Synthesis, purification, and characterization of phosphine oxides and their hydrogen peroxide adducts

Article abstract of DOI:10.1039/c1dt11863c

Reactions of the tertiary phosphines R₃P (R = Me, Bu, Oct, Cy, Ph) with 35% aqueous H₂O₂ gives the corresponding oxides as the H₂O₂ adducts R₃PO·(H ₂O₂)_x (x = 0.5-1.0). Air oxidation leads to a mixture of products due to the insertion of oxygen into one or more P-C bonds. ³¹P NMR spectroscopy in solution and in the solid state, as well as IR spectroscopy reveal distinct features of the phosphine oxides as compared to their H₂O₂ adducts. The single crystal X-ray analyses of Bu₃PO and [Cy₃PO·(H₂O₂)] ₂ show a PO stacking motif for the phosphine oxide and a cyclic structure, in which the six oxygen atoms exhibit a chair conformation for the dimeric H₂O₂ adduct. Different methods for the decomposition of the bound H₂O₂ and the removal of the ensuing strongly adsorbed H₂O are evaluated. Treating R ₃PO·(H₂O₂)_x with molecular sieves destroys the bound H₂O₂ safely under mild conditions (room temperature, toluene) within one hour and quantitatively removes the adsorbed H₂O from the hygroscopic phosphine oxides within four hours. At 60°C the entire decomposition/drying process is complete within one hour.

Full text of DOI:10.1039/c1dt11863c

View Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links
Dalton
Transactions
Cite this: Dalton Trans., 2012, 41, 1742
www.rsc.org/dalton
PAPER
Synthesis, purification, and characterization of phosphine oxides and their
hydrogen peroxide adducts†
Casie R. Hilliard, Nattamai Bhuvanesh, John A. Gladysz and Janet Blu¨mel*
Received 3rd October 2011, Accepted 21st October 2011
DOI: 10.1039/c1dt11863c
Reactions of the tertiary phosphines R3P (R = Me, Bu, Oct, Cy, Ph) with 35% aqueous H2O2 gives the
corresponding oxides as the H2O2 adducts R3P O·(H2O2)x (x = 0.5–1.0). Air oxidation leads to a
mixture of products due to the insertion of oxygen into one or more P–C bonds. 31P NMR spectroscopy
in solution and in the solid state, as well as IR spectroscopy reveal distinct features of the phosphine
oxides as compared to their H2O2 adducts. The single crystal X-ray analyses of Bu3P O and
[Cy3P O·(H2O2)]2 show a P O stacking motif for the phosphine oxide and a cyclic structure, in which
the six oxygen atoms exhibit a chair conformation for the dimeric H2O2 adduct. Different methods for
the decomposition of the bound H2O2 and the removal of the ensuing strongly adsorbed H2O are
evaluated. Treating R3P O·(H2O2)x with molecular sieves destroys the bound H2O2 safely under mild
conditions (room temperature, toluene) within one hour and quantitatively removes the adsorbed H2O
from the hygroscopic phosphine oxides within four hours. At 60 ◦C the entire decomposition/drying
process is complete within one hour.
dependent (see below), and the chemical shift range is fairly un-
characteristic. Therefore, they can, for example, easily be confused
Introduction
with phosphonium salts.10 However, recently a systematic and
mechanistic study has provided a deeper insight into the oxidation
process of phosphines with molecular oxygen.6a Phosphine oxides
are also becoming increasingly important as probes for the surface
acidity of oxide supports, such as silica or alumina, due to their
strong interactions with surface hydroxyl groups. Most of these
studies have been conducted using trialkyl phosphine oxides such
as Me3P O11a–e and Oct3P O11f as the probes. Additionally,
interesting metal complexes have been characterized with one,12
two,13 or three14 oxygen-bound tertiary phosphine oxide ligands.
Furthermore, there is a great potential in secondary phosphine
oxides as pre-ligands in catalysis.15
Phosphines represent one of the most important and ubiqui-
tous classes of substances. Due to their unique properties they
play prominent roles as ligands in coordination chemistry and
catalysis.1,2 Phosphine oxides are equally important but receive
less attention due to various factors. In some cases they are
unavoidable stoichiometric side-products. This is, for example, the
case for the Wittig reaction and its variations,3 the Staudinger
reaction,4 or the Appel5 reaction. Separating the products from
the phosphine oxides after these reactions can be rather tedious.
In other cases, phosphine oxides might be the sign of insufficient
exclusion of oxygen when the phosphine was the actual target
molecule of any synthesis or application.
Every chemist working with phosphine ligands has to deal with
phosphine oxides to some extent and these, in turn, can be oxidized
to form a variety of products.6 Therefore, it is surprising that
they do not receive more attention in the literature. Much of
the knowledge about phosphine oxides within research groups
working with phosphines is anecdotal. This is in part because they
are elusive species regarding their nature and characterization.
For example, they form adducts with H2O,7 boranes,8 or H2O2.9
Furthermore, their 31P NMR chemical shifts are very solvent
The interest of our group in phosphine oxides is multi-fold. We
need clean, adduct-free phosphine oxides with different electronic
and steric characteristics for studying adsorption processes on
oxide surfaces and the dynamics of the adsorbed species.16 Many
of the strategically important phosphine oxides we require, such
as Sn(p-C6H4P(O)Ph2)4,16 cannot be purchased and have to be
synthesized from the corresponding phosphines. Furthermore,
we have to be able to identify quickly and reliably the side-
products that can form during the immobilization of phosphine
linkers incorporating ethoxysilane groups on oxide surfaces.10
Therefore, baseline data regarding the NMR and IR spectroscopic
characteristics of phosphine oxides and their adducts in solution
and in the solid state is essential.
Department of Chemistry, Texas A&M University, P.O. Box 30012, College
Station, TX 77842-3012. E-mail: bluemel@tamu.edu; Fax: (979)845-5629;
Tel: (979)845-7749
† CCDC reference numbers 790886 and 845791 contain the crystallo-
graphic data for Bu3P O (2) and Cy3P O·(H2O2) (4a), respectively.
For crystallographic data in CIF or other electronic format see DOI:
10.1039/c1dt11863c
The main focus of this paper is threefold: (i) to explore quick
and selective methods for the oxidation of phosphines, (ii) to
systematically characterize phosphine oxides in solution and in
1742 | Dalton Trans., 2012, 41, 1742–1754
This journal is
The Royal Society of Chemistry 2012
©

View Online
Table 1 31P NMR chemical shifts d [ppm] of the phosphine oxides 1–5 and their corresponding hydrogen peroxide adducts 1a–5a in C6D6 and CDCl3
and the differences Dd(31P) = d(adduct) – d(phosphine oxide)
R3P O/H2O2 adducts
d(31P) C6D6
Dd(31P) C6D6
d(31P) CDCl3
Dd(31P) CDCl3
a
1/1a
2/2a
3/3a
4/4a
5/5a
32.65/ — a
43.66/51.12
43.93/48.60
46.31/50.28
25.16/27.52
—
38.79/43.23
48.57/52.50
48.48/50.01
49.91/51.53
29.10/30.15
4.44
3.93
1.53
1.62
1.05
7.46
4.67
3.97
2.36
a 1a is not soluble enough in C6D6 to observe a signal.
the solid state by NMR and IR spectroscopy, as well as X-ray
crystallography, and distinguish them from their H2O2 and H2O
adducts, and (iii) to explore methods that allow the safe, quick,
and efficient purification of the phosphine oxides.
experimental section). Table 1 compares the 31P chemical shifts
of 1–5 and their corresponding H2O2 adducts 1a–5a (generated
as described below) in benzene and chloroform. For example,
when 1 is measured in CDCl3 instead of benzene the signal
migrates downfield about 6 ppm. The dependence of d(31P) on
the solvent is less pronounced for the hydrogen peroxide adducts.
Most probably, this reflects the fact that the P O group is strongly
interacting with H2O2 by hydrogen bonding and is less prone to
interactions with solvent molecules. However, even within this
class of compounds, the 31P NMR signal of 5a migrates about
2.6 ppm downfield when changing the solvent from benzene to
The most obvious method to obtain phosphine oxides is the
exposure of the corresponding phosphines to air.6,17 However, this
approach is surprisingly limited since many triaryl phosphines,
such as triphenyl phosphine, even in solution are not efficiently
oxidized in air. This phenomenon has recently been described
mechanistically by Buchwald.6a Another drawback is that the
exposure of alkyl phosphines to air can afford numerous products
involving insertions of oxygen into phosphorus-carbon bonds
(see below).6 Methods to transfer one single oxygen atom per
phosphine have been described,18 but the necessary reagents are
not readily available and would have to be synthesized first.
Therefore, the alternative reaction with aqueous H2O2 will also
be investigated here. Using a selection of representative alkyl and
aryl phosphines, it will be demonstrated that the physical and
spectroscopic characteristics of phosphine oxides and their water
and hydrogen peroxide adducts are distinctly different from each
other. Finally, methods to obtain the adduct-free representative
alkyl and aryl phosphine oxides 1–5 (Scheme 1) are discussed. Two
successful procedures are described, one of which is optimized.
CDCl3. In general, benzene exerts a shielding effect on the P
O
group for both the phosphine oxides and their H2O2 adducts (Table
1).
Most significant analytically is the large change of the chemical
shift when going from the phosphine oxides to their corresponding
H2O2 adducts. Fig. 1 visualizes this dramatic effect for 2 and 2a
in C6D6. As can be seen from the values given in Table 1, the
chemical shift differences can amount to nearly 8 ppm (2/2a) in
benzene. Overall, these shift differences between the substance
classes are more pronounced in benzene than in CDCl3. As a
general trend, the signals of the H2O2 adducts appear downfield
from the phosphine oxides. This is in accordance with expectations
since the hydrogen bonding of H2O2 to the oxygen of the P
group should lead to a deshielding of the phosphorus nucleus.
O
Scheme 1 Representative tertiary phosphine oxides 1–5 used in this work.
Results and discussion
Fig. 1 31P NMR spectra of 2 (bottom trace) and 2a (top trace) in C6D6.
The asterisks denote traces of Ph2ClP O in the capillary, which stem from
oxidation of the standard Ph2PCl (see experimental section).
1. Distinguishing phosphine oxides from their H2O2 adducts
(a) NMR Spectroscopy. Probably the most intriguing part
of working with phosphine oxides and their hydrogen peroxide
adducts is that the chemical shifts are extremely solvent dependent.
In order to obtain accurate chemical shift values with one
measurement, we used ClPPh2 as the chemical shift standard
in a capillary insert centered in the NMR tubes (details: see
In addition to the chemical shift issue, the lines of the H2O2
adducts are often broad due to the non-stoichiometric interactions
of the phosphine oxides with H2O2 and H2O, and exchanges with
solvent molecules. Furthermore, larger aggregates might form due
to the strong intermolecular hydrogen bonding with H2O2.
This journal is
The Royal Society of Chemistry 2012
Dalton Trans., 2012, 41, 1742–1754 | 1743
©

View Online
Therefore, we investigated the dependence of d(31P) of 2, 2a, 5,
and 5a on their concentration in solution. Benzene has been chosen
as the solvent for this quantitative analysis, because it leads to the
largest d(31P) difference between the phosphine oxides and their
H2O2 adducts (Table 1). For 2 and 5, the 31P chemical shift changes
with concentration are minimal. For example, d(31P) of 5 varies
within a 0.12 ppm range without a clearly distinguishable trend
when increasing the concentration from 0.010 to 0.080 mol L-1
(see experimental section for stepwise procedure). A concentration
change from 0.10 to 0.75 mol L-1 for 2 results in a chemical shift
variation of 0.27 ppm. For both the H2O2 adducts 2a and 5a, the
changes of d(31P) are larger, and characteristic curves are obtained.
Fig. 2 shows the graphical display for 2a. Overall the chemical shift
increases for 2a (5a) by 0.87 (1.25) ppm when the concentration
is changed from 0.10 (0.010) to 2.0 (0.080) mol L-1. Since at very
low concentrations the d(31P) values for the adducts show a trend
towards the phosphine oxides 2 and 5, we conclude that the solvent
is, to some extent, able to break up agglomerated molecules and
replace the H2O2. However, even with extreme dilution, the very
different chemical shifts of the adduct-free phosphine oxides are
not reached. Furthermore, the overall chemical shift changes over
a dilution range of at least an order of magnitude are noticeable,
but far from the d(31P) differences between phosphine oxides and
their corresponding H2O2 adducts.
Fig. 3 Water uptake of Me3P O (1) and Bu3P O (2) in air.
state NMR spectra,19 due to the lack of solvent interactions and
exchange equilibria, narrow lines are expected for the polycrys-
talline materials. Unfortunately, most of the phosphine oxides 1–5
and especially the H2O2 adducts 1a–5a are liquids or low-melting
solids (see Fig. 11 below) that would undergo a phase change under
the pressures and elevated temperatures20 of MAS (Magic Angle
Spinning).19 But for the pairs 4/4a and 5/5a, the CSA (chemical
shift anisotropy)19a data and 31P MAS spectra could be obtained.
Fig. 4 and 5 show, for example, the MAS and wideline spectra
of 5 and 5a. In accordance with our expectation for crystalline
material, the residual linewidths of the isotropic peaks in the MAS
spectra are only 166 Hz for 5 and 261/268 Hz for the two lines
of 5a. The isotropic lines can easily be determined by varying the
rotational speed because in contrast to the rotational sidebands,
they are static and do not change their positions.
Fig. 2 Change of the d(31P) of Bu3P O·(H2O2) (2a) with its concentra-
tion in C6D6.
Since the phosphine oxides Me3P O (1) and Bu3P O (2)
are hygroscopic (see below), the dependence of d(31P) on the
concentration of 2·H2O in C6D6 has also been investigated. The 31
P
Fig. 4 31P MAS (top, 4 kHz) and wideline (bottom, 0 kHz) spectra of
polycrystalline 5. The d(31P) value is given for the isotropic chemical shift
in the MAS spectrum; the other lines are rotational sidebands.
chemical shift of Bu3P O·H2O changes from 44.35 to 43.60 ppm
upon dilution of a 0.36 molar to a 0.05 molar solution, which
results again in a noticeable but not crucial overall d(31P) change
of 0.75 ppm.
In contrast to 1 and 2, the phosphine oxides 3–5 are not
The d(31P) of 28.14 ppm for the isotropic line in the solid-state
NMR spectrum of 5 (Fig. 4) lies within the chemical shift ranges
found in solution (Table 1). There is only one isotropic line for 5, as
expected from earlier results21 and tri(p-tolyl)phosphine oxide.21c
This is also in accordance with the single crystal X-ray structures
of 522 or 2 (see below), which imply that all phosphorus nuclei in
the unit cell are magnetically equivalent.
In contrast, the polycrystalline H2O2 adduct 5a displays two
isotropic lines in the 31P MAS spectrum (Fig. 5). The chemical
shifts of 31.01 and 27.90 ppm are close to the values found in
solution (Table 1). Since the intensity ratio of both lines, including
the rotational sideband intensities in the integration, is about 1 : 1,
hygroscopic. Me3P O (1) is even more hygroscopic than Bu3P
O
(2),7 and any exposure to humid air changes the crystalline solid to
an oil within less than 5 min. Fig. 3 illustrates the gravimetrically
determined uptake of H2O with time for 1 and 2. Me3P O not
only absorbs H2O faster but also a larger amount of it, as compared
to 2. Saturation is reached when the ratio of 1 : H2O is about 1 : 3.5.
On the same timeline, 2 is saturated when the ratio of 2 : H2O is
about 1 : 1.
In the previous section we described that with careful chemical
shift referencing, it is possible to distinguish the phosphine oxides
from their H2O2 adducts by 31P NMR in solution. In the solid-
1744 | Dalton Trans., 2012, 41, 1742–1754
This journal is
The Royal Society of Chemistry 2012
©

View Online
147 ppm for 4a. In this case, however, the wideline patterns of both
signals are not as clearly distinct as for 5a.
The general CSA reduction on going from the phosphine oxides
to their H2O2 adducts reflects the fact that with the adduct
formation the P O bond order and its polarity decrease, in
accordance with the IR analysis (see below). For example, in
the extreme case of decreasing the P O bond order on going
from Ph3P O to [Ph3POEt]BF4 the CSA decreases from 195–
200 ppm21a,b to 74 ppm.10c In analogy to H2O2 adduct formation,
substantial reduction in the CSA is also observed when phosphine
oxides are adsorbed on oxide supports, such as silica, as the P
O
groups interact with surface silanol groups and undergo partial
quaternization.16
(b) Crystallography. The crystal structure of a water clathrate
of 2 (2·(H2O)34.5) has been reported,24 but not the pure substance.
Accordingly, suitable crystals of 2 were obtained, X-ray data
were collected, and the structure was solved as described in the
experimental section. This gave the structures displayed in Fig. 6
and 7.
Fig. 5 31P MAS (top, 4 kHz) and wideline (bottom, 0 kHz) spectra
of polycrystalline 5a. The d(31P) values are given for the two isotropic
chemical shifts in the MAS spectrum; the other lines are the rotational
sidebands. The symbols represent the d11 and d33 of the two corresponding
CSA patterns.
the presence of polymorphs is unlikely. The two isotropic lines
most probably stem from two magnetically inequivalent phospho-
rus nuclei in the unit cell. This is corroborated by the single crystal
X-ray structure of the compound (Ph3P O)2·(H2O2),9c where one
H2O2 molecule bridges two Ph3P O molecules, which leads to
two magnetically independent phosphorus nuclei in the unit cell.
Additionally, the unit cell of the single crystal of Cy3P O·(H2O2)
(see below) contains two magnetically inequivalent P nuclei. The
splitting of isotropic lines in the 31P MAS spectra of polycrystalline
compounds is also often observed for chelate phosphine ligands
and metal complexes thereof. For example, (Ph2PCH2)3SiOEt
featured three isotropic lines with equal intensities but with a
large difference in their chemical shifts.23 (Ph2PCH2)2Si(OEt)2
and diverse nickel complexes with chelate phosphine ligands19d,23
always gave two isotropic lines with similar chemical shifts.
Cy3P O (4) and its H2O2 adduct 4a exhibit analogous solid-
state NMR characteristics. The only isotropic signal of 4 has a
chemical shift of 47.28 ppm, which lies within the d ranges found
in solution. The residual linewidth is only 158 Hz, so overlapping
of two or more isotropic signals is unlikely. Analogous to 5a, the 31P
MAS spectrum of the polycrystalline H2O2 adduct 4a displays two
isotropic lines with a 1 : 1 intensity ratio of the signals including the
rotational sidebands. The chemical shifts of 51.53 and 53.61 ppm
are again close to the values found in solution (Table 1).
Fig. 6 Single crystal X-ray structure of Bu3P O (2), view◦showing the
˚
aligned P O bonds. Selected bond lengths (A) and angles ( ): P(1)–O(1)
1.489(2), P(1)–C(1) 1.797(3), P(1)–C(5) 1.798(2), P(1)–C(9) 1.798(2),
O(1)–P(1)–C(1) 113.10(10), O(1)–P(1)–C(5) 112.90(10), O(1)–P(1)–C(9)
113.11(10).
The 31P MAS spectra of the triarylphosphine oxide 5 and
its H2O2 adduct 5a could be obtained with only 64 scans at
the moderate spinning speeds of 4 and 5 kHz. Therefore, the
wideline spectra without rotation could be recorded as well
(Fig. 4 and 5). The chemical shift anisotropy of 5, defined
as the span of the wideline signal,19a is about 200 ppm and
therewith matches the literature values for 5 (CSA 195–200 ppm),21
for tri(p-tolyl)phosphine oxide (CSA 155 ppm),21c or for the
trialkylphosphine oxide (n-C14H29)3P O (CSA 190 ppm).21a The
CSA of the 31P wideline signal of 5a is substantially smaller with
about 170 and 200 ppm for the two 31P wideline signals (Fig. 5).
The reduction of the CSA on going from a phosphine oxide to
its adduct is even more pronounced for the pair 4/4a. The span
of the wideline signal is about 153 ppm for 4 and about 128 and
Fig. 7 Single crystal X-ray structure of Bu3P O (2), view along the P
bond.
O
The bond lengths and angles about phosphorus (Fig. 6)
were routine. Presumably due to dipole–dipole interactions, a
This journal is
The Royal Society of Chemistry 2012
Dalton Trans., 2012, 41, 1742–1754 | 1745
©

View Online
O
P ◊ ◊ ◊ O P stacking motif was observed. This has also been
found with other crystalline phosphine oxides such as 425 and 5.22
These stacking motifs may be relevant for the potential formation
of bilayers on silica surfaces.16 The intermolecular O ◊ ◊ ◊ P distance,
˚
3.587 A, exceeds the sum of the van der Waals radii of the oxygen
and phosphorus atoms (1.52 and 1.80 A). The adjacent P
26
˚
O
linkages within a stack are not perfectly aligned, as reflected by
intermolecular O–P–O and P–O–P angles of 179.2◦ and 174.0◦.
The butyl groups crystallize in all-anti conformations. Fig. 7
illustrates that the butyl groups are slightly staggered between
adjacent molecules in a stack, as quantified by a C1–P–P–
C1 torsion angle of 38.8◦. Importantly, the structure does not
contradict the finding of only one isotropic line in the MAS spectra
of 4 and 5 (see above).
Only one crystal structure of a H2O2 adduct of a phosphine
oxide, the 2 : 1 complex 5·HOOH·5 or (Ph3P O)2·(H2O2), has
been previously reported.9c Suitable crystals of the 1 : 1 complex
4a were obtained as described in the experimental section. X-
ray data were collected and the structure was solved in a routine
manner that included the location and refinement of the oxygen-
bound hydrogen atoms. This gave the cyclic dimeric structure
(Cy3P O·H2O2)2 or (4a)2 displayed in Fig. 8 and 9. The adduct
features two hydrogen bonds to each phosphine oxide oxygen
atom, with an inversion center in the midpoint of the plane formed
by the four H2O2 oxygen atoms.
Fig. 9 Additional views of the single crystal X-ray structure of the dimer
of 4a. Top: tilted side view, bottom: side view.
(90.2(6)◦). Hence, the structure of H2O2 is not appreciably affected
by bonding to 4.
Importantly, Fig. 8 and 9 corroborate the finding of two
isotropic lines in the 31P solid-state NMR spectra of the H2O2
adducts of the phosphine oxides, because there are two magneti-
cally independent phosphorus nuclei in the unit cell.
(c) IR Spectroscopy. IR spectroscopy28 provides unique insight
into the phosphine oxide and H2O2 adduct scenario because it
probes both the P O groups9b,29 and all O–H species9b via their
stretching vibrations. As in solid-state NMR, the IR spectra can
be recorded using the neat liquid or polycrystalline samples, and
therefore, the results are independent of any interactions of the
functional groups with solvents. Fig. 10 shows, as a representative
case, the IR spectra of 2 (top), 2a (bottom), and 2a after partial
decomposition of the bound H2O2 as described below (middle).
Table 2 summarizes key IR data.
Fig. 8 Single crystal X-ray structure of the dimer of 4a. Selected
bond lengths (A) and angles (◦): P(1)–O(1) 1.5045(9), P(1)–C(1)
˚
1.8190(12), P(1)–C(7) 1.8209(12), P(1)–C(13) 1.8258(12), O(1)–P(1)–C(1)
111.03(5), O(1)–H(20) 1.876, H(20)–O(20) 0.896, O(20)–O(21) 1.4504(15),
O(21)–H(21) 0.895, O(1)–P(1)–C(1) 113.03(5), O(1)–P(1)–C(7) 109.92(5),
O(1)–P(1)–C(13) 112.04(5), P(1)–O(1)–H(20) 129.1, O(1)–H(20)–O(20)
176.0, (H20)–O(20)–O(21) 103.2, (O20)–O(21)–H(21) 100.4.
As the predominant analytical features, the spectrum of 2 (Fig.
10, bottom) shows only the bands typical for C–H stretching
absorptions below 3000 cm-1 and the P O stretching band at
1153 cm-1. All bands are very narrow, indicating that the vibrations
within the molecules are well-defined.
In contrast, while the C–H stretching bands are practically
unchanged as compared to 2, the IR spectrum of 2a (Fig. 10,
top) shows a broad O–H stretching band for the bound H2O2
The P O linkage in (4a)2 proves to be somewhat longer
than that in the corresponding phosphine oxide 425 (1.5045(9)
˚
˚
vs. 1.490(2) A). The P O ◊ ◊ ◊ H distance (1.876 A) is in the
typical range for hydrogen bonds, and the O ◊ ◊ ◊ O distance
Table 2 IR absorptions [n] of the phosphine oxides 1–5 and their
corresponding H2O2 adducts 1a–5a and the wavenumber differences
Dn(P O) = n(phosphine oxide) - n(adduct)
˚
in the P O ◊ ◊ ◊ H–O unit (2.743 A) compares with that in
˚
(Ph3P O)2·(H2O2) (2.759 A), for which the oxygen-bound hy-
drogen atom could not be located.9c
R3P O/H2O2
adduct
n(O–H) of
1a–5a/cm-1
n(P O)/cm-1
Dn(P O)/cm-1
The cyclohexyl rings and the assembly defined by the six
oxygen atoms adopt chair conformations. In the latter, the HO–
1/1a
2/2a
3/3a
4/4a
5/5a
1161/1094
1153/1123
1144/1142
1157/1138
1188/1174
67
30
2
19
14
3217
3217
3217
3264
3233
OH distance is within experimental error of that in crystalline
27
˚
H2O2 (1.4504(15) vs. 1.453(7) A). The dihedral angles defined
by the HO–OH and (P O) ◊ ◊ ◊ O—O ◊ ◊ ◊ (O P) units (89.54◦ and
87.10◦) are also very close to the dihedral angle in crystalline H2O2
1746 | Dalton Trans., 2012, 41, 1742–1754
This journal is
The Royal Society of Chemistry 2012
©

View Online
due to the fast oxidation process. The same is true for neat
Ph2PH, which forms Ph2HP O exclusively in a radical oxidation
process when exposed to air.17 The predominant side-reaction
is the additional insertion of oxygen atoms into P–C bonds6
to form mainly phosphinic and phosphonic and occasionally
phosphoric acid esters that can be identified by their d(31P).6,10a,19e
For example, when toluene solutions of Bu3P are exposed to air,
a typical 31P NMR spectrum would show the signal of the desired
product Bu3P O (2, 43.34 ppm, 44%), as well as resonances for
the phosphinic acid ester Bu2(BuO)P O (53.96 ppm, 45%), the
phosphonic acid ester Bu(BuO)2P O (36.28 ppm, 7%), traces of
unidentified oxidation products (around 30 ppm) and unreacted
phosphine (-32.20 ppm). The product ratio does not change with
the reaction time, and we could not optimize this reaction to give
only, or at least predominantly, 2. Even if it was feasible to separate
the product from the reaction mixture, all the side-products still
lead to a diminished yield of phosphine oxide.
As an additional complicating factor, H2O from the humid
air can be adsorbed due to the formation of hydrogen bonds
with the P O group, rendering the analysis of mixtures by
31P NMR difficult (see above). This H2O uptake takes place
preferentially with short alkyl chain substituents R, as discussed
above. Triarylphosphines might not be oxidized by air at all.6a
For example, Ph3P is not transformed into the oxide 5 when
dissolved in a nonpolar solvent and exposed to the ambient
atmosphere for months. Taking all these limitations into account,
the oxidation of phosphines by simple exposure to air is not a
feasible or efficient way to obtain clean tertiary phosphine oxides.
This overall scenario might also be responsible for the frustration
that researchers can experience when handling phosphines without
carefully excluding air.
In contrast, when H2O2 is employed as the oxidant in the
synthesis of phosphine oxides, no byproducts are obtained, and
the reaction proceeds smoothly at room temperature and in a
well-defined manner. However, H2O2 molecules form hydrogen
bonds to the oxygen of the P O group and adducts of the type
R3P O·(H2O2)x with x = 0.5 to 1.0 are obtained. Interestingly,
although the oxidation with hydrogen peroxide is performed in
the presence of water, the latter seems to form weaker adducts and
is not competitive, so that only H2O2 is bound to the products
(Fig. 10, top spectrum). The single crystal X-ray analysis of 5a
proves that two molecules of 5 are bound to one bridging H2O2
molecule.9c On the other hand, in the case of a H2O2 adduct of
Ph3As O, a hydrogen-bonded network with a 1 : 1 ratio of arsine
oxide to H2O2 has been found.30 The hydrogen peroxide adducts
1a–3a most likely have a less well-defined H2O2 to R3P O ratio as
they have an oily appearance and thus can easily be distinguished
from the phosphine oxides 1–3, which are all colorless powders.
This is displayed for 2 and 2a in Fig. 11.
Fig. 10 IR spectra of neat 2 (bottom), 2a (top), and 2a with adsorbed
H2O after partial decomposition of the bound H2O2 (middle).
at 3217 cm-1.9b The intense P O band has shifted to the lower
wavenumber of 1123 cm-1. This indicates that the P O bond order
has decreased, rendering the corresponding stretching vibration
lower in energy. This change goes along with the broadening of the
line because of the less well-defined vibrations due to the hydrogen
bonding interactions with the H2O2 molecules.
When the H2O2 molecule in 2a is partially decomposed ther-
mally, the H2O strongly binds to the P O group. Its O–H
stretching vibration is distinctly different from that of bound
H2O2, but its band at 3408 cm-1 is also broad, as can be seen
in the middle spectrum of Fig. 10. The stretching band of the
residual H2O2 molecules remains practically unchanged at 3219
cm-1. As an additional, analytically useful feature, the overtone of
the O–H absorption of the H2O band appears at about 1647 cm-1,
as described in the literature for adducts of Ph3P O.9b Again,
the P O stretching frequency is broad and found at a lower
wavenumber (1140 cm-1) for 2·(H2O2)x(H2O)y (x,y = 0.5–1.0) than
for 2.
Overall, IR spectroscopy is a powerful tool for quickly analyzing
the nature and purity of phosphine oxides and their various
adducts, and it was primarily used in the following experiments
for characterizing all involved species and optimizing the two most
promising procedures for obtaining clean phosphine oxides.
(b) Possible methods for decomposing the hydrogen peroxide
adducts 1a–5a to give the phosphine oxides 1–5. There are several
known approaches for decomposing H2O2 attached to phosphine
oxides by hydrogen bonds. Bimetallic Pt–Pd surfaces are, for
example, known to decompose H2O2.31 Additionally, copper9a
is reported to achieve the decomposition and also MnO2.9a
Furthermore, the adducts can be heated (> 100 ◦C) to release
the oxygen from the bound H2O2.9a However, with the latter
method, care has to be taken since the oxygen release can happen
2. Controlled decomposition of the hydrogen peroxide adducts
1a–5a to give the phosphine oxides 1–5
(a) Syntheses of 1a–5a. When phosphines are oxidized in
air, especially trialkyl phosphines, a multitude of byproducts are
possible besides the desired phosphine oxides R3P O. Only Me3P
gives Me3P O (1) in a reasonably clean reaction, most probably
This journal is
The Royal Society of Chemistry 2012
Dalton Trans., 2012, 41, 1742–1754 | 1747
©

View Online
Table 3 Melting points and decomposition temperature ranges [◦C] of the phosphine oxides 1–5 and their corresponding H2O2 adducts 1a–5a
R3P O/H2O2 adduct
Decomp. temp. of neat H2O2 adducta
m.p. of neat R3P O, obs.
m.p. of neat R3P Oref.
1/1a
2/2a
3/3a
4/4a
5/5a
63–94
135–139
59–63
48–52
158–160
155–157
137.5–138.533
63–6434
98–149
88–140
120–134
136–183
48–5035
155–15712
156.5–15836
a Values based on the appearance of small bubbles in the capillary during heating, which are attributed to evolution of oxygen from H2O2 decomposition.
behind safety glass because according to our experience, H2O2
adducts of phosphines can lead to explosions when heated in neat
form.
Table 3 summarizes the data obtained from the neat materials
and provides the literature values for comparison. The decompo-
sition of the bound H2O2 happens well below the actual boiling
point of the corresponding adduct-free phosphine oxide.
As illustrated in Fig. 12, the effervescence indicates the release of
O2 gas as H2O2 is broken down. The onset of the decomposition, as
well as its temperature range, varies greatly between the adducts
(Table 3). Naturally, the decomposition process is better visible
for the viscous liquids 1a–3a than for the powders 4a and 5a.
Interestingly, for both 2a and 3a we found that at around 150 ◦C
there is a sudden onset of the formation of larger bubbles. We
assign this phenomenon to the release of bound H2O because this
temperature is still well below the boiling points of 2a and 3a.
Fig. 11 Physical appearance of neat Bu3P
Bu3P O·(H2O2)x (2a, left, x = 0.5–1.0) in a vial.
O (2, right) and
quickly, resulting in violent explosions. We tested and evaluated a
selection of the known procedures for the decomposition of H2O2
attached to phosphine oxides and optimized new methods that
also efficiently remove the resulting H2O quantitatively.
Surprisingly, when we treated 2a in benzene with freshly cut
pieces of a Cu wire, even at higher temperatures and after
prolonged exposure, no H2O2 decomposition occurred as as-
sayed by IR spectroscopy. However, when 2a was treated with
MnO2 powder, O2 evolved immediately and vigorously, and IR
spectroscopy confirmed the complete decomposition of H2O2.
Unfortunately, the H2O formed during this process remains
strongly bound to 2a. Furthermore, the fine MnO2 powder could
not be removed by standard filtration processes, rendering this
process impractical.
Next, we tried silica as a model material for glass with a high
surface area because it is generally known that aqueous H2O2
decomposes with time at the walls of glass bottles.9a Indeed, when
stirring 2a in toluene overnight with a slurry of silica that had
not been dried prior to its application (“wet” silica, containing
adsorbed H2O and a maximal number of surface OH groups),32
H2O2 decomposition was complete. However, IR revealed that the
formed H2O was still attached to 2a, and it could not be removed
by treating the material with Na2SO4 or MgSO4.
Fig. 12 Evolution of small O2 bubbles during decomposition of the neat
materials R3P O·(H2O2)x (x = 0.5–1.0) at the given temperatures (1a–5a).
While the decomposition of the bound H2O2 in neat 1a–5a
is clearly possible, this method suffers from the fact that water
still remains in the samples. Furthermore, high temperatures are
needed, and scaling up this process poses a safety risk. Therefore,
as described in the following paragraph, we optimized two new
methods that can be applied safely in any laboratory and leads to
the clean phosphine oxides in a straightforward manner.
Since we had so far not found a way to obtain the water-
free phosphine oxides from the hydrogen peroxide adducts, we
investigated the option of thermal decomposition because then we
would at least avoid further product separation steps. Heating 2a
in a toluene solution in order to decompose the attached H2O2
to obtain 2a·(H2O)x required 36 h and temperatures of up to
95 ◦C. Besides still being confronted with the task to remove
the bound water, another practical drawback of this method is
the time consuming removal of the relatively high-boiling solvents
required. Therefore, we studied the decomposition of the neat
adducts 1a–5a. We worked on a small scale in melting point tubes
(c) Optimized method for decomposing the hydrogen peroxide
adducts 1a–5a to give the phosphine oxides 1–5. All methods
described in the previous section lead to the water adducts of
the phosphine oxides and not the clean species R3P O. One of
the safest methods consists of the treatment with silica, which can
easily be separated from the supernatant with the product after the
reaction. Therefore, instead of using the “wet” silica, we performed
the same reaction with silica pre-dried at 200 ◦C for seven days in
vacuo. The resulting silica surface should consist largely of siloxane
groups with only a minimal number of residual silanol moieties,32
which might have the potential to compete with the phosphine
1748 | Dalton Trans., 2012, 41, 1742–1754
This journal is
The Royal Society of Chemistry 2012
©

View Online
oxides for the formed H2O by adsorbing it. Being less reactive,
pre-dried silica is our preferred support material for immobilized
catalysts.2,37 The diminished reactivity of a slurry of pre-dried silica
in toluene manifests itself in the prolonged time of about 4 d
needed to decompose the H2O2 in 2a. Fortunately, however, IR
spectroscopy of the supernatant proves the disappearance of O–H
stretching bands for both H2O2 and residual water. The adduct-
free and clean phosphine oxides can be obtained in this way. The
only drawback of using silica is that phosphine oxides are strongly
adsorbed at its surface.16 Therefore, in order to diminish the loss of
product, several wash cycles are required to detach the phosphine
oxides from the silica surface.
Our quest for a material that would decompose the H2O2 in
1a–5a and at the same time quickly remove the resulting H2O
while releasing the phosphine oxides 1–5 easily after the reaction,
led us to molecular sieves. This material seemed promising to us
because it is known to decompose 30% aqueous H2O2 at 80 ◦C.38
Fig. 13 Molecular sieves placed in a tea bag and suspended in toluene in
a round bottom flask for easy removal after the reaction.
teristics of the clean phosphine oxides are distinctly different from
those of the hydrogen peroxide adducts. IR spectroscopy has been
applied to distinguish between the neat phosphine oxides, their
hydrogen peroxide adducts, and to detect adsorbed water. Single
crystal X-ray structures of Bu3P O and Cy3P O·(H2O2) have
been determined, the latter being the first of its type. With IR
spectroscopy, a method using molecular sieves could be optimized
for the safe and quick decomposition of the bound hydrogen
peroxide. The molecular sieves also efficiently and quantitatively
remove the water formed in this process. The yields of the clean
phosphine oxides are nearly quantitative since phosphine oxides
are only weakly adsorbed at the surface of the molecular sieves.
˚
Indeed, pre-dried molecular sieves with 3 A pore diameter allowed
us to obtain the clean phosphine oxides from the corresponding
phosphines in very high yields of up to 94% (Scheme 2). Most of
the loss in the yield occurs in the biphasic oxidation step, as this
requires the separation of organic and aqueous layers, and some
of the H2O2 adducts of the phosphine oxides are rather soluble in
water. The loss of the phosphine oxide products due to adsorption
has been determined by model reactions to be minimal, less than
2 mol% per g of molecular sieves with the amounts and ratios of
solids to phosphine oxides given in the experimental section. The
reason for this is that, in contrast to silica,32 molecular sieves do
not have a large number of acidic protons on the surface, which
could strongly interact with the free electron pairs of phosphine
oxides.
Experimental
(a) General procedures
All reactions involving phosphine starting materials were per-
formed under a purified nitrogen atmosphere. Phosphine oxides
were stored under nitrogen. Chemicals were treated as follows:
toluene (Mallinckrodt Chemicals, ACS grade) was distilled from
Na/benzophenone and CH2Cl2 (Mallinckrodt Chemicals, HPLC
grade) was dried in a commercially available MBraun solvent
purification system. C6D6 (Cambridge Isotope Laboratories) and
˚
CDCl3 (Aldrich) were dried over 3 A molecular sieves (EMD
Chemical Inc.). The latter was also used for obtaining the adduct-
free phosphine oxides, and it has the approximate composition
(weight percentages): silica gel < 50%, Al2O3 < 40%, Na2O <
30%, K2O < 15%, MgO < 5%, and quartz < 3%. The molecular
sieves were activated by heating in vacuum at 120 ◦C for 12 h.
Bu3P (Strem Chemicals, 99%), Oct3P (Alfa Aesar, 90% technical),
and Cy3P (Alfa Aesar) were purified by column chromatography
(alumina, elution with CH2Cl2). Me3P (Alfa Aesar, 99%) and
ClPPh2 (TCI America, 98%) were used as received but kept in
a glove box. H2O2 (Acros Organics, 35% aqueous solution) was
either used as obtained or diluted and stored at 4 ◦C. Ph3P
(Aldrich, 99%), MnO2 (Alfa Aesar, technical, min. 58%), and
Scheme 2 Yields of clean phosphine oxides 1–5 with respect to the
corresponding phosphines after oxidation with hydrogen peroxide to
form 1a–5a, decomposition of attached H2O2, and removal of water with
molecular sieves (x,y = 0.5–1.0).
Using the molecular sieves, both the decomposition of H2O2
and the removal of H2O from the hygroscopic phosphine oxides
were complete within 4 h at room temperature. At the elevated
temperature of 60 ◦C the whole process is complete within 1 h.
The molecular sieves are best placed into a tea bag in order to
facilitate stirring and their retrieval after the reaction (Fig. 13).
This protocol is safe and can easily be scaled up, and amounts of
more than 5 g of molecular sieves pose no problem.
˚
silica (Merck, 40 A average pore diameter, 0.063 to 0.2 mm average
particle size, specific surface area 750 m2 g-1) were used as obtained.
In addition to the latter “wet” silica, a “dry” batch was generated
by heating the silica for 7 d at 200 ◦C at the vacuum line.
Conclusions
1
1
1
The H, 13C{ H}, and 31P{ H} NMR spectra of liquids were
recorded on a 500 MHz Varian spectrometer at 499.70, 125.66,
and 470.17 MHz. The 1H and 13C chemical shifts were referenced
It has been demonstrated that the 31P NMR chemical shifts of
phosphine oxides and their hydrogen peroxide and water adducts
are very different. Furthermore, the 31P solid-state NMR charac-
1
using the solvent signals. For H NMR, the residual protons in
This journal is
The Royal Society of Chemistry 2012
Dalton Trans., 2012, 41, 1742–1754 | 1749
©

View Online
the deuterated solvents (C6D5H, 7.15 ppm; CHCl3, 7.26 ppm)
were used; for 13C NMR the carbon signals (C6D6, 128.02 ppm;
CDCl3, 77.00 ppm). For the accuracy of the 31P NMR chemical
shifts of solutions, ClPPh2 (neat liquid, d(31P) = 81.92 ppm) was
used as a chemical shift standard within a capillary centered in
the 5 mm sample tubes. The signal assignments were based on
previous results2a,39 and confirmed by 2-dimensional 1H,1H COSY
was be extracted from the aqueous layer with CH2Cl2, or the
entire synthesis performed using CH2Cl2 instead of toluene as the
solvent.
Representative procedure for measuring d(31P) depending on the
concentration of 2, 2a, 5, and 5a. The H2O2 adduct of Bu3P O,
2a (0.250 g, 0.991 mmol, assuming one molecule of H2O2 per
one phosphine oxide molecule), was placed into an NMR tube.
2a was dissolved in 0.5 mL of C6D6, and a capillary containing
the neat standard ClPPh2 was centered in the NMR tube. The
31P NMR spectrum was recorded applying 32 scans. Subsequently
the solution was transferred into a 20 mL vial, and an additional
0.5 mL of C6D6 was added. A 0.5 mL aliquot of this solution was
placed in the NMR tube and measured as described above. This
procedure was repeated ten additional times until a total volume
of 6.0 mL was reached. Then 1.5 mL aliquots of C6D6 were added
three times for greater dilution. The final volume of the mixture
was then 10.5 mL.
1
and H,13C COSY NMR correlation spectroscopy experiments.
Since the literature data are fragmentary, all 1H, 13C, and 31P NMR
data obtained in two different solvents (C6D6 and CDCl3) are given
below.
The 31P solid-state NMR spectra were recorded on a 400 MHz
Bruker Avance widebore NMR spectrometer equipped with a
4 mm broadband probehead and ZrO2 rotors. The polycrystalline
substances were densely packed into the rotors. The chemical shifts
were referenced with polycrystalline (NH4)H2PO4 (+0.81 ppm) as
the external standard. A single pulse program with 1H high-power
decoupling was applied, and typically about 3000 scans with a
pulse delay of 10 s were sufficient to obtain a good signal-to-noise
ratio.
Representative procedure for the synthesis of the phosphine
oxides 1–5 from their H2O2 adducts 1a–5a using molecular sieves.
The H2O2 adduct 2a (3.751 g), obtained by treatment of Bu3P
(3.552 g, 17.56 mmol) with aqueous H2O2 as described above, was
placed into a round bottom flask and dissolved in 150 mL of
toluene. Then, a tea bag (obtained by removing the contents of
a commercially available tea bag) was filled with 5 g of activated
molecular sieves, closed by stapling and suspended in the middle
of the flask and fully immersed in the solvent (Fig. 13). The
solution underneath the tea bag was stirred vigorously at room
temperature for 4 h. After retrieving the molecular sieves, the
solvent was removed in vacuo to give 2 as a colorless powder
(3.540 g, 16.21 mmol) in 92.3% yield with respect to the starting
phosphine. At 60 ◦C this reaction is complete after 1 h. All
phosphine oxides 1–5 were colorless crystalline solids, and they
were characterized by 31P NMR and IR spectroscopy, as well as
their melting points (see Tables 1–3). For the sake of completeness,
and because there are noticable differences between the data of the
IR data of the neat powders and viscous liquids were recorded
on a Shimadzu IRAffinity-1 FTIR instrument using a Pike
Technologies MIRacle ATR plate. Melting point videos were
obtained in open capillaries with a Stanford Research Instruments
MPA100 OptiMelt system.
(b) Synthesis and purification of phosphine oxides
Representative procedure for the synthesis of the H2O2 adducts
of the phosphine oxides (1a–5a). For each oxidation the same
procedure was applied, except for minor deviations for 1a and
5a, which are described below. Representative procedure: Bu3P
(3.552 g, 17.56 mmol) was dissolved in 150 mL of toluene in a
Schlenk flask, which was cooled to 0 ◦C. Then ten equivalents
of concentrated aqueous H2O2 solution (35 weight % in H2O,
17.10 mL, 176.0 mmol H2O2) were added dropwise via syringe
with vigorous stirring. The reaction mixture was stirred overnight
and allowed to slowly warm to room temperature. After 12 h,
the toluene layer was transferred into a round bottom flask with
a metal cannula. 31P NMR proved that the oxidation process is
quantitative. The aqueous layer was washed with toluene (3 ¥
20 mL), and all organic phases were combined. The solvent was
removed in vacuo, and 2a was obtained as a colorless viscous
liquid (3.751 g, 84.7% yield, assuming one H2O2 molecule per one
phosphine oxide molecules). All H2O2 phosphine oxide adducts
1a–5a were characterized by 31P NMR and IR spectroscopy, as
well as their melting points (see Tables 1–3 above). The hydrogen
peroxide adducts 1a–3a were colorless oils while 4a and 5a were
white crystalline solids.
The H2O2 adduct 1a was highly water soluble, so it could not be
efficiently extracted into toluene or other polar organic solvents,
such as diethyl ether, THF, CH2Cl2, and ethyl acetate. Therefore,
after the oxidation step, all volatile matter (solvents, excess H2O2,
H2O) was removed in vacuo at 40 ◦C. Excess water remained
hydrogen-bonded to the adduct 1a (see results and discussion
section).
For 5a, no more than ~2 g of 5 per 150 mL of toluene should
be used for the oxidation step because 5a showed a tendency to
precipitate due to its low solubility in toluene. Alternatively, 5a
1
phosphine oxides and their H2O2 adducts, the H and 13C NMR
data are included (see below, (c) in the experimental section).
Representative procedure for decomposition of 1a–5a to form 1–5
using wet silica. Toluene (30 mL) was combined with wet silica
(0.995 g, no pre-conditioning) and 2a (0.313 g, 1.240 mmol, as-
suming one molecule of H2O2 per one phosphine oxide molecules),
dissolved in 20 mL of toluene, was added to the slurry via pipette
with vigorous stirring. The mixture was stirred at RT for 18 h.
After the silica had been allowed to settle down for 1 h, the
supernatant was decanted. The solvent was removed in vacuo to
give 2 as a colorless sticky solid (0.146 g, 53.9% yield). The silica
was washed with CH2Cl2 (2 ¥ 20 mL) to recover residual surface-
adsorbed phosphine oxide 2 (0.089 g, 86.8% combined yield). The
decomposition of the adduct 2a to 2 was confirmed by 31P NMR.
The IR spectrum showed a significant amount of water remaining
hydrogen-bonded to 2 (see results and discussion section).
Representative procedure for decomposition of 1a–5a to form 1–5
using pre-dried silica. The H2O2 adduct of Bu3P O (2a, 0.537 g,
2.128 mmol assuming one molecule of H2O2 per one phosphine
oxide molecules) was added to a slurry of dry silica (1.356 g) in
toluene (30 mL) at RT. The sample was stirred overnight. An
IR spectrum showed a significant amount of H2O2 remaining
1750 | Dalton Trans., 2012, 41, 1742–1754
This journal is
The Royal Society of Chemistry 2012
©

View Online
in the material. Another 1.027 g of silica was added, and the
slurry was stirred for a further 4 d. IR analysis proved that all
H2O2 decomposed, and that the resulting water was quantitatively
removed from the phosphine oxide. The supernatant was decanted,
and the silica was washed with 20 mL of THF (more wash cycles
would improve the yield)16 to recover residual surface-adsorbed 2.
The washes were combined and the solvent removed in vacuum
to give 2 as a white polycrystalline solid (0.181 g, 39.0% overall
yield).
While the phosphine oxide 2 was exposed to air for a further 8 h, it
completely transformed into the oil Bu3P O·(H2O)0.84 (1.240 g),
and no mass increase was noted thereafter.
The result of this experiment is graphically displayed in Fig.
3, together with the similar data obtained for Me3P O (1). For
Oct3P O (3), Cy3P O (4), and Ph3P O (5) no significant mass
changes could be observed when they were exposed to humid air.
Representative procedure quantifying the loss of phosphine oxides
1–5 by adsorption on molecular sieves. The H2O2 adduct of
Bu3P O (2a, 1.010 g) was placed in a Schlenk flask and dissolved
in 50 mL of toluene. A 31P NMR spectrum of the solution
(0.5 mL aliquot) was recorded to determine the mole percent of
2 (incorporated in 2a) with respect to the standard ClPPh2 in a
capillary centered in the middle of the NMR tube (12.755 mole %).
Molecular sieves (2.206 g) in a tea bag (commercially obtained tea
bags are cut upen, stripped of their original contents, filled with the
molecular sieves and sealed with staples, see Fig. 13) were then fully
immersed in the solution, and the mixture was stirred at RT for
18 h. A 31P NMR spectrum of the supernatant solution (0.5 mL
aliquot) was again recorded, and the amount of 2 was 12.183
mol% as compared to the same standard ClPPh2. Therefore, 4.49
mol% of the phosphine oxide 2 had been lost from the solution by
adsorption on the molecular sieves. This corresponded to a loss of
2.03 mol% of 2 per g of molecular sieves. The loss of the phosphine
oxides 1 and 3–5 due to adsorption on the molecular sieves had
been determined to be less than 2 mol% per g of molecular sieves
under identical conditions.
Representative procedure for the decomposition of H2O2 in the
adducts 1a–5a with heating. The H2O2 adduct of Bu3P O (2a,
1.124 g) was placed in a Schlenk flask under nitrogen and dis◦solved
in toluene (100 mL). Then the solution was heated to 90 C for
18 h. After cooling to room temperature an aliquot (20 mL) was
removed and placed in a separate Schlenk flask. The solvent was
removed from this aliquot in vacuum, and a 31P NMR spectrum
showed that the H2O2 in the adduct 2a was not completely
decomposed. This result was confirmed by IR, which shows O–H
stretching bands for H2O and H2O2 bound to the phosphine oxide
2 (Fig. 10). The sample was redissolved in 20 mL of toluene and
added to the original flask for further heating to 95 ◦C for 18
h. After this round of heating the H2O2 was entirely decomposed,
but significant amounts of water remained bound to the phosphine
oxide 2 as shown by IR. Sodium sulfate (ca. 2 g) was added to the
solution, but this was ineffective in removing the residual water.
Representative procedure for decomposition of adducts with
MnO2. The H2O2 adduct of Bu3P O (2a, 0.110 g) was placed
into a 20 mL vial and dissolved in 5 mL of toluene. When MnO2
(0.101 g) was added, vigorous bubbling occurs due to the release of
oxygen during the decomposition of H2O2. The reaction mixture
was stirred for 1 h at RT to complete the decomposition process.
Then the mixture was filtered through a Pasteur pipette filled with
silica gel to remove the MnO2. The latter was a fine black powder
that is not entirely removable by filtration, as indicated by the dark
color of the filtrate. The same result was obtained with filter paper
while Celite improved the process slightly. Finally, the toluene was
removed in vacuo, and the grey product was kept under nitrogen.
31P NMR confirms the decomposition of the H2O2 in 2a, leading
to crude 2·(H2O)x (0.167 g).
Representative procedure for attempted decomposition of H2O2
adducts 1a–5a with Cu. The H2O2 adduct of Bu3P O (2a, 0.110
g) was placed into an NMR tube and dissolved in 0.6 mL of C6D6.
Small copper pieces, cut from a copper wire (ca. 0.1 g overall),
were added directly to the NMR tube. The solution was vigorously
shaken for 0.5 h, then left at RT for 18 h with occasional shaking.
After removal of the copper, the 31P NMR spectrum of the solution
showed that no decomposition of the H2O2 in 2a had occurred.
The same negative result was obtained when a solution of 2a was
heated to 40 ◦C for 18 h in the presence of copper pieces or when
the mixture was stirred for one week at room temperature.
(c) NMR data of the phosphine oxides and their H2O2 adducts
Trimethylphosphine oxide H2O2 adduct (1a). NMR (d,
1
2
CDCl3), 31P{ H} 43.23 (s); 1H 1.57 (d, J(31P-1H) = 13.0 Hz);
13C{ H} 17.56 (d, 1J(31P-13C) = 70.1 Hz).
1
1
Trimethylphosphine oxide (1). NMR (d, C6D6), 31P{ H} 32.65
1
2
1
1
(s); H 0.90 (d, J(31P-1H) = 12.8 Hz); 13C{ H} 17.99 (d, J(31P-
13C) = 69.0 Hz).
NMR (d, CDCl3), 31P{ H} 38.79 (s); H 1.51 (d, J(31P-1H) =
1
1
2
12.8 Hz); 13C{ H} 18.06 (d, 1J(31P-13C) = 70.0 Hz).
1
Tributylphosphine oxide H2O2 adduct (2a). NMR (d, C6D6),
1
31P{ H} 51.12 (s); 1H 1.47–1.34 (m, 12H, PCH2CH2), 1.22 (sextet,
3
3
6H, J(1H-1H) = 7.2 Hz, CH2CH3), 0.80 (t, 9H, J(1H-1H) = 7.3
Hz, CH3); 13C{ H} 27.56 (d, 1J(31P-13C) = 65.1 Hz, PCH2), 24.51
1
(d, 3J(31P-13C) = 14.0 Hz, CH2CH3), 24.00 (d, 2J(31P-13C) = 3.8 Hz,
PCH2CH2), 13.83 (s, CH3).
NMR (d, CDCl3), 31P{ H} 52.50 (s); 1H 1.64–1.56 (m, 6H,
1
PCH2), 1.47–1.38 (m, 6H, PCH2CH2), 1.32 (sextet, 6H, J(1H-
3
1H) = 7.2 Hz, CH2CH3), 0.82 (t, 9H, 3J(1H-1H) = 7.3 Hz,
CH3); 13C{ H} 26.90 (d, J(31P-13C) = 65.2 Hz, PCH2), 23.92 (d,
1
1
3J(31P-13C) = 14.4 Hz, CH2CH3), 23.32 (d, J(31P-13C) = 3.9 Hz,
2
PCH2CH2), 13.29 (s, CH3).
Representative procedure for measuring H2O uptake of the
phosphine oxides 1–5. Polycrystalline Bu3P O (2) (1.160 g,
5.313 mmol) was placed on a watch glass and exposed to air. The
combined mass of 2 and the watch glass was recorded in intervals
of several minutes during the first hour to determine the uptake
of H2O gravimetrically. As H2O was absorbed by 2, the crystals
became more “wet” in appearance and began to turn into an oil.
1
Tributylphosphine oxide (2). NMR (d, C6D6), 31P{ H} 43.66
1
(s); H 1.49–1.38 (m, 6H, PCH2CH2), 1.36–1.29 (m, 6H, PCH2),
1.23 (sextet, 6H, 3J(1H-1H) = 7.3 Hz, CH2CH3), 0.81 (t, 9H, 3J(1H-
1H) = 7.3 Hz, CH3); 13C{ H} 28.13 (d, J(31P-13C) = 65.0 Hz,
PCH2), 24.53 (d, 3J(31P-13C) = 13.6 Hz, CH2CH3), 24.17 (d, 2J(31P-
13C) = 3.8 Hz, PCH2CH2), 13.82 (s, CH3).
1
1
This journal is
The Royal Society of Chemistry 2012
Dalton Trans., 2012, 41, 1742–1754 | 1751
©

View Online
12.5 Hz, 3J(1Hax-1Heq) = 3.0 Hz, PCHax), 1.68–1.62 (m, 6H,
PCHCH2CHeq), 1.57–1.52 (m, 3H, PCH(CH2)2CHeq), 1.46–1.35
(m, 6H, PCHCHax), 1.13–1.00 (m, 9H, PCHCH2CHaxCHax);
1
NMR (d, CDCl3), 31P{ H} 48.57 (s); 1H 1.69–1.64 (m, 6H,
PCH2), 1.57–1.49 (m, 6H, PCH2CH2), 1.42 (sextet, 6H, J(1H-
3
1H) = 7.2 Hz, CH2CH3), 0.92 (t, 9H, 3J(1H-1H) = 7.3 Hz,
1
1
3
13C{ H} 35.49 (d, J(31P-13C) = 61.0 Hz, PCH), 27.13 (d, J(31P-
CH3); 13C{ H} 27.63 (d, J(31P-13C) = 65.0 Hz, PCH2), 24.28 (d,
1
1
2
13C) = 11.6 Hz, PCHCH2CH2), 26.57 (d, J(31P-13C) = 2.9 Hz,
3J(31P-13C) = 14.2 Hz, CH2CH3), 23.75 (d, J(31P-13C) = 3.9 Hz,
2
PCHCH2), 26.42 (d, 4J(31P-13C) = 1.3 Hz, PCH(CH2)2CH2).
PCH2CH2), 13.60 (s, CH3).
1
NMR (d, CDCl3), 31P{ H} 49.91 (s);1H 1.91 (br d, 6H, 2J(1H-
Trioctylphosphine oxide H2O2 adduct (3a). NMR (d, C6D6),
1H) = 12.8 Hz, PCHCHeq), 1.88–1.79 (m, 9H, PCHaxCH2CHeq),
1.75–1.68 (m, 3H, PCH(CH2)2CHeq), 1.48–1.36 (m, 6H,
31P{ H} 48.60 (s); H 1.52–1.37 (m, 12H, PCH2CH2), 1.28–1.12
1
1
(m, 30H, CH2CH3), 0.88 (t, 9H, 3J(1H-1H) = 7.2 Hz, CH3);
1
PCHCHax), 1.30–1.18 (m, 9H, PCHCH2CHaxCHax);13C{ H}
13C{ H} 32.29 (s, CH2CH2CH3), 31.58 (d, J(31P-13C) = 13.6 Hz,
P(CH2)2CH2), 29.62 (s, P(CH2)3CH2)‡, 29.59 (s, P(CH2)4CH2)‡,
28.08 (d, 1J(31P-13C) = 64.8 Hz, PCH2), 23.09 (s, CH2CH3), 22.10
(d, 2J(31P-13C) = 3.7 Hz, PCH2CH2), 14.36 (s, CH3).
1
3
35.37 (d, 1J(31P-13C) = 60.8 Hz, PCH), 26.93 (d, 3J(31P-13C) = 11.6
2
Hz, PCHCH2CH2), 26.34 (d, J(31P-13C) = 2.9 Hz, PCHCH2),
26.14 (d, 4J(31P-13C) = 1.4 Hz, PCH(CH2)2CH2).
NMR (d, CDCl3), 31P{ H} 50.01 (s); 1H 1.71–1.62 (m,
1
Triphenylphosphine oxide H2O2 adduct (5a). NMR (d, C6D6),
1
1
31P{ H} 27.52 (s); H 7.75–7.69 (m, 6H, Ho), 7.06–6.96 (m, 9H,
6H, PCH2), 1.57–1.47 (m, 6H, PCH2CH2), 1.41–1.32 (m, 6H,
PCH2CH2CH2), 1.32–1.28 (m, 24H, (CH2)4CH3), 0.86 (t, 9H,
Hm, Hp); 13C{ H} 134.12 (d, J(31P-13C) = 103.0 Hz, Ci), 132.40
(d, 2J(31P-13C) = 9.7 Hz, Co), 131.61 (d, 4J(31P-13C) = 2.8 Hz, Cp),
128.55 (d, 3J(31P-13C) = 12.1 Hz, Cm).
1
1
3J(1H-1H) = 7.2 Hz, CH3); 13C{ H} 31.61 (s, CH2CH2CH3), 30.98
1
(d, 3J(31P-13C) = 13.8 Hz, P(CH2)2CH2), 28.90 (s, P(CH2)3CH2)‡,
28.88 (s, P(CH2)4CH2)‡, 27.57 (d, J(31P-13C) = 64.6 Hz, PCH2),
1
NMR (d, CDCl3), 31P{ H} 30.15 (s); 1H 7.69–7.64 (m, 6H, Ho),
1
22.44 (s, CH2CH3), 21.46 (d, J(31P-13C) = 2.7 Hz, PCH2CH2),
2
7.57–7.52 (m, 3H, Hp), 7.48–7.44 (m, 6H, Hm); 13C{ H} 132.31 (d,
1
1J(31P-13C) = 102.8 Hz, Ci), 132.06 (d, 2J(31P-13C) = 10.0 Hz, Co),
13.87 (s, CH3).
131.96 (d, J(31P-13C) = 2.8 Hz, Cp), 128.49 (d, J(31P-13C) = 12.3
4
3
Trioctylphosphine oxide (3). NMR (d, C6D6), 31P{ H} 43.93
1
Hz, Cm).
1
(s); H 1.57–1.48 (m, 6H, PCH2CH2), 1.44–1.37 (m, 6H, PCH2),
1.33–1.19 (br m, 30H, (CH2)5CH3), 0.91 (t, 9H, 3J(1H-1H) =
1
Triphenylphosphine oxide (5). NMR (d, C6D6), 31P{ H} 25.16
7.1 Hz, CH3); 13C{ H} 32.24 (s, CH2CH2CH3), 31.57 (d, J(31P-
1
3
1
(s); 1H 7.77–7.71 (m, 6H, Ho), 7.06–6.96 (m, 9H, Hm, Hp); 13C{ H}
13C) = 13.1 Hz, P(CH2)2CH2), 29.58 (s, P(CH2)3CH2)‡, 29.57 (s,
134.00 (d, 1J(31P-13C) = 102.6 Hz, Ci), 132.04 (d, 2J(31P-13C) = 9.7
Hz, Co), 131.17 (d, 4J(31P-13C) = 2.7 Hz, Cp), 128.15 (d, 3J(31P-13C) =
11.9 Hz, Cm).
P(CH2)4CH2)‡, 28.49 (d, J(31P-13C) = 64.6 Hz, PCH2), 23.06 (s,
1
CH2CH3), 22.21 (d, J(31P-13C) = 3.8 Hz, PCH2CH2), 14.35 (s,
2
CH3).
1
NMR (d, CDCl3), 31P{ H} 29.10 (s); 1H 7.70–7.64 (m, 6H, Ho),
NMR (d, CDCl3), 31P{ H} 48.48 (s); 1H 1.67–1.59 (m,
1
1
7.57–7.52 (m, 3H, Hp), 7.49–7.43 (m, 6H, Hm); 13C{ H} 132.54 (d,
6H, PCH2), 1.58–1.48 (m, 6H, PCH2CH2), 1.40–1.32 (m, 6H,
PCH2CH2CH2), 1.32–1.18 (br m, 24H, (CH2)4CH3), 0.85 (t, 9H,
2
1J(31P-13C) = 103.4 Hz, Ci), 132.07 (d, J(31P-13C) = 9.9 Hz, Co),
4
3
131.89 (d, J(31P-13C) = 2.8 Hz, Cp), 128.46 (d, J(31P-13C) = 12.1
3J(1H-1H) = 7.1 Hz, CH3); 13C{ H} 31.78 (s, CH2CH2CH3), 31.39
1
Hz, Cm).
(d, 3J(31P-13C) = 13.6 Hz, P(CH2)2CH2), 29.09 (s, P(CH2)3CH2)‡,
29.04 (s, P(CH2)4CH2)‡, 27.95 (d, J(31P-13C) = 65.0 Hz, PCH2),
1
(d) Crystallography
22.61 (s, CH2CH3), 21.70 (d, J(31P-13C) = 3.7 Hz, PCH2CH2),
2
14.06 (s, CH3).
General procedure. Data were collected using a BRUKER
APEX2 X-ray diffractometer. Cell parameters were obtained from
60 frames using a 0.5◦ scan and refined using the number of
reflections given in sections A and B below. Integrated intensity
information for each reflection was obtained by reduction of the
data frames with the program APEX2.40 Lorentz, polarization,
crystal decay effects, and absorption corrections41 were applied.
The space group was determined from systematic reflection
conditions and statistical tests. The structure was solved by direct
methods using SHELXTL (SHELXS) and refined (weighted least
squares refinement on F2) using SHELXL-97.42 Non-hydrogen
atoms were refined with anisotropic thermal parameters. The
parameters were refined by weighted least squares refinement on
F2 to convergence.42
Tricyclohexylphosphine oxide H2O2 adduct (4a). NMR (d,
1
1
2
C6D6), 31P{ H} 50.28 (s); H 1.87 (br d, 6H, J(1Heq-1Hax) = 12.8
Hz, PCHCHeq), 1.74 (dtt, 3H, J(31P-1Hax) = 11.7 Hz, J(1Hax-
2
3
1Hax) = 12.6 Hz, J(1Hax-1Heq) = 3.0 Hz, PCHax), 1.67–1.61 (m,
3
6H, PCHCH2CHeq), 1.56–1.51 (m, 3H, PCH(CH2)2CHeq), 1.44–
1.33 (m, 6H, PCHCHax), 1.12–1.00 (m, 9H, PCHCH2CHaxCHax);
1
1
3
13C{ H} 35.41 (d, J(31P-13C) = 61.0 Hz, PCH), 27.11 (d, J(31P-
13C) = 11.9 Hz, PCHCH2CH2), 26.50 (d, J(31P-13C) = 2.9 Hz,
2
PCHCH2), 26.39 (d, 4J(31P-13C) = 1.3 Hz, PCH(CH2)2CH2).
NMR (d, CDCl3), 31P{ H} 51.53 (s); H 1.94–1.78 (m, 15H,
PCHaxCHeqCHeq), 1.74–1.67 (m, 3H, PCH(CH2)2CHeq), 1.47–
1.35 (m, 6H, PCHCHax), 1.29–1.17 (m, 9H, PCHCH2CHaxCHax);
1
1
1
1
3
13C{ H} 35.19 (d, J(31P-13C) = 60.8 Hz, PCH), 26.87 (d, J(31P-
A. A nearly saturated toluene solution of 2 was sealed in a
vial under N2 and kept first in a refrigerator (2 ◦C, 1 h) and then a
freezer (-17 ◦C, 12 h). The colorless needles were directly mounted
from the warmed sample. Hydrogen atoms were placed in idealized
positions and were refined using a riding model.
13C) = 11.7 Hz, PCHCH2CH2), 26.23 (d, J(31P-13C) = 2.9 Hz,
2
PCHCH2), 26.08 (d, 4J(31P-13C) = 1.4 Hz, PCH(CH2)2CH2).
Tricyclohexylphosphine oxide (4). NMR (d, C6D6), 31P{ H}
1
46.31 (s); 1H 1.91 (br d, 6H, 2J(1Heq-1Hax)
= 12.7 Hz,
PCHCHeq), 1.73 (dtt, 3H, 2J(31P-1Hax) = 11.5 Hz, 3J(1Hax-1Hax) =
Crystal data: C12H27OP, Mr = 218.31, monoclinic, space group
P21/c, T = 150(2)K, a = 9.326(6), b◦= 15.281(10), c = 10.136(7)
◦
◦
3
˚
˚
‡ assignments are interchangeable
A, a = 90 , b = 100.548(15) , g = 90 , V = 1420.1(16) A , Z = 4,
1752 | Dalton Trans., 2012, 41, 1742–1754
This journal is
The Royal Society of Chemistry 2012
©

View Online
Dc = 1.021 g cm-3, F000 = 488, q range for data collection 2.22◦ to
27.61◦, 15 057 reflections collected, independent reflections 3264
(Rint = 0.0594). Final GooF = 1.062, R indices based on reflections
with I > 2s(I) (refinement on F2) R1 = 0.0595, wR2 = 0.1490, R
indices (all data) R1 = 0.0885, wR2 = 0.1788.
Chem., 1980, 36, 974–977; (d) A. S. Kende, P. Delair and B. E. Blass,
Tetrahedron Lett., 1994, 35, 8123–8126.
10 (a) J. Blu¨mel, Inorg. Chem., 1994, 33, 5050–5058; (b) Y. Yang, B. Beele
and J. Blu¨mel, J. Am. Chem. Soc., 2008, 130, 3771–3773; (c) J. Sommer,
Y. Yang, D. Rambow and J. Blumel, Inorg. Chem., 2004, 43, 7561–
7563; (d) T. Posset, F. Rominger and J. Blu¨mel, Chem. Mater., 2005, 17,
586–595.
¨
B. A toluene solution of 4a was sealed in a vial under N2 and
kept first in a refrigerator (2 ◦C, 12 h) and then a freezer (-17 ◦C,
2 d). The colorless blocks, which dissolved upon warming, were
directly mounted from the cold sample. Carbon-bound hydrogen
atoms were placed in idealized positions. Oxygen-bound hydrogen
atoms were located from the difference Fourier map and refined
using a riding model.
11 (a) S. Hayashi, Chem. Lett., 2009, 38, 960–961; (b) R. Yerushalmi, J. C.
Ho, Z. Fan and A. Javey, Angew. Chem., Int. Ed., 2008, 47, 4440–4442;
(c) B. Alonso, I. Klur and D. Massiot, Chem. Commun., 2002, 804–805;
(d) L. Baltusis, J. S. Frye and G. E. Maciel, J. Am. Chem. Soc., 1986,
108, 7119–7120; (e) H.-M. Kao, P.-C. Chang, Y.-W. Liao, L.-P. Lee and
C.-H. Chien, Microporous Mesoporous Mater., 2008, 114, 352–364; (f) J.
Krˇ´ızˇ, J. Dybal, E. Makrl´ık, J. Budka and Petr Vanˇura, J. Phys. Chem.
A, 2009, 113, 5896–5905.
12 (a) V. W. L. Ng, M. K. Taylor, L. M. R. Hill, J. M. White and C. G.
Young, Eur. J. Inorg. Chem., 2010, 3261–3269; (b) W. Tsai, Y.-H. Liu,
S.-M. Peng and S.-T. Liu, J. Organomet. Chem., 2005, 690, 415–421.
13 (a) L. Li, Z. Wang, X. Song and S. Sun, Spectrochim. Acta, Part A, 2009,
72, 816–818; (b) M. T. Ben Dhia, M. A. M. K. Sanhoury, L. C. Owono
Owono and M. R. Khaddar, J. Mol. Struct., 2008, 892, 103–109; (c) E.
I. Tolis, K. A. Vallianatou, F. A. Andreadaki and I. D. Kostas, Appl.
Organomet. Chem., 2006, 20, 335–337.
¯
Crystal data: C18H35O3P, Mr = 330.43, triclinic, space group P1,
˚
˚
T = 110(◦2)K, a = 8.579(2), b = 9.584(3), c = 12.260(3) A, a =
◦
◦
3
95.445(3) , b = 97.065(3) , g = 112.824(3) , V = 910.8(4) A , Z =
2, Dc = 1.205 g cm-3, F000 = 364, q range for data collection 1.69◦ to
27.48◦, 10409 reflections collected, 4067 independent reflections
(Rint = 0.0169). Final GooF = 1.044, R indices based on reflections
with [I > 2s(I)] (refinement on F2) R1 = 0.0327, wR2 = 0.0843, R
indices (all data) R1 = 0.0376, wR2 = 0.0880.
14 (a) C. Popovici, I. Fernandez, P. Ona-Burgos, L. Roces, S. Garcia-
Granda and F. L. Ortiz, Dalton Trans., 2011, 40, 6691–6703; (b) K.
Miyata, Y. Hasegawa, Y. Kuramochi, T. Nakagawa, T. Yokoo and T.
Kawai, Eur. J. Inorg. Chem., 2009, 4777–4785.
15 (a) L. Ackermann, S. Barfu¨ßer, C. Kornhaaß and A. R. Kapdi, Org.
Lett., 2011, 13, 3082–3085; (b) L. Ackermann, A. R. Kapdi and C.
Schulzke, Org. Lett., 2010, 12, 2298–2301.
16 C. Hilliard, J. A. Gladysz and J. Blu¨mel, Chem. Mater., 2011, in
preparation.
17 M. M. Rauhut and H. A. Currier, J. Org. Chem., 1961, 26, 4626–4628.
18 (a) W. Peng and J. M. Shreeve, J. Fluorine Chem., 2005, 126, 1054–
1056; (b) D. Macikenas, E. Skrzypczak-Jankun and J. D. Protasiewicz,
J. Am. Chem. Soc., 1999, 121, 7164–7165; (c) S. Porcel, G. Bouhadir,
N. Saffon, L. Maron and D. Bourissou, Angew. Chem., Int. Ed., 2010,
49, 6186–6189; (d) G. C. Tucci, J. P. Donahue and R. H. Holm, Inorg.
Chem., 1998, 37, 1602–1608; (e) J. M. Berg and R. H. Holm, J. Am.
Chem. Soc., 1985, 107, 925–932.
Acknowledgements
This material is based upon work supported by the National
Science Foundation under CHE-0911207, CHE-0719267, and
CHE-0840464. We thank The Welch Foundation (A-1706) for
financial support, the ACS Division of Inorganic Chemistry, the
Texas A&M Office of Graduate Studies, the NSF LSAMP Bridge
to Doctorate program, and the NSF Graduate Research Program
(C. R. Hilliard) for fellowships.
19 (a) T. M. Duncan, A Compilation of Chemical Shift Anisotropies,
Farragut Press, Chicago, IL, 1990; (b) C. A. Fyfe, Solid-State NMR
for Chemists, C.F.C. Press, Guelph, Canada, 1983; (c) M. J. Duer,
Introduction to Solid-State NMR Spectroscopy, Blackwell Publishing,
Oxford, 2004; (d) S. Reinhard and J. Blu¨mel, Magn. Reson. Chem., 2003,
41, 406–416; (e) J. Mason, Ed., Multinuclear NMR, Plenum Press, New
York, London, 1987.
Notes and references
1 (a) J. A. Gladysz, Ed., Recoverable Catalysts and Reagents, Special Issue
of Chem. Rev., 2002, 102; (b) J. Blu¨mel, Coord. Chem. Rev., 2008, 252,
2410–2423; (c) F. R. Hartley, Supported Metal Complexes, D. Reidel
Publ. Co., Dordrecht, The Netherlands, 1985; (d) D. E. DeVos, I. F. J.
Vankelecom and P. A. Jacobs, ed., Chiral Catalyst Immobilization and
Recycling, Wiley-VCH, Weinheim, 2000; (e) G. Rothenberg, Catalysis:
Concepts and Green Applications, Wiley-VCH, Weinheim, 2008; (f) P.
Barbaro and F. Liguori, ed., Heterogenized Homogeneous Catalysts
for Fine Chemicals Production, Springer Science and Business Media,
Heidelberg, 2010.
2 (a) J. Guenther, J. Reibenspies and J. Blu¨mel, Adv. Synth. Catal., 2011,
353, 443–460; (b) T. Posset, J. Guenther, J. Pope, T. Oeser and J. Blu¨mel,
Chem. Commun., 2011, 47, 2059–2061; (c) B. Beele, J. Guenther, M.
Perera, M. Stach, T. Oeser and J. Blu¨mel, New J. Chem., 2010, 34,
2729–2731.
3 (a) J. Boutagy and R. Thomas, Chem. Rev., 1974, 74, 87–99; (b) C. J.
O’Brien, J. L. Tellez, Z. S. Nixon, L. J. Kang, A. L. Carter, S. R. Kunkel,
K. C. Przeworski and G. A. Chass, Angew. Chem., Int. Ed., 2009, 48,
6836–6839.
4 Y. G. Gololobov, I. N. Zhmurarova and L. F. Kasukhin, Tetrahedron,
1981, 37, 437–472.
20 J. Blu¨mel, M. Herker, W. Hiller and F. H. Ko¨hler, Organometallics,
1996, 15, 3474–3476.
21 (a) L. Bemi, H. C. Clark, J. A. Davies, C. A. Fyfe and R. E. Wasylishen,
J. Am. Chem. Soc., 1982, 104, 438–445; (b) N. J. Clayden, C. M. Dobson,
L.-Y. Lian and D. J. Smith, J. Magn. Reson., 1986, 69, 476–487; (c) J.
B. Robert and L. Wiesenfeld, Mol. Phys., 1981, 44, 319–327.
22 (a) A. L. Spek, Acta Crystallogr., Sect. C: Cryst. Struct. Commun.,
1987, 43, 1233–1235; (b) C. P. Brock, W. B. Schweizer and J. D. Dunitz,
J. Am. Chem. Soc., 1985, 107, 6964–6970.
23 M. Bogza, T. Oeser and J. Blu¨mel, J. Organomet. Chem., 2005, 690,
3383–3389.
24 V. I. Alekseev, Yu. V. Gatilov, T. M. Polyanskaya, V. V. Bakakin, Yu.
A. Dyadin and L. A. Gaponenko, Zh. Strukt. Khim., 1982, 23, 86–91.
25 J. A. Davies, S. Dutremez and A. A. Pinkerton, Inorg. Chem., 1991, 30,
2380–2387.
26 M. Mantina, A. C. Chamberlin, R. Valero, C. J. Cramer and D. G.
Truhlar, J. Phys. Chem. A, 2009, 113, 5806–5812. See Table 12.
27 W. R. Busing and H. A. Levy, J. Chem. Phys., 1965, 42, 3054–3059.
28 P. Larkin, Infrared and Raman Spectroscopy, Principles and Spectral
Interpretation, Academic Press, Elsevier, 2011.
29 (a) L. W. Daasch and D. C. Smith, J. Chem. Phys., 1951, 19, 22–25;
(b) G. Aksnes and L. J. Brudvik, Acta Chem. Scand., 1963, 17, 1616–
1622; (c) W. Schneider, W. Thiel and A. Komornicki, J. Phys. Chem.,
1988, 92, 5611–5619.
5 R. Appel, Angew. Chem., Int. Ed. Engl., 1975, 14, 801–811.
6 (a) T. E. Barder and S. L. Buchwald, J. Am. Chem. Soc., 2007, 129,
5096–5101; (b) S. A. Buckler, J. Am. Chem. Soc., 1962, 84, 3093–3097;
(c) H. D. Burkett, W. E. Hill and S. D. Worley, Phosphorus Sulfur Relat.
Elem., 1984, 20, 169–172.
7 S. Hayashi, Anal. Sci., 2009, 25, 133–136.
8 M. A. Beckett, D. S. Brassington, M. E. Light and M. B. Hursthouse,
J. Chem. Soc., Dalton Trans., 2001, 1768–1772.
30 G. V. Howell and R. L. Williams, J. Chem. Soc. A, 1968, 117–118.
31 M. A. Hasnat, M. M. Rahman, S. M. Borhanuddin, A. Siddiqua,
N. M. Bhadur and M. R. Karim, Catal. Commun., 2010, 12, 286–
291.
9 (a) R. D. Temple, Y. Tsuno and J. E. Leffler, J. Org. Chem., 1963, 28,
2495; (b) D. B. Copley, F. Fairbrother, J. R. Miller and A. Thompson,
Proc. Chem. Soc., London, 1964, 300–301; (c) D. Thierbach, F. Huber
and H. Preut, Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst.
This journal is
The Royal Society of Chemistry 2012
Dalton Trans., 2012, 41, 1742–1754 | 1753
©

View Online
32 (a) R. K. Iler, The Chemistry of Silica, John Wiley, New York, 1979;
(b) R. P. W. Scott, Silica Gel and Bonded Phases, John Wiley and Sons,
New York, 1993.
Soba, F. Rominger and J. Blu¨mel, Adv. Synth. Catal., 2003, 345, 589–
602.
38 T. Nishimura, T. Onoue, K. Ohe and S. Uemura, J. Org. Chem., 1999,
33 A. B. Burg and W. E. McKee, J. Am. Chem. Soc., 1951, 73, 4590–
64, 6750–6755.
4591.
39 R. Fetouaki, A. Seifert, M. Bogza, T. Oeser and J. Blu¨mel, Inorg. Chim.
Acta, 2006, 359, 4865–4873.
34 L. I. Kudryavtseva, Zh. Obs. Khim., 1990, 60, 74–79.
35 N. G. Feshchenko, I. K. Mazepa, Zh. K. Gorbatenko, Yu. P. Mack-
ovetskii, V. P. Kukhar and A. V. Kirsanov, Zh. Obs. Khim., 1969, 39,
1219–1223.
36 A. Burger and N. D. Dawson, J. Chem. Soc., 1951, 16, 1250–1254.
37 (a) C. Merckle and J. Blu¨mel, Chem. Mater., 2001, 13, 3617–3623; (b) C.
Merckle and J. Blu¨mel, Top. Catal., 2005, 34, 5–15; (c) S. Reinhard, P.
40 APEX2, Program for Data Collection on Area Detectors, BRUKER
AXS Inc., 5465 East Cheryl Parkway, Madison, WI 53711-5373, USA.
41 G. M. Sheldrick, SADABS, Program for Absorption Correction of Area
Detector Frames, BRUKER AXS Inc., 5465 East Cheryl Parkway,
Madison, WI 53711-5373, USA.
42 G. M. Sheldrick, Acta Crystallogr., Sect. A, 2008, 64, 112–122.
1754 | Dalton Trans., 2012, 41, 1742–1754
This journal is
The Royal Society of Chemistry 2012
©