations to be the most stable ones within the set of conformations of a semirigid model with three internal rotational degrees of freedom (section 5). The spectra are complicated in both matrices by the occurrence of many multiplets, whose splittings most often exceed the usually observed site splittings. Furthermore, they show irreversible transformation under irradiation by infrared light. Vibrational analysis is further based on empirical argriments and on normal coordinate calculations for the two re structures rel, r,2. cf. fig. 1. The I80 isotope shifts prove as an irGpoZant tool for correct assignment and support the assumption of the conformation rel. The splittings will be traced back to the existence of at least two conformers of type rel, rendered slightly different by the matrix field. A valence type force field will be given, which will be shown to depend on the conformation and considerably differs for the two OH groups, i.e. free and bonded. The force field may be considered typical for molecular conformations containing the fragment

Finally thermodynamic functions of the rel conformation of glycol iri the rigid rotor-harmonic oscillator approximation are reported (section 6).

$4 CH20DCH20H

2. Ab initio SCF calculation

W

will be reported. The extension of earlier ab initio calculations [I] has been prompted by an investigation published by NmIiif and Stymne [7], and mainly consisted (see section 2) in a more detailed structure op

The approximate Hartree-Fock SCF atomic orbitals employed in this work were the gaussian lobe function representation [S]. The SCF calculations for glycol have been carried out‘for two different choices of basis set contractions. The f&t basis set (bar& set .I) consists

213

H. Frei et aL/Ethylene &ol

of three contractions of four, three and three spherical gaussians for s-orbitals and one contraction of five pairs of gaussians for each of the three p-orbitals of carbon and oxygen, respectively. Each hydrogen sorbital is represented by one contraction of five spherical gaussians scaled by a factor of 21/2 [9]. The second basis set (basis set 2) is a slightly extended one, in which additionally the long range s-group was split into the contractions of two and one gaussians and likewise the p-groups were split into two contractions of three and two gaussian lobe functions for C and 0. The two contracted s-basis functions on the hydrogen were built from four short and intermediate range gaussians and one long range gaussian. The quality of the basis sets used in this work may be illustrated by the following results for total energies E of optimized r,, structures: basis set 1 basis set 2

E = -228.6890 au, E = -228.8045 au.

No comparison may be made with the basis set used in ref. [7], since no E value has been reported. Results of extensive ab initio computation of the electronic potential function with the basis set 1 were presented previously [l] from which two different re conformations were predicted_ As an extension and modification of the previous ab initio work we undertook the following further studies: (i) Rotation about rhe C-C&m& In the previous ab initio study [l] we assumed that the C-C torsional .angle (27) of 60” corresponds to a r, value. To verify this assumption we partially optimized the structure with respect to the structural parameters, uo, uI, and 2~ simultaneously using basis set 1. Fig. 2 shows three energy curves which correspond to 27.= 30”, 60” and 75”, respectively as a function of ul. The fmed u. value corresponds to the structure of the minimal 0-0.H distance for a given 2~ value. It is evident from fig. 2 that 27 = 60” nearly corresponds to a re value of 2~. Furthermore, from the energy surface for 2~ = 60” it is also evident that the minimal energy pathway from rel to re2 is characterized by an almost constant u. value which corresponds to the minimal O-*-H distance. The shift in the energy minimum with decreasing C-C torsional angle results from the directional change in the lone pair electrons of the acceptor oxygen atom to the bonded H nucleus.

-2266900

j -120.

180'

12v

60'

0.

v,

Fig. 2. Electronic energy curves for different values of the CC torsional angle. Key: The fixed u,-, value corresponds to the structure of minimal O-H distance for a given 2r value.

(ii) AN-tramconformation: The energy difference between the rel conformation and the H-bond free alltram conformation (27 = uo = u1 = 180a), has been reexamined with respect to the O-H bond distance and the basis set effect. The previous ab initio study [l] predicted that the ail-trans conformation lies about 7.04 k&/mole above the rel minimum. In the present study we first optimized r(OH)o = r(OH)i with L (COH), = L(COH), = 108.53’, cf. fig. 11 and table 6, for the all-trans conformation, using basis set 1. Then we reoptimized the rel structure with respect to r(OH),,, employing a r(OH)l value of 0.965 A, which corresponds to the optimal O-H distance for the alltrans conformation. The calculated energy difference amounts to 7.15 kcal/mol, which indicates that the structural optimization with respect to the O-H bond distance has brought about no significant effect in comparison to the previous study. The calculated optimal O-H distances in the rel structure keep almost the same value of 0.965 A as in the all-trans conformation_ Even though we have not optimized the all-trans conformation with respect to all three relevant dihedral angles it might be safe to assume that the all-trans conformation corresponds to a local energy minimum as has been remarked by Almliif and Stymne 171. As a next step we examined the effect of the basis set contraction. Employing the basis set 2 we followed the same geometry optimization procedure for the rel

.H.Frei et al.fEthylenegIyco1

274

and the all-trans conformation as mentioned above. Practically the same O-9 bond distances for r(OH)c = 0.958 A and r(OH), = 0.952 A were also obtained. However, the energy difference between the rel and the ah-trans conformation amounted to 3.52 kcal/mol; a significant decrease in comparison with the result .from the basis set 1 calculation (7.15 kcal/mol). This value from the basis set 2 calculation lies rather near to the valut of 3.0 kcaljmol reported by Almlof and Stymne [7]. (iii) Effect of basis set contraction: Because of the significant change in the relative energy of the rel and the all-tram structure, due to different basis set contractions as observed in (ii), we undertook further calculations using the basis set 2 for some of the important points on the energy surface. These include the points rf,r, rz,2 and r:,2, i.e. saddle points between the rel and re2 structures, and between rel and re2 (see fig. 1) Table 1 summarizes the results and compares with previous results obtained from the basis set 1. h shown by this table the energy differences are rather sensitive to the basis set contraction. While the rel conformation lies significantIy lower than the re2 conformation by an amount of 1.83 kcaljmol in the basis set 1 calculation, both conformations have almost the same electronic energy in the calculation using the more flexible basis set 2. Since rel and re2 structures are not optimized additionally with respect to u. and Table 1 Energy difference among some important points in the energy surface a) AC (cm-* ) (kc!al/mole) basis set 1 b)

ul the small energy difference of -0.19 kcal/mol obtained from the basis set 2 calculation is not to be-considered as significant. It should be mentioned that in contrast to the basis set 1 calculations reported previously [ I] the exact positions of the saddle points have not been evaluated in the basis set 2 calculation. Those results from the ab initio SCF.calculations employing the basissets 1 and 2, which are consistent with each other, will furtheron be used as a framework for the interpretation of the experimental vibrationti spectra. In brief these are: (i) there exist two isomeric structures rel and re2, which at the present tune are to be considered as the most stable conformations both having approximately the same Born-Oppenheimer energy; (ii) both the rel and re2 structure feature an intramolecular H-bond and are more stable by epproximately 3 kcal/mol than the H-bond free all-tram conformation_ Restriction to those results of the basis set 1 and 2 calculations which are consistent with each other may serve as a precautionary measure against modifications possibly emerging from ab initio calculations with even more extended basis sets.

3. Experimental 3.1. Preparation of isotopic species Purification of glycoldo and d4 and preparation and purification of the Od, have been reported earlier [l]. GIyc01-~~0~ was prepared according to [IO] CH2:CH2 + H2180 + Br2 5

(1)

basis set 2 c)

CH2BrCH2180H+H2180= E (all-tranS) - E(ret) E(‘e2) - &,,I Vf,t = &,I) - E(ret) Vz’;z=e(rZ2) - E(re2) v:,, = fj,;;

1 :@a; >*

2464 (7.04) 640 (1.83) 726 (2.07) 260 (0.74)

6;; g-8;;

1232 (3.52) -66 (-0.19) 1210 (3.46) 616 (1.76) 286 (0.82) 352 (1.01)

a) For description of the basis set, see text. b) Structurti parameters are reported in ref. [I]. c) Structural parameters:r(OH)I =0_952A, r(OH)o =0.958 A, L (COH)o = L (COH)l = 108.53”, other parameters are the sameasinref. [l].

CH2BrCH2 180H f HBr,

+HBr.

160 h

CH2180H-CH2180H -(a

Both reactions were carried out in a vacuum manifold. For preparation of the bromohydrin-‘a0,6.4 g Br2 (40 mmol) were added in portions of 300 mg to 20 g H2180 (99.8%, 1 mol). The solution was kept at 0°C and stirred.vigorousiy under an ethyiene atmosphere, whose pressure was kept as closely as possible at 200 torr. Under these conditions an amount of 4.5 g 1.2-dibromoethane is formed, which separates itself in a lower liquid phase and was removed mechanically.

215

H. Frei et aI./Ethyleneglycol

The aqueous phase was neutralized with pyridine and separated from the latter by trap to trap distillation. Then the aqueous mixture of H2180 and bromohydrin was heated to 70°C for 160 h under a 400 torr Ar gas atmosphere. HBr formed according to (2) was neutralized with pyridine and the major part of liquid phase transferred by trap to trap distillation at 3 torr. The glycol was separated in vacua from the residual and purified by GLC. Yield of the latter was 0.7 g with an isotopic purity of 99.8%, most of the original amount of H2180 was recovered (18 g). The sample was then dried over molecular sieve A3 for 8 days. Commercially available matrix gas samples were used without further purification (Xe research grade, 99.995%, Matheson Gas Products, Chicago). 3.2. Instnmental

IR irradiation. Qualitative measurements of the kinetics of the relaxation process were carried out by observation of IR spectra without and with IR irradiation as a function of time. Particular band patterns were recorded after 2,4, and 6 h of irradiation. Peak intensities were then plotted for growing and fading bands versus time.

4. Experimental results Fig. 3 shows the absorption spectra of the four species in Ar matrix in the v(OH) and Y(OD) region, 3700-3580 and 2720-2650 cm-l, respectively. Similarly in fig. 4 the band pattern in the v&H) and Y(CD) regions are reproduced. The mid infrared regions of all four species are reproduced in fig. 5 and fig. 6

The matrix spectra (Ar and Xe) were taken with a Perkin Elmer Model 325 spectrophotometer between 4000-200 cm-l at spectral slit width of 3-0.5 cm-l. The matrices were produced in a cryostat of our own design [ll], equipped with separate inlet systems for sample and matrix gas. The Ar-flow was regulated by an Edwards Needle valve to 15 mmol/h. The sample flow of 8 pmol/h was regulated by Brooks valves, with the samples kept in break off seal ampoules at room temperature. These settings yield M/A ratios of =2000 with deposition times of 6-8 h and layer thicknesses of 400-550 p. The deposition rates were controlled by a small laser [12]. These deposition conditions were required in order to obtain sufficient isolation of glycol in both Ar and Xe. For generation of sufficiently transparent Xe matrices within 6-8 h deposition time, Xe flux of approximately 5.5 mmol/h and layers of only 65-85 w could be used. Sample flows of 8 mol/h yielded M/A = 450. Matrix experiments with thermal glycol molecular beams were carried out by means of a heatable Knudsen cell [133. Glycol molecular beams from 120,160 and 250°C were deposited simultaneously with room temperature Ar and Xe in a ratio M/A a 2000 and 450, respectively.

..,

Yy y. CH2OHCH20H.

CH2WCH200.

Ar

sdmcd

5-oncm-’ 37clocnd

3580

h”.‘,..

2740

2640

I.--

CD20HCD2W CH:6OHCH~‘OH.

I

3?OOcd Fig.

_ Ar

Ar

3.3. Kinetics of rehzxation of matrix spectra As mentioned in the introduction, matrix spectra of glycol exhibit pronounced irreversible changes under

Ar

MIA - 2000

35ED

3700

.

.

I

3600

3. Ar-matrixinfrared spectra of glycol: absorption bands

in the

v(OH) and v(OD) region.

276

H. Frei et al./Ethyleneglycol

discussion of the Irreversible’chariges of gIyco1 matrix s+ctra as Wellas for the question of -the e$&nce‘of further conformations. -. ,: 5. Analysis of vibrationA_ s&t&

3OOOCK'

2860

30110ur-’

2860

: and discussion

The analysis of the experimental results will be presented in the. folIowh@otder: (1) Qualitative discu_ssionof irreversible changes of spectra and Knudsen celIexperime&. (2) Assignment based on ab initio calculations, empirical arguments and normal coordinate analysis. (3) Splitting patterns.

5.1. Irreversible spectral changes and Knudsen cell experimen ts

Fig. 4. Ar-matrix infrared spectra of GlycolrCH and CD stretching range. Broken lines denote growing bands.

(methylene modes, skeletal stretching, and bending, and OH torsion modes). By fig. 7 the gIyco1 spectra in Xe matrix are documented as a typical example for various matrix effects. In tables 2-5 the observed absorption frequencies for the four species in both matrices are collected, together with the assignment, the calculated frequencies for both rel and re2 structures and the approximate description of the normal -modes expressed by potential energy distribution @‘ED).The peculiar irreversible changes under IR irradiation are exemplified by fig. 8, which shows the behavior in the 1200-1000 cm-l range of the do: Ar spectrum. Exposure to the IR radiation did not produce reproducibly measurable temperature changes by the thermocouples controlling the temperature of the copper block of the cryostat carrying the crystal window. Similar-relaxation processes were observed in ah other Ar and Xe spectra also. In fig. 9 segments of the spectrum do : Xe generated.from glycol thermal molecular beams at Knudsen cell temperatures 295 K and 520 K, respectively, are reproduced. These spectra wilI prove important in the

In all spectra reported in this work irreversible changes upon IR irradiation have been observed. It.is obvious that this phenomenon will play an important role in the interpretation of the vibrationa! spectra. The irradiation effects also bear a relation to the Knudsen celi experiments. As a consequence, this aspect of the glycol spectra will be discussed first, aiming at an interpretation of the relaxation process and use of this for separating out assignable spectra. The following statements appear relevant: -(i) In both Ar and Xe matrices the irradiation process reduces most of the initially present absorption bands to a limiting value. This.& shown in fig 10a for two decreasing bands. (ii) All bands increasing in intensity upon irradiation are initially present in the spectra. These bands. are marked by broken Ii& in the spectra (figs. 3-7) and by “f” in tables.2-5. From this data it is obvious, that the number of growing bands is sur&singIy small, e.g. 5-6 in the fmgerprint region of & isotopic species. Fig.. 10b shows the kinetics of growth for one increasing band of do (1172 cmyL). (iii) Knudsen experiments with increasing source temperature produce spectra with decreasing intensity of decreasing bands and increasing Intensity of increasing bands before irradiation as shown by tig. 9. The kinetics of increase with irradiation duration shown in fig:lOb for the 1.172 cm-l band_of d,-,:.Ar reveals that the Intensity of increasingbands .tend to a lhi&g‘vahre. The growth of the band is the&y&; the higher the

H. Frei et al.fEthylene glycol

1500cm-’

: 1200

1000

1000

277

900

BOO

700

w

CH2%HCH21BOq.

CH2OOCH200

_ Ar

CO2OHCO2OH

. Ar

Ar

1

1500cm-’

I200

1000

1000

900

BOO

700

Fig. 9. Ar-matrix .jnfrared spectra of glycol: methylene and skeletal mode region. Broken lines denote growing bands. dl : glycol-OdI impurity. D; dimers.

:

..

CHf,HCHf,H Ar-

._-

Matrix

MIA - 2000

Xe- Matrix

TV.

t

CH200CH200 Ar- Matrix

. 53omi’

Fig.

6. &-matrix

4so

infrared

4SO

400

300.

350

.’

.2

-

i50

uo

spectra-of glycol: skeletalbending tid OH.torsionjange. Broken lines _.~...

&iote

..~;

:

.:I

200

gr&ing

.. ,.

*: ,. ._1;

&dc. -:..-:

:_--

.:

. . .. ;.

-. -.. _:..

.-.

._

: _.:

..;_..:

:

:_..

279

.

..

-J---v-

:

i

I’

I

CH~~HCH~OCI Xe-Malrir

1%

MIA- 450

1

s -073 cm-l

3680cm“

3560

S=OOOcmm'

s-06scm-’ I

.1

.I

l..

.,,,,J

1200

1500 cm-’

1000

,1,.

1000

900

s.107n-n-

LJOcm-'

LOO

Fig.7. Xe&&k

350

5.090

,

-

840

550

S-300

300

250

infraredspectrumof glycol. Brokenlines denote growingbands.

Cm-’

200

cm-’

520

:

Table 2 Matrix

!R ~picti

o~C&0HCH20H

Assiinmeit fundamental ai, PED

Frequency (cm-l)

.

cahhited C)

observed b) Xe-matrix

Ar-matrix

vp(CH),[26] -v(CH)s[221 - v(CH)g [28] + Y(CH)IO[231 vsv(CH),[lS] + vKWs[131 + v(CH)g [351 + dCH)10[321 vg’v(CH)7]32] + v(CH)8[361 - v(CH)g[l31 - v(CHho[l81

ZJ$[CH2)o[57]

f 6(CH2)1[15]

w6KHh I571 - 6(CH2)0[201 vq7w(CHz)o[451 + w(CH2h [lOI - v(CC)[171 QOT~(CHZ)I I531 - ~~w(CH2)0[17l vn’rt(CH2)o [321 + rt(CHzh [281 +rwKHzlo [91 - rw(CH2h [81

vlrrt(CHt~o[491

-~t(CHzh

1211

i;

rel. int.

3681.0 ‘=) 3673.0 e, 3667.0 e, 3635.0 3629.0 3624.5 3617.0 3531.0 h) 3496.oW 2980.0f) 2968.0 2959.0fJ 2951.0 0 2940.0 2902.0 0 2891.0

w

i885.0 1494.0 0 1492.0 0 1470.0 1468.0 1461.5 1459.5 i415.5

Tel

*ez

ml. int.

ii

0

W

3649.0

s m W W w VW VW

3644.0

s m

3680.9

3681.0

3593.0

s

3640.0

3640.1

m

2964.0

2963.6

m

2955.1

2956.0

m

2887.5

2887.8

W

2885.2

2886.0

1487.0 f) VW w W W W

1383.0 1350.0 1347.0 1284.0 1271.0 1255.0 1246.0 1239.0 1172.0 f) 1163.0 1160.0 1159.0 1102.0 1100.~ 1098.0

m w w yw m W W W

1079.5 1073.5

VW m

1465.0

W

1474.7

1471.5

1458.0 1414.0

W W

1456.5 1403.9

1462.2 1407.8

1382.0

m

1386.3 1353.2

1378.4 1354.8

1343.0

m

1272.0 1258.0 1237.5 g) 1234.0 1170.0 t-J 1160.0 1156.5

W

1269.3

1267.9

m w

1240.7

1235.3

W

1171.3

1173.6

1102.0 1099.5

m m

1101.5

1106.3

1091.0 f+) 1081.5 1075.0.

W

m m

VW

m

_z-

‘.

:

.:.

H.Rei et al./Ethyleneglycol

:

281

-Table 2 (co&rued) ~Assi&rmknt fundamental a), PED

:.._ ..-

Frequency (cm-r) calculated c)

obsirved b)

Ar-matrix

Xe-matrix

vllv(c0)0[341 - v(CO), 1171 + rt(CHz)o I91 - rt(CHs)r [51 -G(COH)a[13]

1069.0 1062.0 1054.5 1043.0 1041.5 1040.0 1038.0 1027.0

j361 - 7r(CH2)0 I361 + V(CO)I r121 - v(CO)o[71 rJt9~u(co)o]251 +v(co)i]10] •t P(CC)[331 - -rr(CHzh I71 -rr(CH2)0]51 y~s(cco)1[391

- 6(cc0)0]371

vzl’7(OH)e[58] + 8(CCO), [12] +s(CCD)1[6] ~~~~s(CCO),[271 +s(CCO)c[191 --7(OH)o[311 V23*~(DH)r[811 v24*7(CC)[67]+s(CCO)o[8] + 6(cc0)1[71

1071.5

1042.5

1059.5

1059.0

s

VS

104?.0 f) 1040.0

S

s

1037.0

"S

1003.o h)

889.4

885.7

864.6

859.1

508.7

511.4

368.7

367.1

305.8

305.8

267.9

267.1

191.8

194.5

s vf

r)

1009.0 h)

VW

1004.5 b)

VW

880.0

m

886.0

VW VW m

865.0 863.0

m

873.0

S

m

866.0h)

WI

533.0 529.0 391.0 383.5 369.0 324.0

w

891.0h) Wj'7r(CHZh

1078.2

rel. int.

m

f) e) 9

re2

i;

rel. int. Yr6VJ(CO)r~211+rJ(co)c[13] --v(CC)[301 - 7rKH211 I71 -7rm-h)or61

re.1

519.0 514.0 367.0 361.0 320.0 309.0 293.0 9 268.0

w m s s W

w S s

m w

m w

212.0 207.0

s S

a) Discussion in’terms of structure rcr (covering symmetry Cr). Symmetry species of all fundamentals ia a. Numbers in square brackets indicate percentage of potential energy distribution. Signs reprasent relative phases. Symbols are defmed in section 5.2.1. The fragment CHsOH with internatty bonded OH group is denoted by 0, the fragment CHsOH with free OH group is denoted by 1. b) Accuracy i- 0.5 cm-t. C)Byuk(k=1,2 , _..,24) denoted frequencies (dominant site in AI) were used for least squares fit. v24: an estimated Ar matrix frequency 25 cm-r lower than the Xe matrix frequency was used for the tit. d) The sMfts~((CHs160H)2) - ii((CH2180H)2) from Ar matrix spectra were used for least squares fit_ e, Decreasing band, overlapped by an increasing band. r) .Increasing band. 9) Was not reproducrble. b) Dimer. beam source temperature, for the number of molecules at time to increases with increasing cell temperature. (iu) Ihe relaxation process induced by IR is slower by a factor 3-4 for do: Xe than for however-the spectra experience’in both cases

relaxed Knudsen radiation do: Ar, the same

changes within error limits. (u) AU isotopic species showed the same kinetics Within error limits, however detection of the relaxation rate constant is bound to large uncertainties. The following interpretation is based on the kinetic behavior of the typical OH group modes, v(OH) and

.. +j:;:

i ;--...- -.. .:.. .. i

H. ki

.-.,

tit al~/Eihyleneg&ii

..

-: : ---

Table 3 .-

&k&ix IR ;

spectra of CH2180HCH2180H

As&Gent :f&damental

.~

Calculate&

Observed frequency (cm-t).-

u3v(CH)7[23] + v(CH)9[22]

- v(CH)g[28] - v(CH)to[25]

v,,v(c~),[26]

-

- v(CH)9 [28] vg~~(CH)7[l8] + v(CH)~ 1351 ~~w(CkI)~[32] - ~(CH)~[131

+ dcH)IO [231 +~(CH)e[l3] + &xi)10 (321 + u(CH)s[361 - x4CE0,,[~81

dCU3t221

wo*rW(CHz)1ISll

W32)012~1 il21

+ rw(CH2)1

- rW(CH2)0[2ll

v11w(CHz)o[331 i- dCHzhj301~ +TW(CHZ)O~~I -7w(CH2)1171

Y13’%(CH2)0[50]

VM~~(C’*OH)I

- 7:(CH2)1 t2l1

1421

l

7r(CHz)l]141

+-~r(CH2)0[51 + -dCH2hP11

~~ss7r(CH2h1351l

-

6(C180H)1

MWA[111

[31] + v(C%),

[S]

u(c180)or131 - v(CC)[28] - 6(C’*OH)1[ 101 v~7’u(c’*o)o[31] - v(c’*o)1[171 + .y&Hz)o 191 - rt(CHz)i [Cl - 6(C’80H)o[13]

v~~w(C’*0)r[301 +

1*0-shit

ir

rel. int.

s

11.5

3638.5 f) 3632.5

m w w

11.0

3583.0

5

ml. int.

3670.0e) 3662.0 e, 3655Se) 3624.0 3618.0 3613.5 3606.0 3520.0h) 3485.0, b) 2979.0 0 2969.0 2959.0 f) 2952.0 fJ 2939.0 2902.0 fJ 2889.0

W W

2885.0

U7*6(cH&~[57] * W.X2)1[151

Vg’?kr(CHz)0[42] - v(CC)[l8]

-d)

fel.

rez’

m

12.5

12.5

s

12-4

12,4

m

0.0

0.0

m

0.0

0.0

m

0.0

0.0

w

0.0

0.0

s

k VW

.

1488.0 f,

1495.0 9 1493.0 f) 1470.0 1468.0 1461.0 1459.0 1413.5

VW w w W W

1381.0 1347.0 1344.0 1276.5 1262.5 1251.5 1243.0

m w w VW m WJ w

2.0 3.0

s m w

1.5

1171.of) 1161.5 1158.5 1157.5 1093.0 1091.0 1066.0 1062.5 105&O 1051.0 1026.5 1022.0. 1014.0 f,

-.

Xe-matrix

Ar-matrix

&‘6(CH2)1[571 -

:

lsO-shifts (cm-t)

a), PED

0.0

1465.0

w

0.4

1.8

0.5

1458.0 1412.5

w W

0.3 1.1

Ll 0.2

1380.5 1378.0

w m

2.6 2.9

1.6 3.0

1339.5

m

1261.0 1252.5 1237.58) 1233.0 1169.0 f) 1157.5 1155.0

m m VW VW

3.3

5.2

2.2

2.5

m m

3.6.

2.6

2.0

8.5 3.0

1094.5 1092.5 1068.0

W

m w. m m .s vs s

9.0

.. 1058%

il.0 -15.0.

_.. :’

._

iO48.0 lO25.0 1022.0 >pJ2.ofj

w w..

8.0

8.9

10.8

8.7

14.8

12.6

w m

s .s :vs:-

.:

.:‘..

. . . .. 1

H. .Frei et aI./Ethylene gfycol

283

Table 3+cxrrinued) Assigonient fundamental

Observed frequency

(cm-*)

Calculated r80-shifts (cm-r) d)

a), PED Ar-matrix rel. int.

3

mos h)

VW

h)

VW

997.5 YIS17~(CH2)1

[34] - 7r(CH2)0[351 +u(C~*O)~[I~] -u(C’*O)~[8] vpJ’v(c’*o)~[291 +v(c’*0)1[111 t v(CC)[301 - rr(CDz)r 171

-

7r(CH2)0

Xe-matrix “O-shift

870.5

m

9.5

854.5 852.0

m m

11.u

512.0 507.0 362.0 355.0 326.0 gl 317.0 306.0 288.0 0 267.0

W

iT

ml. int.

re2

gg4.o W

VW

876.0

m

8.3

10.4

862.0 854.0 h)

s

12.0

11.5

W

8.0

8.1

4.2

4.0

4.8

5.7

3.8 5.3

1.4 6.6

[sl

v20~6(Ccr80)r

[38] - 6(CC’*O)o[371

val&30~),,[61] +6(CCL80)r[6]

+ ~(cc’ao)o[11]

v~~~~(CC’*O)~[~S] + 6(CC’*O)o[18] - &aOH)o [29] v2g~7(‘*0H)r [SO] +4,~(CC)[66] + 6(CC’*O)o[lO] tS(Cc’*O)r [9] For footnotes

Assignment fundamental

m

525.5 521.5 388.0 380.0 365.0 320.0

7.0

s s VW

6.0

W

m

w w 5 s

m W

3.0 291.0 f) 1.0

w

S

210.0

see table 2.

Table 4 Matrix IR spectra

of

CHaODCHaOD Frequency

(cm-‘)

a), PED calculated c,

observed b, Ar-matrix

vr1v(CH)7[23] +u(CH)g[22]

-n(CH)a[28] - dCW10[251

UzrV(CH)7[261

- U(cH)8[221

-v(CH)g[281 +.NCWlo[231 U3dcH)7[181 +mi)g[351 v4w(cH)7[32]

+ Vm)8[141 + mih0~321 + dcH)s[351

-u(CH)g[Ml [l’#l

Xe-matrix

rei

rez

m

2964.1

2963.6

m

2955.0

2956.0

2891.0 2888.0

w m

2887.5

2887.8

2882.0

w

2885.2

2886.0

2714.0 e) 2706.0 e,

W S

2682.9

2682.8

2682.0 2679.5

m VW

i;

vS~u(ODh

rer

2982.0 f, 2970.0 2954.0 0 2950.0 f) 2940.0

rel. int.

iT

rel. int.

2900.0 f) 2896.0 f-J

- dCHhoV91 2694.0 f) 2691.5

S

s

M-matrix

U’&C~~)O[‘#

+

WHzh[321

w~(CH2h [441 - 6(CH2)0:371 wrw(CWoI331 + r&h.), 1201 -v(CC)[l9] ~1owKH2h [471 - av(CWo[361 nrrt(CH2)o[461 f -rtWhh 1441 ww(CWo[441 - 7tKH2h 1431 Q3’7dcH2h +~CO)I

1271 + 7r(cH2)0

PI +

G-%3[81

[251

q4~dCOh [281 - dWo[W _ -YACHI)I WI + ~dCH2)0[7l + dCH2)1161 - 7t(CH2)0[51 vl5~v(co)ow + V~COh [91 -v(CC)[291 -~r(CH2)0[81 -rrCH2)1[41

v&(COD)o[33]

+ WOD)1(9]

+rrKHz)o[l8l -rr(CH2!l[7i -Y(co)I [91 - Y(co)o[71 - v~cc)[lol vlq*WOD), [371 --r&332)1 [I31 +Y(cc)[ll] v~~~~(CCO)~[371-

2678.0. 2674.5 2669.5 1494.0 0 1471.0 1465.5 1462.0 1457.5 1399.5

6(CCO)o[37]

~&(CC0)~[22] + 6(CCO)1(21] -s(CC)[21] + r(OD)o[16] ~a~v(OD)~[52] - 6(CCO), [18] -L(cco)o[1S] V23’T(OD)t[45] - T(CC)[26] -ND), 1201 ~~,+v(OD)~[48] + 7(CC)[41]

2652.0

1144.0

w

1096.0

m

1087.5 1083.0

s s

918.0 915.0 896.0 f)

836.0

814.5 507.0 504.0 500.0 346.0 342.5 276.5 f) 255.0 249.0 241.0

rel.

int.

s

-_

.. .~ 2653.6

2653.5

‘1467.9

i463.3.

1451.5 1396.5

1456.5 1400.3

W

1378.3 1339.7 1239.8 1137.1

1377.1 .X342.4 1240.4 1130.6

m

1096.6

-1089.5

s

1076.8

1084.7

VS

966.8

970.9

s s m

912.5

912.0

833.9

835.1

.1489.0 9

1370.0

1046.0 0 1045.0 f) 966.0 958.0

~17’7dCH2h 1231 - 7dCHdow -~(C0)~[27] f 6(COD)., :I81

.-

1465.0

W

1459.0 1396.0

W

1366$1

w

W

1238.0 -1144.0’ 1108.0 fj 1094.0

VW

1084.0 1079.0 1070.0 1052.0 f) 1043.0 0

VS

VW

m VS s S

m

961.5 958.0 920.5 917.0 895.0 f) 841.0

839.0

m m

835.0

W

821.0 814.5 518.0 514.0 508.0 ._ 351.0 345.0 288.0 f)

..

s m

813.2

-801.6

492.2

497.6

m W. W

m m

338.5

s

248.7

333.6

:

s S W

269.0 256.0

.

246.4

m ..

For footnotes see table 2.

. .

217.0

199.6

160.5

.. 182.3

285

k. _Fkeiet al./Ethylene glycol

&ignment. futidamentala), PED

Frequency (cm-t) -. observed b)

i&kited

Ar-matrix

VULVA

[loo]

~w(CD)~[~O] - v(CD)e[29] +Y(CD)~[~O] -v(CD)te[26] 2UIO v41v(CD),[27] -v(CD)&31] ~5*v(CD)7[18] +v(CD)g[37] q*~(CD)~[31] -v(CD)g[8] 01

-v(CD>~[~O] + ~(CD)~o[20] + v(CD)8[9] •i-v(CD)10[33] + v(CD)e[39] 1~(CD)10[18]

+u13?

VI2 f v22?

v,*G(COH), [78] Y8’V(CC)[29] - ‘~w(CD2)1 [lS] -r&D2)0[151 -1~(C0)1[131 - v(c0)0[141

vg4COH)1[381 + u(CC)[16]

+

6(CD2)1[2’31

tqev(CO)t [28] - i(CO)o[9] +rw(cD311191

-m~(CD2)o[lll

-qcOH)I[16] q1*6(CD2)0[23] f WDz)1[14] -6(COH)1[16] -I~(COH)~[~] +.“(co)o’[l?l ~12~6(CD2)1[lql--6KD2)0 [Xl

_ --rw(CD2)1[221 + 7W(cD2~0[201 v;3.v(co)l[iq

-;6(CDz)1[22] ~14”$D2)1[261

-7#%)1[91

~v(CO)o[l9] + 6(CDz)o[l7] +

rtWdoI2‘U

.~‘Y$Dz)o[~SI

Xe-matrix

ii

rel. int.

3680.0 e, 3672.0 e) 3666.0 e, 3635.5 3632.0 3629.5 3625.0 3618.0 2246.0

w w S

c)

‘el

re2

z

rel. int.

3650.0 0 3646.0

s m

3680.9

3681.0

3594.5

s

3640.0

3640.1

2228.4

2225.5

m VW W W W W

2233.5 2218.0 f) 2195.0

w

2204.7

2205.9

2137.0

m

2122.1

2121.0

2106.5 2095.5 f-J 2091.5 f-J 2076.0 1339.0 1319.0 1313.0 1302.5

m

2114.4

2114.8

1319.0 1294.9

1323.5 1293.4

1171.7

1164.8

“S

1134.2

1118.7

1109.0 1095.0

w

1102.6

1114.1

1061.0 1033.0 f) 1028.0 f)

s

1071.2

1060.8

m

W W W

m m

1279.0 f)

VW

W

.m

1279.0 f) 1269.0 f) 1200.0 9

1165.0 1162.0

nl m

1123.0

vs

1115.0 0 1110.0 1095.0

W

1064.0 1029.0 fI 1027.0 r) 980.5 978.0 963.0 958.5

1340.0 1315.0 1308.0

VW s

1158.0 1130.0 0 1121.0

s

VW

S

m W VW

976.0 964.0 958.0

986.0

984.0

961.7

972.8

S VI”

w W

.

286

..

.- I. ..:. ..

.-

H.’l&i et al./Eihylerie gljcd:

,.-..-

.:

Table 5 (conrinuedj

kssignmi,fit-.

-.

Freqticncy (cm-‘).-

.fundariental a), PEi

observed b)

-.

:

ml. int.

i; + dCO)r[lS]

-r&%z)1IW i- ~r(CWo[161

--7w~CD2)01211

_WT&%)r[191 +7r(CD2)1[111 --6(CD2,?[91

+-rr(CD2)0[~21 -

WD2)~[10]

witKDdoP61 - rt(CDd, 1391 vuvKC)[27] + ~&D2)1[13] - ~r(cbh[~l + 7w(CD2)0[131 -7rKD2)0[131 ~lg~~r~c~zh

1421 - ~~~~~~~~

1321

v~~~~KCO)~[33] - 6(CCOlo[321

w7(‘3H),~ [831 2u24? q2,6(CCO)1[34]

+ 6(CCO)o[331

V2J’T(OH)I 1921 ~,dCC)[801

942.5

m

894.0 891.5 886.0 881.0

W

I .. : I

calcidatedc).

: :

Ar-matrix

vrs’v(CO)o[251

.-,.

Xe-matrix ii

Gl. rd.

k2

int. 941.1

930.1

886.7

897.3

847.5

844.6

735.5

747.8

733.1

136.1

450.5

451.2

w

347.0

351.0

288.0 218.0 f)

W

287.8

287.2

263.6

264.1

208.0

S

177.8

176.2

943.0

m

VW

886.0 882.0

W

W

748.0

m

750.5

W

741.5 739.0 444.0 441.5 374.0 368.0 364.0 359.0 351.0 335.0 282.5 274.0 f) 264.0

m

745.5 740.5 457.0.

m

VW

W

VW

m

VW

w

W

388.5 380.0 374.0 367.0 363.0

VW

w xlm S

S m m

W.

VW W W

For footnotes see table 2.

T(OH) and results of the normal coordinate analysis discussed in section 5.2.1. Furthermore, results on internal H-bonds of I ,2 dials published by Kuhn [ 141 and generally accepted structure OH group mode relations will be used 1151. Considering the facts reported aforehead, we arrived at the following conclusions: (i) The y(OH) region of Xe matrix spectra of all isotopic modifications consists of two decreasing bands, e.g. for do:Xe at 3644 and 3593 cm-l. Similarly for alI decreasing spectra the 6(COH), 6(COb) modes exhibit a splitting of 100-150 cm-1 again indicating the existence of one free and one internally bonded OH group. Finally all spectra show two OH torsion modes with frequeutiy difference of 90 cm-l, the lower lying mode having a frequency of approximately 270 cm-’ and much lower intensity.

These facts consistently indicate the existence of one free and one internally bonded OH group in all conformers, whose concentration decreases upon.irradiation. Since ab initio klculations predict conformers of this type to represent the most stable nuclear configurations, we infer the decreasing species to have a : conformation of type rel or re2. .For the sake of brevity the,fragment CH20H with‘ free and intemahy bonded OH group will hence forward be.denoted by subscript 1 and 0, respectively. The decreasing spectra in both.Ar and Xe matrices exhibit numerous multiplets. The origin of ‘Aese will be discussed below, but it might be:useful to mention already now that they will b@nterpreted in .terms of. : slightly--distorted conformations of type rel. -. (ii) As stated the increasing spectra feature only I few bands in the fmgerprint region. This phenomenon

-_-

H. Frei et aI./Ethyleneglycol

287

9

“yi i

1

CH2OHCHFH Xe

CH2OHCH20lj

MIA-L50

Ar MIA-2000 +.+240

T=520K

T-295K

s = 0.69 cm-’ I

.

1200cm-'

I

I

:I

1100

1000

i

1200

I 1100

-I

I

Fig. 8. Glycol: Ar IR spectrum: Irreversible change by IR Irmadiation. Key: to: spectrum taken after deposition to + 240: spectrum taken after 4 h of irradiation. Irradiation by Globar (T= 1200 K) of spectrometer, sample at LHe temperature.

*

!

12oocrn-' 1100

1000

.

I

I

1000

1200

.

I

1100

.

8

1000

Fig. 9. GlycokXe IR spectrum: Knudsen cell experiments. Key: 295 K: spectrum deposited with Knudsen cell at 295 K, 520 K: Knudsen cell at 520 K.

CH20HCH20H

IA-Matrix

E(t)

I

E(t)

1172 cm-'

E&I

E&J

L35 K

a

b

Fig. 10. Glycoi:Ar: Kinetics of the relaxation process. (a) Kinetics for two decreasing bands. (b) Kinetics of the growth for one increasing band for various tiolecular beam source temperatures. E = log f/IO.

288

H_fiei et aL/EthyIene glycot

excludes any identification of the conformation of the particles produced by the relaxation process. However, -from.the fact that the increasing spectra in all cases lack consistently bands typical for.bonded OH, they may be assumed to originate from a continuous set of conformations which all feature no internal H-bond. This would explain the survival of only few bands, name& those which are nearly independent on conformation. 5.2. Analysis and assignment of vibrational spectra The vibrational analysis will be kept within the following methodological framework: (1) From the set of experimental data of all isotopic species only those subsets will be subject to vibrational analysis which may be associated with decreasing spectra, i.e. spectra which decrease under IR irradiation. (2) The analysis &ill be restricted to conformations of the types rel and re2, i.e. to the conformations with internal H-bridge predicted as the most stable ones by ab initio calculations. (3) The analysis will be based on empirical arguments for qualitative assignment of group frequencies and on normal coordinate analysis (NCA). (4) As is obvious from the data, all vibrational spectra (decreasing spectra) are complicated by the existence of many multiplets. In the analysis below they will be treated by the following approach: (i) in the starting phase of NCA the centers of the multiplets were taken as zeroth order fundamental frequencies; (ii) independent NCA’s were carried out for rel and re2 conformation separately, aiming at a harmonic force field by fitting the data sets defined sub 1. In the case of the IgO, species the isotope frequency shifts rather than the fundamental frequencies were considered as basic data. (5) The force fields resulting from the ret and re2 fit were then used to predict the spectra of re2 and rel conformations, respectively, in order to get the frequency shifts due to the change in conformation (G-matrix effect). Since most predicted frequency shifts widely exceed the observed matrix multiplet splittings, in particular in case of the group modes of free OH, this procedure has been used to exclude the possibility that the observed multiplets originate from .superpositton of spectra of the rel and re2 conforma-

tions. (6) Using again the harmonjc~force, fields. obtained according to (4). NCA was applied to conformations the internal coordinates of which were given the following.values.(cf. figs. 1 and 11) 27=2ref5,&, i=O, 1,

vi =uje+ 10,i20,+30°, riO= rbe + 0.01 - 0.05 A.

The predicted spectra were again compared with observed data. The comparison then served as a basis to trace back the observed splittings to conformational changes. 5.2.1. Nomal coordinate analysis (NCA) NCA was carried by the aid of a computer program published by Hunziker [ 161 and a sin&Red version by Meyer [17]. Geometrical structural parameters and internal coordinates used in the NCA’are depicted in fig. Il. In table 6 numerical values of the parameters common to both rel and re2 conformations are given. These parameters have been taken from the rs structures of 2-chloroethanol [I81 and methanol [19J_ The data imply that both CHI groups retain local C2v symmetry independently of conformation. The re values of the internal rotation angles were chosen the same as resulting from ab lnitio calculations: rel: 2re = 60”, uOe = -3O”, uIe = +60 0., re2: 2re = 60”, uoe = -45”, ule = -160” .

s

Fig. 11. Glycol structural model. For numericalvaluesof struttural parameterssee table 6.

289

H. Fr.4 et aI./Erkyteneglyool Table 6 Structural

parameters of g&co1 (common to reI and re, conformation) Bond

length (Aj a)

c-c

6)

Bond angle 1.519 1.411 1.093 0.945 0.945

C-O (R, = Ro) C-H (r, = r8 = r, = rto) Os-Hs (ri) 04

-&

C$

LCCO 61 LCCH (0, LHCH (01 LCIO~HS

= PO) = 0s = @g = @to) = @o) (6 I)

LC204He

60)

112.77=’ lllAOO 108.67“ 108.53” 1o853o

a) Symbols see fii. 11.

Internal coordinates typical for CH, groups and the torsional modes were defmed aa follows: Y,(CH~)~ = $(A07 + A@, - Av7 -40s> -Y,(CH~)~ = ;(A@, + A%+

,

Force constant a)

A(ps- Acpd >

KJ%

AO, - Acp7+ Apa) ,

rt(CH2)r

= 2-3/2(-A07

y,(CH&

= 2-3~2(-A0g + AOlo -A(p, I- Apt,),

Table 7 Valance force field of rel and re2 conformation (mdyn/A)

T,(CH~)~ = ;(-A07 + A@ + Atp7- A& ,

KR~=KR~

49784 (940)

KS H&I

4.2366 (1230) 0.6297 (170) 0.7705 (140) 0.6639(1i0)

0.7235 (130)

0.7326 (140)

Kr

HSO

yr(CI-Qo = ;(-A@g rCC = T3124 *

+ AOlo + Alp, - Aplo) > wH)o

= Q-6421

2 (4)

+H)l=753r2

(ref.

[201) -

The harmonic potential therefore is expressed in terms of 24 linearly independent valence type internal coordinates. Initial values for the force constants were taken-from methanol

[21] and 1,2-difluoroethane

WI.

5.3.Assignment of vibrational spectra

-

Ha, = Ho0 H .%t, %_ r%to =Hrwo H-ir - HYro H 'pl'=H90 H UO H “I

7.5612 7.3932 4.6467

rez (850) (830) (270) 4.9768(1030) 4.1419 (1410) 0.6493 (160) 0.7590 (160) 0.6576 (110)

QO

(3)

rel

1.4341 0.9609 1.3661 0.0497 0.0293 H2r 0.2341 ‘S.%VL =kYwo 0.0944 ks, -0.1998 ~R1,O1=~Ro,Oo -0.2517 [R =~Ro,r,vo-0.4417 'RI,&, L,rwl 0.2818 0.5006

(760) (750) (240)

(260) (ZOO) (520) (IO) (10) (50) (260) (440) (300) (270) (300) (540)

The results of the vibrational analysis are collected in tables 2-5, together with the experimental data. In the following discussion only points of specific interest.are included.

53.1. Harmonic force jield The procedure elucidated in section 5.2 leads to the VFF constants listed in table 7. The following comments should be made: (i) Interaction constants between internal coordinates, whose S-vectors have no nuclei in common, have been yeglected a priori in the VFF. Similarly in-

of glycol

%OPO

-0.0334 0.0667

(270) (300)

0.0158 -0.0161

(90) (40)

7.5621 7.3942 4.6491

1.4525 0.9266 1.4327 0.052 1 0.0325 0.2052 0.1321 0.0580 -0.2974 -0.4790 0.4705 0.2790 0.0716

(300) (200) (700) (10). (10) -(40) (320) (330) (320) (320) (500) (540) (80)

-0.3588 -0.0596

(470) (320)

0.0178

(170)

0.0549 (220) 0.0925 (190) -0.0 14 1 (60) -0.0265 (50)

a) Angular coordinates are multiplied by 1 A. For definition of coordinates see fig. 11, table 6 and eqs. (3) and (4).

H. Frei et aL/Ethylme gIycbI

teraction constants between OH and CH stretching coordina& and all pther coordinates were assumed to .be.?ero. Ftrthermore, for the fragment 0CH2CH20 localsYmmetry C, for the force field was maintained fOr.all conformations. ($ The VFF consists of 15 diagonal and 16 off diagonal elements for each conformation rel and re2. It is based on 6X observed frequencies (below 1500 cm-l) and I80 isotope shifts, respectively, and repiesents a best fit for the dominant component of the multiplets. Tables 2-5 give specifically the multiplet component which has been used in the fitting process. -(iii) Interaction constants with values not exceeding their RMS value were discarded. It should be remarked that for reproduction of observed spectra the force constant values have to be taken with all digits given, though not all of them are significant. (iv) As far as comparison is meaningful, the VFF is surprisinglysimilarto that of 1,2-difluoroethane [lo] _ (u) Table 8 gives a global view of the accuracy with which the observed spectra may be reproduced. It indicates a slight superiority for the rel conformation. Further support for this conformation will be obtained from a detailed discussion of the spectra. (ui) An interesting aspect follows from table 7 by comparing the VFF of the two CHzOH group 0 and 1. According to the assumptions defmed in section 5.2, differences in VFF may occur only within the two CH20H fragments. Indeed such differences may be quite large. The effect of the H-bond concerns diagonal constants and will be discussed below. A typical example for off diagonal elements is presented by the which are well defmed and even hrt0.60 =dhw1 Table 8 RMS deviations for rel and re, valence force fields Confor-

RMS deviation (cm-l) a) data sets

mation

rtx rez

31 force const. 92 freq.

28 force const_ 68 freq. b)

‘80-shifts

7.42 1.97

5.43 590

1.87 1.90

a) RMS values refer to zdc - zobs for do, Od,, d4 and to [iXt60) - ii(‘80)]cdc[Z(“O) - i;c’80)]obs. Minimized quantity: Z(AF/@ +Z[A(iT- i?)/lSO]*. b) VFF with K;1, K&, and K, kept fixed from observed frequencies below 1500 cm-t.

have opposite sign. These differences may be considered significtit and may serve as A fust &tipl& for dependence of force fields on conformat&. 5.3.2. Interpretationof multiplets In section 5.2 the procedure has been outlined according to which the multiplets occurring in all spectra observed in this work will be inter&ed. First the fact should be mentioned *at the free OH group modes 6(COH) and T(OH) do not show notice- -. able splittings, though according to NCA th&frequenties should depend sensitively on ul. Hence, the conformation of the free OH group (OH)1 may be considered as fued and causing no splittings by the presence of conformations differing w.r.t. ul_ Splittings predicted by NCA with varying values of T and rb agree well with the observed patterns for [2r--2r,l=

5-loo

)

I(, -rbel = 0.02-0.03 i%_ Table 9 documents this analysis for do : Ar by which the observed multiplet splittings are indeed reproduced quite accurately. It should be remarked that the m-dtiplets may also be explained by the same approach within the framework of the re2 structure. This interpretation of the multiplet splittings is supported by the fact that they are not highly reproducible and that the intensity distribution is different for Ar and Xe spectra. Fig. 12 demonstrates this fact in a striking way. -Basing on this body of arguments we consider the interpretation of the multiplets as originating from conformational changes caused by the matrix crystal field as sufficiently reliable. 5.3.3. Assignment The following assignments start from the results given in sections 5.2 and 5.3, i.e. from the interpretation of multiplets as related to conformational changes. It is specifically aimed at the dominant component of the multiplets of the glycol spectra in .Ar matrices in terms of the conformation rei. Arguments favoring r,, will be included in the discussion of the assignment. 5.3.;3.1_OHgroUpmodes, internalH-bondingand conformationalisotope effects Assignment-to free and bonded v(dH) mod&, 36.67 and 3626,cm-ldO:Ar

H. Frei et d/Ethylene

glycol

291

Table 9 do: Ar: Relation between multiplet splittings and conforma-

tion& changes of re, Fundamental no. v

Y,’ 1468 van 1460 I$’ 1415 v,o’ 1383 Yrt’ 1350 ~,a* 1271 1113’ I246 vt4’ 1163 Yrr’1100 vte’ 1067 Yt,’ 1040 qs 880 I+, 864 vso’ 514 ~21’ 361 V22’ 309 nas’ 268 vx -

Multiplet splittings cm-t obs.

talc. a)

2.0 2.0 0.0 b) 0.0 b) 3.0 14.5 7.0 3.0 2.0 9.0 3.5 0.0 2.0 5.0 6.0 11.0 0.0. -

6.3 0.8 4.2 2.2 4.0 14.0 9.4 4.0 0.4 6.2 2.1 5.7 0.8 86 7.3 10.0 0.6 1.1

a) Calculated for: re1(2te = 60”, voet, uret, rbe = 0.945 A) and rr(2r = 53’, uoet, uret.rb = 0.970 A). b) Broad band, Airr12= 5 cm-t.

1% CCl$lHC0~0H XL? MIA-B0

I-

1

900

1000cm-’

is straightforward, but deserves the following comments: (i) Internal H-bonding produces a relatively smaII shift v(OH)~-V(OH), (5 1 cm-l do : Xe and 41 cm-1 do: Ar) for centers of the multiplets. v(OH), 6(COH) and r(OH) modes show up more or less complicated multiplets, e.g. d,:Ar where v(OH), 3635.0,3629.0, 3624.5,3617.0, 6(CCH)o 1284.0, 1271.0, 1255.0 and TV 367.0,361.0 cm-l, whereas u(OH), 3667, 6(COH)l 1163, TV 268 cm-l. In relation to the shift v(OH), -v(OH& which would indicate weak internal H-bonding, the shifts for 6(COH) and T(OH) are surprisingly large. Apparently the latter two types of OH modes are considerably more sensitive to internal H-bonding than v(OH). This relation also expresses itself strikingly in the force constants associated with the typical OH modes which are given in tables 7 and 10. It should however be menti0ne.d that the 6(COH) mode is for alI species strongly mixed with other group modes. Consideration of frequency shifts therefore is less meaningful than of force constants. A

’

1

.

700

800

Fig. 12. Glycold4: Matrix dependence of maltiplets. Corresponding members are connected by vertical lines. Table 10 Glycol:Ar: effect of internal H-bonding and conformation on OH-group VFF OH group mode

OH valence force constant bonded

4X0 6 (COH) NH)

7.39 0.77 0.050

Shift (%) a)

free 7.56 0.63 0.029

-2 22 72

a) Shift relative to the force constant of the free OH group. further striking effect of the internal H-bonding in the vibrational spectrum is exhibited by the intensities of the T(OH) modes. In ah cases
is much more intense than TV, the latter being even difticult to identify in do : Xe and d4 : AL (ii) The modification Od, offers a possibility for

292

..:...

..

.

H. I+ei et a:./Eihylene glycol

S - 073 cm-’

3700 cm-’

S-072

3600

s=o65d

).

2640 1..

,

.

s=oso cm-’

, 1200

1500cni’

cm-’

2740

8.

_:.

1000

L.

IOOil

f.

900

.

..I..,

800

740

s=c9ocm-’

540 Cd

480

Fig. 13. do, Od,, 04:Ar:

conformational

isotope effect. Ke~i X bands of Odl. Broken lines denote growing bands.

..

H. Frei et al.lEthylene &co1

&scussi& of the existence of a conformational isotope effect involving two ftite coordinates [2 I]. The spectrum-of this species cannot be studied in pure form and’will be a superposition of the spectra of do, Od2 a&of the two conformers

Table 11 Predicted zero point energies (cm-‘) and approximate of population at 10 K of isomers of glycol-OdI ‘el ~~I(OD)I@H)OI

J%~(OH)~(OD)OI .&W%W)ol -J%WH)IW-VOI ~[(oD),(oH),l/~[(OH),(OD)ol

The available spectra give evidence for predominance of the form (OH)l(OD)o. First, the same frequencies of v(OH), and Ye are observed for do and Od, -and for Od, and 0d2, respectively. The same applies for v(OH)o and ma. The lack of splittings proves the coupling between v(OH)&(OD)~) and v(OH)~ (v(OD)r) to be very small. Second, fig. 13 shows in the spectrum d,, Od,, Od,: At the intensity ratio I{v(OD),,}~I{u(OD),} to be remarkably greater than ~{v(OH)~}/I{V(OH),}. Similar conclusions fo!low from discussion of intensities of the modes r(OH)r, r(OH)o, r(OD),, in the spectrum fig. 13 which can be located for do, (OH)I(OD),-,, 0d2 (very weak) as accurately predicted by NCA, while a torsional absorption of (aD)r(OH),, cannot be detected. Corresponding analysis of the G(COH)-region does not allow conclusive decision, however the behavior of the 6(COD),, 6(COD)1 modes confirms the foregoing conclusions. These experimental fmdings are confirmed by NCA. Table 11 shows the zero point energies of (OD),(OH)o and (OH)l(OD)o, predicted by the rel and re2 VFF’s of the four symmetrically substituted isotopic species, respectively. The predicted approximate ratios of population of the two isomers of glycol-Odr at 10 K agree very well with the observed matrix spectrum. The zero point energy difference of (OD)l(OH),-, and (OH),(OD), originates in the differences between stretch, bend and torsional force constants of the free and bonded OH(OD) group (cf. table 10). The 5.3.3.2. Fingetprin~region (i.500-200 cm-l) assignment m &ii region is best explained in terms of fig. 14, in which the dependence of frequencies of group modes,.the latter character&d according to the PED picture of normal modes, is shown graphically as

293

ratio

re2

17554.4 17552.7 17532.3 17535.9 22.1 16.8 0.04 0.09

a function of isotopic substitution. The following comments appear in order: (i) Though the covering symmetry of the ral conformation is C,, all vibrational spectra follow to a remarkably high degree the behavior characteristic for C, covering symmetry of the OCH,CH,O fragment: all typical grottp modes classify themselves as in and out of phase combinations (approximate a and b type fundamentals). (ii) The modes vll and v13, which consist mainly of Tt(CH2)0 and yt(CH2)r, in the d0 spectrum apparently do not experience isotope shifts by going from do to Od,, though the lower lying mode ur3 crosses with g(COH),-,. (iii) A drastic effect between do and Od, is observed for the modes IJ~~,v14 and vls which are composed of fi(COH) and the two yr(CH2) modes. By the substitution d0 * Od2 ~~2 and v14 experience a drastic change in composition. Nevertheless, in the Od, spectrum there occur two normal modes ur6 and vrg, which are related to v12 and PI4 by a nearly full (2)fj2-frequency shift. (iu> The normal modes 2~~4,v15 and VI8 (do) involve the -r,(CH$ group modes. Though ur4 and vr5 lies near 1163 and 1100 cm-l, respectively, i.e. at relatively high frequency, the assignment leaves little doubt about strong involvement of Tr(CHZ) (18% and 33% PED respectively). There are relatively few cases where yr(CH2) modes have been located reliably at such high frequencies. In the case of glycol this also is reflected in the high force constant, 0.96 mdyn/8 = 0.96 X lo- l1 erg rad-2. Further confirmation will come from the analysis of 160-180 isotope shifts. (u) The lower part of fig. 14 illustrates the isotope shifts within the set do-0d2-d4 of the normal modes below 500 cm-l. The most characteristic feature is the strong intermixing of the group modes S(CCO),,

500

cm

-1

Fig. 14. Glycoldo, Od,, d,: Ar: correlationdiagram

ofnormal

frequencies and broken lires calculated frequencies.

S(CCO), and T(OEI)~, whereas TV (free OH group) shows much less interaction with other group modes,

except intheOd,

system.

..

300

4cu

modes in the 1500-200

:

200

..

cm-t region:Solid

..

.

lines indicate observed

-.

-...

(Vi) The 7(CCj mode has been obsekd-in I& ma-. trices.near .2lO_c~~~_&th the sp&iesko;d4: 1802 !. only. O&q-to the matrixshift y$_~v~,~.25 cm+

,

II. Frei et al.fEthyIenegIyco1.

295

: ‘(CHZOH)~

--

.*r

-::

(CH2”OH)2

-(CHZOH)~, Xe

(CH:‘OH)2

Ar

Xe

:

-_I--

I

s- 069 cm-’ EU’

1100 1000 1100 1000

1100

1000

1100

1000crn“

Fig. 15. GlycoMo and Glycol-‘*02 matriv spectra (IlOO1000 cm-t section) t60-‘*0 isotope frequency and intensity shifts. Broken limes denote growing bands.

in the FIR region, this mode was not observable in the Ar spectra with the equipment available in this work. isotope shifts Though isotope 5 .3 .3 .3. r6&80 shifts?(160)-r(180) do not exceed 20 cm-r, they have been proven powerful in several respects. Such shifts are connected with only very slight changes in the normal vectors in general and therefore allow often much simpler interpretation of the vibrational isotope effects in comparison with 1H-2H effects. Fig. 15 shows sections of the spectra of the lsOspecies in Ar and Xe matrices, where the normal modes Table 12 Glyc~l-do and Glycol-‘*02

Internal

matrix spectra. t60-‘*O

vIS (do:Ar 1100, 1802:Ar 1091 cm-l).andvr6 (do: Ar 1067,180,~ or 1058 cm-r) are located. The PED composition of this pair of fundamentals is given in table 12. It may serve as an extraordinary example for a pair of two adjacent fundamentals whose eigenvectors are changing considerably upon 160 + l*O isotopic substitution, in contrast to the rule just mentioned. This 1 change in composition is coupled with strong changes in intensity, cf. fig. 1.5.The latter, in turn, allow to conclude that the intensity of the two fundamentals v15 and VI6 is determined essentially by the dipole derivatives associltted with the group modes v(CO)l, v(CO)~, 6(COH)1. From earlier analysis of the matrix spectra of methanol isotopic species it has been found that the same dipole derivatives determine intensity and their isotope effects to a large extent [2 11. An analogous effect is observed with Xe matrix spectra. As documented by fig_ 15, the effect in this matrix is even more pronounced. Furthermore this figure shows that there also exists a remarkable matrix effect on the intensities of the v15-v16 pair for both isotopic species. This intensity effect is explained by the differ: ence in conformatidn of glycol iu the two matrices. As discussed in detail in section 5.3.2 a change of 2re by 2-j’ is sufficient for explanation of both (relative) frequency and inter& changes observed. The spectra discussed in this work do not indicate the existence of Fermi resonances, in contrast to other polyatomics with 7-10 nuclei and 2 or 3 oxygen atoms, e.g. CH&OOH [22] CH,COCOOH [23]. 5.4. Assignmentin terms of the re2 sflz(cture In section 5.3 it has been remarked that the set of

frequency and intensity shifts of fundamentals

*is

v15 and v16 (PED)

*16

coordinate

WOH)I dCO)L

v(CO)o VKC) %(‘32)1

r&Hdo

(CH2GH)z 1100 cm-’

(CH2 ‘*OH)* 109 1 cm-l

(CHzOH)z 1067 cm-l

(CH2”OH)z 1058 cm-’

37 22 1 0 5.

31 8 0 4 11

2 21 13 30 7

10 30 13 28 4

28

35

6

0

-296

H. Frei et al.fEthyiene &co1

available glycol matrix spectra may be interpreted in terms of the conformation rel or re2 with a nonsignificant differ&ice in the RMS deviation. Since the assignments proposed in section 5.3 have been made for the rel structure, it appears appropriate to give some arguments for the preference of this conformation over the re2 structure. One important argument comes from the composition of the interrelated fundamentals v14, y15, vl6 and v17 for a0 and v13, v14 fGr Odp The four fundamentals of d, have rather different compositions for the conformations rel and re2, but these in both cases appear acceptable. For re2, the three Fundamentals of 0d2 would assume compositions which appear highly unplausible: v13: am

f v(CO),,(16)

+ Tr(CHZ)1(6) + T*(CH2)oW ~14: 7,W2)1W

by a similar analysis of the spectrum of c& , though the crowding of fundamentals in the 13002900 cm-l range renders conclvsions less direct in this case.

6. Statistical thermodynamical functiok The available molecular data of glyc01~~ in.rel conformation allow calculation of the molecular thermodynamic functions in the rigid rotor-harmonic oscillator approximation. Using ab initio results foi structural parameters one may calculate principal moments of inertia as follows:

Ia = 3 1.8086 uR2, I,=

3

+ T,P&JW

Ib = 94.1813 d2

,

111.2755~2.

For vibrational contributions the fundamental frequencies of glycol: Ar (cf. table 2) have been used. The results are collected in table 13.

+ vP-3 (26) I

q5: v(CO)l(8)-v(CO)o(31) -yr(CH2)I(9)+yr(CH2)0(11)-

7. Conclusions

This composition would suggest intensities in the orderI{v13} >I{vlS} S>I{V~~},in conflict to observation. The same argument applied for rel predicts the observed order of intensities qualitatively correctly, cf. table 4 and fig. 5. A further argument is furnished by the 160-180 isotope shifts for the same fundamentals v15, v16, v17 of do and 1802. Predicted isotope shifts deviate significantly from observed shifts for the re2 conformation, whereas the prediction for rel is accurate. This is clearly seen from the numbers collected in table 3. Preference of rel is also supported

The foregoing discussion should have made clear that the glycol molecule offers a sensitive probe for matrix effects on conformations. The conformational matrix effects concern mainly changes in the structural parameters of the internal hydrogen bridge in a rather defmite way, apparently leading to two frequent slightly different conformations whose nuclear configurations lie in a neighbourhood of the r,, structure. In an earlier paper [l] the possibility has been left open that the complex multiplet structure originates

Tab!e 13

Thermodynamic

functions of glycol (re,structure)

a) (JK-’ mol-‘)

T WI

-lga(T)-hO@WT

[ho(T)-hO(0)]/T

so

200 298.15 300 400 500 750 1000 1200

220.703 240.252 240.582 257.241 272.189 305.253 334.181 355.089

44.994 53.503 53.664 62.629 71.651 92.229 109.098 120.293

265.697 293.755 294.246

a) Ideal gaseous state, rigid rotor-harmonic

oscillator approximation.

319.870 343.840 397.482 443.218 475.381

co P 62.380 79.698 80.044

98.924 116.135 148.305 169.774 182.192

H. Frei et al. IEthyleneglycol from simultaneous presence of both r,i and r,2. The results of this work show that the spectra may be interpreted in a natural way in terms of rer alone. This does not contradict the approximately equal electronic energy of rel and re2 obtained by ab initio calculation. The experiment with glycol-Odi has demonstrated that in an Ar matrix the glycol molecule may pass over the barrier Vf,t = 2-3.5 kcal/mol between two r,, minima (fig. 1). Therefore, we have to expect that glycol may cross the barrier pz,2 in the matrix, too. This possibility of OH groups of changing their conformation in an inert gas matrix may explain the absence of rez at liquid helium temperature, provided that the vibrational ground state of re2 lies at least 20 cm-l above the ground state of rel. This (partial) equilibration of the conformers of glycol in a low temperature matrix has to be contrasted with a study of 1,2-dlfluoroethane, which showed no change of conformation of CH,F groups in an Ar matrix at 4 K [24]. A further important conclusion which may be drawn from the experiment with glycol-Od, concerns the correct prediction of the zero point energy difference between the two isomers of this isotopic modification by NCA. It confirms the assignment of the OH and OD modes in the various spectra as well as the significance of the difference of the stretch, bend and torsional force constants of the free and bonded OH group, respectively. Comparison shows that the two systems CH2FCH,F (gauche conformation) and CH,OH-CH20H feature closely analogous vibrational spectra w.r.t. VFF’s, frequencies and normal coordinates, except of course the typical OH-modes [lo]. This fact may be considered as a support for the essential correctness of the analysis of the matrix glycol vibrational spectra in terms of normal modes. A priori a molecular model with 2-3 finite torsional modes is suggested by the ab initio results, leading to splittings of vibrational levels as a consequence of the multiminimum potential. At the present time the normal mode description with inclusion of conformational matrix effects is felt to allow a satisfactory interpretation of the experimental facts. It, however, should be clear that high resolution and multiple resonance experiments might require extension of the molecular model by taking large amplitude motions into account. Regarding the fact that even in noble gas matrices the crystal field produces conformational changes it

297

appears rather questionable that IR and Raman studies of glycol in liquid phase could yield information about conformations. The delicateness of the conformational problem of the “free” glycol molecule expresses itself most clearly in the conformational isotope effect of the Od, modification. It remains an open problem whether by thermal molecular beam experiments further conformations without internal H-bond may be generated and detected.

Acknowledgement The authors wish to thank the Swiss National Science Foundation (Projects No. 2.110-0.74,2.?030.75 and 2.519-0.76) and to Messrs. Sandoz AG., Basle, for financial support. We wish to thank the ETHZ Computation Center for granting free computer time, the Swiss Federal Institute for Reactor Research for the supply of isotopic species and Miss V. Culkova, Mr. P. Matrell and Mr. P. Nyffeler for valuable assistance.

References [I] Tae-Kyu Ha, H. Frei, R. Meyer and Hs.H. Giinthard, Theoret. Chim. Acta 34 (1974) 277. [t] H. Frei, R. Meyer, A. Bauder and Hs.H. Giinthard, Mol. Phys. 32 (1976) 443. [3] L. Hickmann, J. Phys. Chem. 34 (1930) 635. [4] P. Buckley, P.A. GiSuere, Can. J. Chem. 45 (1967) 397. [S] W.K. Bustield, M.P. Ennis and I.J. McEwen, Spectr. Chim. Acta 29A (1973) 1259. [6] H. Matsuura. hl. Hiraishi and T. hliyazawa, Spectr. Chim. Acta 2tA (1972) 2299. [7] J. Almliif and H. Stymne, Chem. Phys. Letters 33 (1975) 118. [8] J.L. Whitten, J. Chem. Phys. 44 (1966) 359. [Y] W.H. Fink and L.C. Allen, J.Chem. Phys.46 (1967)2261. [IO] H. Frei, Thesis Nr. 5905, ETH-Zurich (1977). [ II] R.D. Werder, Thesis Nr. 3970, ETH-Zurich (1967). [12] P. Groner, I. Stolkii and Hs.H. Ciinthard, J. Phys. E, Scient. Instr. 6 (1973) 122. [ 131 R. Frey, Thesis Nr. 4391, ETH-Zurich (1969) [14] L.P. Kuhn, J. Am. Chem. Sot. 74 (1952) 2492. [ 151 CC. Pimentel and A.L. McClellan, The hydrogen bond (Freeman, San Francisco, 1960) p. 118. [16] H. Hunzilcer,1. hlol. Spec?ry.17 (1965) 131. [17] R. Meyer, private communication. [IS] R.G. Azralc and E.B. Wilson, J. Chem. Phys. 52 (1970) 5652. [19] R.K. Lees and J.G. Baker, J. Chem.Phys. 48 (1968) 5299.

2g8_:

. .. . :::

_. .:-:‘;:

.-...._’ ..I._.

~...

-.. :.

H. FIeiet~;.iEthylenkglycol

: :. --&I .E.Br. -W&n,:J;d.-Decius and P.C. Cross, Molecular vii+-: _jii, j ~.:‘ti&~.~c.ik$wHi& New York, 1955) p. 60.. ’.. ~[2i]:A%ietialla~h, _. _- .-. .:-:_ R. Meyer and Hs.H. Giinthard, J. Mbl. :. .r -,~pectry. 52 (1974) 94. .:. . . .. :

-.

..- _. I .:

.. :.

[2i] R. Meyer, T.-K. H&-H. l%eiand HnH. Giinthxd, Chemj: Phys; 9.(1$75) 393.:-. :.;; ;.:- .. .::. .;: : [23] H;H$lenstein,,private commuhicaeo?. ._-’ ._. [24 ] P. Huber-W&hIi~d Hs.~~j_n$a@, Chem. Phis. Letters30(1975)347. _,_ -.. ’._. ..

:

.;

..

,-