Infrared spectroscopy and ab initio computation in conformer determination of keto ester and diketo triphenylphosphonium ylides

Article abstract of DOI:10.1016/j.molstruc.2009.07.028

For stabilized triphenylphosphonium keto ester ylides acyl stretching frequencies of the anti ester group from HF/6-31G(d) calculations fit experimental values with Scale Factor, SF = 0.866, as estimated earlier for the diester ylides, but Scale Factor =

Full text of DOI:10.1016/j.molstruc.2009.07.028

Journal of Molecular Structure 936 (2009) 132–136
Contents lists available at ScienceDirect
Journal of Molecular Structure
journal homepage: www.elsevier.com/locate/molstruc
Infrared spectroscopy and ab initio computation in conformer determination
of keto ester and diketo triphenylphosphonium ylides
Clifford A. Bunton a, Fernando Castañeda b,
*
a Department of Chemistry and Biochemistry, University of California, Santa Barbara, CA 93106, USA
b Departamento de Química Orgánica y Fisicoquímica, Facultad de Ciencias Químicas y Farmacéuticas, Universidad de Chile, Casilla 233, Santiago, Chile
a r t i c l e i n f o
a b s t r a c t
Article history:
For stabilized triphenylphosphonium keto ester ylides acyl stretching frequencies of the anti ester group
from HF/6-31G(d) calculations fit experimental values with Scale Factor, SF = 0.866, as estimated earlier
for the diester ylides, but Scale Factor = 0.834 has to be used to fit data for the syn-keto group and for syn–
anti diketo ylides. The DFT functional BLYP/6-31G(d), with the literature Scale Factor = 0.9945, generally
gives good fits for both keto ester and diketo ylides. For both methods agreement between observed and
predicted frequencies is lowest for the syn–anti di-t-butyl keto ester ylide, although, unlike the situation
for diester ylides, one bulky alkyl group does not significantly degrade the fits. Comparison of predicted
and observed acyl stretching frequencies is useful in establishing conformations of these stabilized phos-
phonium ylides.
Received 28 April 2009
Received in revised form 20 July 2009
Accepted 21 July 2009
Available online 25 July 2009
Keywords:
Triphenylphosphonium ylides
Acyl stretching frequencies
HF and DFT calculations
Ylidic conformation
Ó 2009 Elsevier B.V. All rights reserved.
1. Introduction
Theclassicalstructuresin Scheme1 are inadequatebecause there
is free rotation about the P–C bond on the NMR time scale and con-
Phosphonium ylides with a single keto or ester substituent can
adopt two conformations [1–3] and syn and anti indicate orienta-
tions of the acyl group with respect to phosphorus. The syn con-
former is generally energetically preferred over the anti, due to
interactions between anionoid oxygen and cationoid phosphorus
and in favorable cases interconversion is slow on the NMR time
scale [4,5].
Ylides with two identical acyl groups can adopt three conforma-
tions and with different acyl groups four conformers could form,
but in most of the stabilized ylides with alkyl acyl groups exam-
ined to date one conformer is dominant, usually with syn–anti acyl,
but sometimes anti–anti groups, [6–11]. These generalizations only
apply to ylides with alkyl keto or ester groups and Aitken et al. in
considering ylides with phenyl or very bulky groups in the ylidic
moiety observed a number of systems where two acyl groups
adopted the syn–syn conformation [12]. There would be resonance
between these phenyl groups and the ylidic moiety and such inter-
actions and non-bonding interactions of very bulky groups should
not be important in the alkyl ketones and esters that we examined.
In the present work we examined triphenylphosphonium deriva-
tives and designated conformations as syn or anti depending on
orientations of the two acyl groups with respect to phosphorus
(Scheme 1).
siderable double bond character between the ylidic and acyl car-
bons. Classical structures can be written as zwitterionic [13] with a
P–C single bond and positive charge on phosphorus and negative
charge on the ylidic carbon but there should be extensive ylidic res-
onance with partial charges on these atoms. For alkyl diester ylides
syn–syn conformers have not been detected, but both syn–anti and
anti–anti conformers have been identified and in the syn–anti diester
ylides the bulkier alkoxy group is preferentially oriented towards
phosphorus in the crystal and in solution [9,11]. In the diesters only
one conformer is usually observed, but the crystalline dimethyl dies-
ter, X = OMe, is a 1:1 mixture of syn–anti and anti–anti conformers
[10,11], although only one set of 1H or 13C NMR signals are observed
in solution, possibly because conformers interconvert rapidly on the
NMR time scale. The use of Infrared, IR, spectroscopy does not have
this limitation, but prediction of frequencies is a problem, and this
method lacks the quantitative power of NMR spectroscopy.
Infrared spectra of phosphonium ylides are complicated [14],
but stretching frequencies of acyl and cyano groups can usually
be identified with limited overlap with other signals. For given
structures IR frequencies can be estimated by using a variety of
theoretical models, but predicted values are generally too high
and empirical corrections are used. Scott and Radom [15a], in a
very extensive study of organic and inorganic compounds, showed
that the calculated values can be corrected by use of
method-specific Scale Factors, SF. The estimated SF value for the
HF/6-31G(d) method was 0.8953, and higher level methods gave
SF values closer to unity, but overall results were not necessarily
* Corresponding author. Address: Departamento de Química Orgánica y Fis-
icoquímica, Facultad de Ciencias Químicas y Farmacéuticas, Universidad de Chile,
Santiago, Casilla 233, Santiago-1, Chile. Tel.: +56 (2) 9782856; fax: +56 (2) 9782868.
E-mail address: fcastane@ciq.uchile.cl (F. Castañeda).
0022-2860/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.molstruc.2009.07.028

C.A. Bunton, F. Castañeda / Journal of Molecular Structure 936 (2009) 132–136
133
R'
R'
R'
Keto ester ylides
1: R=R’=Me;
2: R=Me, R’=Et;
C
C
O
R
C
C
O
R
C
C
O
R
3: R=Me, R’= i-Pr;
5: R=Et, R’=Et;
4: R=Me, R’= t-Bu;
6: R= i-Pr, R’=Et;
8: R= t-Bu, R’= t-Bu;
+
+
Ph3P
C
Ph3P
C
O
Ph3P
C
CO2R'
Ph3P=C
7: R= t-Bu, R’=Et;
CO
R
O
O
Scheme 1. Conformation of syn–anti diketo or keto ester ylides, R and R0 = alkyl,
alkoxy.
Diketo ylides
CO R'
better than from this simple method [15b]. However, the SF value
for HF/6-31G(d), as applied to triphenylphosphonium diester
ylides, gave frequencies which were too high and a value of
SF = 0.866 gave reasonable results for a variety of diester and cyano
ylides [16]. The literature SF = 0.9945 for the pure DFT functional
BLYP/6-31G(d) was used for several triphenylphosphonium ylides
of known conformation, and this SF correction could be neglected
within the limitations of experimental error, although we apply it
to all the results from this method. Silverstein et al. [17] noted that
experimental wave numbers can generally be observed to within
10 cmꢀ1, which sets a limit to the required accuracy of a prediction.
For a number of triphenylphosphonium diester and cyano
ylides estimated acyl stretching frequencies from both the DFT
functional BLYP/6-31G(d), and HF/6-31G(d) with our revised SF va-
lue, agreed reasonably well with experiment [16]. However, for
syn–anti conformers with a small (Me or Et) syn ester group and
a large (isoPr or t-Bu) anti ester group the BLYP/6-31G(d) method
gave good results for the anti acyl group, but frequencies for the
syn acyl group were too low by 33–61 cmꢀ1 and well outside the
limits of experimental error. Corresponding values from the hybrid
DFT functional B3LYP/6-31G(d), with the literature SF = 0.9614,
were slightly higher than experimental values for the larger ester
groups and lower for the smaller ester groups, but here the SF va-
lue cannot be neglected. The HF/6-31G(d) method with the revised
SF gave reasonable fits with the large ester groups, but this SF value
was estimated from data on all the diester ylides [16].
The ability of a large alkoxy group in an anti ester residue to af-
fect the predicted acyl stretching frequency of the small syn ester
group indicates that there are interactions between these groups,
despite their separation by the ylidic carbon, and they are probably
not treated fully by BLYP/6-31G(d). We cannot transpose these
groups in diesters where the smaller ester group is always syn
[16], and were unable to isolate diester ylides with two bulky alk-
oxy groups, but it is possible to make these structural changes in
keto ester ylides where earlier evidence from X-ray crystallography
and NMR spectroscopy showed that the keto acyl group is always
syn and the ester group is anti, regardless of the size of its alkoxy
residue [6,7,18]. Experimental and predicted stretching frequen-
cies had been compared for some keto esters with a small (methyl)
syn-keto group and a variable anti ester group (Me, Et, isoPr, t-Bu)
[16]. For this series of ylides (Chart 1) the DFT functional BLYP/6-
31G(d) with the near unity SF gave good fits, unlike the situation
with some diesters, and the HF/6-31G(d) method, with our revised
SF = 0.866, gave good fits for the ester acyl group, but predicted
values for the keto acyl group were too high [16]. The ylidic syn–
anti dimethyl ketone [8a] was also examined and here fits were
good with BLYP/6-31G(d), but both HF/6-31G(d) and B3LYP/6-
31G(d) gave predicted values for the keto groups which were too
high, as discussed later. Examination of a series of keto ester ylides
with different keto alkyl groups and various sized ester groups
(Chart 1), should complement earlier results [16] on interactions
between syn and anti groups and extend the earlier limited exam-
ination of diketo ylides which had indicated that the SF value for
HF/6-31G(d), estimated for the ester groups, is inappropriate for
keto groups. Behaviors of keto and ester groups may differ because
of the geometries of the alkyl and alkoxy groups and the marked
electronic delocalization in the latter.
Ph3P=C
9: R=R’=Me;
10: R=R’=Et;
CO
R
Compounds 1-4, and 9 have conformations from X-ray crystallography [9b, 18].
Chart 1. General structures for keto ester and diketo ylides.
Conformations of most of the ylidic diesters and the methyl
keto ester ylides examined to date have been determined by X-
ray crystallography [9b,10,11b,18] and bond lengths and angles
are similar to those from the HF/6-31G(d) optimizations.
Observation of
p
-shielding [6,19] in the 1H NMR signals of a ter-
minal methyl hydrogen by a phenyl group indicates that an alkoxy
group in an anti ester residue is oriented towards phosphorus. Re-
sults of this test have been consistent with evidence from X-ray
crystallography and we used it in establishing conformations of es-
ter groups [11,20], but the sensitivity is low for t-butoxy groups
where the shielding effect is spread over nine methyl hydrogens.
2. Experimental
2.1. Materials
Some of the ylides examined in this work (Chart 1) were known
compounds described earlier [7,8a,b], and most of the keto esters
with small ester groups and various keto alkyl groups were pre-
pared by standard methods.
The new keto ester ylides, 5–8, were prepared by classical tran-
sylidation [20] of a alkoxycarbonylmethylenetriphenylphosphora-
ne, Ph3P = CH–CO2R0, with acyl chlorides, R–COCl, in dry benzene,
at 25° C, under an inert atmosphere. The reaction is an acid–base
interaction with a second equivalent of ester ylide, as described
for keto ester ylides 1–4 by Castañeda et al. [7]. Elemental analyses
confirmed the structures of ylides 5–8.
Ethyl 2-triphenyl phosphoranylidene-3-oxovalerate, 5. Yield 92%,
m.p. 124 °C, 1H NMR (CDCl3) dppm: 0.66 (t, 3H, CH3, J = 7 Hz); 1.10
(t, 3H, CH3, J = 7.5 Hz); 2.96 (q, 2H, CO–CH2, J = 7.5 Hz); 3.77 (q,
2H, O–CH2, J = 7 Hz); 7.5 (m, 15H). IR (KBr): 1558, 1658 cmꢀ1
.
Ethyl 2-triphenylphosphoranylidene-4-methyl-3 -oxovalerate, 6.
Yield 72%, m.p. 180–183 °C, 1H NMR(CDCl3) dppm: 0.63 (t, 3H,
CH3, J = 7 Hz); 1.04 (d, 6H, CH(CH3)2); 3.47 (m, 1H, O–CH); 4.03
(q, 2H, O–CH2, J = 7 Hz); 7.45 (m, 15H). IR (KBr): 1556, 1654 cmꢀ1
.
Ethyl 2-triphenylphosphoranylidene-4,4-dimethyl-3-oxovalerate, 7.
Yield 85%, m.p. 154 °C, 1H NMR (CDCl3) dppm: 0.65 (t, 3H, CH2–CH3,
J = 7 Hz); 1.30 (s, 9H, C(CH3)3); 3.60 (q, 2H, O–CH2, J = 7 Hz); 7.4 (m,
15H). IR (KBr): 1531, 1658 cmꢀ1
.
t-Butyl 2-triphenylphosphoranylidene-4,4-dimethyl-3-oxovalerate,
8. Yield 76%, m.p. 130–131 °C, 1H NMR (CDCl3) dppm: 1.03 (s, 9H, O–
C(CH3)3); 1.32 (s, 9H, CO–C(CH3)3); 7.35 (m, 15H). IR (KBr): 1577,
1661 cmꢀ1
.
2.2. IR spectroscopy
Spectra were examined on Bruker IFS 56 FT or on Leitz III-G
spectrometers.

134
C.A. Bunton, F. Castañeda / Journal of Molecular Structure 936 (2009) 132–136
Table 2
Unless specified, spectra were measured with a KBr disk. Some
Acyl stretching frequencies of syn–anti keto ester ylides with SF values for HF/6-
IR spectra were of new compounds prepared in the course of this
work, but some acyl stretching frequencies were for keto ester
ylides synthesized earlier and of known conformation. A few fre-
quencies were from the literature with compounds whose chemi-
cal composition was known, but whose conformations were not
established, but are consistent with our new evidence.
31G(d)a.
Ylides
Predicted
Observed
Calculated
SF Correctedb
1 Me keto Me ester
2 Me keto Et ester
3 Me keto i-Pr ester
4 Me keto t-Bu ester
5 Et keto Et ester
6 i-Pr keto Et ester
7 t-Bu keto Et ester
8 t-Bu keto t-Bu ester
1865
1865
1866
1865
1859
1849
1846
1856
1920
1615
1615
1616
1615
1610
1602
1599
1610
(1555)
(1555)
(1556)
(1555)
(1550)
(1542)
(1540)
(1550)
1663
1661
1652
1648
1658
1655
1647
1633
1560
1555
1552
1543
1558
1556
1531
1577
1681
1918
1908
1903
1915
1911
1902
1986
1654
1652
1662
1658
1654
1658
1658
2.3. NMR spectroscopy
The 1H NMR spectra were monitored at 60 or 300 MHz in acid-
free CDCl3 for characterizing structures of novel ylides and in
establishing orientations of CH3 alkoxy groups with respect to
the triphenylphosphonium group. Signal multiplicity allowed dis-
tinction between primary, secondary and tertiary alkyl groups.
a
With SF = 0.866 unless specified.
Values in parentheses are for syn-keto groups with SF = 0.834.
b
2.4. Computation
group in the anti ester group in the methyl keto t-butyl ester ylide,
4, is predicted to only modestly increase the stretching frequency
of the syn-keto group by 32 cmꢀ1, and have a smaller increasing ef-
fect on that of the anti acyl group, and for the t-butyl keto ethyl es-
ter ylide, 7 predicted and observed frequencies are similar. For
most of the keto ethyl ester ylides in Table 1 agreements between
observed and predicted acyl stretching frequencies are not very
sensitive to the bulk of the alkyl groups. Conformations of some
of the keto ester ylides were known from X-ray crystallography
[18] and 1H NMR spectroscopy shows that the ester groups have
the anti conformation (Table 1). For the diester ylides fits between
observed and predicted frequencies were significantly worse when
a small alkoxy group in the anti ester residue was replaced by an
isopropoxy or t-butoxy group [16], but for the keto esters such size
effects of a single group are much less important. Except for t-butyl
derivatives predicted and observed frequencies are in good agree-
ment and the frequency of the keto acyl group is always lower than
that of the ester acyl group. The hybrid DFT functional B3LYP/6-
31G(d), with the literature SF [15], gives frequencies of the ylidic
methyl keto ethyl ester, 2, which are too high for the keto and ester
acyl groups, (Table 1 and Ref. [16]), consistent with the HF contri-
bution in this method.
Although the revised SF value of 0.866 for the HF/6-31G(d)
method gives satisfactory results for diester ylides [16], it has lim-
itations for keto ester ylides. For the series with a methyl keto
group and a variable ester group observed and predicted stretching
frequencies are in satisfactory agreement for the ester acyl group,
but predicted frequencies with this SF are consistently too high
for the syn methyl keto acyl group, regardless of the size of the anti
ester group. The situation is the same with the small anti ethyl es-
ter group and variable sized syn alkyl keto groups, where the calcu-
lation gives too high frequencies for the keto acyl groups (Table 2).
The agreement would be worse with the literature SF = 0.8953
[15], but a revised SF = 0.834 gives reasonable fits for the keto
groups, while retaining SF = 0.866 for the ester groups. These gen-
eralizations break down for the di t-butyl keto ester, 8, where this
The general approach followed methods described earlier [16]
with use of Spartan ‘06 or ‘08 for Windows. Structures were opti-
mized by using the same method as that for prediction of the IR
spectra with no structural constraint. Acyl stretching frequencies
are rounded off to the nearest whole number. Convergence condi-
tions were as described earlier [16]. For a few compounds compu-
tations are with conformations assumed or known to be incorrect.
Computations are for isolated molecules, although spectra were
monitored in KBr disks, and later we consider evidence on acyl
stretching frequencies for some keto ester ylides which had been
examined in chloroform [21]. These results, were with compounds
with longer keto alkyl groups than those of ylides prepared here.
3. Results and discussion
3.1. Keto esters and diketone ylides
The DFT functional BLYP/6-31G(d) with the literature SF value
gave satisfactory agreement for all the keto ester ylides, except
the di t-butyl derivative, 8, where the predicted frequency for the
(assumed) syn-keto acyl group is significantly too low, but that of
the anti ester acyl group is in better agreement with experiment
(Table 1). Unlike the situation for the diester ylides [16] a t-butyl
Table 1
Acyl stretching frequencies of syn–anti keto ester ylides from BLYP/6-31 G (d) and 1H
NMR chemical shifts of the terminal CH3 of R0 showing
p-shielding.
R'
O
C O
Ph3P
C
C R
1H NMR
BLYP/6-31G(d)
Observed
O
a,c
1. R = Me, R0 = CH3
3.10(s)
0.63 (t)
0.75 (d)
1.10 (s)
0.66 (t)
0.63 (t)
0.65 (t)
1.03 (s)
1574
1566
1571
1572
1556
1548
1517
1523
1667
1560
1555
1552
1543
1558
1556
1531
1577
1681
1654
1652
1662
1658
1654
1658
1661
a,b
a
2. R = Me, R0 = CH2–CH3
1663
1652
1648
1659
1654
1652
1639
3. R = Me, R0 = CH(CH3)2
a
4. R = Me, R0 = C(CH3)3
Table 3
c
5. R = Et, R0 = CH2–CH3
Observed and predicted acyl stretching frequencies for diketo ylides.
c
6. R = i-Pr, R0 = CH2–CH3
7. R = t-Bu, R0 = CH2–CH3
8. R = t-Bu, R0 = C(CH3)3
Diketo ylides
Predicted
Observed
HF/6-31G(d)a
BLYPb
a
Conformations of ylides 1–4 are also obtained from X-ray crystallography [18].
For the syn methyl anti ethyl derivative, 2, predicted frequencies are 1604 and
9 Syn–anti di Me keto
10 Syn–anti di Et keto
9 Syn–syn di Me ketoc
9 Anti–anti di Me ketoc
1539
1533
1521
__
1577
1550
1543
1549
1625
1602
1590
1573
1718
1557
1535
__
1603
1590
__
b
1574
1559
__
1686 cmꢀ1 from B3LYP/6-31G(d).
c
For syn–syn and anti–anti dimethyl derivatives 1 (incorrect conformations)
__
__
predicted frequencies are 1569–1624 and 1625–1718 cmꢀ1, respectively. For anti–
anti diethyl derivative 5 (incorrect conformation) predicted frequencies are 1625
and 1712 cmꢀ1. For anti isopropyl keto anti ethyl ester 6 (incorrect conformation)
a
SF = 0.834.
SF = 0.9945.
Incorrect conformation.
b
c
predicted frequencies are 1623 and 1710 cmꢀ1
.

C.A. Bunton, F. Castañeda / Journal of Molecular Structure 936 (2009) 132–136
135
method modestly underpredicts the frequencies of both acyl
groups (Table 2).
oriented anti to phosphorus in a diester ylide can be close to the
sum of the van der Waals radii and interactions between these
oxygens should also affect the orientation of a syn acyl oxygen with
respect to phosphorus. Therefore evidence in the crystal and com-
putations for isolated molecules indicates that the geometry of one
ester group may affect that of the other, depending on the size of
the anti ester alkoxy group and the relayed effect on the acyl oxy-
gen of the small syn ester group.
Only two diketone ylides, the dimethyl and diethyl ylides, 9 and
10, were examined (Table 3).The DFT functional BLYP/6-31G(d)
functional gives good agreement in both diketone ylides, but HF/
6-31G(d) with SF = 0.866, as for ylidic diesters, gives frequencies
that are too high by ca. 50 cmꢀ1 and the fit is much better with
SF = 0.834. As expected from earlier work [16] incorrect conform-
ers give poor agreement between observed and predicted frequen-
cies (Tables 1 and 3).
Interactions between ester and keto groups are apparently
much smaller than in the diesters, because, except for the di-t-bu-
tyl keto ester ylide, 8, the DFT functional BLYP/6-31G(d) satisfacto-
rily predicts keto and ester acyl stretching frequencies (Table 1).
The HF/6-31G(d) method, with the revised keto specific and size
independent SF, also gives reasonable results (Table 2). For a syn-
keto group interactions between the alkyl group and oxygen or a
(saturated) alkyl residue of the other ester or keto group should
be less important than those between the oxygens in ylidic carbox-
ylic ester groups, so that factors that control acyl stretching fre-
quencies are apparently different in ylidic keto esters and
diesters and, for the former, should be less sensitive to size of alk-
oxy groups The pure DFT functionals have limitations in treating
London dispersion forces [15b,23], which should be important in
these stabilized triphenylphosphonium ylides, and to that extent
the potential ability of the BLYP/6-31G(d) method to fit observed
stretching frequencies for acyl groups of many of our phosphonium
ylides, and usually within experimental error, may be due to can-
cellation of errors which is more important in some keto ester
ylides than in the diesters. This DFT functional method [15b,24]
tends to slightly overestimate lengths of C@O bonds, but it has lit-
tle effect on bond angles and therefore on conformations. Although
the hybrid DFT B3LYP/6-31G(d) was developed to avoid weak-
nesses in the pure DFT functionals it does not perform well in pre-
dicting acyl stretching frequencies for keto ester and diester
phosphonium ylides, and the literature SF [15a] is apparently not
satisfactory for these ylides.
Conformations of some keto ester and diketone ylides are
known in the crystal [9b,18], and for most keto ester and diester
ylides the anti conformation of an ester group is established by
observation of p
-shielding of an 1H NMR signal in the alkoxy group
(Refs. [6], [7] and [22], and Section 2).These conclusions agree with
those from theoretical treatments. For the syn–anti diketone and
diester ylides, the syn acyl group has a lower stretching frequency
than the anti, because favorable interaction between an anionoid
syn acyl oxygen and the cationoid phosphorus slightly increases
the CO bond length, reducing the double bond character and there-
fore the force constant in the stretching vibration.
Some structures from the computations had earlier been com-
pared with those from X-ray crystallography, [9b,18], and bond
lengths and angles for some of the ylides are given as Supporting
material. For most of the simple ylides that we examined the lengths
and angles are not very sensitive to changes in the alkyl groups.
3.2. Use of Scale Factors
The application of simple computational methods in the esti-
mation of IR frequencies involves use of SF, although for the DFT
functional BLYP/6-31G(d), perhaps fortuitously, SF can be taken
as unity [15]. We only used the hybrid functional B3LYP/6-
31G(d) with a few phosphonium ylides and this method, with
the literature SF value, does not give good agreement with exper-
iment (Table 1 and Ref. [16]).
Some of the compounds used earlier in the establishment of
method-specific SF values had simple structures [15], but in the
phosphonium ylides there is extensive electronic delocalization
involving the whole molecule, so that motions in an acyl group, for
example, are coupled with other motions, such as those of phenyl
hydrogens and associated alkyl groups. In the keto ester ylides pre-
dicting frequencies by the HF/6-31G(d) method requires use of SF
values that are group specific, and a single SF is inadequate in treat-
ing acyl group IR spectra, with different values for keto and ester acyl
groups (Table 2). The selection of SF values would be a major prob-
lem if IR spectra were used in conformational assignments of ylides
with complicated structures such as those treated in Ref. [12].
Recent articles cover in detail the use of various theoretical
methods for the prediction of molecular properties [23–26], but
some are limited to small molecules [27].
3.4. IR spectroscopy of ylides and determination of structure
Examination of IR spectra is useful in identification of functional
groups and comparison of signals in the ‘‘finger-print” region al-
lows identification of structures [14,17]. Results given here and
earlier [16] show that comparison of observed and predicted acyl
stretching frequencies is useful in identifying conformations of sta-
bilized phosphonium ylides. The DFT functional BLYP/6-31G(d)
provides useful evidence on conformation, although we note its
limitations as applied to diester ylides with a secondary or tertiary
alkoxy group [16]. The HF/6-31G(d) method is also useful,
although structure specific SF values, lower than literature values
[15], have to be used for keto and ester acyl groups of phospho-
nium ylides.
The use of IR spectra for conformer identification has limita-
tions in examining ylides of similar chemical composition and con-
formation, because acyl stretching frequencies for diketones and
keto esters of a given conformation are generally not very sensitive
to changes in the alkyl groups and are easily identified (Tables 1–
3). As a test we examined the isomeric keto ester ylides, 11 and 12.
3.3. Limitations of the DFT functional BLYP/6-31G(d)
Provided that we discount results for the di t-butyl keto ester
ylide, 8, this DFT functional satisfactorily predicts acyl stretching
frequencies of keto ester ylides, even those with a single bulky al-
kyl group, although for syn–anti diester ylides with one bulky anti
ester group predicted frequencies for the smaller syn ester acyl
group are too low [16]. There should be non-covalent interactions
between the two ester groups in these conformers. In the crystal
the syn acyl oxygen is oriented between two phenyl groups, with
interactions between the oxygen and phenyl hydrogens, while
for the anti ester residue the alkoxy group is oriented towards
the face of the third phenyl group. These orientations are consis-
tent with structural predictions for isolated molecules and the
geometry of a large anti ester group may affect that of the small
syn ester group by an effect relayed through the phenyl groups
[16]. In addition the center–center distances between the oxygens
n
CO2
CO2Et
C5H11
Ph3P=C
Ph3P=C
n
CO
CO Et
C5H11
11
12

136
C.A. Bunton, F. Castañeda / Journal of Molecular Structure 936 (2009) 132–136
electronic delocalization in the ylidic system and steric effects have
to be considered.
The structure of 11 is known from its synthesis [21] and the ob-
served acyl stretching frequencies in CHCl3 are 1550 (keto) and
1640 (ester) cmꢀ1. The corresponding predicted values from BLYP/
6-31G(d) are 1558 and 1661 cmꢀ1 and HF/6-31 G(d), with the
appropriate SF, gives 1547 and 1640 cmꢀ1. So far as we know 12
has not been synthesized, but the corresponding predicted stretch-
Acknowledgments
The authors are grateful to Paul Silva and Ximena Ramírez for
their valuable work in production of this paper.
ing frequencies from BLYP/6-31G(d) are 1562 and 1661 cmꢀ1
.
Therefore comparison of observed and calculated acyl stretching
frequencies would not distinguish between these (hypothetical)
isomers.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.molstruc.2009.07.028.
3.5. Treatment of incorrect conformations
References
Observed acyl stretching frequencies for diester and cyano
ylides of incorrect conformations [16] estimated largely by the
HF/6-31G(d) method with SF = 0.866, do not agree with observed
values. In applying this test to the present results we note that this
SF value does not give agreement between observed and predicted
frequencies of keto acyl groups (Table 3), but the DFT functional
BLYP/6-31G(d) does not have this limitation, and we used it for
keto ester ylides with incorrect conformations (Table 1). An exam-
ple of the treatment of a diketone with incorrect conformations is
given in Table 3.
[1] (a) I.F. Wilson, J.C. Tebby, J. Chem. Soc. Perkin Trans. 2 (1972) 31;
(b) I.F. Wilson, J.C. Tebby, Tetrahedron Lett. (1970) 3769.
[2] (a) H.I. Zeliger, J.P. Snyder, H.J. Bestmann, Tetrahedron Lett. (1969) 2199;
(b) H.I. Zeliger, J.P. Snyder, H.J. Bestmann, Tetrahedron Lett. (1970) 3313.
[3] D.M. Crouse, A.T. Wehman, E.E. Schweizer, Chem. Commun. (1968) 866.
[4] (a) G.A. Gray, J. Am. Chem. Soc. 95 (1973) 7736;
(b) F.J. Randall, A.W. Johnson, Tetrahedron Lett. (1968) 2841.
[5] T.A. Albright, M.D. Gordon, W.J. Freeman, E.E. Schweizer, J. Am. Chem. Soc. 98
(1976) 6249.
[6] A.D. Abell, B.M. Clark, W.T. Robinson, Aust. J. Chem. 42 (1989) 1161.
[7] F. Castañeda, C.A. Terraza, C.A. Bunton, N.D. Gillitt, M.T. Garland, Phosphorus,
Sulfur, Silicon 178 (2003) 1973.
[8] (a) F. Castañeda, G.I. Recabarren, J. Hu, N.D. Gillitt, C.A. Bunton, Phosphorus,
Sulfur, Silicon 178 (2003) 2505;
4. Conclusions
(b) M.P. Cooke Jr., R. Goswami, J. Am. Chem. Soc. 95 (1973) 7891.
[9] (a) F. Castañeda, C. Aliaga, M.T. Garland, J. Hu, N.D. Gillitt, C.A. Bunton,
Phosphorus, Sulfur, Silicon 182 (2007) 151;
(b) F. Castañeda, C. Aliaga, C.A. Bunton, M.T. Garland, R. Baggio, Acta
Crystallogr. C61 (2005) 496.
[10] F. Castañeda, C. Jullian, C.A. Bunton, M.T. Garland, R. Baggio, Acta Crystallogr.
C63 (2007) 267.
Examination of acyl stretching frequencies of keto groups in a
series of keto ester ylides by the HF/6-31G(d) method shows that
different SF values are required for keto and ester acyl groups in
triphenylphosphonium ylides. The SF estimated in this way with
keto esters also gives satisfactory fits of predicted and observed
frequencies for di keto ylides with the expected conformation.
The DFT functional BLYP/6-31G(d), with a near unity SF, and a sim-
ple basis set, gives good agreement with experiment for acyl
stretching frequencies in di keto ylides and most keto ester ylides.
For the di t-butyl keto ester ylide, 8, the difference between ob-
served and predicted frequencies of the keto group is beyond the
experimental error, although differences between observed and
predicted frequencies with only a single bulky substituent in the
keto or ester group are much less than those for diester ylides with
a bulky alkoxy group in the anti ester residue. The non-covalent
interactions involving syn–anti diester groups which may degrade
frequency predictions [16] are apparently much less important in
keto ester ylides than in diester ylides. Comparison of observed
and predicted acyl stretching frequencies is useful in identifying
conformations in the simple triphenylphosphonium ylides that
we examined, but, as applied to keto ester ylides, it may not distin-
guish between isomers with the same conformation and trans-
posed keto and ester groups.
Reliance on the use of empirical SF values, which may be struc-
ture specific, means that stretching frequencies for given com-
pounds are not predicted, but are compared with those of
standards. The conclusion that SF for BLYP/6-31 G(d) can be ne-
glected, within the limits of experimental error, is apparently sen-
sitive to the size of substituents in diester ylides. Reliance on SF
values allows the use of economically attractive computational
methods in relating molecular spectra to structure for given classes
of organic compounds. These generalizations regarding the use of
IR spectroscopy may not apply to phenyl keto ylides where both
[11] (a) F. Castañeda, P. Silva, M.T. Garland, A. Shirazi, C.A. Bunton, Phosphorus,
Sulfur, Silicon 184 (2009) 19;
(b) F. Castañeda, P. Silva, C.A. Bunton, M.T. Garland, R. Baggio, Acta Crystallogr.
C 64 (2008) 405.
[12] R.A. Aitken, N. Karodia, P. Lightfoot, J. Chem. Soc. Perkin Trans. 2 (2000) 333.
[13] (a) S.M. Bachrach, C.I. Nitsche, in: F.R. Hartley (Ed.), The Chemistry of
Organophosphorus Compounds, vol. 3, Wiley, New York, 1994, p. 273
(Chapter 4);
(b) M. Kobayashi, F. Sanda, T. Endo, Macromolecules 33 (2000) 5384;
(c) H. Schmidbaur, W. Bucher, D. Schentzow, Chem. Ber. 106 (1973) 1251.
[14] (a) L.C. Thomas, Interpretation of the Infrared Spectra of Organophosphorus
Compounds, Heyden, London, 1974;
(b) W. Luttke, K. Wilhelm, Angew. Chem. Int. Ed. Engl. 4 (1965) 875.
[15] (a) A.P. Scott, L. Radom, J. Phys. Chem. 100 (1996) 16502;
(b) C.J. Cramer, Essentials of Quantum Chemistry, Wiley, New York, 2002
(Chapter 9).
[16] C.A. Bunton, P. Silva, F. Castañeda, J. Mol. Struct. 921 (2009) 6.
[17] R.M. Silverstein, G.C. Bassler, T.C. Morrill, Spectrometric Identification of
Organic Compounds, fourth ed., Wiley, New York, 1981 (Chapter 3).
[18] F. Castañeda, C.A. Terraza, M.T. Garland, C.A. Bunton, R.F. Baggio, Acta
Crystallogr. C57 (2001) 180.
[19] (a) M. Nishio, Y. Umezawa, M. Hirota, Y. Takeuchi, Tetrahedron 51 (1995)
8665;
(b) M. Nishio, M. Hirota, Tetrahedron 45 (1989) 7201.
[20] (a) H.J. Cristau, F. Plenat, in: F.R. Hartley (Ed.), The Chemistry of
Organophosphorus Compounds, vol. 3, Wiley, New York, 1994, p. 108
(Chapter 2);
(b) H.J. Bestmann, Angew. Chem. Int. Ed. Engl. 4 (1965) 583.
[21] M.P. Cooke Jr., J. Org. Chem. 47 (1982) 4963.
[22] F. Castañeda, P. Silva, M.T. Garland, A. Shirazi, C.A. Bunton, Phosphorus, Sulfur,
Silicon 184 (2009) 2152.
[23] A.J. Cohen, P. Mori-Sanchez, W. Yang, Science 321 (2008) 792.
[24] K.E. Riley, B.T. Op’t Holt, K.M. Merz Jr., J. Chem. Theory Comput. 3 (2007) 407.
[25] J. Tirado-Rives, W.L. Jorgenson, J. Chem. Theory Comput. 4 (2008) 297.
[26] J.P. Perdew, A. Ruzsinszky, L.A. Constantini, J. Sun, G.I. Csonka, J. Chem. Theory
Comput. 5 (2009) 675.
[27] K.S. Raymond, R.A. Wheeler, J. Comput. Chem. 20 (1999) 207.