Catalytic dehydrogenation of cyclooctane and triethylamine using aliphatic iridium pincer complexes Dedicated to John Bercaw on the occasion of his 70th birthday

Article abstract of DOI:10.1016/j.poly.2014.06.016

The majority of the known pincer iridium based catalysts for dehydrogenation of alkanes has arene-based backbones. Here, the catalytic activity of aliphatic iridium pincer complexes, viz. the cyclohexane-based phosphine complex (PCyP)IrHCl (4) (PCyP = {cis-1,3-bis-[(di-tert-butylphosphino)methyl]cyclohexane}^-) and the 2-methylpropane-based phosphinite complex (POCOP)IrHCl (5) (POCOP = 1,3-bis-(di-tert-butylphosphinito)-2-methylpropane^-), in dehydrogenation of cyclooctane and triethylamine was studied. They give TONs that are in the range of 0-200. In addition, improved procedures for synthesis and metallation of the PCyP ligand (3) are presented.

Full text of DOI:10.1016/j.poly.2014.06.016

Polyhedron xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Polyhedron
journal homepage: www.elsevier.com/locate/poly
Catalytic dehydrogenation of cyclooctane and triethylamine using
aliphatic iridium pincer complexes
⇑
Alexey V. Polukeev, Roman Gritcenko, Klara J. Jonasson, Ola F. Wendt
Centre for Analysis and Synthesis, Department of Chemistry, PO Box 124, 22100 Lund, Sweden
a r t i c l e i n f o
a b s t r a c t
Article history:
The majority of the known pincer iridium based catalysts for dehydrogenation of alkanes has arene-based
backbones. Here, the catalytic activity of aliphatic iridium pincer complexes, viz. the cyclohexane-based
phosphine complex (PCyP)IrHCl (4) (PCyP = {cis-1,3-bis-[(di-tert-butylphosphino)methyl]cyclohexane}À)
and the 2-methylpropane-based phosphinite complex (POCOP)IrHCl (5) (POCOP = 1,3-bis-(di-tert-butyl-
phosphinito)-2-methylpropaneÀ), in dehydrogenation of cyclooctane and triethylamine was studied.
They give TONs that are in the range of 0–200. In addition, improved procedures for synthesis and met-
allation of the PCyP ligand (3) are presented.
Received 8 April 2014
Accepted 6 June 2014
Available online xxxx
Dedicated to John Bercaw on the occasion of
his 70th birthday
Ó 2014 Elsevier Ltd. All rights reserved.
Keywords:
Dehydrogenation
Iridium
Pincer complexes
Homogeneous catalysis
Hydrogen transfer
1. Introduction
2. Experimental
The conversion of cheap and abundant, but relatively inert,
alkanes into more reactive olefins is an industrially important
process [1]. Significant efforts have been made to develop cata-
lysts that provide selective alkane dehydrogenation under mild
conditions [2–5], and iridium PCP pincer complexes have been
dominating the field for the last two decades [6]. The majority
of the ligands studied are composed of an arene backbone, e.g.
benzene [7–13], anthracene [12,14] or a 7-6-7 fused-ring based
systems (Fig. 1) [15]. In contrast, iridium pincer complexes with
aliphatic backbones have received relatively little attention and
their performance in catalytic dehydrogenation of alkanes has
not been reported. To fill this gap, we here present a study on
the catalytic activity of a cyclohexane-based (PCyP)IrHCl com-
plex (4) (PCyP = {cis-1,3-bis-[(di-tert-butylphosphino)methyl]
cyclohexane}À) [16], and a 2-methylpropane-based phosphinite
complex (POCOP)IrHCl (5) (POCOP = 1,3-bis-(di-tert-butylphos-
phinito)-2-methylpropaneÀ) [17]. We also report on an improved
synthesis of compound 4.
2.1. General experimental procedures
All manipulations were carried out under an Ar or N2 atmo-
sphere using standard Schlenk or glovebox techniques, unless
otherwise stated. All catalytic experiments were performed under
Ar. Hydrocarbon solvents were degassed and distilled from Na/
benzophenone. Chlorinated solvents and triethylamine were
degassed and distilled from CaH2. Commercially available reagents
were used as received. NMR spectra were recorded on a Varian
Unity INOVA 500 MHz instrument and referenced to the residual
solvent peaks for 1H and 13C measurements, and to external 85%
H3PO4 for 31P measurements. Elemental analyses were performed
by H. Kolbe Microanalytisches Laboratorium, Mülheim an der Ruhr,
Germany.
2.2. Synthesis of cis-1,3-bis(iodomethyl)cyclohexane (2)
Iodine (10.24 g, 40.4 mmol) was added portion-wise to a stirred
mixture of Ph3P (10.59 g, 40.4 mmol) and imidazole (2.64 g,
38.8 mmol) in CH2Cl2 (200 mL) at 0 °C. The mixture was stirred
for 30 min at 0 °C and 1 h at RT, before a solution of cis-1,3-
bis(hydroxymethyl)cyclohexane (1) (2.82 g, 19.6 mmol) in THF
(40 mL) was added dropwise at 0 °C. The reaction mixture was
allowed to warm to RT and was stirred for an additional hour.
Further steps could be done in air. After addition of pentane
⇑
Corresponding author. Tel.: +46 46 2228153.
E-mail address: ola.wendt@chem.lu.se (O.F. Wendt).
http://dx.doi.org/10.1016/j.poly.2014.06.016
0277-5387/Ó 2014 Elsevier Ltd. All rights reserved.
Please cite this article in press as: A.V. Polukeev et al., Polyhedron (2014), http://dx.doi.org/10.1016/j.poly.2014.06.016

2
A.V. Polukeev et al. / Polyhedron xxx (2014) xxx–xxx
into a pre-heated oil bath with the specified temperature for
24 h. After that, the flask was cooled by a stream of air and the
sample was analysed by NMR spectroscopy. Two runs were per-
formed to determine average TONs. No compounds other than
COA, cyclooctene (COE), TBE and tert-butylethane (TBA) could be
detected by 1H NMR spectroscopy.
2.6. Acceptorless dehydrogenation of cyclooctane by complex 4
Fig. 1. Arene-based pincer complexes, used in alkane dehydrogenation (14e
fragments are shown).
Complex 4 (0.0073 g, 0.0116 mmol) and tBuONa (0.0017 g,
0.0177 mmol, 1.5 eq) were placed into a Schlenk flask and COA
(1.56 ml, 11.6 mmol, 1000 eq) was added. The flask was connected
to a reflux condenser, immersed into an oil bath pre-heated to
170 °C and the mixture was refluxed for 18 h while passing a slow
flow of argon above the reflux condenser. After cooling with a
stream of air, the sample was analysed by NMR spectroscopy.
(200 mL) the mixture was filtered through a pad of silica and con-
centrated in vacuum. The residue was re-dissolved in pentane
(60 mL), filtered through another pad of silica and washed out with
additional pentane. Concentration in vacuum afforded cis-1,3-
bis(iodomethyl)-cyclohexane as a colourless oil which readily
solidified slightly below RT. Yield: 6.32 g (89%). (Rf = 0.49 in hex-
ane). Anal. Calc. for C8H14I2: C, 26.40; H, 3.88. Found C, 26.41, H,
3.89. 1H NMR (CDCl3): d 3.12 (d, J = 6.2 Hz, 4H, –CH2I), 2.03 (d of
m, J = 12.6 Hz, 1H, 1-CHAHB), 1.86–1.78 (m, 3H, 3-CHAHB +
4-CHAHB), 1.52–1.44 (m, 2 H, –CH–), 1.34 (m, 1H, 4-CHAHB), 0.89
(m 2H, 3-CHAHB), 0.70 (apparent q, J = 11.9 Hz, 1H, 1-CHAHB). 13C
NMR (CDCl3): d 40.16 (1-CH2), 39.73 (–CH–), 33.08 (3-CH2), 25.48
(4-CH2), 15.18 (–CH2I).
2.7. Dehydrogenation of triethylamine in the presence of
tert-butylethylene
In a typical experiment, complex 4 (0.0054 g, 0.0086 mmol) and
tBuONa (0.0012 g, 0.0125 mmol, 1.5 eq) were placed into a Straus
flask, and the specified amounts of NEt3 (10 or 100 eq), TBE (20
or 200 eq) as well as 1.5 ml of toluene were added. The flask was
sealed and fully immersed into a pre-heated oil bath at 120 °C
for 18 h. Subsequently, the flask was cooled by a stream of air
and the sample was analysed by NMR spectroscopy. An average
of two runs were performed to determine TONs. No compounds
other than NEt3, N,N-diethylvinylamine, N,N-divinylethylamine,
TBE and TBA could be detected by 1H NMR spectroscopy.
2.3. Synthesis of cis-1,3-bis-[(di-tert-
butylphosphino)methyl]cyclohexane, (PCyP)H (3)
To a À78 °C solution of cis-1,3-bis(iodomethyl)-cyclohexane
(5.37 g, 14.8 mmol) in Et2O (100 mL) a solution of tBuLi in pentane
(1.6 M, 46.2 mL, 74.0 mmol) was slowly added. The reaction mix-
ture was stirred for 1 h at 0 °C and 2 h at RT. tBu2PCl (8.43 mL,
44.4 mmol) was added dropwise at À78 °C and the reaction was
stirred at RT overnight. The solvent was removed under vacuum
before hexane (400 mL) and degassed water (100 mL) were added,
and the resulting mixture was stirred for 10 min. The organic phase
was separated, dried over MgSO4 and concentrated in vacuum. The
residue was dried for ca. 5 h at 50 °C under 10À3 mbar vacuum to
give a pale yellow oil, which solidifies upon standing. This material
was crystallized from CH2Cl2/MeOH and dried under vacuum to
give a white powder. Yield: 4.0 g (68%). NMR spectra are consistent
with the literature [18].
3. Results and discussion
3.1. Improved synthesis of 4
While the previously reported procedure [18] can give good
yields of the cyclohexane-based pincer ligand 3, difficulties in the
handling of the highly unstable cis-1,3-bis[(trifluoromethylsulfo-
nyloxy)methyl]cyclohexane intermediate [19] makes the synthesis
inconsistent in the reproducability of the yields. For the same rea-
son it is also impractical for large-scale synthesis. Metallation of 3
according to the literature procedure gives 4 in a moderate yield
(62%) [16]. In addition, 1H and 31P{1H} NMR spectra indicate that
this protocol always gives trace impurities in the resulting samples
of 4, and thus we wanted to develop an over-all more robust and
convenient route to 4.
2.4. Synthesis of (PCyP)IrHCl (4)
The (PCyP)H ligand 3 (1.500 g, 3.74 mmol) and [Ir(COD)Cl]2
(1.258 g, 1.87 mmol) were placed into a Straus flask inside a
nitrogen glovebox and 35 ml of toluene was added. The flask
was sealed, fully immersed into an oil bath and heated for 6 h
at 170 °C. After reaching RT, the solution was degassed, cooled
to À196 °C and the flask was refilled with H2. The reaction mix-
ture was heated for 5 h at 160 °C under H2 atmosphere. The vol-
atiles were evaporated and the resulting red powder was
washed with cold hexane and dried in vacuum to give 2.261 g
(96%) of 4. NMR spectra are consistent with the literature data
[16].
2.5. Dehydrogenation of cyclooctane in the presence of
tert-butylethylene
In a typical experiment, the catalyst (0.0116 mmol) and tBuONa
(0.0017 g, 0.0177 mmol, 1.5 eq) were placed into a Straus flask and
specified amounts of cyclooctane (COA) as well as tert-butylethyl-
ene (TBE) were added. The flask was sealed and fully immersed
Scheme 1. New synthetic route to (PCyP)IrHCl (4).
Please cite this article in press as: A.V. Polukeev et al., Polyhedron (2014), http://dx.doi.org/10.1016/j.poly.2014.06.016

A.V. Polukeev et al. / Polyhedron xxx (2014) xxx–xxx
3
To obtain a more stable intermediate in the synthesis of ligand 3
we wanted a good leaving group other than triflate. The diol 1 was
made through reduction of commercially available cis-1,3-cycloh-
exanedicarboxylic acid with an excess of BH3 in THF, according
to the previously published procedure [19]. The hydroxyl groups
were then iodinated by reaction with PPh3/I2/imidazole in anhy-
drous CH2Cl2 [20] (Scheme 1) affording the diiodo compound 2
in 89% yield. In contrast to the corresponding triflate, 2 is air and
moisture-stable and can be stored in the freezer for at least several
months without any signs of decomposition. Attempts to directly
substitute the iodides with HPtBu2 failed, but lithiation of 2 fol-
lowed by reaction with ClPtBu2 gave the desired ligand 3 in a
68% yield. The improved metallation protocol includes treatment
of the reaction mixture with hydrogen at elevated temperatures.
At these conditions, all by-products are converted to complex 4.
Hence very pure 4 is obtained and no further purification is
required, except for washing with hexane in order to get rid of
traces of grease.
Scheme 2. Transfer dehydrogenation of COA by iridium pincer complexes with
aliphatic backbones.
3.2. Dehydrogenation reactions
A common test reaction used to compare catalytic activity in
dehydrogenation of alkanes is the transfer dehydrogenation of
cyclooctane (COA) in the presence of tert-butylethylene (TBE),
which acts as hydrogen acceptor. The reaction is known to be
inhibited by both the hydrogen acceptor (TBE) and the dehydroge-
nation product (cycloctene, COE) [6], and thus it is important to
compare activity using the same ratio of catalyst, COA and TBE. Ini-
tially, we chose COA/TBE/cat = 3030/3030/1, the same conditions
Brookhart and coworkers used for their (POCOP)Ir catalysts such
as [2,6-(tBu2PO)2C6H3]IrHCl (6) (cf. Fig. 2) [8], since most literature
data refers to this work. tBuONa (1.5 eq) was also added in order to
remove HCl from 4 and 5 and generate an active 14e species. All
catalytic results are summarised in Table 1. Unfortunately, no
products of dehydrogenation were observed when complex 5
was used, indicating a lack of activity (Scheme 2). This result is
in line with the poor thermal stability reported for 5 [17] and it
is likely that the stability of the catalytically active species is even
lower. Complex 4 is more robust (however, dehydrogenation of the
Scheme 3. Acceptorless dehydrogenation of COA by complex 4.
Fig. 2. Benzene-based pincer complexes used in alkane dehydrogenation.
Scheme 4. Transfer dehydrogenation of triethylamine by complex 4.
Table 1
cyclohexyl ring (to mono-ene or diene species) in the ligand is
noticeable around 200 °C) and thus was expected to show better
performance. Indeed, after heating the mixture for 24 h at 200 °C,
complex 4 demonstrated a turnover number (TON) of 50, with
more than a half (30) of the turnovers observed within the first
10 min of the reaction. Longer heating, as well as raising the tem-
perature to 240 °C did not improve this result. At lower tempera-
ture (150 °C) we obtained almost the same TON (45). In
comparison, catalyst 6 [8] is reported to give a TON of 1583 in
40 h. To validate our results we performed the reaction using our
conditions and 6 as a catalyst, obtaining comparable results. It is
known that the catalytic activity can be inhibited by traces of N2
[21] and given the higher electron-donating nature of aliphatic
pincer ligands this effect can be even stronger for these systems.
Catalytic dehydrogenation using different substrates and iridium catalysts. COA/TBE/
cat ratio was 3030/3030/1 unless indicated.
Entry
Catalyst
Substrate
Acceptor
TON
T/°C
Reference
1
2
3
4
5
6
7
8
4
5
6
6
4
4
4
7
4
4
7
7
COA
COA
COA
COA
COA 7200 eq
COA 7200 eq
COA
TBE
TBE
TBE
TBE
TBE 300 eq
TBE 300 eq
–
50
0
200
200
200
200
200
120
150
150
120
120
90
this work
this work
[8]
this work
this work
this work
this work
[22]
1583
1918
60
200
5
105
18.9
4.3
6.5
11.4
COA
–
9
NEt3 100 eq
NEt3 10 eq
NEt3 10 eq
NEt3 10 eq
TBE 200 eq
TBE 20 eq
TBE 10 eq
TBE 30 eq
this work
this work
[24]
10
11
12
90
[24]
Please cite this article in press as: A.V. Polukeev et al., Polyhedron (2014), http://dx.doi.org/10.1016/j.poly.2014.06.016

4
A.V. Polukeev et al. / Polyhedron xxx (2014) xxx–xxx
Indeed, carrying out the catalysis under high vacuum (10À3
4. Conclusions
mbar) instead of an Ar atmosphere raised the TON, but only
slightly; 70 equivalents of COE were observed. The reason behind
the inferior performance of 4 is the poor thermal stability of the
catalytically active species derived from it; monitoring the reaction
mixture by means of 31P NMR-spectroscopy after activation with
tBuONa indicate complete decomposition of the catalytically active
species after several hours at 150 °C. It seems likely that some of
the activity of 4 with tBuONa belongs to an unsaturated, more
active species. Removal of HCl from 4 is fairly slow, and at 200 °C
some amount of 4 seems to undergo an intramolecular dehydroge-
nation prior to dehydrochlorination. Indeed, when a mixture of 4,
COA and TBE were heated for 1 h at 210 °C prior to activation with
tBuONa, the overall turnover number increased to 130. A thorough
investigation of the decomposition routes of complex 4 is currently
in progress.
A direct comparison of the activity of saturated species derived
from 4 with their aromatic counterparts is possible using 7200/
300/1 COA/TBE/cat ratio, for which literature data for low temper-
ature exists. Under these conditions, the result of complex 4 at
200 °C is almost unchanged (60 TON), reflecting the rapid decom-
position. At 120 °C, however, the active species are more stable.
Thus, an initial rate of 10 TON/h was observed, slowly diminishing
to give a total turnover number of 200 after 36 h, whereupon the
reaction was almost over. At slightly lower temperature (100 °C)
and the same reactant ratio the benzene-based complex [2,6-(tBu2
PCH2)2C6H3]IrH2 (7) was reported to demonstrate nearly twice the
reaction rate (20.5 TON/h) [7]. Thus, under all conditions the activ-
ity of 4 seems to be lower than that of arene-based catalysts.
In conclusion we have presented an improved synthetic route to
the cyclohexane-based PCyP ligand 3, suitable for multigram scale,
and improved the metallation procedure for the complex
(PCyP)IrHCl (4). The latter, together with the aliphatic phosphinite
complex 5, was tested for several catalytic dehydrogenations. The
activity was found to be relatively low compared to the corre-
sponding aromatic complexes, and this is primarily due to the
low thermal stability of the catalytically active species.
Acknowledgements
Financial support from the Swedish Research Council, the Knut
and Alice Wallenberg Foundation, The Crafoord Foundation and
the Royal Physiographic Society in Lund is gratefully acknowl-
edged. A.P. thanks the Swedish Institute (Visby program) for a
fellowship.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.poly.2014.06.016.
References
[1] K. Weissermel, H.-J. Arpel, Industrial Organic Chemistry, Wiley-VCH,
Weinheim, Germany, 2003.
[2] R.H. Crabtree, J. Chem. Soc., Dalton Trans. 17 (2001) 2437.
[3] B.A. Arndtsen, R.G. Bergman, T.A. Mobley, T.H. Peterson, Acc. Chem. Res. 28
(1995) 154.
[4] J.A. Labinger, J.E. Bercaw, Nature 417 (2002) 507.
[5] R.G. Bergman, Nature 446 (2007) 391.
[6] J. Choi, A.H.R. MacArthur, M. Brookhart, A.S. Goldman, Chem. Rev. 111 (2011)
1761.
[7] M. Gupta, C. Hagen, R.J. Flesher, W.C. Kaska, C.M. Jensen, Chem. Commun.
(1996) 2083.
Acceptorless dehydrogenation of COA by
4 resulted in 5
turnovers (Scheme 3), which can be compared to 105 for
benzene-based complex [2,6-(tBu2PCH2)2C6H3]IrH2 (7) (cf. Fig. 2)
[22]. Probably the decomposition of 4 is quite fast already at the
COA boiling point of 149 °C [23].
Considering the low thermal stability of active species derived
from 4, we decided to test it in a reaction that does not require high
temperature such as the dehydrogenation of tertiary amines,
which was reported to proceed at 90 °C with 7 as a catalyst [24].
A toluene solution instead of neat reactants was used to suppress
inhibition by the generated products. In the case of complex 4,
the reaction was too slow at 90 °C, but heating a 100/200/1 NEt3/
TBE/cat mixture in toluene at 120 °C afforded a TON of 18.9, with
the yields of N,N-diethylvinylamine and N,N-divinyl-ethylamine
being 15.7 and 1.6%, respectively (Scheme 4). Complex 7 has been
reported to give a TON of 6.5 and 65% yield of N,N-diethylvinyl-
amine using a 10/10/1 NEt3/TBE/cat ratio and a TON of 11.4 with
64% yield of N,N-diethylvinylamine and 25% yield of N,N-divinyl-
ethylamine using a 10/30/1 NEt3/TBE/cat ratio [24]. For a more
direct comparison with literature data, we carried out the dehydro-
genation using 10/20/1 NEt3/TBE/4 ratio, and 4.3 turnovers with
43% yield of N,N-diethylvinylamine was observed. NMR monitor-
ing showed that in this case interactions with the toluene lead to
formation of a catalytically inactive species prior to complete
decomposition of the catalyst.1 Thus, the low TON presumably
reflects a kinetic competition between these unwanted processes
and interaction with NEt3. When COA was used as a solvent, compe-
tition between dehydrogenation of COA and NEt3 was observed; on a
per mol basis, selectivity is around 1:10. However, in view of the
large excess of COA, high yields of vinylamines could not be
achieved.
[8] I. Göttker-Schnetmann, P. White, M. Brookhart, J. Am. Chem. Soc. 126 (2004)
1804.
[9] K. Zhu, P.D. Achord, X. Zhang, K. Krogh-Jespersen, A.S. Goldman, J. Am. Chem.
Soc. 126 (2004) 13044.
[10] S. Kundu, Y. Choliy, G. Zhuo, R. Ahuja, T.J. Emge, R. Warmuth, M. Brookhart, K.
Krogh-Jespersen, A.S. Goldman, Organometallics 28 (2009) 5432.
[11] B. Punji, T.J. Emge, A.S. Goldman, Organometallics 29 (2010) 2702.
[12] R. Ahuja, B. Punji, M. Findlater, C. Supplee, W. Schinski, M. Brookhart, A.S.
Goldman, Nature Chem. 3 (2011) 167.
[13] J.J. Adams, N. Arulsamy, D.M. Roddick, Organometallics 31 (2012) 1439.
[14] M.W. Haenel, S. Oevers, K. Angermund, W.C. Kaska, H.-J. Fan, M.B. Hall, Angew.
Chem., Int. Ed. 40 (2001) 3596.
[15] Y. Shi, T. Suguri, C. Dohi, H. Yamada, S. Kojima, Y. Yamamoto, Chem. Eur. J. 19
(2013) 10672.
[16] A. Arunachalampillai, D. Olsson, O.F. Wendt, Dalton Trans. (2009) 8626.
[17] K.J. Jonasson, N. Ahlsten, O.F. Wendt, Inorg. Chim. Acta 379 (2011) 76.
[18] S. Sjövall, O.F. Wendt, C. Andersson, J. Chem. Soc., Dalton Trans. (2002) 1396.
[19] S. Sjövall, M.H. Johansson, C. Andersson, Eur. J. Inorg. Chem. (2001) 2907.
[20] P.J. Garegg, B. Samuelsson, J. Chem. Soc., Perkin Trans. 1 (1980) 2866.
[21] D.W. Lee, W.C. Kaska, C.M. Jensen, Organometallics 17 (1998) 1.
[22] F. Liu, A.S. Goldman, Chem. Commun. (1999) 655.
[23] NIST Chemistry WebBook, NIST Standard Reference Database Number 69, P.J.
Linstrom, W.G. Mallard (Eds.), National Institute of Standards and Technology,
Gaithersburg MD, 20899, http://webbook.nist.gov, (retrieved April 2, 2014).
[24] X. Zhang, A. Fried, S. Knapp, A.S. Goldman, Chem. Commun. (2003) 2060.
[25] M. Kanzelberger, B. Singh, M. Czerw, K. Krogh-Jespersen, A.S. Goldman, J. Am.
Chem. Soc. 122 (2000) 11017.
[26] I. Göttker-Schnetmann, P. White, M. Brookhart, Organometallics 23 (2004)
1766.
1
NMR spectra indicate the formation of
a tolyl hydride presumably through
oxidative addition of toluene to the 14e Ir(I) fragment as previously reported [25,26],
which further undergoes transformations leading to catalytically inactive species.
Please cite this article in press as: A.V. Polukeev et al., Polyhedron (2014), http://dx.doi.org/10.1016/j.poly.2014.06.016