On the interactions of alkyl 2-hydroxycarboxylic acids with alkoxysilanes 2. Complexation and esterification of di- and tricarboxylic acids

Article abstract of DOI:10.1039/b814488e

The selective esterifications of l-malic, l-tartaric and citric acids with tetramethoxysilane Si(OMe)₄, in methanol (MeOH) have been demonstrated for the first time. The interactions between these acids and Si(OMe)₄, and also between

Full text of DOI:10.1039/b814488e

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/dalton | Dalton Transactions
On the interactions of alkyl 2-hydroxycarboxylic acids with alkoxysilanes 2.
Complexation and esterification of di- and tricarboxylic acids†
Richard J. Ansell,* Jonathan E. Meegan,‡ Simon A. Barrett and Stuart L. Warrinner
Received 20th August 2008, Accepted 14th November 2008
First published as an Advance Article on the web 16th January 2009
DOI: 10.1039/b814488e
The selective esterifications of L-malic, L-tartaric and citric acids with tetramethoxysilane Si(OMe)4, in
methanol (MeOH) have been demonstrated for the first time. The interactions between these acids and
Si(OMe)4, and also between oxalic acid and Si(OMe)4, were investigated using 1H, 13C and 29Si solution
phase NMR, and electrospray mass spectrometry (ES-MS). Si(OMe)4 acts as a catalyst/reagent in the
selective esterification of simple 2-hydroxycarboxylic acids (2HOAs), but with the di- and tri-carboxylic
acids more complex selectivities are observed: the esterification of oxalic acid proceeds slower than in
MeOH alone, L-malic acid is selectively esterified ~1000 times faster at the 2-hydroxy acid, L-tartaric
acid is esterified ~1000 times faster to the mono- and diester, while citric acid is selectively methylated at
the terminal (3-hydroxycarboxyl) groups ~1000 times faster than in the absence of Si(OMe)4.
Esterification is associated with the condensation of silane units to form oligomers. A mechanism is
proposed in which 2HOAs attach to silicon via the alkoxy group and carboxyl groups to form various
reactive cyclic intermediates. These intermediates may lead to accelerated esterification via nucleophilic
attack of MeOH at the ligated carboxyl group, while a separate reaction pathway leads to condensation
of silicon centres leading to oligosilanes. The mechanism has implications for the use of 2HOAs as
templates in sol–gel silica preparation.
Diols, catechols and bis(carbonyls) form complexes with silicon
Introduction
in protic solvents or aqueous solutions at high pH, often with an
-
The interactions of small organic ligands with silicon are of great
increase of the coordination number to five (monoanionic SiX5 )
or six (dianionic SiX62-).14 Monocarboxylic acids are poor ligands
for silicon, but dicarboxylic acids may benefit from additional sta-
bility via the chelate effect: thus, tris(oxalato)silicate (Fig. 1, 1) was
prepared in 1969 from SiCl4 and Ag2(C2O4) in refluxing acetone.15
interest for the design of new hybrid organic-inorganic or tem-
plated inorganic materials, and for increasing our understanding
of the role of silicon in bioinorganic and prebiotic chemistry.
Although alkyl 2-hydroxycarboxylic acids (2HOAs) have been
used in sol–gel syntheses of structured silicas for many years,1–4
and a number of 2HOA-silicon complexes have been isolated
from solution and characterised,5–12 the nature of the interactions
of 2HOAs with siloxanes and silicates in solution is not well
understood.
It was shown in a preceding paper13 that the simple 2HOAs gly-
colic, lactic and 2-hydroxybutyric acid react with tetramethoxysi-
lane, Si(OMe)4, in methanol (MeOH) at room temperature with
rapid formation of methyl esters (by contrast, esterification of
acetic and 3-hydroxybutyric acids is not accelerated under these
conditions). This selectivity for 2HOAs suggests the process may
be synthetically useful. In the current work the investigation
has been extended to the di- and tri-carboxylic acids oxalic,
L-malic, L-tartaric and citric acids (OA, MA, TA, CA, Table 1).
Studying the reactions of these acids with Si(OMe)4 in MeOH is
also important because they are widely used as templates in the
synthesis of structured silica from tetraalkoxysilanes Si(OR)4.
School of Chemistry, University of Leeds, Leeds, UK. E-mail: chmrja@
leeds.ac.uk; Fax: +44 (0)113 3436565; Tel: +44 (0)113 3436415
† Electronic supplementary information (ESI) available: NMR of siloxane
signals, algorithms for quantifying esters, FTIR data, data for TA-
Si(OEt)4reaction. See DOI: 10.1039/b814488e
‡ Present address: Cytec Engineered Materials, Abenbury Way, Wrexham,
UK LL13 9UZ.
Fig.
1
Structures of silicon-HOA complexes identified by Tacke
et al.6–9,11,12,17 or Kinrade et al.20,21
1460 | Dalton Trans., 2009, 1460–1470
This journal is
The Royal Society of Chemistry 2009
©

View Article Online
bipyramidal with carboxyl ligands at the axial positions.10 The
complex of CA with (dimethylamino)methyltrimethoxysilane fol-
lows this pattern (Fig. 1, 2), with the 3-hydroxycarboxyl groups
not ligated.11 Complexes with SiO5- or SiO62- cores were obtained
by treating Si(OMe)4 with 2HOAs in MeCN or tetrahydrofuran
and recrystallising the precipitated product. Glycolic acid (GA)
gave an octahedral complex with three bidentate glycolates,7 MA
gave an analogous tris(malato) complex (Fig. 1, 3),8 while CA gave
bis(citrato) complexes with tridentate CA (Fig. 1, 4).8,9 NMR data
could not be obtained for 3 and 4 since they disproportionated in
MeCN or DMSO, and dissociated completely in H2O. MA and CA
have been used as structure directing agents in silica production:
mixtures such as CA-Si(OEt)4-EtOH-H2O-HNO3, MA-Si(OEt)4-
EtOH-H2O-HCl or MA-Si(OEt)4-dioxane-H2O where water is the
major solvent give silicas with large surface areas and pore sizes
proportional to the amount of acid used, suggesting MA-silicon or
CA-silicon complexes play a role in silica precipitation, although
the authors generally suggest the interactions between the acids
and silanol groups are simply hydrogen bonding.4
Tacke et al. did not describe complexes analogous to 3 or
4 forming from Si(OMe)4 and TA, possibly because intractable
mixtures of products are formed. However, 5 was obtained from
TA and aminomethyltriethoxysilane, and characterised by X-ray
crystallography.12 As with the simpler complexes made by Tacke
having RSiO4- cores (such as 2), silicon in 5 is trigonal bipyramidal
and carboxyl groups occupy the axial positions. TA has been
especially widely used as a structure directing agent or template
in sol–gel silica syntheses,3,22 and unlike MA and CA there is
more support for the formation of covalent silicon-TA complexes,
particularly where TA and Si(OEt)4 are preincubated before the
addition of water.2 However, it is unclear whether TA interacts with
silicon via the carboxy groups, or just via the two alkoxy groups.
The latter possibility arises since the TA diester, dibenzoyl tartrate,
has also been shown to direct the sol–gel process with Si(OEt)4 to
give highly porous silica.23 Further, there is ample evidence that
silicon-diol complexes such as 6 are stable in solution, even under
aqueous conditions.21
Table 1 Carboxylic acids used in the current study. Mixtures of acids
(0.42M) with Si(OMe)4 (0.85 M) left in CD3OD at room temperature for
24 h. Yields calculated from 1H NMR spectra as described in Results
section
Acid
Products
Yield (%)a
1
2
=
=
=
H
R
R
Me, R
R
26 (64)
14 (16)
1
2
=
Me
1
2
1
2
=
=
=
=
H
=
H
R
R
R
Me, R
66 (1)
0 (0)
14 (0)
1
Me, R
2
=
R
Me
1
1
2
=
=
=
H
R
R
Me, R
R
26 (7)
73 (0)
2
=
Me
1
2
1
1
1
2
1
3
=
=
=
=
=
=
=
=
H
=
H
=
H
=
H
R
R
R
R
R
Me, R
R
R
26 (5)
2 (0)
2 (0)
47 (0)
4 (0)
3
3
2
Me, R
2
=
=
=
R
R
R
Me, R
3
2
Me, R
3
=
R
Me
a Figures in parentheses indicate yields in absence of Si(OMe)4.
In previous studies reacting glycolic, lactic, and 2-hydroxy-
butyric acids13 with Si(OMe)4 in methanol (MeOH) at room
temperature, complexes of the HOAs with silicon were not
detected by NMR but were found using ESI-MS. A mechanism for
the observed esterification was proposed involving the formation
of a 5-membered cyclic intermediate in which both the alkoxy and
carboxy groups are bonded to silicon, which is reactive and labile
so cannot be detected spectroscopically, yet can react with MeOH
More recently Tacke et al. prepared the triethylammonium salt of 1
from OA, tetraethoxysilane, Si(OEt)4, and triethylamine in toluene
under reflux, confirmed the structure by X-ray crystallography,
and showed 1 decomposes in H2O.16,17 Sugahara et al. studied the
reaction of OA with Si(OEt)4 in ethanol (EtOH) and observed
oligomerisation of siloxanes and esterification of OA, without
identifying any OA-silicon complex,18 but crucially they did not
consider ester formation in the absence of Si(OEt)4, so it is
unclear whether the reaction is accelerated by Si(OEt)4 or not.
OA has also been used in silica production from Si(OEt)4 in
aqueous conditions: the mixture OA-Si(OEt)4-H2O gave silica with
higher surface area than when formic or acetic acids were used,19
which might suggest OA-silicon complexes play a role in silica
precipitation.
∫
to yield the methyl 2HOA, and with SiOH to yield oligomers.
=
Si(OMe)2 + HOOCCHROH ꢀ SiOOCCHRO +
2 MeOH (1.1)
=
=
SiOOCCHRO + 2 MeOH → Si(OH)(OMe) +
MeOOCCHROH (1.2)
Complexes of silicon with 2HOAs were first produced from
Si(OMe)4 and Si(OEt)4 by Mehrotra et al. in the 1960s,5 but
structures were assigned based only on the quantities of alcohol
eliminated and elemental analyses of the residues. Tacke et al.
prepared a series of pentacoordinate silicon complexes with two
chelating 2HOAs and an alkylamine as a fifth ligand: crystal
=
∫
∫
=
+
SiOOCCHRO + SiOH → SiOSi(OH)
HOOCCHROH (1.3)
The intermediates hypothesized in step (1.1) are clearly sim-
ilar to 1–5, and their formation under such conditions would
have implications for the mechanism when 2HOAs are used as
templates in the preparation of structured silica. Thus, it was
-
structures show that in each case the RSiO4 core is trigonal
This journal is
The Royal Society of Chemistry 2009
Dalton Trans., 2009, 1460–1470 | 1461
©

View Article Online
of interest to determine whether OA, MA, CA and TA react
with Si(OMe)4 in MeOH to form complexes, and whether similar
reactions occur as with the simpler 2HOAs. Since esterification of
2-hydroxybutyric acid is accelerated under these conditions but
that of 3-hydroxybutyric acid is not,13 regioselective esterification
of MA and CA was expected.24,25
Results
13C and 1H NMR of acid/Si(OMe)4 mixtures in CD3OD – the acid
and ester signals
When OA, MA, TA or CA are added to Si(OMe)4 in CD3OD
at room temperature ligand exchange (OCD3/OCH3), hydrolysis
and condensation reactions of Si(OMe)4 are accelerated, leading
to changes in the H and 13C siloxane signals and the growth of
1
Materials and methods
signals due to CH3OD (see ESI†). Also, conversion of certain
carboxyl groups into methyl esters was observed:
NMR
1H NMR were recorded on a Bruker Avance 500 spectrometer
at 500 MHz. Proton-decoupled 13C NMR were recorded on a
Bruker DPX 300 spectrometer at 75.5 MHz. Acid (0.375 mmol)
was dissolved completely in CD3OD (0.75 mL), and 1H and
13C spectra were recorded. Si(OMe)4 (0.750 mmol) was added
immediately before placing the sample into the magnet and further
R1COOD + R2OC[D3/H3] ↔ R1COOC[D3/H3] + R2OD (2.1)
(assuming complete exchange of the acidic protons for the
more abundant deuteriums of the solvent. The methoxy group
transferred to the acid may come from free solvent (R2 = D), or
2
1
∫
siloxane species (R = Si)). Fig. 2 shows selected regions of H
and 13C spectra after mixing Si(OMe)4 + TA for 10 min, whence
approximately 30% of the acid groups have been esterified. The
spectra are more complex than for monocarboxylic acids because
both the monoester and diester are formed. The COOCH3 and
COOCH3 peaks are very small because most of the ester is formed
as COOCD3. Interestingly an additional species appears, labelled
as ‘transient complex’ in Fig. 2, which is present for only ~1 h after
mixing.
1
spectra of the mixtures recorded at increasing intervals. H and
13C spectra were recorded in 0.5 ¥ 18 cm borosilicate glass tubes.
Proton-decoupled 29Si NMR were recorded on a Bruker DRX
500 spectrometer at 99.4 MHz, an acquisition time of 0.43 s
with a delay of 5 s was used and the number of scans was 256.
Amounts for 29Si NMR were four times the quantities above and
spectra were recorded in 1.0 ¥ 21 cm Teflon tubes. 29Si NMR
were recorded only after mixing acid + Si(OMe)4. All shifts are
reported relative to tetramethylsilane (SiMe4, TMS) as internal
standard.
The esterifications of OA, MA, TA and CA at room temperature
in CD3OD in the presence or absence of Si(OMe)4 were compared
(Si(OMe)4 or CH3OH added, Fig. 3). Signals attributed to the
esters were confirmed by spiking the reactions after 1 week with
authentic samples where commercially available. For assignments
of the observed signals see ESI† Table 1.
In the absence of Si(OMe)4, the rate of esterification increases in
the order MA < CA ~ TA << OA suggesting it depends on acid
pKa and steric accessibility. Only for OA is the background rate
appreciable, with >10% of the acid esterified after 2 h. For TA +
MeOH the background rate is very low and only monomethyl
TA is produced. For MA + MeOH only the 2-hydroxycarboxyl
group undergoes esterification, as shown by comparison with the
spectrum of purified product, also very slowly. For CA + MeOH
only the asymmetric monoester forms significantly, as shown in
the splitting of the –CH2-signal.
ES-MS
Spectra were recorded on a Bruker micrOTOF instrument using
direct injection electrospray introduction in negative ion mode
with a total capillary voltage of 4500 V and a capillary exit
voltage of -100 V. Prior to injecting sample, LiOOCH (10 mM)
in MeOH/PriOH (9:1 v/v) was injected at 9 mL min-1 to
provide a calibration signal. The capillary was then flushed
with H2O/MeCN (1:1 v/v) at 0.3 mL min-1 before sample was
introduced at 4 mL min-1. Data were collected for at least 2 min or
until a constant signal was obtained.
When the acids are mixed with Si(OMe)4 in CD3OD, ~1000
fold acceleration of esterification is observed for MA, TA and
CA. For OA, the overall rate of esterification falls, especially the
formation of the monoester (the diester forms at a similar rate as in
the absence of Si(OMe)4). For TA + Si(OMe)4 the distribution of
products is approximately statistical, suggesting TA is converted to
monomethyl TA at the same rate as monomethyl TA is converted
to dimethyl TA.
Quantifying the esters is easiest for TA and can be done using
the CHO[H/D] signals (Fig. 2a) where the signals due to the acid,
monoester and diester are fairly distinct. For MA the CHO[H/D]
signals can again be used but quantification is potentially far more
difficult. Fortunately there is a strong preference for esterification
of the acid adjacent to the hydroxy group, such that only two
products, the 2-hydroxyester and the diester, need be considered.
Product concentrations were calculated from the integrals of
regions A–D in Fig. 4a (for details see ESI†).
Preparative scale reaction
Si(OMe)4 (5.6 mL, 37.5 mmol) was added to MA (5.03 g,
37.5 mmol) diss◦olved in CH3OH (20 mL), and stirred in a sealed
vessel at 15–20 C for 30 h. The solvent was evaporated and the
product suspended in EtOAc (20 mL). HCl (aq, 1 M, 1 mL) was
added, the mixture shaken and filtered through a plug of silica,
followed by 50 mL EtOAc to collect all of the product. The solvent
was evaporated and the product recrystallised repeatedly from
EtOAc/petroleum ether at -20 C.
Yield = 3.16 g, 56.8%; mpt: 79–80 C; [a] = -5.32◦ (c = 0.94,
MeOH); Found: C, 40.3; H, 5.3; calc. for C5H8O5: C, 40.5; H, 5.4%;
dH (500 MHz; CDCl3; Me4Si) 2.85 (1H, dd, J = 17.0, 6.4 Hz),
2.93 (1H, dd, J = 16.8, 4.2 Hz), 3.82 (3H, s), 4.53 (1H, dd, J =
6.3 Hz, J = 4.3 Hz); dC (75.5 MHz; CDCl3; Me4Si) 39.9, 53.8,
-
68.7, 174.00, 175.6; m/z (-ve ESI): 147.0294 (calc. for C5H7O5
147.0299).
1462 | Dalton Trans., 2009, 1460–1470
This journal is
The Royal Society of Chemistry 2009
©

View Article Online
Fig. 2 Selected regions of the 1H and 13C spectra after mixing Si(OMe)4 (0.85 M) + TA (0.42 M) for 10 min. a) CHO[D/H] region, b) –OCH3 region, c)
COOMe region, d) CHOH region and e) OCH3 region.
Esterification of CA must be followed via the –CH2-signal
and analysis is yet more complicated. By contrast with MA, the
acids distant from the hydroxy group are esterified preferentially,
such that in MeOH alone only the asymmetric monoester forms
appreciably, and with Si(OMe)4 this and the symmetric diester
need to be considered. Concentrations were calculated using the
integrals of regions A–I in Fig. 4b (for details see ESI†). Some
of the alternative esters do form in very small amounts; the
peak I is attributed to one of these. Calculated concentrations
of monomethyl and dimethyl CA are therefore approximate.
For OA, esterification was quantified by comparing the integral
of the –COOCH3 signals with all other –OCH3 signals (since the
ester is produced as both –COOCH3 and –COOCD3, the total
amount of ester is estimated by assuming complete interconversion
of –OCH3 and –OCD3 so that [ester] = int(-COOCH3)/int(all
OCH3) ¥ [OCH3 + OCD3]t = 0/[OCH3]t = 0. This approach contains
several sources of error, so is avoided where it is possible to quantify
the ester by measuring signals due to other protons). The actual
concentrations of all products could be assessed more accurately
by a chromatographic separation.
The selectivity for MA is as might be expected, with preferen-
tial esterification of the 2-hydroxycarboxyl group, although the
3-hydroxycarboxyl group is also esterified faster than in the
absence of Si(OMe)4. For CA, the selectivity is contrary to what
might be expected with esterification of the 3-hydroxycarboxyl
groups most accelerated. In the previous work no evidence was
seen for acid-siloxane complexes in the 1H and 13C NMR spectra,
although such complexes were seen in ES-MS experiments.13
Again, here, there was no evidence for such complexes in the NMR
spectra of mixtures of Si(OMe)4 with OA, MA or CA, but there
was clear evidence for a TA-siloxane complex which is present for
only 1 h after mixing.
29Si NMR of acid/Si(OMe)4 mixtures in CD3OD – evidence for
TA-Si(OMe)4 intermediate
29Si NMR was also used to probe the species present after
the acids were mixed with Si(OMe)4 in CD3OD (Fig. 5). The
assignments shown are based on reported studies of the hydrolysis
of alkoxysilanes using 29Si NMR:26 Q0 corresponds to Si with four
1
The results confirm the esterification of 2HOAs by
Si(OMe)4/MeOH at room temperature observed in the previous
paper,13 but the patterns of selectivity have additional implications
in each case. The inhibition of esterification of OA in the
presence of Si(OMe)4 suggests OA interacts with Si(OMe)4, and
the mechanism may differ from that for the ‘true’ 2HOAs.
∫
–OC or –OH ligands, Q to Si with one of these ligands replaced
2
∫
∫
by –OSi , and Q to Si with two ligands replaced with –OSi .
In the previous study, mixing AA or 3HBA with Si(OMe)4
resulted only in the formation of a minor Q0 peak at -76.0 ppm
which could be attributed to Si(OMe)3OH, whilst mixing the
2HOAs glycolic acid (GA), lactic or 2-hydroxybutyric acid with
This journal is
The Royal Society of Chemistry 2009
Dalton Trans., 2009, 1460–1470 | 1463
©

View Article Online
Fig. 4 Selected regions of the 1H NMR spectra for MA + Si(OMe)4
and CA + Si(OMe)4 in CD3OD, showing assignments of peaks used to
quantify the products. a) the CHOH region for MA + Si(OMe)4, 192 min
after mixing, b) the CH2 region for CA + Si(OMe)4, 1041 min after mixing.
to Si(OMe)4. Spectra were collected between 0 and -200 ppm, but
only the regions containing peaks are shown in Fig. 5. None of the
acids except TA show peaks upfield of -100 ppm which could be
Fig. 3 Concentrations of product monoesters (diamonds) and diesters
(triangles) versus time for the mixtures of dicarboxylic acids (0.42 M) with
Si(OMe)4 (0.85 M, filled symbols and bold lines) or CH3OH (3.3 M, open
symbols and dotted lines) in CD3OD. a) OA b) MA, c) TA, d) CA.
assigned to a pentacoordinate SiO5 centre (P)21,27 or hexavalent
-
SiO62- centre (H). TA + Si(OMe)4 shows a P signal at -107.4 ppm.
For Si(OMe)4 + OA, Q1 and Q2 species are observed, but
also Si(OMe)3OH. A signal near those at -176.6 ppm17 (in
d6-DMSO) or -173.3 ppm14 (in C6F6) reported for the hexacoor-
dinate trisoxalato complex 1 in Fig. 2 is not observed. For MA +
Si(OMe)4 only a slight amount of Q1, and no hydrolysed species,
are observed, suggesting the mixture behaves like GA + Si(OMe)4
but reaction is slower. CA + Si(OMe)4 gives a spectrum similar to
OA + Si(OMe)4, with Q1, Q2 and hydrolysed species.
Si(OMe)4 resulted in the immediate formation of a Q1 peak,
and the gradual diminishment of Q0 and growth of Q1 and Q2,
indicating the appearance of oligosiloxanes. With the 2HOAs no
∫
peak at -76.0 ppm was observed, indicating any SiOH that forms
may react very fast to give the oligomer species.
Interestingly the di- and tri-carboxylic acids all exhibit distinct
behaviour. In each case the major peak is still the Q0 Signal due
1464 | Dalton Trans., 2009, 1460–1470
This journal is
The Royal Society of Chemistry 2009
©

View Article Online
16 h these have multiplied to three Q1 and at least six different
Q2 signals, indicating the evolution of a more complex array
of oligosiloxane products. Most interestingly of all the peak at
-107.4 ppm is observed immediately after mixing but disappears
after approximately 1 h, which corresponds to the lifetime of the
1
transient complex observed in the H and 13C spectra (Fig. 2).
It is suggested that this peak indeed belongs to a TA-Si(OMe)4
complex, and that the silicon in this species is pentacoordinate.
MS analysis of acid/Si(OMe)4 mixtures in CH3OH
As in the previous work, samples were directly injected unadulter-
ated into the capillary of an ES-microTOF instrument capable of
delivering accurate mass (5 ppm). They were ionised by negative
electrospray and the structures of the resulting ions were assigned
on the basis of accurate masses. Spectra are shown in Fig. 6,
Fig. 5 29Si NMR of various carboxylic acids (0.42 M) + Si(OMe)4
(0.85 M) in CD3OD: a) OA, b) MA, c) and d) TA, e) CA. All recorded
between 5 and 45 min after mixing except d) after 16 h. Shifts relative to
TMS.
Fig. 6 ES-MS spectra obtained in negative ion mode with direct infusion
of samples. Si(OMe)4 (0.85 M) + acid (0.43 M) in CH3OH. a) OA, b) MA,
c) TA, d) TA, e) CA. All collected ~2 h after mixing except c) collected
10 min after mixing.
For TA + Si(OMe)4 Q1 and Q2 signals are observed, and no
hydrolysed species. In the spectrum recorded immediately after
mixing there are just two Q1 signals and two Q2 signals, but after
This journal is
The Royal Society of Chemistry 2009
Dalton Trans., 2009, 1460–1470 | 1465
©

View Article Online
proposed structures in Fig. 7 and detailed assignments in ESI†
Table S2.
Fig. 6b (MA + Si(OMe)4, 2 h) can be accounted for by the
free acid and monoester, and structures of type F, again similar
to D and to the complexes in the previous paper, except that one
chelating cycle is 5-membered and the other 6-membered. This is
because the carboxyl group without the 2-hydroxy group is not
significantly esterified, yet the ion at m/z = 337.0227, for instance,
must have two –OCH3 groups and only one –OH, so at least one of
the non-reactive carboxyls must be complexed to Si. 6-Membered
chelating cycles could also have formed with 3-hydroxybutyric acid
in the previous work,13 but did not undergo esterification since the
Fig. 6a (OA + Si(OMe)4, 2 h) is dominated by monosilox-
ane complexes which can only be assigned by invoking cyclic
chelated pentavalent silicon structures, similar to those previously
proposed for mixtures of the simple 2HOAs with Si(OMe)4,
but here including also the hexavalent complex B (1 in Fig. 1).
As with the previous work, these structures are seen in the
MS even though they were not detected via 29Si NMR. The
oligomeric ions D could have various alternative structures
but the structure written is consistent with the chelated struc-
tures of A–C and agrees with the proposed structures of the
oligomeric ions formed with the simple 2HOAs in the previous
paper.
∫
ring is less strained, making –OSi a less good leaving group.
Fig. 6c (TA + Si(OMe)4, 10 min) includes the free acid, no
free ester at that stage, but an array of different silicon-TA
complexes. The ion H (m/z = 204.9810) is interesting despite
Fig. 7 Proposed structures for the ions detected by ES-MS. See also ESI† Table S2.
1466 | Dalton Trans., 2009, 1460–1470 This journal is The Royal Society of Chemistry 2009
©

View Article Online
its very low intensity because it could have a structure with two
5-coordinate silicons as in 5 (Fig. 1) (H has a M+12 peak showing
it is a dianion). Highly substituted monosiloxane complexes I
are observed which might have the structures shown, with one
tridentate and one bidentate TA, either via the diol group or
one hydroxy and one carboxy group. The tridentate TA has the
same Si–O–C moiety involved in both a 5- and a 6-membered
chelating cycle. This arrangement invokes too much strain when
the silicon is tetracoordinate, but can be accommodated when it is
5-coordinate (or 6-coordinate, as for structure 4 in Fig. 1). Ions J
can be accounted for by structures analogous to D and F or by
structures with two chelating diols. The highly substituted ions
K, L and M, which might be obtained from H, could again be due
to structures similar to 5 in Fig. 1 (although these would be very
strained) or to more symmetric structures as shown.
to methyl groups initially present, 16:84, indicating that ROCD3
reacts more readily than ROCH2CH3, but providing no clear
=
evidence whether the alkyl groups come from solution (R H, or
=
∫
from siloxanes (R Si )). The formation of mixed products reflects
the rapid exchange of ligands on the siloxane via a process similar
to eqn (3.1):
∫
∫
Si–OCH2CH3 + CD3OD → Si–OCD3 + CH3CH2OD (3.1)
Formation of total ester (methyl and ethyl) proceeds ~10 times
slower than with Si(OMe)4 in CD3OD. Again a transient interme-
diate TA complex, which appears to be symmetric, is visible in the
1H and 13C NMR spectra for ~1 h.
The ES-MS spectrum of TA + Si(OEt)4 in CH3OH was recorded
several days after mixing and confirms the formation of both
methyl and ethyl esters.
Fig. 6d (TA + Si(OMe)4, 2 h) shows some differences with
the spectrum recorded after only 10 min: It includes the free
monoester; the J ions are more intense than the L (while at
10 min the L ions are much more intense than J); and the highly
substituted complexes H, K and M are completely absent. The
dominance of J ions at longer times is probably because once the
TAs are esterified, they are no longer able to form the complexes
which are more highly substituted with TA.
Fig. 6e (CA + Si(OMe)4, 2 h) includes the free acid and
monoester, ions O which can be accounted for by structures
analogous to D, F and H, including 5-membered and/or
6-membered cycles, and ions P which include one silicon and two
CAs, must be chelated, and could be drawn either as shown or with
one free carboxylate carrying the charge. The structure drawn is
closer to 6 in Fig. 1, where a single CA acts as a tridentate ligand.
Scale-up of the MA/Si(OMe)4 reaction
The only published methods for esterification of MA selectively at
the 2-hydroxycarboxyl group are by treatment with trifluoroacetic
anhydride, generating the trifluoroacetate of malic acid anyhy-
dride, followed by MeOH,24,25 or with MeOH containing boric
acid as a selective catalyst, the postulated mechanism involving a
5-membered cyclic intermediate with a chelated boron.28
To determine the synthetic utility of the esterification with
Si(OMe)4, the reaction of MA + Si(OMe)4 was performed on
37.5 mmol scale, in technical grade CH3OH. The reaction vessel
was sealed but not purged of air, and no especial measures were
taken to exclude water or oxygen. After 30 h at ambient temper-
1
ature H NMR suggested the speciation was ~10% MA, ~70%
monomethyl ether, and ~20% dimethyl ether. The isomerically
pure CH3OOCC(OH)HCH2COOH could be purified by adsorb-
ing starting materials onto silica, followed by recrystallisation from
ethyl acetate, although owing to a non-optimised work-up the final
yield was only 57%. Characterisation data are in agreement with
previous reports,24,29 two-dimensional 1H-13C NMR (heteronu-
clear multiple quantum correlation and heteronuclear multi-bond
correlation) spectra confirm the isomeric structure.
13C and 1H NMR and ES-MS of TA/Si(OEt)4 mixtures
TA was also incubated with Si(OEt)4, both in C2D5OD and
in CD3OD, to demonstrate the versatility of the esterification
reaction, and to attempt to shed light on whether the OR group
which adds to the acid originates from the siloxane starting
material or from the solvent.
The esterification reaction with Si(OEt)4 in C2D5OD was
quantified in the same way as with Si(OMe)4 in CD3OD (see
Fig. S1†). It proceeds 10–100 times slower. Again a transient
intermediate TA complex, which appears to be symmetric, is
visible for ~1 h.
The ES-MS spectrum of TA + Si(OEt)4 in C2H5OH was
recorded ~2 h after mixing (see ESI†) and contains no ester,
and predominantly K, L and M type, highly substituted ions,
unlike the spectrum of TA+ Si(OMe)4 which contained the less
substituted H-type ions. This can be explained since much less
TA is esterified in the mixture with Si(OEt)4 in C2H5OH, so more
can act as multidentate ligands. After several days, the ES-MS
spectrum shows J-type ions, and the mono- and diethyl TA esters
as expected.
Discussion
A mechanism was previously proposed for the reaction of
Si(OMe)4 with simple 2HOAs (Scheme 1). The complexes, whose
formation is suggested by the NMR and ES-MS data in the present
work, might further explain the activity and selectivity of the
esterification reactions with the di- and tri-carboxylic acids.
With Si(OMe)4 + OA, the formation of the monoester is slowed
down, suggesting that complexes with Si, such as those revealed by
ES-MS, are less reactive toward esterification than free OA. This
may be because of steric hindrance of the electrophilic carboxyl C
atoms, or because the Si atom is more electron donating than an H
-
atom (especially in its pentavalent (formally SiX5 ) or hexavalent
The esterification reaction with Si(OEt)4 in CD3OD was also
studied by 1H and 13C NMR. It was anticipated that both
methyl and ethyl esters would be formed (methoxy ligands
originating from the solvent and ethoxy from the Si(OEt)4) and
that quantification of each might help to reveal the mechanism.
Some –OCH2CH3 ester was formed, but mostly –OCD3 ester (the
ratio was about 7:93, which is less than the ratio of ethyl groups
(formally SiX62-) forms). Once some of the OA molecules have
been esterified to the monoester, however, they can no longer act
as ligands to Si, and further esterification is not hindered: thus, the
monoester converts to diester as fast as in the absence of Si(OMe)4.
The MA + Si(OMe)4 reaction can be explained via Scheme 1
where X = CH2COOH. A five-membered chelating cycle as in
4 and 5 is formed, and esterification proceeds at the chelating
This journal is
The Royal Society of Chemistry 2009
Dalton Trans., 2009, 1460–1470 | 1467
©

View Article Online
Scheme 1
2-hydroxy carboxyl C atom, driven by its increased electrophilicity,
methoxy ligands were cleaved: this would provide a driving force
for esterification and cleavage of the 5-membered chelating cycle.
Experiments with Si(OEt)4 cannot confirm whether the alcohol
group transferred during esterification comes from solution (as
suggested in Scheme 1) or from the siloxane: other mechanisms
are possible.
∫
and the enhanced leaving group –OSi (steps 4.8 to 4.9). The ions
observed in the ES-MS spectrum for MA + Si(OMe)4 can be
explained by assuming that many other complexes are present in
solution, but only some give stable ions under the ES conditions
(as proposed in13). Thus, complexes with just one 2HOA attached
to Si (3, 4) can cyclise to give neutral species (5), and 2HOAs can
also be displaced by MeOH ligands from solution. However if two
2HOAs are attached to the same Si atom (e.g. 8), they will cyclise
to give a stable ion (9). 6-Membered chelating cycles also form,
but do not lead to significantly accelerated ester formation since
In the case of CA the lack of esterification at the 2-hydroxy
carboxyl carbon may be due to complexes with 6-membered
chelating cycles being more stable and predominating, or alterna-
tively if 5-membered chelating cycles do form the carboxyl carbon
may be too hindered to react. This however cannot explain why
esterification at the 3-hydroxycarboxyl groups is accelerated (in
the case of 3-hydroxybutyric acid, 6-membered chelating cycles
were also postulated to form, and evidenced by ES-MS, yet no
esterification was observed13). Such activity might arise if chelated
complexes form such as Q in Fig. 8. The arrangement with two
6-membered cycles is less strained than with one 5-membered and
one 6-membered, but sufficiently strained that there would be a
greater driving force for esterification at the complexed carboxyls
than in the case of complexes with just one 6-membered cycle.
Also, in Q the carboxyls are hindered from one face but quite
open to attack from the other.
∫
the ring is less strained and consequently –OSi is a less effective
leaving group.
The reaction with TA can again be explained via Scheme 1,
though many different complexes are possible in this case, a
number of which appear in the ES-MS spectra and at least one
in the 1H NMR spectra. The best candidate for the transient
intermediate observed by NMR in the first hour after mixing,
which appears to be symmetric with respect to TA and contain
pentacoordinate Si, is the dianion H. This structure is similar
to that of 5 in Fig. 1, and although only detected at very low
level in the ES-MS spectrum is a precursor to other structures
such as K, L and M by elimination and/or hydration. The
structure is fairly strained and would be yet more so if one of the
Again the ions formed under the conditions of the MS
experiment are proposed not to be significant intermediates in the
1468 | Dalton Trans., 2009, 1460–1470
This journal is
The Royal Society of Chemistry 2009
©

View Article Online
Acknowledgements
The authors are grateful to Tanya Marinko-Covell for technical
assistance with the MS, and acknowledge the support of the
EPSRC via grant number EP/C010973/1.
Notes and references
1 C. B. Hurd, J. Phys. Chem., 1936, 40, 21; E. I. Klabunovskii, A. E.
Agronomov, L. M. Volkova and A. A. Balandin, Izv. Akad. Nauk SSSR,
Ser. Khim., 1963, 228; H. Izutsu, F. Mizukami, T. Sashida, K. Maeda,
Y. Kiyozumi and Y. Akiyama, J. Non-Cryst. Solids, 1997, 212, 40; J. B.
Pang, K. Y. Qiu and Y. Wei, J. Non-Cryst. Solids, 2001, 283, 101; J.-B.
Pang, K.-Y. Qiu and Y. Wei, Chem. Mater., 2001, 13, 2361; J. -Y. Zheng,
J. B. Pang, K. Y. Qiu and Y. Wei, J. Inorg. Organometal. Polym., 2001,
10, 103; H. Nakamura and Y. Matsui, J. Am. Chem. Soc., 1995, 117,
2651; F. Miyaji, S. A. Davis, J. P. H. Charmant and S. Mann, Chem.
Mater., 1999, 11, 3021; L. Z. Wang, S. Tomura, F. Ohashi, M. Maeda,
M. Suzuki and K. Inukai, J. Mater. Chem., 2001, 11, 1465; F. Miyaji,
Y. Watanabe and Y. Suyama, Mater. Res. Bull., 2003, 38, 1669.
2 H. Izutsu, F. Mizukami, Y. Kiyozumi and K. Maeda, J. Mater. Chem.,
1997, 7, 1519; H. Izutsu, F. Mizukami, Y. Kiyozumi and K. Maeda,
J. Amer. Ceram. Soc., 1997, 80, 2581.
Fig. 8 Neutral CA-silicon complexes which may be present.
3 F. Mizukami, Y. Akiyama and H. Izutsu, Supramol. Sci., 1998, 5, 433;
F. Mizukami, H. Izutsu and T. Osaka, Adv. Mater., 1994, 6, 854; J. B.
Pang, K. Y. Qiu and Y. Wei, Micropor. Mesopor. Mater., 2000, 40, 299;
J.-B. Pang, K.-Y. Qiu, Y. Wei, X.-J. Lei and Z.-F. Liu, Chem. Commun.,
2000, 477.
4 R. Takahashi, S. Sato, T. Sodesawa, M. Kawakita and K. Ogura,
J. Phys. Chem. B, 2000, 104, 12184.
esterification reaction, since there is no evidence for their existence
in the NMR spectra. Neutral precursors such as R and S, which
could lead to ions such as O and P may be present, but there is no
driving force for esterification in R, and in S the 2-hydroxycarboxyl
carbon is hindered.
5 R. C. Mehrotra and B. C. Pant, Jour. Indian Chem. Soc., 1963, 40, 623;
R. C. Mehrotra and B. C. Pant, Indian J. Chem., 1963, 1, 380; R. C.
Mehrotra and R. P. Narain, Indian J. Chem., 1968, 6, 110.
6 R. Tacke, C. Burschka, I. Richter, B. Wagner and R. Willecke, J. Am.
Chem. Soc., 2000, 122, 8480.
7 I. Richter, M. Penka and R. Tacke, Inorg. Chem., 2002, 41, 3950.
8 R. Tacke, R. Bertermann, C. Burschka and S. Dragota, Z. Anorg. Allg.
Chem., 2004, 630, 2006.
Conclusion
A mechanism was previously proposed for the reaction of
Si(OMe)4 with simple 2HOAs, involving uncharged cyclic interme-
diates. The chelating cycle is a 5-membered ring, and the resulting
ring strain provides a driving force for nucleophilic attack at
the carboxy carbon atom. The same mechanism can explain the
results achieved here: OA forms more stable complexes with less
strain than for a true HOA, because of the wider bond angle
at the second sp2 hybridised carbon, hence its methylation is
actually hindered. MA can form complexes with 5-membered
or 6-membered chelating cycles, however only the former is
strained, so methylation occurs preferentially at the 2-hydroxy
carboxyl group. TA udergoes ready methylation as expected. CA
can form complexes with either a 5-membered or 6-membered
chelating cycle, however the most stable neutral complex is
structure Q with two 6-membered chelating cycles. This structure
promotes methylation at the 3-hydroxycarboxyl groups. Anionic,
pentavalent silicon complexes may form under the conditions
of the electrospray experiment but are not inherently present at
significant concentrations in methanolic solution, since they are
not detected in 29Si NMR. Therefore it is not considered that
these are involved in the methylation mechanism, although they
provide some hints as to what neutral complexes may be present.
The exception is TA, which does form a stable pentavalent Si
complex under the experimental conditions, which is suggested
to be H.
9 R. Tacke, M. Penka, F. Popp and I. Richter, Eur. J. Inorg. Chem., 2002,
1025.
10 M. Muehleisen and R. Tacke, Organometallics, 1994, 13, 3740; R.
Tacke, A. Lopez-Mras and P. G. Jones, Organometallics, 1994, 13, 1617;
R. Tacke and M. Muehleisen, Inorg. Chem., 1994, 33, 4191; R. Tacke,
R. Bertermann, C. Burschka, S. Dragota, M. Penka and I. Richter,
J. Amer. Chem. Soc., 2004, 126, 14493–14505.
11 M. Muehleisen and R. Tacke, Chem. Ber., 1994, 127, 1615–1617.
12 R. Tacke, M. Muehleisen and P. G. Jones, Angew. Chem. Int. Ed., 1994,
33, 1186.
13 R. J. Ansell, S. A. Barrett, J. E. Meegan and S. L. Warriner, Chem.
Eur. J., 2007, 13, 4654.
14 J. A. Cella, J. D. Cargioli and E. A. Williams, J. Organomet. Chem.,
1980, 186, 13.
15 P. A. W. Dean, D. F. Evans and R. F. Phillips, J. Chem. Soc. A: Inorg.
Phys. Theor., 1969, 363.
16 K. E. Bessler and V. M. Deflon, Z. Anorg. Allg. Chem., 1994, 620, 947.
17 O. Seiler, C. Burschka, m. Penka and R. Tacke, Z. Anorg. Allg. Chem.,
2002, 628, 2427.
18 Y. Sugahara, S. Sato, K. Kuroda and C. Kato, J. Non-Cryst. Solids,
1992, 147–148, 24.
19 T. A. Gallo, B. L. Simms and V. T. Brice, Mat. Res. Soc. Symp. Proc.,
1990, 180, 269.
20 K. Benner, P. Kluefers and J. Scuhmacher, Z. Anorg. Allg. Chem., 1999,
625, 541.
21 S. D. Kinrade, E. W. Deguns, A.-M. E. Gilson and C. T. G. Knight,
Dalton Trans., 2003, 3713.
22 H. Izutsu, F. Mizukami, P. K. Nair, Y. Kiyozumi and K. Maeda,
J. Mater. Chem., 1997, 7, 767–771; H. Izutsu, P. K. Nair, K. Maeda, Y.
Kiyozumi and F. Mizukami, Mater. Res. Bull., 1997, 32, 1303.
23 Y. Wei, D. Jin and T. Ding, J. Phys. Chem. B, 1997, 101, 3318; Y. Wei,
D. L. Jin, T. Z. Ding, W. H. Shih, X. H. Liu, S. Z. D. Cheng and Q. Fu,
Adv. Mater., 1998, 10, 313.
The results suggest that Si(OMe)4 is an effective reagent for
selective synthesis of MA monomethyl ester and CA dimethyl
ester. Future work will investigate the conditions required to
stabilise the cyclic intermediates in solution, and their possible
role as templates in sol–gel silica synthesis.
24 M. J. Miller, J. S. Bajwa, P. G. Mattingly and K. Peterson, J. Org. Chem.,
1982, 47, 4928.
This journal is
The Royal Society of Chemistry 2009
Dalton Trans., 2009, 1460–1470 | 1469
©

View Article Online
25 K. Hirota, H. Kitagawa, M. Shimamura and S. Ohmori, Chem. Lett.,
1980, 191.
26 F. Brunet, B. Cabane, M. Dubois and B. Perly, J. Phys. Chem., 1991,
95, 945; J. Sanchez and A. McCormick, J. Phys. Chem., 1992, 96, 8973;
M. Mazur, V. Mlynarik, M. Valko and P. Pelikan, Appl. Magn. Reson.,
1999, 16, 547.
27 B. Herreros, S. W. Carr and J. Klinowski, Science, 1994, 263, 1585; S. D.
Kinrade, R. J. Hamilton, A. S. Schach and C. T. G. Knight, J.Chem.
Soc. Dalton Trans., 2001, 961.
28 T. A. Houston, B. L. Wilkinson and J. T. Blanchfield, Org. Lett., 2004,
6, 679.
29 Y. Hayashi, J. Yamaguchi and M. Shoji, Tetrahedron, 2002, 58, 9839.
1470 | Dalton Trans., 2009, 1460–1470
This journal is
The Royal Society of Chemistry 2009
©