ARTICLE

pubs.acs.org/JPCA

A Revisited Picture of the Mechanism of Glycerol Dehydration

Teodoro Laino,*, Christian Tuma, Alessandro Curioni, Evan Jochnowitz, and Steen Stolz,z

IBM Research
Zurich, Saumerstrasse 4, CH-8803 Ruschlikon, Switzerland

Philip Morris International R&D, Quai Jeanrenaud 5, CH-2000 Neuch^atel, Switzerland
z
University of Twente, Faculty EEMCS, P.O. Box 217, NL-7500 AE Enschede, The Netherlands

bS Supporting Information
ABSTRACT: The dehydration mechanism of neutral glycerol
in the gas phase was investigated by means of metadynamics
simulations. Structures, vibrational frequencies, Gibbs free energy
barriers, and rate constants at 800 K were computed for the
dierent steps involved in the pyrolytic process. In this article,
we provide a novel mechanism for the dehydration of neutral
glycerol, proceeding via formation of glycidol with a barrier of
66.8 kcal/mol. The formation of glycidol is the rate limiting step
of the overall decomposition process. Once formed, glycidol
converts into 3-hydroxypropanal with a barrier of 49.5 kcal/mol.
3-Hydroxypropanal can decompose further into acrolein or into formaldehyde and vinyl-alcohol with barriers of 53.9 and 35.3 kcal/mol,
respectively. These ndings oer new insights to available experimental data based on glycerol pyrolysis studies performed with
isotopic labeling and on the interpretation of the chemistry of glycerol and sugars in pyrolytic conditions.

INTRODUCTION
Carbohydrates are the primary constituents of most biomass
materials and due to their ubiquitous presence their chemistry
is of paramount importance. However, while the reactivity of
sugars at ambient conditions was explored in the last century, the
understanding of the fundamental underlying reaction mechanisms involved in the pyrolysis of the sugars is still basically
undiscovered. The lack of knowledge about chemical processes
involved in the pyrolysis of carbohydrates is surprising since
thermal degradation processes occur widely wherever biomasses
are combusted1 or food is cooked at high temperatures,2 not to
mention its relevance in the formation of cigarette smoke.3 As
already outlined in several studies,4
7 the reason for such a lack
of understanding can be explained by the inherent diculties in
studying the combustion mechanisms in sugars due to the
complexity arising from the large number of adjacent hydroxyl
groups. Despite chemical dierences, glycerol (propan-1,2,3-triol)
has been largely established as the smallest compound possessing
much of the functionalities of a carbohydrate but still small
enough to make its behavior at high temperature more readily
understood. Moreover, since it is a small molecule, it allows the
application of high level and time-consuming potential energy
exploration tools to characterize thermal decomposition pathways. Last, it is worth noting that the thermal decomposition of
glycerol is an important topic itself. In fact, its wide presence
in animal fats and vegetable oils exposes glycerol contained in
food to possible thermal decomposition processes during heat
treatments.8
Glycerol pyrolysis has been the subject of several experimental
studies. The formation of acrolein (2-propenal), formaldehyde,
r 2011 American Chemical Society

and acetaldehyde depends on the temperature6 and on the

presence of water either as steam6,9 or as supercritical liquid.10,11
Recently, Paine et al.7 investigated pyrolysis mechanisms using
isotopic labeling and found that the decomposition pathway is
entirely unimolecular in nature. They proposed three dierent,
highly competitive mechanisms depending on the presence of
catalysts or impurities.
Two studies recently published4,5 contributed to the understanding of glycerol pyrolysis using theoretical approaches. Nimlos
et al.4 studied the dehydration of neutral and protonated glycerol
reporting a high barrier for 1,2-dehydration (71 kcal/mol), which
was lower for a pericyclic 1,3-dehydration (65 kcal/mol) and
considerably lower in the case of the protonated glycerol. They
concluded their study by stating that the decomposition of neutral
glycerol can occur only at a relatively high temperature such as in
pyrolysis or combustion. From a mechanicistic point of view, they
inspected, based on chemical intuition, dierent reaction mechanisms: 1,2-dehydration and the pericyclic 1,3-dehydration for
neutral glycerol and the formation of oxirane or oxetane intermediates for protonated species. Nimlos et al.4 did not consider the
possible formation of oxirane intermediates for the neutral glycerol
decomposition. More recently, Sun et al.5 investigated the possibility for neutral glycerol to dehydrate into glycidol (an oxirane
intermediate). The results of their calculations suggest a barrier of
59 kcal/mol, also requiring high temperatures for glycerol to
decompose into glycidol.
Received: February 1, 2011
Revised:
March 17, 2011
Published: March 31, 2011
3592

dx.doi.org/10.1021/jp201078e | J. Phys. Chem. A 2011, 115, 35923595

The Journal of Physical Chemistry A

ARTICLE

In this article, we employ the metadynamics12 technique in

order to explore in a much more thorough way the potential
energy surface related to the pyrolytic decomposition of glycerol in
vacuum. This exploration provides an alternative reaction mechanism for glycerol pyrolysis based on the possibility of glycerol to
dehydrate into glycidol. We discuss the thermodynamic and
kinetic properties of the dierent reaction steps with particular
attention to the eect of temperature, providing rate constants
k(T) obtained from canonical transition-state theory.13
17

METHODS
Metadynamics simulations,12 in contrast to standard quantum
chemical approaches where results strongly depend on the chemical intuition employed to build the model, sample in a more
unbiased way the potential energy space with respect to a set of
internal (collective) variables. Therefore, it can be considered a
useful exploratory tool oering the advantage of providing reactive
trajectories that follow the minimum energy path projected in the
space of collective variables.
Free-energy surfaces were sampled according to the following
strategy: the dierent systems have been rst thermalized by
molecular dynamics in an NVT ensemble at the temperature of
800 K using a density functional theory (DFT, PBE18) Hamiltonian.
Upon completion of the thermalization process, determined
by the averages and the uctuations of the potential energy, we
have started the metadynamics12 exploration, employing a set of
appropriately chosen collective variables to sample the free energy
space of glycerol, glycidol, and 3-hydroxypropanal, searching for
low-barrier reactive events. Collective variables employed were
all related to the coordination of oxygen with hydrogens and
carbons, describing a water molecule either connected to the
glycerol moiety or a dehydrated glycerol. The functional form
employed to describe the coordination variable, as implemented
in the CPMD19 code is
sO1 , C1 , C2 , C3

8








 9
>
RO1 C 1 6
R O 1 C2 6
R O 1 C3 6 >
>
>
>
>
1

1

1

=
1<
R0
R0
R0

12


12


12
>
2>
R O 1 C1
R O 1 C2
R O 1 C3
>
>
>
>
1

1

;
:1

R0
R0
R0

1
Here ROiCj are the distances between a selected oxygen (i) and the
carbon atoms (j) of the glycerol moiety. R0 is the equilibrium
distance of an oxygen
carbon bond in glycerol, equal to 1.55 .
Reactive trajectories were optimized using a climbing image
nudged elastic band (NEB)20 approach with 20 images sampling
the reactive path. These calculations were performed using the
CP2K code21 and a semiempirical PM622 Hamiltonian.
Starting from selected points of the PM6 optimized reactive
paths, stationary points corresponding to reactants, transition
structures, and products were rened using density functional
theory (DFT) rst with the PBE18 and, subsequently, the PBE023
parameter free exchange-correlation functionals. The choice of
rening the PBE results with PBE0 is driven by the higher
systematic accuracy of hybrid functionals in estimating barrier
heights compared to pure gradient corrected functionals.24 In
these calculations a plane wave basis set with a 100 Ry kinetic
energy cuto was employed combined with norm conserving
atomic pseudopotentials.25 Periodic images of the molecules were
decoupled using a Poisson solver26 with cubic computational

boxes of 15 edge length. Convergence was achieved in structure

relaxations when all Cartesian force components became smaller
than 2.0  10
5 atomic units. PBE0 force constant matrices
(Hessians) were computed by numerical dierentiation of analytical forces. Translations and rotations were projected from the
Hessian. Stationary points were characterized on the basis of
corresponding vibrational frequencies. Reactants and products of
each reactive step had no imaginary frequencies while transition
structures had exactly one imaginary frequency. In addition,
transition structures were inspected by intrinsic reaction coordinate (IRC) calculations27 in both directions in order to retrieve
either the reactants or the products. These calculations were
performed using the CPMD code.19
The evaluation of reaction rate constants k(T) was done
within canonical transition-state theory formalism
kB T
G
exp

kT
h
RT

!
2

where kB is the Boltzmann constant, T is the temperature

(800 K), h is the Planck constant, G is the activation Gibbs
free energy of the reaction, and R is the gas constant. Finite temperature entropy and energy contributions were obtained from
standard statistical mechanics approaches.

RESULTS AND DISCUSSION

In addition to the three decomposition pathways proposed by
Nimlos et al.,4 we complete the picture with an energetic
competitive scheme, leading either to acrolein or formaldehyde/
acetaldehyde. It is worth noting that in our simulations we
sampled the free energy landscapes according a set of collective
variables describing the elimination of water only; for this reason,
although similar in transition state energies, we were sampling
regions of the phase space dierent from the ones inspected by
Nimlos et al.4 A similar reactivity4 can be achieved by enforcing
a free energy sampling involving both water elimination and
the corresponding carbon
carbon bond breaking. The new
decomposition mechanism of glycerol can be sketched into four
dierent reaction steps: formation of glycidol via 1,2-water
elimination (i); conversion of glycidol into 3-hydroxypropanal
(ii); decomposition of 3-hydroxypropanal into vinyl-alcohol
and formaldehyde (iii) or into acrolein and water (iv). In the
following paragraphs we will review these steps.
Glycerol can form glycidol via 1,2-elimination of water
(Scheme 1, reaction 1). This reaction path has been recently
investigated by Sun et al.5 who computed kinetics and thermodynamics for several structures with dierent relative congurations of the atoms involved. Rening the reactive trajectory using
DFT, we computed the thermodynamic and kinetic quantities
reported in Table 1 in good agreement with previously published
results.5 With a Gibbs free energy barrier of 66.8 kcal/mol (T =
800 K), this reaction step is to be considered the rate limiting step.
Upon opening of the epoxy ring, glycidol converts into 3-hydroxypropanal via the transition structure depicted in Scheme 1,
reaction 2. This reaction step is extremely fast; it proceeds via
formation of a transient carbocationic species (secondary carbocation). Hydrogen present on atom C1 of glycidol is transferred
to the secondary carbon atom concurrently to the formation of
the carbonyl group. Metadynamics runs provided no evidence of
the possibility to form a keto group via formation of a primary
3593

dx.doi.org/10.1021/jp201078e |J. Phys. Chem. A 2011, 115, 35923595

The Journal of Physical Chemistry A

ARTICLE

Scheme 1. Reaction Prole for the Pyrolysis of Glycerol

Table 1. DFT (PBE0) Total Energy, E, Zero-Point

Vibrational Energy Corrected, E0, Enthalpy, H(T), and
Gibbs Free Energy, G(T), Barrier Heights (in kcal/mol) As
Well As Rate Constants k(T) (in 1/s) Computed for the
Dierent Reaction Steps of Glycerol Pyrolysis Shown in
Scheme 1 (T = 800 K)
step

E0

H(T)

G(T)

k(T)

70.9

67.1

66.9

66.8

9  10
6

53.2

48.4

47.7

49.5

5  10
1

36.7

33.4

32.7

35.3

4  103

58.3

53.8

54.0

53.9

3  10
2

carbocation which is well-known to be less stable than a secondary

carbocation.
3-Hydroxypropanal may decompose into two dierent paths.
The most favorable in terms of barrier height is the conversion
into formaldehyde and the enolic form of acetaldehyde (Scheme 1,
reaction 3). This reaction represents the fastest process in the
decomposition of glycerol and is driven by the acidity of the
hydroxyl group.
Although energetically less favorable, 3-hydroxypropanal can
follow a dierent decomposition pathway. It can decompose into
water and acrolein (Scheme 1, reaction 4). Because of the larger
barrier height, the formation of acrolein should be observed only
at high temperatures where this process competes with the
decomposition into formaldehyde and vinyl-alcohol. The formation of acrolein is driven by the acidity of the hydrogen atom at
the R-carbon atom, which is known to be less acidic than a
hydroxyl group.
The pyrolytic studies of glycerol in steam6,9 and supercritical
water10,11 cannot be directly compared to the present computational results. In fact, either supercritical water or steam at high
temperature may aect the decomposition pathways, because
water itself is more acidic at high temperature.
Only recently, Paine et al.7 studied the decomposition of pure
glycerol in the gas phase. Unfortunately, while investigating
extensively the mechanisms they did not focus on any measurement of kinetic data. Interestingly enough, from a quantitative point of view, our ndings complete the energetics for

mechanism B proposed by Paine7 in which the center carbon in

glycerol ultimately ends up as the methyl group in acetaldehyde,
thus providing strong support to their outlines.

CONCLUSIONS
We have proposed a completely new decomposition pathway
for glycerol via formation of glycidol. The present computational
study shows that the mechanism is isoenergetic and therefore
highly competitive with the ones already published in literature
and oers a complete picture of its pyrolytic chemistry without
assuming any enol-type intermediate. Since no experimental
studies have ever hypothesized the pyrolytic decomposition of
glycerol via glycidol formation, this work provides a substantial
improvement to the interpretation of the chemistry of glycerol
and sugars in pyrolytic conditions.
ASSOCIATED CONTENT

bS

Supporting Information. A comparison between PBE/

PBE0 barrier heights, Cartesian coordinates, and vibrational
frequencies obtained for the transition structures and intermediates are available. This material is available free of charge via the
Internet at http://pubs.acs.org.

AUTHOR INFORMATION
Corresponding Author

*E-mail: teo@zurich.ibm.com. Phone: 41 (0) 44 724 8933.

Fax: 41 (0) 44 724 8958.

REFERENCES
(1) Kaltschmitt,M.; Bridgewater, A. W. Biomass Gasication and
Pyrolysis: State of the Art and future prospects; CPL Scientic Limited:
Newbury, U.K., 1997.
(2) Tomasik, P.; Palasinski, M.; Wiejak, S. Adv. Carbohydr. Chem.
Biochem. 1989, 47, 203.
(3) Burton, H. R.; Childs, G., Jr. Beitr. Tabakforsch. 1975, 8, 174.
(4) Nimlos, M. R.; Blanksby, S. J.; Qian, X.; Himmel, M. E.; Johnson,
D. K. J. Phys. Chem. A 2006, 110, 61456156.
(5) Sun, W.; Liu, J.; Chu, X.; Zhang, C.; Liu, C. J. Mol. Struct.:
THEOCHEM 2010, 942, 3842.
3594

dx.doi.org/10.1021/jp201078e |J. Phys. Chem. A 2011, 115, 35923595

The Journal of Physical Chemistry A

ARTICLE

(6) Stein, S. Y.; Antal, M. J., Jr. J. Anal. Appl. Pyrolysis 1983,
4, 283296.
(7) Paine, J. B., III; Pithawalla, Y. B.; Naworal, J. D.; Thomas, C. E.,
Jr. J. Anal. Appl. Pyrolysis 2007, 80, 297311.
(8) Landin, H.; Tareke, E.; Rydberg, P.; Olsson, U.; Tornqvist, M.
Food Chem. Toxicol. 2000, 38, 963.
(9) Antal, M. J.; Mok, W. S. L.; Roy, J. C.; Raissi, A. T.;
Anderson, D. G. M. J. Anal. Appl. Pyrolysis 1985, 8, 291.
(10) Buhler, W.; Dinjus, E.; Ederer, H. J.; Kruse, A. J. Supercrit. Fluids
2002, 22, 37.
(11) Ramayya, S.; Brittain, A.; Dealmeida, C.; Mok, W.; Antal, M.
J. Fuel 1987, 66, 1364.
(12) Laio, A.; Parrinello, M. Proc. Natl. Acad. Sci. U.S.A. 2002,
99, 1256212566.
(13) Eyring, H. J. Chem. Phys. 1935, 3, 492.
(14) Evans, M. G.; Polanyi, M. Trans. Faraday Soc. 1935, 31, 875.
(15) Evans, M. G.; Polanyi, M. Trans. Faraday Soc. 1937, 33, 448.
(16) Laidler, K. J.; King, M. C. J. Phys. Chem. 1983, 87, 2657.
(17) Truhlar, D. G.; Hase, W. L.; Hynes, J. T. J. Phys. Chem. 1983,
87, 2664.
(18) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996,
77, 38653868.
(19) CPMD, Copyright Max Planck Institut fur Festkorperforschung
Stuttgart 1997
2001, Copyright IBM Corp. 1990
2011. http://www.
cpmd.org.
(20) Henkelman, G.; Uberuaga, B. P.; Jonsson, H. J. Chem. Phys.
2000, 113, 9901.
(21) The CP2K developers group, code released under GPL license,
freely available at http://cp2k.berlios.de.
(22) Stewart, J. J. P. J. Mol. Model. 2007, 13, 1173.
(23) Adamo, C.; Barone, V. J. Chem. Phys. 1999, 110, 6158.
(24) Zhao, J.; Gonzalez-Garcia, N.; Truhlar, D. G. J. Phys. Chem. A
2005, 109, 20122018.
(25) Troullier, N.; Martins, J. L. Phys. Rev. B 1991, 43, 1993.
(26) Martyna, G. J.; Tuckerman, M. E. J. Chem. Phys. 1999,
110, 2810.
(27) Gonzalez, C.; Schlegel, H. B. J. Chem. Phys. 1989, 90, 2154.

3595

dx.doi.org/10.1021/jp201078e |J. Phys. Chem. A 2011, 115, 35923595