Silicon chelation in aqueous and nonaqueous media: The furanoidic diol approach

Article abstract of DOI:10.1002/chem.200500147

Anhydroerythritol (AnEryt) shares some of its ligand properties with furanosides and furanoses. Its bonding to silicon centers of coordination number four, five, and six was studied by X-ray and NMR methods, and compared to silicon bonding of related comp

Full text of DOI:10.1002/chem.200500147

DOI: 10.1002/chem.200500147
Silicon Chelation in Aqueous and Nonaqueous Media: The Furanoidic Diol
Approach**
Xaver Kästele, Peter Klüfers,* Florian Kopp, Jçrg Schuhmacher, and Martin Vogt[a]
Abstract: Anhydroerythritol (AnEryt)
shares some of its ligand properties
with furanosides and furanoses. Its
bonding to silicon centers of coordina-
tion number four, five, and six was
studied by X-ray and NMR methods,
and compared to silicon bonding of re-
lated compounds. Diphenyl(cycloalky-
lenedioxy)silanes show various degrees
of oligomerization depending on the
diol component involved. For example,
Ph2Si(cis-ChxdHÀ2) (1) and Ph2Si-
{(R,R)-trans-ChxdHÀ2)} (2; Chxd=cy-
clohexanediol) are dimeric, Ph2Si-
(AnErytHÀ2) (3) is monomeric, and
Ph2Si(l-AnThreHÀ2) (4; AnThre=an-
hydrothreitol) is trimeric both in the
solid state and in solution. Ph2Si(cis-
CptdHÀ2) (5) (Cptd=cyclopentanediol)
is monomeric in solution but dimerizes
on crystallization. Si(AnErytHÀ2)2 (6)
and Si(cis-CptdHÀ2)2 (7) are monomer-
ic spiro compounds in solution but are
pentacoordinate dimers in the crystal-
line state. Pentacoordinate silicate ions
are found in A[Si(OH)(AnErytHÀ2)2]
(A=Na, 8a; Rb, 8b; Cs, 8c). Related
compounds are formed by substitution
of the hydroxo by a phenyl ligand.
K[SiPh(AnErytHÀ2)2]·1/2MeOH (9) is
a prototypical example as it shows the
two most significant isomers in one
crystal structure: the syn/anti and the
anti/anti form (syn and anti define the
oxolane ring orientation close to, or
apart from, the monodentate ligand,
respectively). syn/anti Isomerism gener-
ally rules the appearance of the NMR
spectra of pentacoordinate silicates of
furanos(id)e ligands. NMR spectro-
scopic data are presented for various
pentacoordinate bis(diolato)silicates of
adenosine, cytidine, methyl-b-d-ribofur-
anoside, and ribose. In even more basic
solutions, hexacoordinate silicates are
enriched. Preliminary X-ray analyses
are presented for Cs2[Si(CydHÀ2)3]·
21.5H2O (10) and Cs2[Si(cis-InsHÀ3)]·
cis-Ins·8H2O
(11)
(Cyd=cytidine,
Keywords: carbohydrates
· NMR
Ins=inositol).
spectroscopy · Si ligands · silicates ·
spiro compounds
Introduction
problem is particularly obvious with the reducing sugars—
the aldoses and ketoses (glycoses)—as they lack configura-
tional stability. Thus, the question of (semi)metal bonding
continues: which functional group of which glycose isomer
provides a suitable binding site? Our knowledge of this sub-
ject is rather limited, particularly with respect to X-ray anal-
yses. The metal-binding sites of aldoses with special O-atom
patterns were the first to be determined. Along with Taylor
and Waterꢀs pioneering workon the structure of a lyxose–
molybdenum complex,[1] the binucleating ability of mannose
towards trivalent metal ions is a further example of strong
metal binding that is, however, restricted exclusively to one
special sugar.[2,3] More general rules for assessing the (semi)-
metal-binding sites of a variety of glycoses have been gained
by NMR investigations on boron and molybdenum com-
plexes.[4] However, firm knowledge in terms of both crystal-
structure workand NMR spectroscopy in solution has re-
mained elusive. Presently, this criterion is met by only two
classes of glycose complexes, one of which comprises a
series of aldose–palladium(ii) complexes in aqueous solu-
Carbohydrates are polyfunctional molecules. Synthetic
chemistry thus focuses on the regioselectivity of carbohy-
drate transformations. Although metal catalysis appears to
be a promising way to achieve selectivity, the basic problem
remains the same when designing a catalytic reaction: which
functional group is the prevalent metal-binding site? This
[a] X. Kästele, Prof. Dr. P. Klüfers, Dipl.-Chem. F. Kopp,
Dr. J. Schuhmacher, Dr. M. Vogt
Department Chemie und Biochemie der Ludwig-Maximilians Univer-
sität
Butenandtstrasse 5–13, 81377 München (Germany)
Fax : (+49)89-2180-77407
E-mail: kluef@cup.uni-muenchen.de
[**] Polyol Metal Complexes, Part 51. For Part 50 see: A. Geißelmann, P.
Klüfers, C. Kropfgans, P. Mayer, H. Piotrowski, Angew. Chem. 2005,
117, 946–949; Angew. Chem. Int. Ed. 2005, 44, 924–927.
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
6326
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6326 – 6346

FULL PAPER
tion, with the second containing a number of glycose-phe-
nylsilicates in methanolic solution and includes some aldo-
pentoses as well as two ketoses.[5–7] From the viewpoint of
the carbohydrate, palladium is a “conservative” bonding
partner as bonding to palladium alters the equilibrium mix-
ture of aldose isomers in only a limited way—the prevalence
of the pyranoses is maintained on deprotonation and bond-
ing by the rather large metal center.[5,6] Silicon binding, on
the other hand, induces a radical shift of the isomer mixture
in the protic methanol solvent towards the furanose that
bears its anomeric hydroxy group cis to the epimeric one.
Thus, the most acidic cis-furanose group of a monosaccha-
ride acts as a bidentate silicon chelator in the isolated bis-
(diolato)(phenyl)silicates, as shown in the synthetic strategy
that has been derived from studies on glycose model com-
pounds.[7]
To master the difficulties of the conformationally unstable
and polyfunctional glycose ligands in a rational way, carbo-
hydrate model compounds that provide isolated binding
sites are a prerequisite for evaluating the metal-binding sites
in question. The simpler such compounds are, however, the
less representative they are as carbohydrate models. Hence
the diol function of a carbohydrate is badly modeled, for ex-
ample, by ethylene glycol (1,2-dihydroxyethane), which re-
sembles the diol function(s) of a sugar neither in terms of
acidity nor in terms of steric restrictions. In this situation,
anhydroerythritol (cis-oxolane-3,4-diol, abbr.: AnEryt,
Scheme 1) is an optimum “in-between” molecule as, on the
one hand, AnEryt is closely related to the carbohydrates as
it can be considered to be a molecular cut-out of such im-
portant glycosides as the nucleosides (Scheme 1), and, more-
over, its acidity is about the same as that of the glycosides,
including the polysaccharides.[8] On the other hand, AnEryt
is easily obtainable, configurationally stable in the entire pH
range of interest, and chemically stable under alkaline con-
ditions. AnEryt shares the latter two properties, which are
critical for the glycoses, with the glycosides. Contrary to
even glycosides, AnEryt is achiral, a property that we found
to be beneficial for the crystallization of its complexes. Fi-
nally, as a meso compound, AnEryt gives rise to particularly
simple 13C NMR spectra, which often exhibit a typical “coor-
dination-induced shift” (CIS), usually a downfield shift of
the signals of those carbons that bear metal-binding oxygen
atoms. These properties have led to extensive use of AnEryt
as a bridge to carbohydrate–(semi)metal chemistry, both in
the area of transition-metal as well as in main-group chemis-
try.
The fact that AnEryt models a furanoside and not a pyra-
noside is of particular importance if one is dealing with
small central atoms. For silicon, it has been shown that no
complexes are formed in aqueous media upon reaction with
trans-pyranoidic diol moieties. Instead, hydrogen-bonded
oligosilicate–oligosaccharide assemblies can be isolated, as
has been shown for a-cyclodextrin as the carbohydrate com-
ponent.[9] Contrary to trans-pyranoidic diols, silicon-chelat-
ing properties have been found with sugar alcohols—glycon-
ic and glycaric acids—whose open-chain diol groups are
more flexible.[10] However, to open the way to a diolatosili-
cate chemistry in aqueous solution for a sugar alcohol, sup-
port by a pattern of spatially suited hydrogen bonds is re-
quired.[11] cis-Furanoidic diols like AnEryt, but also the ribo-
furanosides, including the nucleosides and the furanoses, in-
trinsically provide the property of silicon chelation in aque-
ous media without the need for support by secondary
interactions such as hydrogen bonds.
For these reasons, AnEryt plays a leading role in the rap-
idly growing field of the carbohydrate chemistry of silicon:
À À
the first hydrolytically stable Si O C linkage was prepared
with AnEryt.[12] The privileged position of furanoidic diols
as silicon chelators has been derived by comparing AnEryt
and open-chain polyols.[11] NMR workby Kinrade et al. con-
cerning the question of isomers at pentacoordinate silicon
centers has focused on AnEryt,[13] and the latest published
workconcerns the attempts of Lambert et al. to use
13C NMR shift differences derived from AnEryt–silicate sol-
utions as the key to aqueous glycose–silicate chemistry.[14] In
this work, an unusual number of misleading errors accumu-
late, and these are discussed in some detail in the Support-
ing Information. Most of them originate in an uncritical ap-
plication of shift-difference concepts. Thus, as a methodolog-
ical leitmotif, the NMR spectroscopic properties of the re-
spective compound classes are deduced here in order to pro-
vide reliable data in particular on shift-difference rules,
which are a valuable tool when applied carefully.
The chemical focus of the present workis the astonishing-
ly multifaceted silicon chemistry of AnEryt. On the one
hand, AnEryt is a powerful tool for surmounting the obsta-
cles of carbohydrate–silicon chemistry but, on the other, and
like its carba analog cis-cyclopentanediol, AnEryt exhibits a
unique and unexpected alkoxysilane chemistry which ap-
pears to be untransferable to a polyfunctional carbohydrate.
Scheme 1. Diols referred to in this work(hydroxyl hydrogen atoms omit-
ted), including the atomic numbering scheme; AnEryt numbering is de-
rived from the parent polyol erythritol, not from oxolane numbering.
Chem. Eur. J. 2005, 11, 6326 – 6346
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6327

P. Klüfers et al.
Results
xylopyranose) are ruled out as efficient silicon chelators in
view of the high energy needed for torsion, in line with the
results in the a-cyclodextrin/silicate system mentioned
above. As expected, the same result is obtained for trans-
furanoses: although comprising a much larger range of flexi-
bility at low energy, 08 torsion is far outside the energetically
accessible region (Figure 1 top: the perfluoro derivative of
a-l-threofuranose). Torsion of a cis-pyranose proceeds with
Furanoses—better ligands for silicon: Structural analyses
that show five-membered chelate rings of the SiO2C2 type
always reveal small torsion angles of the O-C-C-O moiety.
This finding is to be expected for the C2O2 units of, say, cat-
echolate or oxalate,[15] but is also the case for diolates and
the dianions of hydroxycarboxylic acids. In turn, the more
cheaply an approximately 08 torsion can be achieved in
terms of energy, the better silicon chelators are the diolate
ligands derived from carbohydrates. A cis-furanoidic diol,
that is, a diol group attached cis to an oxolane ring, thus ap-
pears to be a better choice than open-chain diols and, partic-
ularly, pyranoidic diol functions. The reason for the confor-
mational flexibility of a furanose ring is its balance of strain.
While open-chain diols experience Pitzer strain when twist-
ed into an eclipsed conformation, and pyranoses are bur-
dened with ring strain on being twisted towards 08 torsion, a
furanose is characterized by a balance of the various types
of strain over a pronounced range of torsions.
To give an impression of numbers, the torsional energies
of some molecules that are related to the carbohydrates
have been calculated at the B3LYP/6–31G(d) level of
theory. To prevent, in terms of X-ray structures, irrelevant
intramolecular hydrogen bonding within the diol group, the
hydroxy groups were replaced by fluorine atoms in the cal-
culations. Figure 1 shows the energetic costs for reaching the
range of torsion angles of less than about 308 that appears
to be typical for silicon–diol bonding. As a result, trans-
pyranoses (Figure 1 bottom: the perfluoro derivative of b-d-
4
less resistance on a path that connects the starting C1 chair
4
and the B1 boat conformation. In terms of energy, however,
cis-pyranose (Figure 1 bottom: the perfluoro derivative of a-
d-xylopyranose) and open-chain (Figure 1 bottom: the per-
fluoro derivative of threitol) torsions are significantly more
expensive, in energy terms, than torsion of a cis-furanoidic
molecule (Figure 1 top: the perfluoro derivatives of b-l-
threofuranose and anhydroerythritol). For the latter, the
typical large range of low-energy conformations of about
1008 for furanose includes the 08 point.
To derive the rules of glycofuranose ligation, the cis-oxo-
lane diols thus appear to be the first choice. Two diols may
be considered. To model the most acidic C1/C2 binding site
of a furanose, cis-2,3-oxolane diol appears suitable, whereas
for modelling the cis-furanosides, AnEryt may be a good
choice. Of these two diols, AnEryt has benefits for practical
workin terms of stability, particularly in alkaline solution
(2,3-oxolane diol is much more reactive as it is the hemiace-
tal of 2,4-dihydroxybutanal).
The applicability of the model calculations on fluorine
compounds to the oxo compounds of interest has been
proven for some AnEryt conformations. The structures of
two AnEryt and one AnEryt·3H2O conformations were re-
fined at the same level of theory. The O-C-C-O torsion
angles are in the range of 30–408, depending on the actual
conformation, and include the F-C1-C2-F angle of the fluoro
derivative of a-threose of 39.18. About the same torsion
angle, 41.48, is obtained for the minimum structure of the di-
fluoro derivative of AnEryt. These calculated values are in
agreement with the four angles found in the X-ray crystal
structures: 40.18 and 41.08 in AnEryt·NaClO4,[16] and
38.1(2)8 and 42.0(2)8 in two symmetrically independent mol-
ecules in the structure of pure AnEryt (Figure 2, left). As a
result of both computational and X-ray methods it may
therefore be stated that AnEryt does not reside at 08 torsion
Figure 1. Top: B3LYP/6–31G(d) torsion energies of various F-C-C-F
functions in the perfluoro derivatives of a- and b-l-threofuranose
Figure 2. ORTEP diagrams (50% probability ellipsoids) of the two oxo-
lane diols. Left: one of the two symmetrically independent molecules in
crystals of AnEryt. The diol torsion angle is 38.1(2)8 in the depicted mol-
ecule (the other value is 42.0(2)8). Right: one of the three symmetrically
independent molecules in crystals of l-AnThre. The diol torsion angle is
166.7(2)8 in the depicted molecule (the other values are 166.9(2) and
167.2(2)8).
À
(torque about the C1 C2 bond), and anhydroerythritol (torque about the
À
C2 C3 bond). Bottom: B3LYP/6–31G(d) torsion energies of various F-
C-C-F functions in the perfluoro derivatives of a- and b-d-xylopyranose
À
À
(torque about the C1 C2 bond), and d-threitol (torque about the C2 C3
bond). The colored area indicates the range of abscissa values that is
spanned in silicon-bonded diols.
6328
www.chemeurj.org
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6326 – 6346

Silicon Chelation in Aqueous and Nonaqueous Media
FULL PAPER
in its stable conformation, but 08 torsion is available at low
cost.
chelators according to the DFT treatment above. This find-
ing was supported by experiments in nonaqueous media,
where silicon bonding to such a diol can be forced due to
the absence of thermodynamic traps like ortho-silicate.
Thus, the cis and trans isomers of cyclohexane-1,2-diol
(Chxd) react with dichlorodiphenylsilane in aprotic media
to give products of the net formula Ph2Si(ChxdHÀ2). The
29Si NMR solution spectra show a single resonance for both
diols. The actual values are those for unstrained tetrahedral
coordination of the central atom: d=À34.4 and À35.3 ppm
for Ph2Si(cis-ChxdHÀ2) (1) and Ph2Si{(R,R)-trans-ChxdHÀ2)}
(2) respectively. Crystal-structure analysis revealed the
origin of the missing strain: both 1 (Figure 3) and 2
(Figure 4) are in fact dimers. The fact that a single 29Si
NMR signal is observed with almost the same value for
both isomers can be interpreted in terms of the dimers
being the only solution species. Silicon chelation by the cis-
diol hence does not occur even to a minor extent in the so-
The conformers of the trans isomer anhydrothreitol
(AnThre) exhibit much larger diol torsion angles in the
solid state. The respective mean angle of the l enantiomer
(Figure 2, right) is 166.98, and is also close to the F-C1-C2-F
angle in the DFT structure of the fluoro derivative of b-l-
threose at minimum energy (162.28). Keeping in mind the
respective curve in Figure 1, chelation of any central atom
by a trans-furanose, an example being the 1,2-site of b-ribo-
furanose, is clearly ruled out.
Owing to the repeatedly incorrect assignment of the
13C NMR signals of AnEryt in recent work,[13,14] let us recall
the well-known fact that AnEryt, as opposed to AnThre, ex-
hibits an irregular order of its 13C NMR signals, with C1/C4
being downfield of C2/C3 in the usual solvents (Table 1).[17]
Table 1. Signal order in the 13C NMR spectra of AnEryt (c=0.6 molLÀ1
in various solvents. Dd=d(C1/C4)Àd(C2/C3).
)
d(C1/C4)
d(C2/C3)
Dd
CDCl3
CD3OD
72.9
73.3
72.5
72.9
71.4
72.5
71.1
71.4
1.5
0.8
1.4
1.5
1m [K([18]crown-6)]OMe/MeOH
D2O
In D2O, the C1/C4 position is 1.5 ppm downfield of C2/C3, a
result that can be rationalized by DFT calculations. At the
PBE1PBE/6–311++G(2d,p)//B3LYP/6–31G(d) level, the
mean signal positions of the AnEryt·3H2O aggregate are
d=78.6 and 77.3 ppm relative to tetramethylsilane for C1/
C4 and C2/C3, respectively, and show a 1.3 ppm downfield
shift for the carbons adjacent to the diol group, the differ-
ence being close to the experimental value for an aqueous
solution. It should be mentioned that AnEryt shares this ir-
regularity with some compounds with the AnEryt partial
structure. Thus, the C3/C4 signals show a reversed order in
the 13C NMR spectra of d-erythrose and the methyl-d-eryth-
rofuranosides as well.[17]
Figure 3. ORTEP diagram (50% probability ellipsoids) of the Ci-symmet-
À
rical dimers in crystals of 1. Bond lengths [Š] and angles [8]: Si O1
i
1.631(1), Si O2 1.637(2); O1-Si-O2i 111.97(7); diol torsion: 61.0(2)8.
À
Symmetry code: i: 1Àx, 1Ày, 1Àz.
To rationalize the systematic findings derived from com-
puter chemistry, compounds with the diol patterns in ques-
tion were treated with various silicon-based starting materi-
als to give centers with coordination numbers four, five, and
six. The focus of the following sections is on AnEryt, which
gives rise to the most widespread chemistry.
Tetracoordinate silicon centers—oxysilanes derived from
furanoidic diols: Penta- and hexacoordination of silicon to-
wards polyols is consistently prevented by introducing two
organyl substituents at the silicon atom. To examine the
structures and NMR shift differences for tetracoordinate sil-
icon centers, the diphenylsilylene moiety was bonded to var-
ious diols.
Figure 4. ORTEP diagram (50% probability ellipsoids) of the dimers in
1,2-Cyclohexylenedioxy(diphenyl)silanes: Diol functions at-
tached to six-membered rings like pyranoses should be
unable (trans-diols) or hardly able (cis-diols) to act as silicon
À
À
crystals of 2. Bond lengths [Š] and angles [8]: Si1 O11 1.619(1), Si1 O22
À
À
1.638(1), Si2 O12 1.634(1), Si2 O21 1.637(1); O11-Si1-O22 114.28(7),
O12-Si2-O21 113.57(7); torsion: À61.1(2)8 for both diol groups.
Chem. Eur. J. 2005, 11, 6326 – 6346
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6329

P. Klüfers et al.
lution equilibrium. The chemical shifts and shift differences
in the 13C NMR spectra are given in Table 2. It should be
noted that shift differences on H/Si exchange are hardly sig-
nificant for these nonchelated four-coordinate silanes.
of 4 is not that of a monomeric chelate. Instead, an unstrain-
ed molecule with all bonding angles close to their ideal
values is observed in a trimeric structure (Figure 6). Al-
though 08 torsion is outside the range of achievable diol tor-
sion angles, the great flexibility of furanoidic rings is obvious
from the AnThre structure as well. To build up the trimer,
the diol torsion angles almost span the entire available 1008
range of a furanoidic diol, with the actual values being be-
tween 828 and 1658.
Table 2. Chemical shifts in the 13C NMR spectra of the Ph2Si(ChxdHÀ2
)
isomers 1 and 2, and shift differences (d(1/2)Àd(free diol)). Bold: Dd of
those carbon atoms that bear the silicon-binding oxygens.
C1/C2
C3/C6
C4/C5
1
2
d [ppm]
Dd [ppm]
d [ppm]
77.5
1.6
76.9
1.4
30.4
0.4
34.1
1.1
21.7
0.2
24.2
0.2
Dd [ppm]
3,4-Oxolanylenedioxy(diphenyl)silanes
and
1,2-
cyclopentylenedioxy(diphenyl)silane: The structural and
spectral properties of four-coordinate Si centers bonded to
an alkylenedioxy substituent derived from a furanoidic diol
were investigated with the two isomeric oxolane-3,4-diols.
Thus, in addition to AnEryt, anhydrothreitol (AnThre) was
included in this part of the investigation to demonstrate the
inability of furanoidic trans-diols to form five-membered
chelate rings.
Dichlorodiphenylsilane reacts with AnEryt or l-AnThre
in trichloromethane, in the presence of pyridine as a base, to
form Ph2Si(AnErytHÀ2) (3) and Ph2Si(l-AnThreHÀ2) (4), re-
spectively. The crystal structure of 3 is shown in Figure 5.
Figure 6. ORTEP diagram (50% probability ellipsoids) of the C1-sym-
metrical trimers in crystals of 4. Mean bond lengths [Š] and angles [8]:
À
Si O 1.638; O-Si-O 112.1. Diol torsion O2n-C2n-C3n-O3n: 81.6(4)8 for
n=1, 164.6(2)8 for n=2, and 163.3(2)8 for n=3.
The 29Si NMR spectra verify the assumption that the mol-
ecules found in the solid state are also the solution species.
Thus, a single resonance is observed for solutions of both
oxolanediols. The value for 4 (d=À29.6 ppm) resembles the
value for the also unstrained but dimeric (cyclohexylene-
dioxy)silanes. The AnEryt derivative, however, shows a dis-
tinct downfield shift (d=À1.4 ppm). The 13C NMR spectra
of the same solutions exhibit larger shift-differences for H/Si
exchange than the Chxd derivatives. For the chelating
silane, the numerical values in Table 3 show that the
13C NMR signals of those carbon atoms that bear the sili-
Figure 5. ORTEP diagram (50% probability ellipsoids) of one of the two
symmetrically independent monomers in crystals of 3. Bond lengths [Š]
À
À
and angles [8]: Si1 O21 1.649(2), Si1 O31 1.642(2); O21-Si1-O31 97.9(1);
diol torsion: À5.4(2) (À6.7(2)8 for the second molecule in the asymmetric
unit).
Table 3. Chemical shifts in the 13C NMR spectra of Ph2Si(DiolHÀ2) de-
rived from diols attached to five-membered rings, and shift differences
(d(Si-bonded)Àd(free diol)). Bold: Dd of those carbon atoms that bear
the silicon-binding oxygens. The atomic numbering of the cyclopentane-
diol has been adapted to the oxolanediols (cf. Scheme 1).
The molecular structure is that of a monomer. Contrary to
the case of the cyclohexylenedioxy derivatives, chelation is
possible with the furanoidic diol. As expected, the five-
membered chelate ring is almost planar, with a diol torsion
angle close to 08. Such geometrical parameters cannot be
met by AnThre. However, this diol provides another exam-
ple that the inability to form a chelate ring must not be con-
fused with a lackof reactivity. Thus, the solid-state structure
C2/C3
C1/C4
C5
5
3
4
d [ppm]
Dd [ppm]
d [ppm]
Dd [ppm]
d [ppm]
Dd [ppm]
80.1
7.2
79.2
7.8
80.0
3.7
35.2
4.3
75.2
2.3
73.0
0.0
22.5
2.7
6330
www.chemeurj.org
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6326 – 6346

Silicon Chelation in Aqueous and Nonaqueous Media
FULL PAPER
con-binding oxygens are shifted downfield by almost 8 ppm.
However, mere H/Si exchange is obviously not the only
factor responsible for this significant shift difference, as is
shown by the AnThre values, which are about half of the
AnEryt values even though the structures are isomeric.
The body of shift-difference data was enlarged by investi-
gating the carba analog of AnEryt, namely cis-cyclopentane-
1,2-diol (cis-Cptd). The usual synthetic procedure yielded
solutions of Ph2Si(cis-CptdHÀ2) (5), which contain mono-
meric molecules according to the 29Si NMR spectra (d=
À3.7 ppm). The 13C NMR spectroscopic data verify the as-
sumption of monomeric molecules in solution as well, as the
shift differences closely resemble the high values of the
AnEryt case and not those of the AnThre-derived trimer
(Table 3). On crystallization, however, a significant change
of the 29Si NMR spectrum was observed. In the solid, the re-
spective signal is shifted more than 30 ppm upfield. The mo-
lecular origin of this large shift difference was unraveled by
an X-ray analysis (Figure 7): dimerization has occurred and
Tetrachlorosilane and a double molar amount of either
AnEryt or cis-Cptd react, upon heating in toluene, with for-
mation of hydrogen chloride. The 29Si and 13C NMR spectra
indicate formation of the expected spiro compounds. For
both diols, the 13C NMR signals are doubled according to
the spiro pattern and are shifted downfield. In the case of
AnEryt, the downfield shift in toluene solution roughly re-
sembles that of the analogous diphenylsilane (Dd=6.2 and
6.3 ppm for C2/C3, and 2.0 and 2.1 ppm for C1/C4; 29Si
NMR: d=À36.7 ppm). The respective 13C NMR values for
cis-Cptd are shifted to a smaller extent (Dd=3.5 and
3.7 ppm for C1/2, 2.0 and 2.2 ppm for C3/5, and 0.9 for C4;
29Si: d=À36.8 ppm). The 29Si NMR spectroscopic data,
which are typical for strained chelate rings, again show a
pronounced downfield shift when compared to an unstrain-
ed Si(OR)4 reference like tetramethoxysilane (d=
À78.0 ppm).
Attempts to crystallize the silicic acid ester of anhydroery-
thritol yielded a total of four polymorphic forms of Si-
(AnErytHÀ2)2 (6). When sorted according to the molecular
structures, three polymorphs are formed from the spirocyclic
molecules whose synthesis had been attempted. The trimor-
phic spirosilane is termed a-6, b-6, and g-6 according to de-
creasing density. The molecular structure of a-6, which is de-
picted in Figure 8, is representative of all the modifications
Figure 7. ORTEP diagram (50% probability ellipsoids) of the Ci-symmet-
À
rical dimers in crystals of 5. Bond lengths [Š] and angles [8]: Si O2
i
i
À
À À
1.629(1), Si1 O3 1.636(2); O2 Si O3 112.51(8)8; diol torsion: 52.4(2)8.
Symmetry code: i: 1Àx, 1Ày, 1Àz.
an unstrained structure with bonding angles at the Si center
close to the tetrahedral angle has formed. Thus, even a diol
group attached to a five-membered ring, which is capable of
chelating a silicon atom in principle, is obviously able to
support the formation of unstrained oligomers as well.
The data in the preceding section stress the fact that in
13C NMR spectra a substantial downfield shift occurs on H/
Si exchange provided the discussion is restricted to chelating
silanes.
Figure 8. ORTEP diagram (50% probability ellipsoids) of the monomeric
À
spirosilane in crystals of a-6. Bond lengths [Š] and angles [8]: Si O21
À
À
À
1.628(2), Si O22 1.629(2), Si O31 1.619(2), Si O32 1.625(1); O21-Si-
O31 99.05(9), O22-Si-O32 98.73(7); diol torsion: O21-C21-C31-O31:
À0.8(2)8; O22-C22-C32-O32: À1.2(2)8.
of 6. The common features include, again, almost flat
SiO2C2 chelate rings with approximately zero torsion of the
oxolanylenedioxy moiety (i.e. almost ideal local D2d symme-
try of the silacycles), and bending of the oxolane O-atoms
towards the silicon atom. Although the differences with re-
spect to molecular symmetry are small, the packing patterns
of the molecules are completely different in the polymorphs.
Hence, no group–subgroup relationships connect the poly-
morphs nor are there phase transitions between the mono-
mer forms on heating. Thermal analysis, optical inspection
on heating, and X-ray powder diffraction at various temper-
atures show partial melting and re-solidifying until eventual-
Bis(cycloalkylenedioxy)silanes derived from anhydroerythri-
tol or cis-cyclopentane-1,2-diol: As the next step towards
carbohydrate–silicate complexation, the steric and electronic
restrictions introduced by the two organyl substituents were
lifted. Starting with SiCl4 as an organyl-free silicon source,
the acyclic diol pinacol (2,3-dimethylbutane-2,3-diol) forms
a spirocyclic bis(alkylenedioxy)silane.[18] Furanoidic diols,
however, were not used in this context. Since not many data
on such compounds are available, both AnEryt and its carba
analog cis-Cptd were also included in this part of the investi-
gation.
Chem. Eur. J. 2005, 11, 6326 – 6346
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6331

P. Klüfers et al.
ly all the samples melt at the melting point of b-6. On cool-
ing, the melts solidify uniformly to yield only b-6.
tivity of the substance. After thorough grinding, the main
signal is observed at d=À80 ppm, which is indicative of un-
strained tetrahedral Si(OR)4 coordination at silicon (cf. the
d=À78 ppm signal of tetramethoxysilane) in a tentative
oligomer or polymer. Attempts to grow crystals of the mon-
omer failed; even sublimation yielded a dimer instead of a
monomer according to solid-state 29Si NMR spectroscopy
(d=À94.6 ppm). A structure analysis confirmed a second
modification of the dimeric pentacoordinate Cptd ester. The
molecular structures of the two modifications are very simi-
lar, hence only the structure of b-72 (crystals grown from tol-
uene) is depicted in Figure 10.
This latter property has also been found in a fourth poly-
morph, whose crystallization follows a special protocol: crys-
tallization from toluene solutions succeeded at about 48C
after the solutions were saturated at the same temperature
with respect to the fourth polymorph (i.e., saturate at room
temp., crystallize at 48C for about four days, remove crystals
of a-6 completely, and allow the solution to stand at 48C for
a further 2–3 weeks). 29Si solid-state NMR spectra indicate
that the molecular structure of this polymorph is different
(d=À36.7, À37.9, and À36.6 ppm for a-, b-, and g-6, respec-
tively; d=À94.3 ppm for the fourth polymorph). X-ray anal-
ysis revealed an unexpected molecular structure. As shown
in Figure 9, the crystals are made up of dimers of the spirosi-
Figure 10. ORTEP diagram (50% probability ellipsoids) of the Ci-sym-
À
metrical dimers in crystals of b-72. Bond lengths [Š]: Si O21 1.651(2),
i
À
À
À
À
Si O22 1.929(1), Si O22 1.726(1), Si O31 1.672(1), Si O32 1.641(2);
diol torsion: O21-C21-C31-O31: 11.8(2)8; O22-C22-C32-O32: À27.3(2)8.
Symmetry code: i: Àx, Ày, Àz.
Figure 9. ORTEP diagram (50% probability ellipsoids) of the Ci-symmet-
À
rical dimers in crystals of Si(AnErytHÀ2)2. Bond lengths [Š]: Si O21
The structures of the Si2O8 cores of 62 and 72 are the
i
À
À
À
À
1.650(2), Si O22 1.929(2), Si O22 1.732(2), Si O31 1.667(2), Si O32
1.641(2); diol torsion: O21-C21-C31-O31: À2.8(3)8; O22-C22-C32-O32:
À16.8(3)8. Symmetry code: i: 1Àx, 1Ày, 1Àz.
À
same. Two Si O distances are substantially longer than the
À
other ones, although these longer Si O bonds are not the
new bonds that are formed on dimerization. Thus, the new
structures appear to be stabilized transition states of dimeri-
zation towards molecules like 52. Such aspects were investi-
gated by DFT methods. These results, which include the
close resemblance of calculated and solid-state structures,
the almost equal energy of a pair of monomers and the re-
spective dimer, and the low activation energy for formation
and cleavage of the dimer, will be published in a separate
paper. Although the focus of this worklies in a critical eval-
uation of NMR shift differences in the diol/silicon field, the
solid-state structures of 62 and 72 show that the chemistry of
simple model diols is not free of surprising results, thus indi-
cating that not only the carbohydrate–silicon interaction is
waiting to be uncovered, but also the basic chemistry behind
it.
lane but not of the tetracoordinate type, whose formation
might be plausible as a means of reducing steric strain in the
chelate rings as with 5. Instead, the coordination number of
silicon is five, which is not unusual for anionic Si centers but
is a new and unexpected structural motif for a simple silicic
acid ester.
When AnEryt is replaced by its carba analogue cis-Cptd,
the solutions of the monomeric spirosilane show a different
crystallization behavior, as monitored by 29Si solid-state
NMR spectroscopy. Under various crystallization conditions,
the Cptd ester always shows downfield-shifted resonances
typical for pentacoordination. Thus, an approximate 60 ppm
shift difference of the 29Si NMR signal is obtained for gently
ground crystals of Si(CptdHÀ2)2 (7) that had been grown
from a toluene solution (solid state: d=À94.7 ppm; solu-
tion: d=À36.8 ppm). Gentle grinding is essential to record
the solid-state signal correctly due to the mechanical sensi-
What about the typical shift differences in the NMR spec-
tra? Having arrived at this point, things seem to develop
promisingly. Chelation—not mere binding—of tetracoordi-
nate silicon centers by alkylenedioxy groups is indicated
6332
www.chemeurj.org
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6326 – 6346

Silicon Chelation in Aqueous and Nonaqueous Media
FULL PAPER
both by a typical downfield shift in the 29Si NMR spectra
and by a typical coordination-induced shift in the 13C NMR
spectra of 6–8 ppm downfield. Note, however, the lower
values for 7.
whereas the other oxolane ring and the hydroxo ligand are
on different faces of the respective chelate ring (anti). As
for the lithium and potassium analogs, the sodium com-
pound contains both syn and anti isomers. Another structur-
al motif is found in the isotypic Rb (8b) and Cs (8c) salts.
The anion structure of the Cs salt 8c is shown in Figure 12.
Pentacoordinate silicates with AnErytHÀ2 ligands
Structures of the alkali-metal salts of the [Si(OH)-
(AnErytHÀ2)2]À ion: The current focus of the discussion is
doubtless the aqueous chemistry of the silicate/AnEryt
system. When introducing water as the solvent, protolytic
and hydrolytic equilibria have to be considered in addition
to the chelation/oligomerization chemistry discussed above.
Thus, hydrolysis of the spirocyclic silane Si(AnErytHÀ2)2 in
neutral or acidic aqueous solution may start by the addition
of a water molecule to form a transient pentacoordinate
species [Si(H2O)(AnErytHÀ2)2]. At higher pH, the mono-
deprotonated anion of this hypothetical acid is the predomi-
nant species, which is obtained directly by the action of
AnEryt on silica in alkali lye.
Figure 12. ORTEP diagram (40% probability ellipsoids) of the C2-sym-
À
À
metrical anions in crystals of 8c. Bond lengths [Š]: Si O2 1.730(4), Si
À
O3 1.720(3), Si O8 1.652(5); diol torsion (O2-C2-C3-O3): À3.5(5)8.
As discussed both in Lambertꢀs workand in a workof the
Kinrade group, such aqueous solutions show three signals in
the 29Si NMR spectra in the region of pentacoordinate sili-
con.[13,14] Thus, the question of isomerism arises. To get an
idea of what kinds of isomers have to be considered for
these pentacoordinate silicates, a lookat solid-state struc-
tures is helpful. Presently, a total of five crystal structures
are available. In addition to the published structures of Li-
[Si(OH)(AnErytHÀ2)2]·H2O and K[Si(OH)(AnErytHÀ2)2],[12]
the anhydrous Na, Rb, and Cs salts have been crystallized
from aqueous solutions and structurally resolved by single-
crystal X-ray analysis. The structure of the silicate ion in Na-
[Si(OH)(AnErytHÀ2)2] (8a) is shown in Figure 11. In a
With these larger cations the anti/anti isomer is isolated,
with the coordination at silicon being sp as with the lithium
and sodium salt. The only exception to sp coordination is
found with potassium as the counterion. The hydroxo ligand
takes an equatorial position in this case, and each diolato
ligand is bonded axially/equatorially, which is the only possi-
ble position since, again, the bite of a diolate ligand is too
small to span an equatorial/equatorial chelate. Table 4 gives
a comprehensive survey of significant data on the five
alkali-metal diolatosilicates.
Table 4. Solid-state properties of the pentacoordinate silicates A-
[Si(OH)(AnErytHÀ2)2] (A=Li–Cs; the Li salt crystallizes as a monohy-
drate). Reference denotes the workwhere the structure is described; all
NMR spectroscopic data are from this work.
A
Isomer[a]
Si[b]
% tbp[c]
29Si[d]
Reference
[12]
Li
Na
K
Rb
Cs
syn/anti
syn/anti
syn/anti
anti/anti
anti/anti
sp
sp
tbp
sp
sp
2.0
7.8
62.7
13.9
13.3
À95.5
À97.6
À98.3
À92.7
À92.9
this work
[14]
this work
this work
[a] Orientation of the oxolane ring relative to the hydroxy ligand. [b] Co-
ordination at Si. [c] Percentage distortion of sp towards tbp (cf. R. R.
Holmes, Prog. Inorg. Chem. 1984, 32, 119–235). [d] Solid-state 29Si NMR
chemical shift.
Figure 11. ORTEP diagram (50% probability ellipsoids) of the anions in
À
À
À
crystals of 8a. Bond lengths [Š]: Si O21 1.730(2), Si O22 1.734(2), Si
À
À
O31 1.728(2), Si O32 1.720(2), Si O8 1.662(2); diol torsion: O21-C21-
C31-O31: 14.1(2)8; O22-C22-C32-O32: 9.7(2)8.
As a result, two kinds of isomers have to be considered at
the pentacoordinate silicon centers: syn/anti and sp/tbp.
Prior to a search for such isomers in solution by NMR spec-
troscopy, it would be helpful to have an idea what activation
barriers may be expected for the mutual transformation of
such isomers in order to decide whether these isomers can
be expected to be resolved on the NMR time scale. For this
purpose, DFT calculations were performed on the isolated
[Si(OH)(AnErytHÀ2)2]À ion, not for modeling the reaction
path for syn/anti isomerization, which surely cannot be de-
slightly distorted square-pyramidal (sp) coordination at the
silicon center, the hydroxo ligand takes the apical position.
Each diolate is bonded in the basal plane, which is the only
possible position since the “bite” of a diolate ligand is not
large enough to span a basal/apical chelate with its larger
angle at silicon. Two different orientations are taken by the
oxolane rings. One of them is located on the same face of
the corresponding chelate ring as the hydroxo ligand (syn),
Chem. Eur. J. 2005, 11, 6326 – 6346
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6333

P. Klüfers et al.
scribed in a realistic way by considering the isolated anion
only, but to learn about sp/tbp isomerism.
former oligomerization hypothesis, which was based on the
assumption that the syn/syn case is the only stable one (see
above).[13,14]
However, transformations at the silicon centers other
than syn/anti have to be considered. Examples include reac-
tions at the hydroxo ligand such as condensation. Thus, a
The question was dealt with using an overall anti/anti ge-
ometry first. At the B3LYP/6–31G(d) level of theory for
structure optimization, the sp structure found in the crystal-
line state refined to minimum energy on changing to tbp,
whereas starting with overall syn/anti geometry and sp coor-
dination at silicon, a local minimum structure of sp geome-
try is only achieved if a hydrogen bond from the hydroxo
ligand as the donor towards the oxolane O-atom of the syn
substituent is provided in the starting geometry. Directing
condensation
reaction
À À
may
yield
products
like
2À
[(AnErytHÀ2)2Si O Si(AnErytHÀ2)2] , which appear possi-
ble in view of related species published by the Tacke group,
which has used hydroxycarboxylic acids as ligands.[19] An ex-
perimental hint regarding the occurrence of such condensa-
tion comes from thermal analysis of the potassium and
cesium compounds. Both salts lose half a mole of water per
formula unit at a temperature of 2808C, and they then de-
compose without melting at about 4008C.
À
the O H vector away from the ether oxygen results in a
transformation of the starting sp geometry to the tbp struc-
ture. When starting the refinement of a tbp structure with
the hydrogen bond towards the oxolane-O acceptor, this
bond is destroyed in favor of a bond to one of the alkoxo
O-atoms. From this refinement behavior, we conclude that
no considerable activation barriers should exist between the
sp and tbp forms of the anion in question. sp/tbp Isomerism
thus should not be detectable in real-world experiments,
thus leaving syn/anti isomerism as the only type of isomer-
ism to be considered. It should be recalled at this point that
in the crystalline alkali-metal salts only the syn/anti and the
anti/anti geometries have been found. The syn/syn case has
never been observed in solid-state work, despite both Lam-
bert and Kinradeꢀs discovery of this geometry in a workby
our group which, however, does not deal with silicates at
all.[12–14] As a result of this section, the interpretation of the
three signals in the 29Si NMR solution spectra in terms of
syn/anti isomerism appears straightforward, in agreement
with Lambertꢀs suggestion and Kinradeꢀs revision of his
To gain deeper insight into the significance of condensa-
tion of hydroxo ligands in solution, analogous phenylsili-
cates of the formula [SiPh(AnErytHÀ2)2]À were included in
the investigation.
The [SiPh(AnErytHÀ2)2]À ion and its relation to the hydroxo
analog: Bis(diolato)(phenyl)silicates were obtained by
switching to trimethoxyphenylsilane/methanol instead of sili-
cate/water. When both condensation and hydrolysis are pre-
vented by this modification, syn/anti isomers of the [SiPh-
(AnErytHÀ2)2]À ion should remain as the only species. The
29Si NMR spectra of such solutions exhibit the signals of two
major and one minor pentacoordinate species. The similarity
of these spectra to those recorded from aqueous silicate sol-
utions is obvious (Figure 13). In agreement with the exis-
tence of two main species, crystallization with potassium as
Figure 13. 29Si and 13C (DEPT-135) NMR spectra of pentacoordinate AnEryt–silicate species prepared from the diol, SiPh(OMe)3 or Si(OMe)4, and base
at the molar ratio given in a)–c). Color code: red: anti/anti; blue: syn/anti; green: syn/syn isomer of the respective [Si(R)(AnErytHÀ2)2]À ion; violet: equi-
librating isomer mixture. a) R=OH; base: LiOH; solvent: water; molar ratio: 3:1:1; total Si concentration: 0.54 molkgÀ1. The methanol signal stems
from hydrolysis of the Si(OMe)4 starting material and may be used as a reference for DEPT assignment. Dashed line: free AnEryt. b) R=Ph; base:
KOMe/[18]crown-6; solvent: methanol; molar ratio: 2:1:2; total Si concentration: 0.38 molkgÀ1. AnEryt region in DEPT mode referenced to the
[18]crown-6 signal. c) Same as (b), but with a molar ratio of 3:1:1; total Si concentration: 0.61 molkgÀ1. Dashed line: free AnEryt.
6334
www.chemeurj.org
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6326 – 6346

Silicon Chelation in Aqueous and Nonaqueous Media
FULL PAPER
the counterion yields crystals of the formula K[SiPh-
(AnErytHÀ2)2]·1/2MeOH (9) which are made up of the anti/
anti and the syn/anti phenylsilicate in equal parts
(Figure 14). Thus, the two main signals in the 29Si NMR
nals are close together; it may be assumed that this syn/anti
signal stems from the anti part of this isomer. The corre-
sponding spectrum of an aqueous diolatosilicate solution is
shown in Figure 13a. The C1/C4 region is less structured but
the C2/C3 signals again show the pattern of a single strong
signal, a pair of signals of about equal intensity, and a single
[13]
weakresonance, in agreement with Kinradeꢀs finding.
From the isomer distribution in the crystalline salts we
assume that the minor component resembles the syn/syn
isomer in this system also. Moreover, the fact that both the
29Si and the 13C NMR spectra show such a close resemblance
leads us to believe that the chemistry of the aqueous system
is also adequately described by taking into account syn/anti
isomers of a mononuclear bis(diolato)hydroxosilicate as the
only relevant type of isomerism.
A special feature of the phenylsilicate spectra will gain
significance in more complicated cases. Figure 13 shows the
13C NMR signals of the phenyl residue in addition to the
diol signals. It should be noted that both the signals stem-
ming from the ipso-carbon atom as well as the signals from
the two ortho-carbon atoms mirror the 29Si NMR spectrum.
These signals appear to be as suitable as the silicon nucleus
as a probe, and thus provide a tool for resolving an acciden-
tal overlap of 29Si NMR signals.
What about characteristic shift differences in the
13C NMR spectra? Figure 13a shows Si-bonded and free
AnEryt. The result is opposite to that of the tetracoordinate
chelating silanes. The signals of C2/C3—the carbon atoms of
the silicon binding site—are shifted about 1 ppm downfield
(about 6–8 ppm in chelating oxolanylenedioxysilanes),
whereas the signals of the adjacent carbons experience an
about 4 ppm downfield shift (about 2 ppm in the respective
silanes). The shift differences are thus sensitive to at least
the coordination numbers and possibly to the actual struc-
tures. It remains to be clarified whether or not these new
values are representative of related silicates of the furano-
ses.
This result may be compared with published work: by er-
roneously switching the CH and CH2 coordinate in their
DEPT spectrum, the Lambert group has wrongly assigned
both the 13C signals of free and Si-bonded AnEryt and thus
derived wrong shift differences, which are then the basis for
interpreting the spectra of aldoses and ketoses, which are
wrongly assigned as well.[14]
In attempts to record phenylsilicate NMR spectra with
excess AnEryt as an internal standard, an interesting obser-
vation was made. In the 29Si NMR spectrum (Figure 13c), a
single resonance at an intermediate value is found (d=
À87.1 ppm from a 3:1:1 solution of AnEryt, PhSi(OMe)3,
and KOMe/18-crown-6; cf. d=À87.1 and À88.6 ppm for the
stronger signals in Figure 13b, and d=À89.3 ppm for the
weaksignal) instead of a triad of signals. Accordingly, the
13C NMR spectrum shows only four signals for AnEryt: two
for the free diol and two downfield-shifted signals for Si-
bonded AnEryt (Figure 13c). The shift differences in these
methanol solutions are slightly smaller than in aqueous solu-
tion: 1.6 ppm downfield for Si-bonded C2/C3 and 3.2 ppm
Figure 14. ORTEP diagram (40% probability ellipsoids) of the anions in
crystals of K[SiPh(AnErytHÀ2)2]·1/2MeOH. Distances [Š] in the anti/anti
À
À
À
À
isomer (top): Si O21 1.719(2), Si O31 1.749(2), Si O22 1.733(2), Si O32
À
1.725(2), Si C 1.887(3); diol torsion: O21-C21-C31-O31: À16.4(3)8; O22-
À
C22-C32-O32: 6.2(3)8. Distances [Š] in the syn/anti isomer (bottom): Si
À
À
À
À
O24 1.711(2), Si O34 1.752(2), Si O25 1.710(2), Si O35 1.766(2), Si C
1.899(3); diol torsion: O24-C24-C34-O34: À3.1(3)8; O25-C25-C35-O35:
À22.2(3)8. Percentage distance from the square-planar coordination
along the Berry pseudorotation coordinate: anti/anti isomer: 23.0; syn/
anti isomer: 64.5.
spectrum are assigned to the anti/anti and the syn/anti iso-
mers. In light of this assumption, interpretation of the
13C NMR spectrum is straightforward. Neglecting sp/tbp iso-
merism, both the anti/anti and the syn/syn isomer have ap-
parent C2v symmetry and should thus give rise to one signal
for C2/C3 and one signal for C1/C4. The syn/anti isomer on
the other hand, with its apparent Cs symmetry, will show
twice the number of signals, a phenomenon that has been
pointed out by Kinrade for the analogous hydroxosilicate.[13]
The 13C DEPT-135 NMR spectrum is shown in Figure 13b.
Both the C1/C4 region (consider the [18]crown-6 signal as a
reference for DEPT assignment) and the C2/C3 region at
higher field show the expected pattern of one strong signal,
two signals of medium intensity, and one weaksignal for the
anti/anti, the syn/anti, and the syn/syn isomer, respectively.
A distinct feature that is observed in the spectra of glyco-
sides with the AnEryt partial structure as well may be
noted, namely that the anti/anti and one of the syn/anti sig-
Chem. Eur. J. 2005, 11, 6326 – 6346
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6335

P. Klüfers et al.
for C1/C4. Upon varying the composition of the solutions, a
simple spectrum is always obtained if free AnEryt is detect-
ed in the 13C NMR spectrum, whereas split signals are only
present if no signals of free AnEryt are visible. We con-
clude, then, that rapid equilibration of the three silicate iso-
mers occurs in the presence of excess AnEryt. Catalytic
amounts of free AnEryt appear to be sufficient since the
AnEryt signals themselves are not affected by equilibration;
that is, there are not two signals at a medium position, as
often observed with diolato-metal complexes of high kinetic
lability.
There is a hint that catalysis/inhibition of syn/anti isomeri-
zation is not only meaningful on the NMR timescale but
also drives crystallization. Instead of the solid-state 1:1 mix-
ture of the isomers in 9, the pure anti/anti form is obtained
when small amounts of free AnEryt are present in the oth-
erwise unchanged crystallization batches of 9. The anti/anti
isomer thus appears to be the thermodynamically most
stable one for the specific counterion/silicate combination of
9. The structure is not presented here, however, because the
quality of the crystals is still low.
thoxysilane and adenosine in methanol, and trimethoxyphe-
nylsilane/methanol or aqueous silicate and cytidine. With
the Si(OMe)4/OMeÀ/MeOH system, a step is inserted be-
tween the inert phenylsilicon moiety and the reactive hy-
droxysilicon function. The synthetic rationale behind the use
of methoxysilicates is commented upon in the ribose section
below.
Adenosine: Adenosine (Ado) provides the most complete
set of spectra, with little overlap of signals, for the study of
syn/anti isomerism of pentacoordinate bis(diolato)silicon
centers. Figure 15 shows the respective spectra, which can
be interpreted completely, without leaving signals unas-
signed, by assuming two major components and one minor
species. We assumed by analogy that the minor species is
the syn/syn isomer, since no stabilization of the syn/syn
isomer and, at the same time, destabilization of the anti/anti
species is obvious with adenosine. In light of this assump-
tion, an assignment is straightforward. Table 5 shows the po-
sitions of the signals of those carbons that are closest to the
diol group. The 1 ppm/4 ppm AnEryt pattern of shift differ-
ences for the diol and adjacent carbons obviously cannot be
confirmed entirely. The largest shifts are observed for two of
the four signals of the anomeric carbon, that is, at one of the
two carbons adjacent to the diol group. The signal of the
second adjacent carbon, C4, however, is shifted to a lesser
extent. Moreover, a large C1’ shift difference is observed for
only two of the four C1’ signals, thus showing nicely that the
connectivity contributes to the shift difference but the par-
ticular structure in question does so as well. Although the
shift differences are thus of less diagnostic use, it should be
noted that a signal shows a particularly distinct isomeric
splitting if the carbon atom is close to the silicon center.
Thus, the signals of all the diol carbons are well resolved
into four components.
It should be noted that rapid syn/anti equilibration is not
observed on the NMR timescale with the pentacoordinate
hydroxosilicates, regardless of whether free diol is detected
in the spectra.
Extending the anhydroerythritol core—pentacoordinate sili-
cates with b-d-ribofuranosides: One could still hope at this
point to recognize the silicon-binding site(s) of a carbohy-
drate by shift differences in the 13C NMR spectrum. A
knowledge of siliconꢀs coordination number appears to be a
prerequisite, which is acceptable since 29Si NMR spectrosco-
py will provide this number. That the signals of the carbons
adjacent to the diol function are shifted to a larger extent
than the diol carbons may be accepted as well, as long as
this is constant behavior at least for AnEryt derivatives. The
fact that there is a marked shift at all is a positive aspect
that should not be underestimated.
Attempts to prepare aqueous silicate solutions at the
same molar ratio failed. No clear solutions were obtained;
instead, precipitates formed.
b-d-Ribofuranosides such as methyl-b-d-ribofuranoside
(Me-b-d-Ribf) or the nucleosides were then used as probes.
Their structures can be derived from that of AnEryt by
adding the additional hydroxymethyl and methoxy or nucle-
obase residue to the oxolane face opposite the diol group so
that the diol group does not experience steric hindrance
from the additional functional groups. Similar ligand proper-
ties should then be expected for AnEryt and the b-d-ribo-
furanosides. However, some experimental peculiarities of
the nucleosides have to be considered: first, they often form
precipitates upon preparation of the required 2:1:1 molar
ratio of diol, silicon starting material, and base. Second,
there is a marked tendency to form hexacoordinate silicates.
Both increased solubility in aqueous solution and hexacoor-
dination are effected by large excesses of base, as has been
shown by Kinrade for adenosine and guanosine.[13] The ef-
fects of pure pentacoordination are presented for two cou-
ples of reaction mixtures that yielded clear solutions with
the 2:1:1 molar ratio: trimethoxyphenylsilane or tetrame-
Cytidine: The reaction of twice the molar amount of cytidine
(Cyd) with trimethoxy(phenyl)silane and methoxide in
methanol yields exclusively pentacoordinate silicon accord-
ing to the 29Si NMR spectra, which have a pattern similar to
the analogous AnEryt and Ado systems. The minor compo-
nent of the three is hardly visible due to more-pronounced
signal overlap. The signal pattern of the two major species,
however, is in agreement with the assumption of a syn/anti-
anti/anti mixture without rapid exchange. With the exception
of signal overlap in the C1’/C5 region at d=94–95 ppm, all
the riboside signals can be unambiguously assigned to their
respective carbon atoms. The isomeric splitting is shown in
Figure 15c. However, a problem occurs regarding the molar
ratio of the isomers. Since the syn/anti:anti/anti ratio seems
to approach 2:1, all strong signals are of approximately
equal intensity. Attempts were made to shift the isomeric
ratio by altering the solution composition. Thus, replace-
ment of methanol as the solvent by a 1:1 mixture of meth-
6336
www.chemeurj.org
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6326 – 6346

Silicon Chelation in Aqueous and Nonaqueous Media
FULL PAPER
Figure 15. syn/anti Isomerism governing the 29Si (left) and 13C NMR spectra of the ribose region in methanolic solutions containing pentacoordinate bis-
(nucleosidato)silicates. Molar ratio of nucleoside:[SiR(OMe)3]:methoxide=2:1:1; counterion: [K([18]crown-6)]+ (Ado) or K+ (Cyd); total Si concentra-
tion: 0.20 and 0.36 molkgÀ1 for Ado and Cyd, respectively. a) [SiPh(Ado2’,3’HÀ2)2]À; b) [Si(OMe)(Ado2’,3’HÀ2)2]À; c) [SiPh(Cyd2’,3’HÀ2)2]À; the C1’ sig-
nals of Cyd are not shown due to overlap with the C5 signals. Color code: red: anti/anti; blue: syn/anti; green syn/syn.
anol and N,N-dimethylformamide yielded 13C NMR spectra
which show the isomer of third-highest abundance enriched
in terms of the most-upfield phenyl signals. At the same
time, one of the three main signals of each carbon atom
shows increased intensity. Hence, the enriched isomer is as-
signed as anti/anti (Figure 16c). It should be noted that this
assignment resembles the AnEryt pattern in the phenyl
region. A detailed table of shift differences of individual iso-
mers would be less complete than for Ado. To show the gen-
eral trend that the largest shift differences are found for the
carbons adjacent to the diol group, the mean values of the
three strong signals are used. The individual shift differences
in this case are 3.4 for C1’, 2.3 for C2’, 2.1 for C3’, 2.8 for
C4’, and 1.4 ppm for C5’.
Attempts to prepare tetramethoxysilane/methanol solu-
tions as with Ado failed due to the formation of precipitates.
On the other hand, aqueous solutions with a 2:1:1 ratio of
nucleoside, silica, and hydroxide were possible with Cyd. In
comparison with AnEryt, the nucleoside appears to be less
suited to forming pentacoordinate silicon than the parent
diol in terms of free diol and oxosilicate left. The 13C NMR
signals show the usual downfield shift but to a somewhat
lesser extent than with the phenylsilicate. The individual
shift differences are 3.0 for C1’, 1.7 for C2’, 1.6 for C3’, 2.6
for C4’, and 0.8 ppm for C5’.
As regards 13C NMR shift differences, the N-glycosides
Ado and Cyd have watered down the rule emerging from
AnEryt but have not completely invalidated it in showing
the tendency that the highest values are found for the car-
bons adjacent to the diol function.
Methyl-b-d-ribofuranoside: Bearing in mind the aim of de-
termining rules for the analysis of ribofuranose–silicate
signal patterns, methyl-b-d-ribofuranoside (Me-b-d-Ribf)
should add particularly valuable data, since as an O-glyco-
side it more closely resembles the sugar than a nucleoside
does. Reaction of a double molar amount of Me-b-d-Ribf
with trimethoxy(phenyl)silane/methoxide in methanol yield-
ed the three expected pentacoordinate silicate species
(Figure 16). The 13C NMR signals are not as well resolved as
in the case of Ado, but the typical syn/anti split can be ob-
served for the diol carbons C2’ and C3’ (Figure 16). The ten-
dency for the signals of the carbon atoms adjacent to the
diol carbons to shift downfield to the largest extent is the
Chem. Eur. J. 2005, 11, 6326 – 6346
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6337

P. Klüfers et al.
Table 5. Chemical shifts [ppm] in the 13C NMR spectra of solutions of
[K([18]crown-6)][SiR(Ado2’,3’HÀ2)2], and shift differences (d-
(silicate)Àd(free diol)). Bold: Dd of those carbon atoms that bear the sili-
con-binding oxygens. For atomic numbering of Ado see Scheme 1. n.d.=
not determined due to signal overlap.
R=Ph
R=OMe
d Dd
d
Dd
anti/anti
syn/anti 1
syn/anti 2
syn/syn
93.6
4.0
4.1
2.3
2.3
93.0
93.3
3.4
3.7
93.7
91.9
91.9
C1’
C2’
C3’
C4’
C5’
C8
92.3
2.7
anti/anti
syn/anti 1
syn/anti 2
syn/syn
76.4
77.0
76.3
77.1
2.3
2.9
2.2
3.0
76.4
76.6
76.3
76.8
2.4
2.6
2.3
2.8
anti/anti
syn/anti 1
syn/anti 2
syn/syn
72.5
72.3
72.2
72.3
1.4
1.2
1.1
1.2
72.8
72.6
72.2
72.3
1.7
1.5
1.1
1.2
anti/anti
syn/anti 1
syn/anti 2
syn/syn
88.1
88.0
1.5
1.4
88.0
87.9
1.4
1.3
87.9
1.3
anti/anti
syn/anti 1
syn/anti 2
syn/syn
62.8
0.9
62.7
62.6
0.8
0.7
62.6
62.5
0.7
0.6
Figure 16. 29Si and 13C NMR spectra (C2 and C3 signals) of 2:1:1 solu-
tions of Me-b-d-Ribf, SiR(OMe)3, and base. Red: anti/anti; blue: syn/
anti; green: syn/syn isomer of the respective [Si(R)(Me-b-d-
Ribf2,3HÀ2)2]À ion. a) R=Ph; base: KOMe; solvent: methanol. b) R=
OMe; base: KOMe; solvent: methanol. c) R=OMe; base: CsOH; sol-
vent: water. Total Si concentration: 0.72, 0.47, and 0.75 molkgÀ1 for a–c,
respectively. The ratio of oxosilicate and pentacoordinate silicate is ap-
proximately 3:1.
anti/anti
syn/anti 1
syn/anti 2
syn/syn
140.8
140.8
n.d.
n.d.
140.7
140.7
140.4
140.4
n.d.
n.d.
n.d.
n.d.
140.1
n.d.
same as with AnEryt, but the numbers are again different.
Using mean values including all three isomers, the shift dif-
ferences in question are those given in Table 6. Again, the
rule is weakened: the C1 signal is no longer the most shifted
as the C4 signal shows the largest shift difference.
As an overall result it can be stated that a moderate
downfield shift of about 2 ppm of the diol carbons is observ-
able on binding to pentacoordinate silicon. At the same
time, a more pronounced downfield shift of one of the adja-
cent carbons of up to about 5 ppm may occur.
Table 6. Shift differences (d(silicate)Àd(free diol)) in the 13C NMR spec-
tra of solutions of A[SiR(Me-b-d-Ribf2,3HÀ2)2] prepared with a 2:1:1
molar ratio of the riboside, the silicon starting material, and base. Bold:
Dd of those carbon atoms that bear the silicon-binding oxygens. For
atomic numbering of Me-b-d-Ribf see Scheme 1.
R
solvent
A
C1
C2
C3
C4
C5
Ph
OMe
OH
MeOH
MeOH
H2O
K
K
Cs
2.4
1.9
2.3
2.4
2.0
2.3
1.9
1.5
1.8
5.2
4.5
4.1
0.2
0.0
0.5
More-pronounced linking of equilibria—ribose in silicate
solutions: Up to this point, there has been no doubt about
the configuration of the diol in question. A known total con-
centration of the ligand thus enters the equilibria that de-
scribe complex formation and dissociation. This point be-
comes different when changing from simple diols and glyco-
sides to glycoses. The free ligand is now withdrawn to a
large extent from the complex equilibrium by the linked
pyranose/furanose and anomeric equilibria of the carbohy-
drate itself. Compared to AnEryt, the situation is worse for
a pure NMR treatment if a glycose is employed as a silicon
ligand in aqueous solution.
employed into penta- and hexacoordinate diolatosilicates is
possible.[13,14] The experimental conditions for achieving this
result include a large excess of aldose and high total concen-
trations (it should be noted that the 3.3m sugar concentra-
tion of reference [14] resembles a syrup, with more than a
kilogram of sugar per liter in the case of a disaccharide). In
terms of 13C NMR spectra, a proper analysis becomes more
difficult due to large amounts of free sugar beside small
amounts of Si-bonded glycose. To demonstrate the specific
problems, let us focus on the less basic solutions that have
been used by Lambert to detect ribose ligands in pentacoor-
dinate bis(diolato)hydroxosilicate species.[14] From the view-
point of glycose-bonded silicate, these solutions belong to
the regime of pentacoordination, and the shifting between
Starting with 29Si NMR spectroscopy, it has been demon-
strated that the entire transformation of all the oxosilicate
6338
www.chemeurj.org
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6326 – 6346

Silicon Chelation in Aqueous and Nonaqueous Media
FULL PAPER
penta- and hexacoordinate species is effected by the molar
amount of base. A maximum of pentacoordinate silicate, to-
gether with small amounts of oxosilicate, is found close to a
molar ratio of 1:1 for silicon and base. The residual oxosili-
cate vanishes on removing some 20% of base. Hexacoordi-
nate silicates become enriched on drastically increasing the
amount of base towards a 1:3 Si:OHÀ ratio.[13] The 13C NMR
spectra of solutions of both kinds are different and both are
rather complicated.
Prior to looking for Si-bonded ribose in the respective
spectra, the coordination ability of ribose towards silicon
centers is assessed in order to have a rough clue of what
should be searched for in addition to the established a-
Ribf1,2 silicon chelator.[7]
Inspection of Scheme 2 suggests the consideration of tri-
dentate triolate binding as well, since ribose is particularly
rich in isomers that provide O3 patterns for a facial binding
site. More generally, the question arises whether a pyranoi-
dic cis,cis-1,2,3-triol moiety may act as a silicon chelator,
i.e., whether the unsuitable, but possibly borderline, Si-bind-
ing properties of a pyranoidic cis-diol are over-compensated
by providing a third binding site. Even more generally, is a
pyranoidic 1,3-diol a silicon chelator? Since tridentate sili-
con binding appears to be a particularly unexplored area, it
will be assessed very roughly for the current purpose. If pyr-
1
anoidic triolate binding is of considerable significance, C4-
Me-b-d-Ribp should be able to enrich a silicate solution
with six-coordinate species under the usual aqueous alkaline
conditions. Experiment shows that Me-b-d-Ribp does not.
Thus, pyranoidic triolate binding will not be considered in
this discussion. It should, however, be kept in mind that this
point needs further clarification, particularly because Me-b-
d-Ribp does not include the anomeric hydroxy group in its
O3 pattern and the required 1C4 conformation is not pre-
dominant in solution. In the same, very preliminary, way,
the significance of tridentate a-d-Ribf1,2,3 binding is rough-
ly assessed. If these ligands play a dominant role, hexacoor-
dinate [Si(a-d-Ribf1,2,3HÀ3)2]2À species will be formed from
stoichiometric solutions under nonhydrolytic conditions.
However, solutions in methanol of an n:1:2 molar ratio of
Rib, Si(OMe)4, and OMeÀ show hexacoordinate silicate only
for n=3 but not for n=2. We can therefore conclude that
the hexacoordinate silicates are mainly tris(diolato) species
instead of bis(triolato)silicates, that is, tridentate chelation
by ribose does not appear to be a predominant binding
mode. (The rationale for using the tetramethoxysilane/meth-
anol system is explained in the next section.)
The tentative silicon binding sites of ribose: Scheme 2 sum-
marizes the furanose and pyranose forms that have to be
When restricting the discussion of silicon binding to fura-
nose diol groups, the main contribution should stem from
1,2-deprotonated a-ribofuranose. 1,2-Deprotonated b-ribo-
furanose, which contains a trans-furanoidic diol like anhy-
dro-threitol, is not suited as a chelator. The two remaining
silicon-chelating binding sites, a-Ribf2,3 and b-Ribf2,3, are
derived from diol functions of lower acidity. The b-furanose,
however, is more abundant in aqueous equilibria than the a-
furanose, which may compensate the lower acidity of its 2,3
site to some extent.
Scheme 2. O-atom patterns of the various ribose isomers (H atoms omit-
ted); fac-tridentate patterns are highlighted in red.
considered. Since there is no experimental evidence yet that
open-chain ribose may play a role, this form will not be con-
sidered further. Restricting our discussion to diolate binding
first, the experimentally substantiated a-Ribf1,2 form ap-
pears the most suitable in terms of acidity and stereochemis-
try. For the stereochemical reasons discussed above, pyra-
nose diol functions seem to be less suitable for silicon chela-
tion. a-Ribf2,3 and b-Ribf2,3 appear suitable in terms of
stereochemistry but the most acidic anomeric hydroxy group
would be left protonated on 2,3-complexation. It should be
noted at this point, however, that increasing the acidity of
the ligand is by no means a guarantee of obtaining a chelat-
ed silicate of hydrolytic stability. Silicon complexes of the
hydroxycarboxylic acids, for example, are hydrolytically sen-
sitive, as has been demonstrated recently by Tacke et al. for
hexacoordinate silicates like citratosilicate or oxalatosili-
cate.[20] It may thus be important that the ligand of choice
bears a sufficient amount of nondelocalized negative charge
on its oxygen atoms to compete with the oxo ligands in
aqueous silicate solutions.
Steps from a well-established phenylsilicate of ribose towards
aqueous silicate solutions: The use of AnEryt as a ligand has
demonstrated that there is a close resemblance between sol-
utions of the phenylsilicates in methanol and those in water.
To derive a sensible signal assignment in the more compli-
cated ribose case, one more step will be taken to arrive at
the aqueous medium, as in the case of Me-b-d-Ribf above.
Figure 17a shows the spectrum of ribose in methanol, which
is close to the aqueous one (Figure 17f). On adding trime-
thoxyphenylsilane and methoxide according to a final 2:1:1
molar ratio of Rib, SiPh(OMe)3, and OMeÀ, the a-furanose
binds to silicon and becomes the prominent isomer (Fig-
ure 17b).[7] Due to its bonding in the major species, positive
Chem. Eur. J. 2005, 11, 6326 – 6346
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6339

P. Klüfers et al.
signal assignment is possible with certainty using 2D NMR
techniques. In terms of shift differences, the result is disas-
trous in light of the rules derived up to now. The resonance
positions of the 13C NMR signals of a-Ribf in methanol are
shifted downfield by 0.9 (C1), 0.5 (C2), 0.9 (C3), À4.6 (C4),
and À0.9 ppm (C5) on formation of the [SiPh(a-d-
Ribf1,2HÀ2)2]À complex ion. The largest shift is that of C4—
the carbon with the largest distance from the reaction center
within the furanose ring—of more than 4 ppm upfield. Al-
though silicon binding through the 1,2-chelate was con-
firmed by X-ray analysis,[7] this binding site clearly cannot
be recognized from AnEryt-derived shift differences, either
from those in reference [14] or from corrected values that
are unaffected by the wrong DEPT assignment.
To take a step towards the aqueous silicate solution,
methanol is kept as the solvent for the moment but trime-
thoxyphenylsilane is replaced by tetramethoxysilane. The
spectrum with a 2:1:1 ratio of Rib, Si(OMe)4, and OMeÀ, is
shown in Figure 17c. Again, there is a main species with the
same stoichiometry as employed, just as with the phenyl de-
rivative. Signal assignment by two-dimensional techniques is
possible as well. Not surprisingly, the order of signals re-
mains the same due to the obvious similarity of the spectra.
Keeping to the discussion of the main species, the difference
between the phenylsilicate and the tentative methoxosilicate
is obvious: each of the five main signals is accompanied,
within a 2 ppm range, by weaker signals which adopt a pat-
tern that recalls the spectral features of the syn/anti isomers
of AnEryt and the b-d-furanosides. It should be noted that
the [SiPh(a-d-Ribf1,2HÀ2)2]À ion is anti/anti in the solid
state.[7] Hence, stability relationships such as those for
AnEryt, where anti/anti and syn/anti isomers are the major
species and the syn/syn isomer is the minor one, do not
appear to be divergent. The spectral properties of the
isomer mixture were evaluated as above: one signal for each
unique carbon of the ligand for anti/anti and syn/syn, two
signals of about equal intensity for the syn/anti isomer, and
all the signals of one carbon in a 2 ppm range. Bearing in
mind these characteristics, a well-resolved signal such as
that for C1 seems to support the idea that syn/anti isomer-
ism may occur with ribose as well.
The next step appears to be the most critical. The addition
of water to these solutions causes partial hydrolysis and oxo-
silicate formation. However, the methanol solutions do not
appear to be moisture-sensitive in a narrower sense. The ad-
dition of a few mols of water per mol of silicon does not
alter the spectra significantly, an observation that is in line
with Kinradeꢀs and Harrisꢀs reports on Si-OH/Si-OMe ex-
change in water/methanol mixtures.[21] Complete replace-
ment of methanol by water, however, drives oxosilicate for-
mation close to completeness in dilute solutions with a 2:1:1
stoichiometry. At the same time, the equivalent amount of
ribose, which occurs as a mixture of its four isomers in the
alkaline solution, is set free. Increasing the total concentra-
tion allows increasing amounts of pentacoordinate silicate to
survive. At the same time, there are no longer equilibrium
conditions. For example, 2:1:1 solutions containing
Figure 17. 13C NMR spectra of the ribose region of various aqueous and
methanolic solutions. Color codes: red: a-Ribf; blue: b-Ribf; violet: equi-
librating a-/b-Ribf; green: b-Ribp; yellowish green: a-Ribp. a) Rib in
MeOH, c=0.2m. b) Reaction mixtures containing a 2:1:1 molar ratio of
Rib, SiPh(OMe)3, and KOMe/[18]crown-6 in MeOH, with the [PhSi(a-d-
Ribf1,2HÀ2)2]À ion as the main species (red). Total Si concentration:
0.21 molkgÀ1. Dashed line: [18]crown-6. c) Reaction mixtures containing
a 2:1:1 molar ratio of Rib, Si(OMe)4, and KOMe/[18]crown-6 in MeOH
and containing isomeric forms of the [Si(OMe)(a-d-Ribf1,2HÀ2)2]À ion as
the main species (red). Total Si concentration: 0.48 molkgÀ1. Dashed
line: [18]crown-6. d) Aqueous solution of Rib, Si(OMe)4, and NaOH at a
2:1:1 molar ratio. Total Si concentration: 0.46 molkgÀ1; pH 11.8. e) Rib in
NaOH, molar ratio: 1:0.12, total ribose concentration: 0.46 molkgÀ1
;
pH 11.5. f) Same as (e), but pH adjusted to 6.5 after about 18 h at
pH 11.5 at room temperature. Note the weak, unassigned signals around
the position of b-Ribf-C2 that indicate some degree of decomposition.
6340
www.chemeurj.org
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6326 – 6346

Silicon Chelation in Aqueous and Nonaqueous Media
FULL PAPER
0.46 molkgÀ1 Si remain clear for a day but colorless precipi-
tates appear after that time. More-concentrated batches of
the 2:1:1 stoichiometry cannot be prepared as clear solu-
tions. A 2:1:1 batch of 0.67 molkgÀ1 Si, for example, shows
the formation of precipitates during preparation. Lowering
the amount of base causes gelation, and increasing the
amount of ribose retards gelation. Thus, a solution which re-
sembles the conditions of reference [14] (ca. 3:1:0.8 at a
total Si concentration of 1.1m), gels within half an hour in
our laboratory. To come close to this reported Si concentra-
tion of 1.1m, but at the same time to avoid gelation, the
spectrum of the 2:1:1 solution with 0.46 molkgÀ1 Si was
chosen for Figure 17d. Comparison with reference [14]
shows the same features for both spectra. To separate the
signals of the minor amount of ribose that is still bonded to
silicon, the signals of the free sugar have to be recognized as
such. Figure 17e shows a ribose spectrum taken at pH 11.5.
A comparison with Figure 17d reveals a few signals that are
unique to the silicate spectrum, therefore assignment of
these signals to the Si-bonded a-Ribf1,2 ligand appears sen-
sible. Comparison of Figures 17c and d shows that the
C2,3,5 signals of Si-chelating a-Ribf are prone to overlap
with signals of free ribose. Si-bonded a-Ribf-C1 and C4,
however, are expected to resonate slightly upfield from their
positions in methanol, in regions free of signals of ribose
itself. In fact, broad signals are visible at these positions
which indicate the presence of the pentacoordinate
[Si(OH)(a-Ribf1,2HÀ2)2]À species in aqueous solution.
The result of the ribose investigation may be summarized
in terms of available methods. Since there is obviously no
usable system for shift differences on the basis of the pres-
ently available data, the silicon-binding sites of the carbohy-
drates cannot be determined in solutions that do not allow
for unambiguous signal assignment by, preferably, two-di-
mensional methods. The main problems of an NMR meth-
odology in aqueous glycose chemistry are thus the weakand
broad 13C NMR signals, owing to small amounts of com-
plexed glycose species, together with the spreading of their
signal intensity over several isomers, a problem that is
hidden by presenting 29Si NMR spectra that benefit from
using glycose in excess.
cate and the methoxosilicate solutions a signal group at
about d=104 ppm occurs which can be recognized under
aqueous conditions only at higher resolution than that of
the figure. A second group occurs at about d=88 ppm in
methanol solution, but is lacking in water. In light of the
above considerations, a b-Ribf2,3 ligand appears to be a
candidate for a second Si-binding site. Data that can support
or exclude this suggestion have been presented above. In ad-
dition to AnEryt, Me-b-d-Ribf is a suitable ligand for silicon
because it enriches a silicate solution with pentacoordinate
silicate under the same experimental conditions as ribose.
Owing to the greater abundance of b-ribofuranose (13.2%
in aqueous equilibrium) over the a-isomer (7.4%), it does
not appear to be impossible to observe the b-Ribf2,3HÀ2
ligand as a minor species.[22] Moreover, the consideration of
b-Ribf2,3 binding clarifies another reason for signal splitting
apart from syn/anti isomerism and contributes to an under-
standing of the actual appearance of the spectra in Figures
17b and c. As b-Ribf2,3 is the minor ligand, the predomi-
nant solution species which contains this ligand should be
anti/anti-[SiPh(a-Ribf1,2HÀ2)(b-Ribf2,3HÀ2)]À in the phenyl-
silicate case. Signals of the a-Ribf1,2 ligand of this species
should thus emerge to the same extent as the b-Ribf2,3 sig-
nals, and the position of these a-Ribf1,2 signals should be
close to the signals of the same ligand in the main silicate
species. In fact, each signal of the main species rises above a
background which has the same appearance as the signals
stemming from the b-Ribf2,3 ligands. (One point has to be
clarified in future work: is the anti/anti isomer the only spe-
cies in the phenylsilicate solutions or is there rapid exchange
(the latter appears more probable)?).
A final comment on shift differences in the area of penta-
coordinate silicon seems appropriate. On considering b-
Ribf2,3 binding, the large downfield shift of about 5 ppm of
C4 appears to be most questionable. Although we have
demonstrated that shift-difference rules derived from the
AnEryt family clearly do not workfor Rib f1,2 bonding, they
appear to be usable for Ribf2,3-derived ligands in the sense
that a large shift of the C4 signal is not ruled out from the
beginning. The close structural resemblance of the species
to be compared should be noted: AnEryt, the ribosides in-
cluding the nucleosides, and 2,3-bonding ribofuranose com-
prise the same atomic pattern, whereas 1,2-bonding ribofur-
anose does not. With the data available now, shift-difference
rules may be used, in a very restricted sense, to verify
whether an unexpectedly large shift difference is unprece-
dented. It should be noted that, in a narrow sense, only
equal bonding patterns are compared, a fact that is well-
known and frequently used in carbohydrate chemistry.[23]
A strategy to overcome these problems is evident in
Figure 17. By the introduction of chemically related systems
such as the phenylsilicates/methoxosilicates in methanol, iso-
merism and hydrolysis, which are the two most critical pro-
cesses in aqueous solution, may be separated from one an-
other. When using a strategy starting at silicon centers of re-
duced functionality, it should always be kept in mind that
the use of phenylsilicon residues, for example, introduces a
(deliberate) bias towards pentacoordinate bis(diolato) spe-
cies; either pentacoordinate (diolato, triolato) or hexacoor-
dinate bis(triolato) species will be suppressed and have to
be considered separately.
Hexacoordinate silicates—preliminary structural results: We
have pointed out that the growing body of spectroscopic
and structural data regarding pentacoordinate silicon com-
plexes is not paralleled by such data on hexacoordinate sili-
cates. The following section provides some preliminary
structural data on furanoside complexes and contributes to
the above-mentioned question of 1,3-diolate bonding.
Minor species in methanol solutions: In the search for minor
products in the 13C NMR spectra of Figure 17, two groups of
resonances attracted our attention. Both in the phenylsili-
Chem. Eur. J. 2005, 11, 6326 – 6346
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6341

P. Klüfers et al.
Tris-bidentate 1,2-diolate binding—a crystalline cytidine com-
plex: There are no structural data for either pentacoordinate
or hexacoordinate nucleoside–silicon complexes. First X-ray
results were therefore obtained in attempts to crystallize a
pentacoordinate silicate of a nucleoside. With potassium or
cesium as the counterion, thin, weakly diffracting, hexagonal
platelets were obtained from solutions of a 2:1:1 molar ratio
of cytidine, silica, and cesium or potassium hydroxide. The
same crystalline silicate grew from solutions of different sto-
ichiometry, including the 3:1:2 ratio which resembles the for-
mula of a tentative hexacoordinate silicate. Although the
crystallization conditions were varied widely, the crystals
always proved to be the same and were always of very low
quality for X-ray diffraction. Although the structure of the
silicate ion in Cs2[Si(Cyd2’,3’HÀ2)3]·21.5H2O (10) was re-
solved (Figure 18), the structure solution is only preliminary.
plete a second six-membered chair). The finding that a pyra-
noidic 1,2,3-triol like 1C4-Me-b-d-Ribp, with its large tor-
sional angles, is not as efficient an Si-chelator as a furanoidic
diol should be contrasted by the significant enrichment of
aqueous alkaline silicate solutions by hexacoordinate species
on addition of the 1,3,5-triol cis-inositol (cis-Ins). With
cesium as the counterion, crystals of Cs2[Si(cis-InsHÀ3)]·cis-
Ins·8H2O (11) were isolated and structurally resolved.
Figure 19 shows the bis(triolato)silicate ion.
Figure 19. ORTEP diagram (40% probability ellipsoids) of the Ci-sym-
metrical anions in crystals of 11.
Conclusions
Anhydroerythritol: Anhydroerythritol has been introduced
primarily as a model compound. In the spirosilane part of
this work, however, it was demonstrated for the first time
that AnEryt deserves interest in its own right for stabilizing
unexpected bonding patterns of silicon. Pentacoordination
in a simple silicic acid diol ester appears to be an expression
of unsaturated Lewis acidity at the silicon center. Due to
the limited “bite” of a diol, the five-membered chelate rings
Figure 18. SCHAKAL diagram of the C3-symmetrical anions in crystals
À
of 10. Mean Si O distance: 1.79 Š. Diol torsion: À278.
Surprisingly, three Cyd2’,3’HÀ2 ligands bind to a hexacoordi-
nate silicon center, regardless of the solution composition.
The stereochemistry at the silicon atom is L. In accordance
with the larger apparent radius of the six-coordinate central
element, the chelating diol group deviates more strongly
À À
are strained and exhibit O Si O angles of about 108 less
than the tetrahedral angle. Correspondingly, the expanded
À À
tetrahedron edges (O Si O ca. 1158) are prone to attackby
a Lewis base. In the dimerization reaction that occurs
À
À À
from zero torsion. The Si O distances resemble those in the
during the crystallization of 62, the O atoms of the Si O C
hexacoordinate silicates of the open-chain sugar alcohols.[11]
linkages take on the role of the base. This chemistry is thus
unique to diols that lackfurther centers of basicity such as
additional hydroxy functions. Pentacoordinate silicic acid
dimers are thus not expected to occur as a general bonding
pattern in carbohydrate–silicon chemistry as they require
special properties of the diol component. We will support
this view of a special position for anhydroerythritol and cis-
cyclopentanediol in due course by means of DFT calcula-
tions.[24]
Bis-tridentate 1,3-diolate binding—cis-inositol complexes:
The reasons for the furanose-derived diol moieties being the
better ligands for small central atoms are restricted to the
1,2-diolate binding mode. Taking into account 1,3-binding,
the situation reverses. The bite of a furanoidic 1,3-diolato
ligand is relatively large, and thus this coordination mode
has been detected in aldose complexation only with a rela-
tively large central atom like palladium(ii).[5] Even simple
wire models show that the opposite holds for pyranoidic 1,3-
diols, which exhibit a small bite (in a trivial case, a carbon
atom matches the central atom site free of strain to com-
In the case of the penta- and hexacoordinate silicates,
AnEryt is not merely a carbohydrate model of reduced
functionality but occupies a special position as well: current-
ly, it is the best suited furanoidic ligand for silicon chelation
6342
www.chemeurj.org
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6326 – 6346

Silicon Chelation in Aqueous and Nonaqueous Media
FULL PAPER
in aqueous solution. It should be noted that, compared with
AnEryt, reducing sugars such as ribose are silicon chelators
of minor affinity,[13] so preparative use of glycose–silicon
chemistry may require nonaqueous solvents. To continue the
search for chelate silicates of hydrolytic stability, glycosides
with special structural features may be more promising than
the parent glycoses themselves. Thus the glycosesꢀ benefit of
being about an order of magnitude more acidic than AnEryt
or a glycoside appears overcompensated by the problem of
the existence of unsuited pyranose and trans-furanose iso-
mers in the solution equilibria.
mum pH for this species, would be the best choice. In an in-
dependent set-up, one could determine the hydrogenphos-
phate concentration at pH 7 as a significant number for as-
sessing its biochemical relevance. For a newly emerged sub-
stance class like the carbohydrate–silicon complexes, it thus
appears essential to strictly separate the question of their
formulae and structures from the question of their biochem-
ical relevance. It appears more appropriate to us, therefore,
to find conditions that show the species in question at a
maximum concentration, and not to move towards neutral
aqueous solution before having ascertained the structures. A
species that is detectable at pH 11.7 may be as meaningless
at pH 7 as a species investigated at pH 13. However, collect-
ing unambiguous structural information, even at the
“wrong” pH for biochemistry, may unravel bonding princi-
ples that are useful in the search for molecules that combine
them, thereby opening-up a less heuristic approach to new
silicon chelators. In a largely unexplored area such as carbo-
hydrate–(semi)metal chemistry, it is, of course, not straight-
forward to choose the optimum conditions for a species to
be investigated. We have used a simple rationale in this
work. If there is information on the stoichiometry of a spe-
cies, for example from solid-state work, solutions of the
same stoichiometry are used to investigate that species. We
recommend this procedure not as a dogma, but as the most
reasonable starting point for an investigation.
13C NMR spectroscopy: Aqueous solutions of glycoses and
silicate contain pentacoordinate bis(diolato)hydroxosilicate
ions. NMR spectroscopic characterization is complicated
mainly because of two effects. First, the signals are split and
broadened by isomerism. The syn/anti binding of two major
ligands is well documented but a second form should be
considered as well: the exchange of one of the two major
diol ligands for a minor ligand gives rise not only to signals
of the minor ligand, which may be well separated from the
major ligandꢀs shift but, in addition, slightly shifted signals
of the remaining major ligand occur. The result is that the
signals of the main complex appear to emerge out of a
hump in the background. A strategy to assign the signals of
at least the main component has been presented. It is based
on chemical variation by, for example, a reduction of the
functionality at the silicon center or a change of the solvent.
One conceivable method for signal assignment is, however,
clearly impossible and leads to wrong assignments:[14] the Si-
binding sites of a carbohydrate cannot be determined by
using typical and constant shift differences in 13C NMR spec-
tra unless the spectra are taken from closely related com-
pounds whose structures perfectly match the structures in
question.
Experimental Section
General: Reagent-grade chemicals were purchased from Fluka and used
as supplied. All syntheses were carried out under dry nitrogen using stan-
dard Schlenktechniques. Organic solvents were dried and purified ac-
cording to standard procedures and stored under nitrogen. 1H, 13C{1H},
and 29Si NMR spectra were recorded at room temperature on a Jeol EX
400 (1H: 400 MHz; 13C: 100 MHz; 29Si: 79.4 MHz) or an Eclipse 500
NMR spectrometer (1H: 500 MHz; 13C: 125 MHz). The spectra were ref-
erenced to external SiMe4. The 1H and 13C signals of pentacoordinate sili-
cate species in methanol were assigned by means of 1H/1H COSY and
1H/13C HMQC experiments. Shift differences are given as d-
(silicate)Àd(free diol). Both kinds of shifts were taken from the same
spectrum. In the case of insufficient amounts of starting material in nona-
queous solutions, d(free diol) was taken from reaction mixtures with a
slight excess of diol added.
Anhydroerythritol was prepared according to standard procedures.[26] X-
ray: C4H8O3, Mr =104.105 gmolÀ1, orthorhombic, P212121, a=5.4980(1),
b=8.1573(1), c=20.9765(3) Š, Z=8, Rw(F2)=0.0760, S=1.095.
l-Anhydrothreitol was prepared according to the literature as well.[27] X-
ray: C4H8O3, Mr =104.105 gmolÀ1, orthorhombic, P212121, a=5.3944(4),
b=10.3152(7), c=26.904(3) Š, Z=12, Rw(F2)=0.0712, S=0.844.
Towards physiological conditions: A major motive for inves-
tigating aqueous silicon–carbohydrate chemistry is to con-
tribute to biochemical questions regarding the role of silicon
in living organisms. A special topic of current interest deals
with silica biomineralization. The relevance of higher-coor-
dinate silicates of polyols and catechols in the biomineraliza-
tion process has been regarded as unlikely due to the unphy-
siologically high pH values that are needed for the prepara-
tion of such compounds.[25] This opinion is supported by this
workfor the case of simple diols and some common glyco-
ses. The penta- and hexacoordinate silicate species dealt
with here are clearly species of the alkaline regime. Does it
thus make sense to investigate the composition of, and bind-
ing sites in, five-coordinate glycose–silicate anions at a pH
value as low as possible, say 11.7?[14] The more general ques-
tions are: 1) What is the structure and what are the spectro-
scopic characteristics of a species? and 2) what is its signifi-
cance in a physiological environment? Mixing up these two
questions may lead to an experimental set-up that contrib-
utes convincingly to neither. Consider, say, HPO42À. To do
spectroscopy in aqueous solution, pH 9.5, which is the opti-
Ph2Si(cis-ChxdHÀ2) (1): cis-Cyclohexane-1,2-diol (0.09 g, 0.79 mmol) and
pyridine (0.17 mL, 2.0 mmol) were dissolved in trichloromethane
(13 mL). Dichlorodiphenylsilane (0.17 mL, 0.79 mmol) in trichlorome-
thane (13 mL) was then added dropwise and the suspension was stirred
for four days at room temperature. After this time the volume was re-
duced by three quarters and toluene (20 mL) was added. Precipitated
pyridinium chloride was filtered off and the solution volume was reduced
to 10 mL in vacuo. Colorless crystals of 1 formed in the course of two
days whilst keeping the solution at 48C (0.18 g, 0.30 mmol, 77% yield).
M.p. 1918C; elemental analysis calcd (%) for C18H20O2Si: C 72.93, H
6.80; found: C 72.58, H 6.89; 13C{1H} NMR (toluene, 100.5 MHz): d=
21.7 (C4, C5), 30.4 (C3, C6), 77.5 (C1, C2), 127.7 (phenyl-C3, phenyl-C5),
Chem. Eur. J. 2005, 11, 6326 – 6346
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6343

P. Klüfers et al.
130.2 (phenyl-C4), 133.6, 132.7 (phenyl-C1), 135.4 ppm (phenyl-C2,
phenyl-C6); 29Si NMR (toluene, 79.4 MHz): d=À34.4 ppm; MS (EI+,
70 eV): m/z: 592 [12], 516 [12ÀPh], 418 [12ÀPhÀC6H10O], 337, 296 (1),
132.7 (phenyl-C1), 134.9, 135.1 ppm (phenyl-C2, phenyl-C6); 29Si NMR
(toluene, 79.4 MHz): d=À3.7 ppm; 29Si CP/MAS NMR (99.4 MHz): d=
À34.7 ppm; MS (EI+, 70 eV): m/z: 564 [52], 282 [5], 204 [5ÀPh], 176
218 [1ÀPh]; X-ray: Mr =296.435 gmolÀ1, triclinic, P1, a=9.3456(1), b=
[5ÀPhÀC2H4]; X-ray: Mr =282.409 gmolÀ1, triclinic, P1, a=9.249(1), b=
¯
¯
9.6991(1), c=10.6862(2) Š, a=106.1110(8)8, b=95.1714(9)8, g=
118.4405(6)8, Z=2, Rw(F2)=0.1457, S=1.125.
9.537(1), c=10.048(2) Š, a=83.17(2)8, b=68.72(2)8, g=61.08(1)8, Z=2,
Rw(F2)=0.0793, S=0.933.
a-Si(AnErytHÀ2
) (a-6): Anhydroerythritol (2.97 g, 28.5 mmol) was dis-
2
Ph2Si{(R,R)-trans-ChxdHÀ2)}
(2):
(R,R)-trans-cyclohexane-1,2-diol
solved in toluene (40 mL) at 1008C and then tetrachlorosilane (1.63 mL,
14.2 mmol) in toluene (30 mL) was added dropwise within 15 min. A
transient colorless precipitate dissolved on completion of the silane addi-
tion. After heating for 2 h under reflux, the solution volume was reduced
to 10 mL. On keeping the solution at 48C, colorless platelets of a-6
formed (2.89 g, 12.4 mmol, 88% yield). M.p. 1818C; elemental analysis
calcd (%) for C8H12O6Si: C 41.37, H 5.21, Si 12.09; found: C 41.28, H
5.28, Si 11.90; 1H NMR (500.2 MHz, [D8]toluene): d=2.87 (m, 4H; H1,
H4), 3.79 (m, 4H; H1, H4), 4.06 ppm (m, 4H; H2, H3); 13C{1H} NMR
([D8]toluene, 128.5 MHz): d=74.9, 75.0 (C1, C4), 77.6, 77.7 ppm (C2,
C3); 29Si NMR ([D8]toluene, 79.4 MHz): d=À36.7 ppm; 29Si CP/MAS
NMR (99.4 MHz): d=À36.7 ppm; MS (EI+, 70 eV): m/z: 232 [6], 231
[6ÀH], 202 [6ÀCH2O], 201 [6ÀHÀCH2O], 190 [6ÀC2H2O], 189
[6ÀHÀC2H2O], 171 [6ÀHÀ2CH2O], 159 [6ÀHÀC2H2OÀCH2O]; X-ray:
Mr =232.263 gmolÀ1, orthorhombic, Pca21, a=10.3366(2), b=10.1473(2),
c=9.1532(1) Š, Z=4, Rw(F2)=0.0837, S=1.066.
(0.11 g, 0.96 mmol) and pyridine (0.20 mL, 2.3 mmol) were dissolved in
toluene (13 mL) and heated to 508C. Dichlorodiphenylsilane (0.20 mL,
0.96 mmol) in toluene (13 mL) was then added dropwise; pyridinium
chloride precipitated immediately. The suspension was stirred for 14 h at
508C. The solid was filtered off and the solution volume was reduced to
5 mL in vacuo. Colorless crystals of 2 formed in the course of a few days
whilst storing the solution at 48C (0.21 g, 0.35 mmol, 74% yield). M.p.
2558C; elemental analysis calcd (%) for C18H20O2Si: C 72.93, H 6.80;
found: C 72.53, H 6.86; 13C{1H} NMR (toluene, 100.5 MHz): d=24.2 (C4,
C5), 34.1 (C3, C6), 76.9 (C1, C2), 127.6 (phenyl-C3, phenyl-C5), 129.9
(phenyl-C4), 133.9, 132.7 (phenyl-C1), 135.7 ppm (phenyl-C2, phenyl-
C6); 29Si NMR (toluene, 79.4 MHz): d=À35.3 ppm; MS (EI+, 70 eV):
m/z: 593 [22], 516 [22ÀPh], 418 [22ÀPhÀC6H10O], 337, 319, 296 [2], 218
[2ÀPh]; X-ray: Mr =296.435 gmolÀ1, monoclinic, P21, a=10.9794(1), b=
10.1930(1), c=14.9508(2) Š, b=104.1066(5)8, Z=4, Rw(F2)=0.1128, S=
1.117.
b-Si(AnErytHÀ2)2 (b-6): Compound a-6 (4.17 g, 18.0 mmol) was sublimed
(1 mbar, 1208C, cold finger: 608C). Colorless crystals (3.86 g, 16.6 mmol,
93% yield). M.p.: 1818C; elemental analysis calcd (%) for C8H12O6Si: C
41.37, H 5.21, Si 12.09; found: C 41.12, H 5.47, Si 11.84; 29Si CP/MAS
NMR (99.4 MHz): d=À36.6 ppm; MS (EI+, 70 eV): m/z: 232 [6], 231
[6ÀH], 202 [6ÀCH2O], 201 [6ÀHÀCH2O], 190 [6ÀC2H2O], 189
Ph2Si(AnErytHÀ2) (3): Anhydroerythritol (0.30 g, 2.87 mmol) and pyri-
dine (0.54 mL, 6.3 mmol) were dissolved in trichloromethane (20 mL),
and dichlorodiphenylsilane (0.60 mL, 2.87 mmol) in trichloromethane
(20 mL) was then added dropwise. The suspension was stirred for 16 h at
508C. After this time the volume was reduced by three quarters and tolu-
ene (20 mL) was added. The pyridinium chloride precipitate was filtered
off and the solution volume was reduced to 10 mL in vacuo. Colorless
crystals of 3 formed in the course of a few days whilst keeping the solu-
tion at 48C (0.72 g, 2.54 mmol, 88% yield). M.p. 798C; elemental analysis
calcd (%) for C16H16O3Si: C 67.58, H 5.67; found C 67.72, H 5.74; 13C{1H}
NMR (toluene, 100.5 MHz): d=75.2 (C1, C4), 79.2 (C2, C3), 127.9, 128.0
(phenyl-C3, phenyl-C5), 130.8, 131.1 (phenyl-C4), 131.6, 132.4 (phenyl-
C1), 134.8, 135.7 ppm (phenyl-C2, phenyl-C6); 29Si NMR (toluene,
[6ÀHÀC2H2O],
172
[6À2CH2O],
171
[6ÀHÀ2CH2O],
159
[6ÀHÀC2H2OÀCH2O]; X-ray: Mr =232.263 gmolÀ1, orthorhombic, Pbca,
a=8.9325(1), b=9.1678(1), c=23.7776(3) Š, Z=8, Rw(F2)=0.0828, S=
1.058.
g-Si(AnErytHÀ2
) (g-6): Compound a-6 (1.05 g, 4.5 mmol) was dissolved
2
in boiling acetonitrile (2 mL). Colorless crystals formed in the course of
few days at 48C (0.14 g, 0.6 mmol, 8% yield). Crystallization succeeded
once but has not been reproduced yet! M.p. 1798C; elemental analysis
calcd (%) for C8H12O6Si: C 41.37, H 5.21, Si 12.09; found: C 40.97, H
5.47, Si 11.76; 29Si CP/MAS NMR (99.4 MHz): d=À37.9 ppm; MS (CI,
methane): m/z: 337 [6+AnEryt], 232 [6], 189 [6ÀHÀC2H2O], 105
[AnEryt+H], 87 [C4H7O2]; X-ray: Mr =232.263 gmolÀ1, orthorhombic,
Pbca, a=9.3070(1), b=11.0378(2), c=19.1601(3) Š, Z=8, Rw(F2)=
0.1144, S=1.101.
79.4 MHz): d=À1.4 ppm; MS (EI+, 70 eV): m/z: 284 [3], 206 [3ÀPh],
176 [3ÀPhÀCH2O]; X-ray: Mr =284.382 gmolÀ1
,
triclinic, P1, a=
¯
10.6338(2), b=10.8232(2), c=13.2007(3) Š, a=72.9934(9)8, b=
85.2784(8)8, g=83.764(1)8, Z=4, Rw(F2)=0.1717, S=0.998.
Ph2Si(l-AnThreHÀ2) (4): l-Anhydrothreitol (0.09 g, 0.92 mmol) and pyri-
dine (0.17 mL, 2.0 mmol) were dissolved in trichloromethane (15 mL).
Dichlorodiphenylsilane (0.10 mL, 0.47 mmol) in trichloromethane
(15 mL) was then added dropwise and the suspension was heated for
16 h under reflux. After this time the solution volume was reduced to
7 mL and toluene (20 mL) was added. The pyridinium chloride precipi-
tate was filtered off and the solution volume was reduced to 5 mL in
vacuo. Colorless platelets of 4 formed in the course of a few weeks whilst
keeping the solution at 48C (0.21 g, 0.17 mmol, 56% yield). Elemental
analysis calcd (%) for C16H16O3Si: C 67.58, H 5.67; found C 67.42, H
5.59; 13C{1H} NMR (toluene, 100.5 MHz): d=73.0 (C1, C4), 80.0 (C2,
C3), 128.0 (phenyl-C3, phenyl-C5), 130.7 (phenyl-C4), 132.0 (phenyl-C1),
136.0 ppm (phenyl-C2, phenyl-C6); 29Si NMR (toluene, 79.4 MHz): d=
À29.6 ppm; MS (EI+, 70 eV): m/z: 852 [43], 775 [43ÀPh], 690, 569 [42],
491 [42ÀPh], 319, 303, 283 [4], 259, 241, 199, 183, 163; X-ray: Mr =
Si2(AnErytHÀ2
) (62): Compound a-6 (1.75 g, 7.5 mmol) was dissolved in
4
boiling toluene (5 mL). Colorless crystals of a-6 formed in the course of
few days at 48C (0.79 g, 1.69 mmol, 45% yield). These crystals were fil-
tered off at 48C. Colorless platelets of 62 formed subsequently in the
course of a few weeks. M.p. 1408C; elemental analysis calcd (%) for
C8H12O6Si: C 41.37, H 5.21, Si 12.09; found: C 41.20, H 5.39, Si 11.97; 29Si
CP/MAS NMR (99.4 MHz): d=À94.3 ppm; MS (CI, methane): m/z: 464
[62], 337 [6+AnEryt], 232 [6], 105 [AnEryt+H]; X-ray: Mr =
232.263 gmolÀ1
6.9375(2) Š, b=105.5409(12)8, Z=4, Rw(F2)=0.0852, S=1.052.
b-Si2(cis-CptdHÀ2 (b-72): cis-Cyclopentane-1,2-diol (1.15 g, 11.3 mmol)
, monoclinic, P21/c, a=8.3619(2), b=16.8394(5), c=
)
4
and toluene (25 mL) were heated to 1008C. Tetrachlorosilane (0.593 mL,
5.6 mmol) in toluene (25 mL) was then added dropwise within 15 min. A
transient colorless precipitate dissolved on completion of the silane addi-
tion. After heating for 2 h under reflux, the solution volume was reduced
to 5 mL. On keeping the solution at 48C, colorless platelets of b-7
formed (1.12 g, 2.5 mmol, 88% yield). M.p. 1028C; elemental analysis
calcd (%) for C10H16O4Si: C 52.61, H 7.06, Si 12.30; found: C 52.51, H
7.16, Si 12.20; 1H NMR (500.2 MHz, [D8]toluene): d=1.23 (m, 1H; H5),
1.26 (m, 2H; H1, H4), 1.79 (m, 1H; H5), 4.32 ppm (m, 4H; H2, H3);
13C{1H} NMR ([D8]toluene, 128.5 MHz): d=20.2 (C5), 33.5, 33.6 (C1,
C4), 76.9, 77.1 ppm (C2, C3); 29Si NMR ([D8]toluene, 79.4 MHz): d=
À36.8 ppm; 29Si CP/MAS NMR (99.4 MHz): d=À94.6 ppm; MS (FAB+,
xenon, NBA, 6 keV): m/z: 457 [72], 228 [7], 200 [7ÀC2H4], 199 [7ÀC2H5],
171 [7ÀC2H4ÀC2H5]; X-ray: Mr =228.317 gmolÀ1, monoclinic, P21/c, a=
284.382 gmolÀ1
, monoclinic, P21, a=10.0849(1), b=23.5576(4), c=
10.2639(2) Š, b=114.1764(6)8, Z=6, Rw(F2)=0.1114, S=1.013.
Ph2Si(cis-CptdHÀ2) (5): cis-Cyclopentane-1,2-diol (0.28 g, 2.74 mmol) and
pyridine (0.51 mL, 6.03 mmol) were dissolved in toluene (20 mL) and
heated to 508C. Dichlorodiphenylsilane (0.57 mL, 2.74 mmol) in toluene
(20 mL) was then added dropwise; pyridinium chloride precipitated im-
mediately. The suspension was stirred for 14 h at 508C. The solid was
then filtered off and the solution volume was reduced to 10 mL in vacuo.
Colorless platelets of 5 formed in the course of few days whilst keeping
the solution at 48C (0.71 g, 1.25 mmol, 92% yield). M.p. 1328C; elemen-
tal analysis calcd (%) for C17H18O2Si: C 72.30, H 6.42; found: C 71.96, H
6.10; 13C{1H} NMR (toluene, 100.5 MHz): d=22.5 (C4), 35.2 (C3, C5),
80.1 (C1, C2), 128.0 (phenyl-C3, phenyl-C5), 131.0 (phenyl-C4), 131.6,
6344
www.chemeurj.org
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6326 – 6346

Silicon Chelation in Aqueous and Nonaqueous Media
FULL PAPER
10.5586(2), b=10.3715(2), c=9.9903(2) Š, b=105.4087(8)8, Z=4,
Rw(F2)=0.0837, S=1.056.
atoms. CCDC-262917 (AnEryt), CCDC-262918 (l-AnThre), CCDC-
262919 (1), CCDC-262920 (2), CCDC-262921 (3), CCDC-262922 (4),
CCDC-262923 (5), CCDC-196570 (a-6), CCDC-196571 (b-6), CCDC-
196572 (g-6), CCDC-196573 (62), CCDC-196568 (a-72), CCDC-196569
(b-72), CCDC-262924 (8a), CCDC-262926 (8b), CCDC-262927 (8c), and
CCDC-262928 (10) contain the supplementary crystallographic data for
this paper. These data can be obtained free of charge from the Cam-
bridge Crystallographic Data Center via www.ccdc.cam.ac.uk/data_
request/cif.
a-Si(cis-CptdHÀ2
) (a-72): Compound b-72 (2.05 g, 4.5 mmol) was sub-
2
limed (1 mbar, 608C, cold finger: 158C). Colorless crystals (1.82 g,
4.0 mmol, 89% yield). M.p. 1008C; elemental analysis calcd (%) for
C10H16O4Si: C 52.61, H 7.06, Si 12.30; found: C 52.31, H 7.09, Si 12.17;
29Si CP/MAS NMR (99.4 MHz): d=À94.7 ppm; MS (CI+, isobutane):
m/z: 457 [72], 228 [7]; X-ray: Mr =228.317 gmolÀ1, monoclinic, P21/c, a=
10.0605(4), b=11.2398(5), c=9.2031(4) Š, b=91.582(2)8, Z=4, Rw(F2)=
0.1063, S=1.099.
Computational chemistry: DFT calculations were performed by using the
GAUSSIAN 98 and 03 program packages.[28] The B3LYP/6–31G(d) level
of theory was used for geometrical refinement; NMR shifts were calcu-
lated at the PBE1PBE/6–311++G(2d,p)//B3LYP/6–31G(d) level.
Na[Si(OH)(AnErytHÀ2)2] (8a): Silica (0.060 g, 1.0 mmol) and anhydroery-
thritol (0.312 g, 3.0 mmol) were suspended in a 2m sodium hydroxide so-
lution (1.0 mL, 2 mmol). Ultrasonification (30 min, 258C) yielded a clear
solution. Colorless platelets formed on evaporation of the solvent at 48C
in the course of three months. 29Si NMR (water, 79.4 MHz): d=À97.6,
À98.4 ppm, and oxosilicates; 29Si CP/MAS NMR (99.4 MHz): d=
À97.6 ppm; X-ray: C8H13NaO7Si, Mr =272.260 gmolÀ1, triclinic, P1, a=
5.741(1), b=6.004(1), c=9.076(2) Š, a=78.03(2)8, b=89.70(2)8, g=
61.89(2)8, Z=1, Rw(F2)=0.0565, S=1.043.
Acknowledgments
The authors thankProf. K. Hegetschweiler (Universität des Saarlandes,
Saarbrücken) for providing cis-inositol, Prof. Dr. J. Senker (Universität
Bayreuth) for recording the solid-state NMR spectra, and Prof. Dr. J.
Evers (Ludwig-Maximilians-Universität, München) for collecting the X-
ray powder data. Part of this workwas supported by the focus pro-
gramme Principles of Biomineralization of the Deutsche Forschungsge-
meinschaft.
Rb[Si(OH)(AnErytHÀ2)2] (8b): Anhydroerythritol (1.04 g, 10.0 mmol)
and tetramethoxysilane (0.61 g, 4.0 mmol) were dissolved in water
(10 mL). After the addition of rubidium hydroxide dihydrate (1.05 g,
7.6 mmol) the solution was boiled to evaporate the methanol. Slow evap-
oration yielded colorless crystals in the course of three weeks. X-ray:
C8H13O7RbSi, Mr =334.74 gmolÀ1, orthorhombic, Cmc21, a=18.701(15),
b=6.193(3), c=9.829(4) Š, Z=4, Rw(F2)=0.0955, S=1.100.
Cs[Si(OH)(AnErytHÀ2)2] (8c): Anhydroerythritol (0.78 g, 7.5 mmol) and
tetramethoxysilane (0.38 g, 2.5 mmol) were dissolved in water (10 mL).
After the addition of cesium hydroxide monohydrate (0.88 g, 5.0 mmol)
the solution was boiled to evaporate the methanol. Slow evaporation
yielded colorless crystals in the course of three weeks. X-ray:
[1] G. E. Taylor, J. M. Waters, Tetrahedron Lett. 1981, 22, 1277–1278.
[2] J. Burger, C. Gack, P. Klüfers, Angew. Chem. 1995, 107, 2950–2951;
Angew. Chem. Int. Ed. Engl. 1995, 34, 2647–2649.
[3] A. Geißelmann, P. Klüfers, B. Pilawa, Angew. Chem. 1998, 110,
1181–1184; Angew. Chem. Int. Ed. 1998, 37, 1119–1121.
[4] a) R. van den Berg, J. A. Peters, H. van Bekkum, Carbohydr. Res.
1994, 253, 1–12; b) J.-F. Verchere, S. Chapelle, F. Xin, D. C. Crans,
Prog. Inorg. Chem. 1998, 47, 837–945.
C8H13CsO7Si, Mr =382.18 gmolÀ1
b=6.500(3), c=9.876(2) Š, Z=4, Rw(F2)=0.0874, S=1.103.
K[SiPh(AnErytHÀ2)2]·1/2MeOH (9): solution of anhydroerythritol
, orthorhombic, Cmc21, a=18.629(4),
A
(0.231 g, 2.22 mmol), potassium methoxide (0.078 g, 1.11 mmol), and tri-
methoxyphenylsilane (0.207 mL, 1.11 mmol) in methanol (7 mL) was stir-
red at 258C for 16 h. After concentration of the solution, colorless crys-
tals of 9 formed in the course of a few weeks at 48C. 13C{1H NMR (meth-
anol, 67.9 MHz): d=71.5, 71.9, 72.1 (C2, C3), 74.7, 74.8, 75.3 (C1, C4),
125.6, 126.0, 126.2, 126.5 (phenyl-C3, phenyl-C4, phenyl-C5), 133.7, 134.2,
134.7 (phenyl-C2, phenyl-C6), 143.4, 144.6, 146.2 ppm (phenyl-C1); 29Si
NMR (methanol, 79.4 MHz): d=À86.7, À88.1, À88.8 ppm. X-ray:
C14.5H19KO6.5Si, Mr =364.486 gmolÀ1, monoclinic, P21/n, a=15.9982(3),
b=10.9630(2), c=18.8191(4) Š, b=91.7758(8)8, Z=8, Rw(F2)=0.1163,
S=1.046.
[5] P. Klüfers, T. Kunte, Angew. Chem. 2001, 113, 4356–4358; Angew.
Chem. Int. Ed. 2001, 40, 4210–4212.
[6] P. Klüfers, T. Kunte, Chem. Eur. J. 2003, 9, 2013–2018.
[7] P. Klüfers, F. Kopp, M. Vogt, Chem. Eur. J. 2004, 10, 4538–4545.
[8] W. Burchard, N. Habermann, P. Klüfers, B. Seger, U. Wilhelm,
Angew. Chem. 1994, 106, 936–939; Angew. Chem. Int. Ed. Engl.
1994, 33, 884–887.
[9] K. Benner, P. Klüfers, J. Schuhmacher, Angew. Chem. 1997, 109,
783–785; Angew. Chem. Int. Ed. Engl. 1997, 36, 743–745.
[10] a) S. D. Kinrade, J. W. Del Nin, A. S. Schach, T. A. Sloan, K. L.
Wilson, C. T. G. Knight, Science 1999, 285, 1542–1545; b) S. D. Kin-
rade, R. J. Hamilton, A. S. Schach, C. T. G. Knight, J. Chem. Soc.
Dalton Trans. 2001, 961–963; c) S. D. Kinrade, A. S. Schach, R. J.
Hamilton, C. T. G. Knight, Chem. Commun. 2001, 1564–1565.
[11] K. Benner, P. Klüfers, M. Vogt, Angew. Chem. 2003, 115, 1088–
1093; Angew. Chem. Int. Ed. 2003, 42, 1058–1062.
[12] K. Benner, P. Klüfers, J. Schuhmacher, Z. Anorg. Allg. Chem. 1999,
625, 541–543.
[13] S. D. Kinrade, E. W. Deguns, A.-M. E. Gillson, C. T. G. Knight,
Dalton Trans. 2003, 3713–3716. Addendum: S. D. Kinrade, R. J.
Balec, A. S. Schach, J. Wang, C. T. G. Knight, Dalton Trans. 2004,
3241–3243.
Cs2[Si(Cyd2’,3’HÀ2)3]·21.5H2O (10): Silica (0.060 g, 1.0 mmol) and cyti-
dine (0.73 g, 3.0 mmol) were dissolved in 2m cesium hydroxide (1.0 mL,
2.0 mmol) in an ultrasonic bath. After the addition of acetone (0.2 mL),
colorless hexagonal platelets formed within a weekat 4 8C. 29Si CP/MAS
NMR (79.4 MHz): d=À136.1 ppm. X-ray: C27H76Cs2N9O36.50Si, Mr =
1404.828 gmolÀ1, hexagonal, P6322, a=15.4313(3), c=29.6565(7) Š, Z=
4, R(F)=0.22.
Cs2[Si(cis-InsHÀ3)]·cis-Ins·8H2O (11): A solution of cis-inositol (0.090 g,
0.50 mmol) and tetramethoxysilane (0.037 mL, 0.25 mmol) in 1m cesium
hydroxide (0.50 mL, 0.50 mmol) was stirred at 258C for 16 h. After con-
centration of the solution, colorless crystals of 11 formed in the course of
a few weeks at 48C. X-ray: C18H46Cs2O26Si, Mr =972.439 gmolÀ1, triclinic,
[14] J. B. Lambert, G. Lu, S. R. Singer, V. M. Kolb, J. Am. Chem. Soc.
2004, 126, 9611–9625.
[15] O. Seiler, C. Burschka, M. Penka, R. Tacke, Z. Anorg. Allg. Chem.
2002, 628, 2427–2434.
¯
P1, a=10.957(5), b=12.268(5), c=12.501(5) Š, a=83.356(5)8, b=
78.126(5)8, g=78.011(5)8, Z=2, Rw(F2)=0.0969, S=1.058.
Crystal-structure determination and refinement: Crystals suitable for X-
ray crystallography were selected by means of a polarization microscope,
mounted on the tip of a glass fiber, and investigated either on a Nonius
KappaCCD diffractometer using graphite-monochromated MoKa radia-
tion (l=0.71073 Š) or a Stoe IPDS diffractometer using the same radia-
tion. The structures were solved by direct methods (SIR 97, SHELXS)
and refined by full-matrix least-squares calculations on F2 (SHELXL-97).
Anisotropic displacement parameters were refined for all non-hydrogen
[16] a) R. E. Ballard, A. H. Haines, E. K. Norris, A. G. Wells, Angew.
Chem. 1974, 86, 555–556; Angew. Chem. Int. Ed. Engl. 1974, 13,
541–542; b) R. E. Ballard, A. H. Haines, E. K. Norris, A. G. Wells,
Acta Crystallogr. Sect. B 1974, 30, 1590–1593.
[17] R. G. S. Ritchie, N. Cyr, B. Korsch, H. J. Koch, A. S. Perlin, Can. J.
Chem. 1975, 53, 1424–1433.
[18] D. Schomburg, Angew. Chem. 1983, 95, 52; Angew. Chem. Int. Ed.
Engl. 1983, 22, 65.
Chem. Eur. J. 2005, 11, 6326 – 6346
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6345

P. Klüfers et al.
[19] R. Tacke, C. Burschka, I. Richter, B. Wagner, R. Willeke, J. Am.
Chem. Soc. 2000, 122, 8480–8485.
[20] R. Tacke, M. Penka, F. Popp, I. Richter, Eur. J. Inorg. Chem. 2002,
1025–1028.
[21] a) S. D. Kinrade, K. J. Maa, A. S. Schach, T. A. Sloan, C. T. G.
Knight, J. Chem. Soc. Dalton Trans. 1999, 3149–3150; b) A. Samadi-
Maybodi, R. K. Harris, S. N. Azizi, A. M. Kenwright, Magn. Reson.
Chem. 2001, 39, 443–446.
[22] K. N. Drew, J. Zajicek, G. Bondo, B. Bose, A. S. Serianni, Carbo-
hydr. Res. 1998, 307, 199–209.
B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R.
Gomperts, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y.
Peng, A. Nanayakkara, C. Gonzalez, M. Challacombe, P. M. W. Gill,
B. G. Johnson, W. Chen, M. W. Wong, J. L. Andres, M. Head-
Gordon, E. S. Replogle, J. A. Pople, Gaussian, Inc., Pittsburgh, PA,
1998. b) M. J. Frisch, G. W.Trucks, H. B. Schlegel, G. E Scuseria,
M. A. Robb, J. R. Cheeseman, J. A. Montgomery, Jr., T. Vreven,
K. N. Kudin, J. C. Burant, J. M. Millam, S. Iyengar, J. Tomasi, V.
Barone, B. Menucci, M. Cossi, G. Scalmani, N. Rega, G. A. Peters-
son, H. Nakatsuji, M. Hada, , M. Ehara, , K. Toyota, R. Fukuda, J.
Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai,
M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J. B. Cross, C. Adamo,
J. Jaramillo, R. Gomperts, R. E Stratmann, O. Yazyev, A. J. Austin,
R. Cammi, C. Pomelli, J. Ochterski, P. Y. Ayala, K. Morokuma,
G. A. Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dap-
prich, A. D. Daniels, M. C. Strain, O. Farkas, A. D. Rabuck, K. Ra-
ghavachari, D. K. Malick, J. B. Foresman, J. V. Ortiz, Q. Cui, A. G.
Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A. Liashen-
ko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A.
Al-Laham„ C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W.
Gill, B. Johnson, W. Chen, M. W. Wong, C. Gonzalez, J. A. Pople,
Gaussian03, version B.04, Gaussian, Inc., Pittsburgh PA, 2003.
[23] M. J. King-Morris, A. S. Serianni, J. Am. Chem. Soc. 1987, 109,
3501–3509.
[24] M. Bootz, P. Klüfers, unpublished results.
[25] C. C. Perry, D. Belton, K. Shafran, in Silicon Biomineralization
(Ed.: W. E. G. Müller), Springer, Berlin, 2003, 269–299.
[26] F. H. Otey, C. L. Mehltretter, J. Org. Chem. 1961, 26, 1673.
[27] J. S. Brimacombe, A. B. Foster, M. Stacey, D. H. Whiffen, Tetrahe-
dron 1958, 4, 351–360.
[28] a) Gaussian98 (RevisionA.7), M. J. Frisch, G. W. Trucks, H. B.
Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, V. G. Zakr-
zewski, J. A. Montgomery, R. E. Stratmann, J. C. Burant, S. Dap-
prich, J. M. Millam, A. D. Daniels, K. N. Kudin, M. C. Strain, O.
Farkas, J. Tomasi, V. Barone, M. Cossi, R. Cammi, B. Mennucci, C.
Pomelli, C. Adamo, S. Clifford, J. Ochterski, G. A. Petersson, P. Y.
Ayala, Q. Cui, K. Morokuma, D. K. Malick, A. D. Rabuck, K. Ra-
ghavachari, J. B. Foresman, J. Cioslowski, J. V. Ortiz, A. G. Baboul,
Received: February 10, 2005
Published online: August 3, 2005
6346
www.chemeurj.org
ꢁ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6326 – 6346