Triflimide-catalysed sigmatropic rearrangement of N-allylhydrazones as an example of a traceless bond construction

Article abstract of DOI:10.1038/nchem.576

The recognition of structural elements (that is, retrons) that signal the application of specific chemical transformations is a key cognitive event in the design of synthetic routes to complex molecules. Reactions that produce compounds without an easily identifiable retron, by way of either substantial structural rearrangement or loss of the atoms required for the reaction to proceed, are significantly more difficult to apply during retrosynthetic planning, yet allow for non-traditional pathways that may facilitate efficient acquisition of the target molecule. We have developed a triflimide (Tf 2 NH)-catalysed rearrangement of N-allylhydrazones that allows for the generation of a sigma bond between two unfunctionalized sp 3 carbons in such a way that no clear retron for the reaction remains. This new traceless bond construction displays a broad substrate profile and should open avenues for synthesizing complex molecules using non-traditional disconnections.

Full text of DOI:10.1038/nchem.576

ARTICLES
PUBLISHED ONLINE: 28 FEBRUARY 2010 | DOI: 10.1038/NCHEM.576
Triflimide-catalysed sigmatropic rearrangement of
N-allylhydrazones as an example of a traceless
bond construction
Devon A. Mundal, Christopher T. Avetta Jr and Regan J. Thomson
*
The recognition of structural elements (that is, retrons) that signal the application of specific chemical transformations is
a key cognitive event in the design of synthetic routes to complex molecules. Reactions that produce compounds without
an easily identifiable retron, by way of either substantial structural rearrangement or loss of the atoms required for the
reaction to proceed, are significantly more difficult to apply during retrosynthetic planning, yet allow for non-traditional
pathways that may facilitate efficient acquisition of the target molecule. We have developed a triflimide (Tf2NH)-catalysed
rearrangement of N-allylhydrazones that allows for the generation of a sigma bond between two unfunctionalized sp3
carbons in such a way that no clear retron for the reaction remains. This new ‘traceless’ bond construction displays a
broad substrate profile and should open avenues for synthesizing complex molecules using non-traditional disconnections.
he [3,3] sigmatropic rearrangements are a particularly power- both a strategic and practical level, it has remained undeveloped
ful class of reactions for the synthesis of complex molecules. for over 35 years due to a limited substrate scope, the requirement
T
For example, the Claisen rearrangement is a powerful fragment for high temperatures and low yields. In 2006, we initiated a research
coupling reaction that can create an otherwise difficult-to-form programme aimed at unlocking the potential of N-allylhydra-
carbon–carbon bond by the initial construction of a comparatively zones10,11. We now report the successful development of a triflimide
easily forged carbon–oxygen bond1. Because sigmatropic rearrange- (Tf2NH)-catalysed variant of the Stevens [3,3] sigmatropic
ments typically involve significant structural reorganization and rearrangement (4 ! 5, Fig. 1c) that displays a significantly
generate functional groups not present in the starting material,
they do not often produce clearly identifiable retrons2 that signal
for their application in the forward sense during the planning
stages of any synthetic endeavour. Recognition and implementation
of such reactions can, however, lead to highly inventive and non-tra-
ditional approaches to complex molecules, such as the spectacular
demonstration by Shair and colleagues in their synthesis of (þ)-
CP-263,114 (ref. 3). It was within the context of devising unique
fragment couplings to facilitate synthesis along non-traditional
pathways that we became interested in the sigmatropic rearrange-
ment of N-allylhydrazones. These compounds undergo a unique
[3,3] sigmatropic rearrangement followed by dinitrogen extrusion
to generate products that bear no resemblance to the starting
materials, a process that we term ‘traceless bond construction’.
The general concept of a traceless bond construction is outlined
in Fig. 1a, in which two functional groups are shown to react to
form a new compound that undergoes a subsequent reaction with
loss of the remaining functional group. In essence, a new sigma
bond is formed at a location where there is no remaining function-
ality. In 1973, Stevens and colleagues reported the low yield and
limited thermal rearrangement of N-allylhydrazones (Fig. 1b), a
reaction that has seen virtually no use by synthetic chemists4.
Given that the generation of hydrazones is typically a high-yielding
and facile process, we considered the two-step transformation (that
is, 1 ! 3) to have significant potential for carbon–carbon bond
fragment coupling. As an additional benefit, the usefulness of the
coupling reaction manifests itself further in that there are no post-
reaction functional groups that need to be manipulated as part of
an ongoing synthesis. This general type of reactivity is common
for many hydrazone-based transformations5–9. Despite the signifi-
cant potential impact of this reaction on synthetic chemistry at
a
R2
FG
R2
R1
R1
R2
FG
FG
R1
FG
H
N
b
H
N
H2N
300 °C
–N2
N
R1CHO
R1
–H2O
R1
1
2
3
• Low yields
• Limited scope
R1 = aryl
c
Boc
N
R3
R2
H2N
Boc
N
Tf2NH
(10 mol%)
R3
R2
R3
R2
N
R1CHO
125 °C
–N2, –CO2
–C4H8
R1
–H2O
R1
1
5
4
R1 = aryl or alkyl
Figure 1 | Traceless bond constructions of N-allylhydrazones. a, General
concept of a traceless bond construction, in which functional groups (FG)
present within the starting material are excised during the course of carbon–
carbon bond formation. b, Original rearrangement developed by Stevens and
colleagues in 1973, which required very harsh conditions and had a low
yield. c, Triflimide-catalysed rearrangement, demonstrating its wider scope
and higher yields. Tf, trifluoromethanesulfonyl; Boc, t-butyl carbamate.
Department of Chemistry, Northwestern University, Evanston, Illinois, 60208 USA. e-mail: r-thomson@northwestern.edu
*
294
NATURE CHEMISTRY | VOL 2 | APRIL 2010 | www.nature.com/naturechemistry
© 2010 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.576
ARTICLES
Table 1 | Substrate scope of triflimide-catalysed traceless bond construction.
Entry
Substrate (4)
Product (5)
Yield (%)*
E:Z†
Boc
i-Bu
N
i-Bu
1
4a R ¼ H
4b R ¼ i-Pr
4c R ¼ Br
5a R ¼ H
5b R ¼ i-Pr
5c R ¼ Br
66
74
75
.20:1
.20:1
.20:1
N
R
R
Boc
i-Bu
N
i-Bu
2
3
4d
5d
49
75
.20:1
.20:1
N
N
N
Boc
i-Bu
N
i-Bu
4e
5e
N
O
O
Boc
N
i-Bu
i-Bu
i-Bu
R
i-Bu
4
5
6
7
4f
5f
60
6:1
N
N
Me
(
(
)
)
Me
8
(
(
)
)
8
Boc
N
i-Bu
4g R ¼ TBDPS
4h R ¼ Bn
5g R ¼ TBDPS
5h R ¼ Bn
57
63
12:1
5:1
RO
RO
4
4
Boc
Me
N
i-Bu
4i
5i
65
6:1
Me
N
i-Pr
(
)
i-Pr
3
(
)
3
Boc
N
R
4j R ¼ H
4k R ¼ Me
4l R ¼ Et
4m R ¼ n-Pr
4n R ¼ i-Bu
4o R ¼ i-Pr
4p
5j R ¼ H
5k R ¼ Me
5l R ¼ Et
5m R ¼ n-Pr
5n R ¼ i-Bu
5o R ¼ i-Pr
5p
0
53
54
69
70
63
51
—
N
4:1
5:1
8:1
7:1
8:1
—
Boc
N
8
9
Me
(
)
8
N
Me
Me
(
)
8
Me
Me
Me
Boc
4q R ¼ Me
4r R ¼ OBn
5q R ¼ Me
5r R ¼ OBn
55
51
12:1
17:1
R
N
R
N
1:1
Standard conditions: 0.5 mmol 4, 10 mol% Tf2NH, 0.05 M diglyme, 125 8C, 0.25–2 h. *Isolated yield. †Determined by 1H NMR spectroscopy or GC.
Boc, t-butyl carbamate; Me, methyl; Et, ethyl; Bu, butyl; Pr, propyl; Bn, benzyl; TBDPS, t-butyldiphenylsilyl; Tf, trifluoromethanesulfonyl.
improved substrate scope and holds broad promise for applications has found significant use as a catalyst for a wide range of
in synthesis.
transformations12.
The reaction proved tolerant of hydrazone substrates derived
from aromatic aldehydes (Table 1, entry 1), including heterocyclic
Results
During our initial studies10, we discovered that treating N-allylhydra- systems (entries 2 and 3), providing the coupled products in good
zones such as 4 with stoichiometric copper(II) chloride (CuCl2) yield and in a .20:1 ratio of E- to Z-isomers. Our earlier work on
resulted in a previously unknown carbon–carbon and carbon–chlor- hydrazone rearrangements only worked well for hydrazones
ine bond forming reaction. Although this discovery was interesting, derived from aromatic aldehydes; aliphatic systems resulted in mul-
we wished to return to our original goal of establishing a useful tiple decomposition products. Given this background, we were
variant of the original Stevens process. An investigation of metal anxious to demonstrate a wider substrate scope for the new trans-
complexes other than CuCl2 provided few useful outcomes, but formation. The hydrazone derived from decanal (4f) was therefore
revealed that hafnium(IV) trifluoromethanesulfonate produced prepared and exposed to the triflimide reaction conditions (entry 4).
some of the desired product at 125 8C. Trifluoromethanesulfonic Smooth rearrangement ensued, and the desired adduct 5f was iso-
acid alone was more effective than the metal triflate, and after lated in a yield of 60% (6:1, E:Z). The presence of the t-butyl carba-
further investigations into Brønsted acids we found that 10 mol% tri- mate (Boc) group was crucial to the success of this rearrangement;
flimide in diethylene glycol dimethyl ether (diglyme) at 125 8C pro- aliphatic hydrazones lacking the Boc-group proved difficult to
duced the best yields across a range of substrates (Table 1). Triflimide prepare with high purity and were unstable in the presence of air.
NATURE CHEMISTRY | VOL 2 | APRIL 2010 | www.nature.com/naturechemistry
© 2010 Macmillan Publishers Limited. All rights reserved.
295

NATURE CHEMISTRY DOI: 10.1038/NCHEM.576
ARTICLES
O
A more pressing issue to address was whether a substrate derived
from an enantio-enriched aldehyde having an a-stereocentre would
racemize under the acidic reaction conditions of the sigmatropic
rearrangement (Fig. 2). To answer this question, we prepared
chiral enantio-enriched hydrazone 8 from the corresponding
aldehyde 6 (93:7 e.r.) (ref. 13). The use of a racemic hydrazine
fragment (7) meant that hydrazone 8 was formed as a mixture of
diastereomers. Exposure of this mixture to our standard conditions
for the triflimide-catalysed process resulted in the formation of the
desired product with a yield of 65% and minimal erosion of optical
purity over the two steps from aldehyde 6. Finally, we prepared a
hydrazone with a methyl-bearing stereocentre on each component
(that is, 11). This substrate underwent the desired rearrangement
with a yield of 63% (.20:1, E:Z; 91:9 d.r.), providing a new com-
pound with two stereogenic centres positioned in a 1,6-relationship.
i-Pr
H
Me
Boc
Boc
Me
N
i-Bu
N
H2N
H2N
Et
6 93:7 e.r.
7
10
Me
Boc
Boc
N
i-Bu
N
N
N
Et
i-Pr
i-Pr
Me
Me
8 88%
11 92%
–N2, –CO2
–N2, –CO2
–C4H8
Tf2NH (10 mol%)
125 oC
–C4H8
Me
Discussion
i-Bu
Two reasonable pathways may be considered for this acid-catalysed
rearrangement (Fig. 3). The first (Path A) proceeds with initial
loss of the Boc-group to generate a protonated species such as II,
which would undergo bond reorganization to produce III and
eventually the final product, via diazene IV, which would readily
extrude dinitrogen14. Given the poor results obtained for hydra-
zones lacking the Boc-group, we speculate that carbon–carbon
bond formation might precede decomposition of the Boc-group
(Path B). In this alternative pathway, rearrangement of protonated
hydrazone V would produce VI, which would lose the Boc-group
to generate diazene IV. Formation of the (E)-alkene as the major
isomer in all cases is most likely due to the reaction proceeding
through a chair-like cyclic transition state similar to that postulated
for our previously reported N-allylhydrazone-based transform-
ations10,11. We speculated that perhaps the olefin ratios observed
were due to a thermodynamic equilibration of the alkene isomers
under the acidic reaction conditions, but observed no such
Et
i-Pr
i-Pr
Me
Me
9 65%, 5:1 E:Z
12 63%, >20:1 E:Z
87:13 e.r.
91:9 d.r.
Figure 2 | Impact of resident stereocentres. The extent to which hydrazones
possessing a stereocentre might undergo scrambling under the reaction
conditions was examined. Hydrazones 8 and 11 undergo traceless bond
construction to give products 9 and 12, respectively. In each case, the loss of
stereochemical integrity of the potentially labile methyl stereocentre was
shown to be minimal, indicating that this chemistry could find use in the
synthesis of complex acyclic molecules. Boc, t-butyl carbamate; Me, methyl;
Et, ethyl; Bu, butyl; Tf, trifluoromethanesulfonyl.
On the other hand, aryl aldehyde-derived hydrazones lacking the isomerization when pure (Z)-2-methylhexadec-4-ene was exposed
Boc-group were prepared readily, but resulted in multiple products to the reaction conditions.
and low yields when exposed to 10 mol% triflimide at
elevated temperatures.
At a strategic level, this new bond construction opens new
avenues for synthesis using disconnections that are not traditionally
Functionalized substrates were tolerant of the reaction con- applied by synthetic chemists. For the molecule indicated in Fig. 4,
ditions; hydrazones containing a primary t-butyldiphenylsilyl the most obvious disconnection point is at the p-bond (a) in the
(TBDPS) or benzyl (Bn) ether resulted in 57% and 63% yields, form of an olefination transform. Similarly, if one chose to discon-
respectively (Table 1, entry 5). Use of the more acid-labile p-meth- nect adjacent to the sp2 carbon (b), a metal-based coupling could
oxybenzyl (PMB) ether resulted in significant decomposition, indi-
cating that the current chemistry remains limited to those substrates
Path A
Path B
containing functional groups that are stable under acidic conditions.
b-Substitution on the aldehyde component was well tolerated (entry 6).
Using hydrazones derived from cyclohexanecarboxaldehyde, we
explored the use of various N-allylhydrazide fragments (entry 7).
The parent allyl-based hydrazone 4j underwent significant
decomposition and failed to provide the desired product (5j, R ¼ H).
Alkyl substitution at the 1-position of the hydrazide was compatible
with the conditions (entry 7, 5k–o, R ¼ Me, Et, n-Pr, i-Bu, i-Pr),
providing the (E)-1,2-disubstituted adducts between 53 and 70%
yield. Methyl substitution at the 2-position of the hydrazide
fragment produced the corresponding 1,1-disubstituted alkene 5p
in 51% yield (entry 8). We next synthesized the non-racemic
N-allylhydrazides derived from the (S)-2-methylbutan-1-ol and
H
N
Boc
N
Boc
N
–CO2
–C4H8
H
i-Bu
i-Bu
H
i-Bu
N
N
N
II
I
V
R
R
R
R
Tf2NH
Tf2NH
[3,3]
[3,3]
H
N
H
N
Boc
N
H
i-Bu
i-Bu
H
i-Bu
N
N
N
Tf2N
Tf2N
III
IV
VI
–CO2
–C4H8
R
R
–N2
Product
the Roche ester (that is, methyl (S)-3-hydroxy-2-methylpropionate), Figure 3 | Possible reaction pathways for traceless bond construction. The
which were each condensed with cyclohexanecarboxaldehyde to working hypothesis for plausible reaction pathways is shown. In Path A, the
provide hydrazones 4q and 4r (Table 1, entry 9). The hydrazones Boc-group is lost first to produce II, which then undergoes C–C bond
thus formed were mixtures of diastereomers at the nitrogen- rearrangement to generate III, and then the product via intermediate diazene
bearing stereocentre, but this proved to be inconsequential, IV. In Path B, C–C bond formation precedes loss of the Boc-group (that is, V
because each rearranged in greater than 50% yield (ꢀ12:1, E:Z). to VI to IV). Reactions involving hydrazones lacking the Boc-group produced
These last two entries clearly demonstrate the usefulness of this poor results, providing some circumstantial evidence that Path B may be
new traceless bond construction for the stereoselective synthesis of favoured, but further studies are required to establish this definitively. Boc,
alkenes having allylic stereogenic centres.
t-butyl carbamate; Bu, butyl; Tf, trifluoromethanesulfonyl.
296
NATURE CHEMISTRY | VOL 2 | APRIL 2010 | www.nature.com/naturechemistry
© 2010 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.576
ARTICLES
Traditional disconnections
Non-traditional disconnection
Me
a
• Julia and Wittig-type reactions
• Cross-metathesis
• Allylic electrophile displacement
• Allyl-metal SN2 reaction
• sp3–sp3 metal-mediated cross-coupling
Et
c
a
R
c
b
• Vinyl anion SN2 reaction
• sp3–sp2 metal-mediated
cross-coupling
b
• Hydrazone-based traceless bond
construction
Me
Figure 4 | Analysis of strategic bond disconnections: new avenues for synthesis using a non-traditional disconnection. Analysis of the fictive target shown
reveals three major points of disconnection. Bond a lends itself to construction by means of a Julia- or Wittig-type olefination, or through olefin cross-
metathesis. Carbon bond b may potentially be formed by an SN2 reaction or by the more modern sp3–sp2 metal-mediated cross-coupling. In general, bonds
a and b are typical choices for synthetic chemists, whereas disconnection at bond c is not as common due to less developed or unreliable methods. The
hydrazone-based traceless bond construction enables convergent fragment assembly at bond c and thus opens avenues for alternative syntheses along
non-traditional paths. Me, methyl; Et, ethyl.
occur. Disconnection at the allylic position (c) would traditionally
be a poorer strategic choice, because formation of this bond using
either an allyl electrophile or an allyl nucleophile raises issues of
regiocontrol and product alkene geometry. Recent methods for
transition metal-mediated sp3–sp3 cross-coupling do, however,
offer a solution to this problem for many cases15–17. The reaction
we have developed also enables such a disconnection to be made
with a high level of confidence; the condensation step is straight-
forward, and the sigmatropic rearrangement is completely regio-
selective and displays good levels of stereoselectivity. Furthermore,
the concept of traceless reactions represents a powerful approach
to the design of complex molecules and provides enhanced
options for the synthetic chemist engaged in target directed syn-
thesis. Future efforts in this area will be focused on the requirement
for high temperatures and a strong acid in the reaction, enabling
a significantly enhanced substrate scope, particularly regarding
substrates bearing acid-sensitive functional groups.
References
1. Castro, A. M. M. Claisen rearrangement over the past nine decades. Chem. Rev.
104, 2939–3002 (2004).
2. Corey, E. J. & Cheng, X.-M. The Logic of Chemical Synthesis (Wiley, 1989).
3. Chen, C., Layton, M. E., Sheehan, S. M. & Shair, M. D. Synthesis of (þ)-CP-
263,114. J. Am. Chem. Soc. 122, 7424–7425 (2000).
4. Stevens, R. V., McEntire, E. E., Barnett, W. E. & Wenkert, E. Thermally induced
[3,3] sigmatropic rearrangement of N-allylhydrazones. Chem. Commun.
662–663 (1973).
5. Vedejs, E. & Stolle, W. T. Reductive alkylation of aldehyde tosylhydrazones with
organolithium reagents. Tetrahedron Lett. 18, 135–138 (1977).
6. Bertz, S. H. The preparation of hindered cuprates from aldehyde
tosylhydrazones. Tetrahedron Lett. 21, 3151–3154 (1980).
7. Myers, A. G. & Kukkola, P. J. Stereoselective synthesis of olefins from silylated
sulfonylhydrazones. J. Am. Chem. Soc. 112, 8208–8210 (1990).
8. Myers, A. G. & Movassaghi, M. Highly efficient methodology for the reductive
coupling of aldehyde tosylhydrazones with alkyllithium reagents. J. Am. Chem.
Soc. 120, 8891–8892 (1998).
´
´
9. Barluenga, J., Tomas-Gamasa, M., Aznar, F. & Valdes, C. Metal-free carbon–
carbon bond-forming reductive coupling between boronic acids and
tosylhydrazones. Nature Chem. 1, 494–499 (2009).
The reaction we have developed is applicable to a wide range of
substrates, including aromatic and aliphatic hydrazones, and those
possessing stereogenic centres. From a synthetic standpoint, the
transformation offers a unique means for constructing a s-bond
between two unfunctionalized sp3 carbons, while simultaneously
generating a stereo-defined alkene. Moreover, because this reaction
forms products that do not contain an obvious retron, its synthetic
application has the potential to lead to novel and creative pathways
that are not immediately obvious. Continued research in this area
should unveil more such traceless bond constructions and facilitate
synthesis along non-traditional routes.
10. Mundal, D. A., Lee, J. J. & Thomson, R. J. Tandem carbon–carbon and carbon–
chlorine bond formation by Cu(II) chloride-promoted [3,3] sigmatropic
rearrangement of N-allylhydrazones. J. Am. Chem. Soc. 130, 1148–1149 (2008).
11. Mundal, D. A., Lutz, K. E. & Thomson, R. J. Stereoselective synthesis of dienes
from N-allylhydrazones. Org. Lett. 11, 465–468 (2009).
12. Akiyama, T. Stronger Brønsted acids. Chem. Rev. 107, 5744–5758 (2007).
13. Taber, D. F. & Houze, J. B. The 2-hydroxycitronellols, convenient chirons for
natural products synthesis. J. Org. Chem. 59, 4004–4006 (1994).
14. Myers, A. G., Movassaghi, M. & Zheng, B. Mechanistic studies of the free-radical
fragmentation of monoalkyl diazenes. Tetrahedron Lett. 38, 6569–6572 (1997).
15. Saito, B & Fu, G. C. Alkyl–alkyl Suzuki cross-couplings of unactivated secondary
alkyl halides at room temperature. J. Am. Chem. Soc. 129, 9602–9603 (2007).
16. Saito, B & Fu, G. C. Enantioselective alkyl–alkyl Suzuki cross-couplings of
unactivated homobenzylic halides. J. Am. Chem. Soc. 130, 6694–6695 (2008).
17. Son, S. & Fu, G. C Nickel-catalyzed asymmetric Negishi cross-couplings of
secondary allylic chlorides with alkylzincs. J. Am. Chem. Soc. 130,
2756–2757 (2008).
Methods
General procedure for the traceless bond construction. Triflimide (0.25 M in
CH2Cl2, 200 ml, 0.05 mmol) was added to a flame-dried 25-ml round-bottomed
flask containing a magnetic stir bar under a N2 atmosphere. The CH2Cl2 was
removed by vigorously purging the flask with N2 until dry crystalline triflimide was
visible in the reaction flask. The N2 flow rate was decreased and diglyme (5.0 ml) was
added to the reaction flask. A solution of the requisite hydrazone (0.50 mmol) in
diglyme (2.5 ml þ 2.5 ml rinse) was then added to this, via a cannula, at room
temperature under a N2 atmosphere. The reaction flask was fitted with a reflux
condenser and purged with N2. The reaction was then stirred at 125 8C until deemed
complete by thin layer chromatography (30% EtOAc/hexanes, p-anisaldehyde
stain). After cooling to room temperature the reaction was washed into a separatory
funnel with hexanes (20 ml) and diluted with sat. NaHCO3 (15 ml) and H2O
(100 ml). The aqueous phase was removed and the organic phase was washed with
sat. NaCl (15 ml) and H2O (100 ml). The combined aqueous phases were extracted
with hexanes (10 ml) and the combined organic phases were dried over Na2SO4 and
concentrated to produce a yellow oil. Flash column chromatography on silica gel
resulted in the alkene product.
Acknowledgements
This work was supported by the National Science Foundation (CHE0845063), the donors
of the American Chemical Society Petroleum Research Fund (grant no. 46778-G) and
Northwestern University.
Author contributions
R.J.T. conceived the idea and wrote the manuscript. D.A.M. and C.T.A. performed the
experiments. All the authors analysed the data, contributed to discussions and edited
the manuscript.
Additional information
The authors declare no competing financial interests. Supplementary information and
chemical compound information accompany this paper at www.nature.com/
naturechemistry. Reprints and permission information is available online at http://npg.nature.
com/reprintsandpermissions/. Correspondence and requests for materials should be
addressed to R.J.T.
Received 17 September 2009; accepted 19 January 2010;
published online 28 February 2010
NATURE CHEMISTRY | VOL 2 | APRIL 2010 | www.nature.com/naturechemistry
297
© 2010 Macmillan Publishers Limited. All rights reserved.