Formation of hydrogen peroxide and water from the reaction of cold hydrogen atoms with solid oxygen at 10 K

Article abstract of DOI:10.1016/j.cplett.2008.02.095

The reactions of cold H atoms with solid O₂ molecules were investigated at 10 K. The formation of H₂O₂ and H₂O has been confirmed by in situ infrared spectroscopy. We found that the reaction proceeds very effici

Full text of DOI:10.1016/j.cplett.2008.02.095

Chemical Physics Letters 456 (2008) 27–30
Contents lists available at ScienceDirect
Chemical Physics Letters
journal homepage: www.elsevier.com/locate/cplett
Formation of hydrogen peroxide and water from the reaction of cold
hydrogen atoms with solid oxygen at 10 K
*
N. Miyauchi , H. Hidaka, T. Chigai, A. Nagaoka, N. Watanabe, A. Kouchi
Institute of Low Temperature Science, Hokkaido University, N19-W8, Kita-ku, Sapporo 060-0819, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
The reactions of cold H atoms with solid O2 molecules were investigated at 10 K. The formation of H2O2
and H2O has been confirmed by in situ infrared spectroscopy. We found that the reaction proceeds very
efficiently and obtained the effective reaction rates. This is the first clear experimental evidence of the
formation of water molecules under conditions mimicking those found in cold interstellar molecular
clouds. Based on the experimental results, we discuss the reaction mechanism and astrophysical
implications.
Received 11 January 2008
In final form 26 February 2008
Available online 2 March 2008
Ó 2008 Elsevier B.V. All rights reserved.
1. Introduction
which has an activation energy of 5.16 kcal/mol in the gas phase [7].
For the third process, Tielens and Hagen [8] proposed alternative
Water is the most abundant solid material in space, and has
been observed in various astrophysical environments, such as out-
er planets, satellites, comets, and interstellar clouds [1]. Since the
solar system evolved from an interstellar molecular cloud, icy ob-
jects in the solar system originated from the water ice formed in
the interstellar molecular cloud. Therefore, gaining an understand-
ing of the origin of water molecules in interstellar molecular clouds
is critical not only for discussing the origin of the solar system, but
also for understanding chemical evolution and the origin of life [2].
However, the formation mechanism of water molecules in the
interstellar clouds has not been understood to date. It has been
clarified that the formation of water molecules in the gas phase
is incapable of explaining the observed abundance in molecular
clouds [3,4]. Thus, it has been suggested that water molecules
are synthesized by atomic reactions involving H and O on pre-
existing silicate or carbonaceous grains at around 10 K [3–5].
The first possible process for H2O formation is the sequential
hydrogenation of O atoms on grains
reactions
O þ O ! O2;
O2 þ H ! HO2;
HO2 þ H ! H2O2;
H2O2 þ H ! H2O þ OH:
ð4Þ
ð5Þ
ð6Þ
ð7Þ
Reaction (5) has essentially no barrier; theoretically estimated acti-
vation energies lie between 0.1 and 0.4 kcal/mol [9]. Reaction (7)
has activation energies of 3.6–4.3 kcal/mol [10]. All activation ener-
gies referred here are value in the gas phase. There has been no data
available on a surface. Fig. 1 summarizes the possible water forma-
tion routes from reactions (1)–(7). Cuppen and Herbst [6] simulated
the formation of water molecules in various environments of inter-
stellar clouds, and found that reactions (1) and (2) are the main
routes in cool diffuse clouds and reactions from (3)–(7) are the main
routes in cold molecular clouds. H and O atoms are the major gas-
phase species in the former clouds, whereas H2 molecules are the
major gas-phase species in the latter clouds.
O þ H ! OH;
ð1Þ
ð2Þ
Hiraoka et al. [11] investigated reactions (1) and (2) by spraying
D atoms onto O atoms trapped in an N2O matrix at 12 K. They ana-
lyzed the products by using temperature-programmed desorption
(TPD) spectroscopy and observed D2O. A similar experiment has
been performed by Dulieu et al. [12] although their results are still
very preliminary. However, in the experiments of both groups, it
remains unclear whether D2O is formed at 10–12 K or during heat-
ing. It is essential to perform in situ analysis during H/D atom irra-
diation to confirm the formation temperature of H2O and to obtain
kinetic data. For reactions (5)–(7), Klein and Scheer [13] performed
pioneering experiments on the reaction of H atoms with solid O2 at
OH þ H ! H2O:
These reactions have no activation barriers [6]. The second possible
process is the reaction of OH with hydrogen molecules absorbed on
the surface of grains
OH þ H2 ! H2O þ H;
ð3Þ
* Corresponding author. Fax: +81 11 706 7142.
E-mail address: naoya@lowtem.hokudai.ac.jp (N. Miyauchi).
0009-2614/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.cplett.2008.02.095

28
N. Miyauchi et al. / Chemical Physics Letters 456 (2008) 27–30
Fig. 1. Possible reaction scheme for formation of H2O involving H and O atoms in
interstellar clouds (modified from [5,8]). Atoms or molecules and numerals next to
the arrows indicate the reactants in the reaction and the number of the reaction in
the text, respectively. Broken and solid arrows indicate reactions with and without
activation barriers, respectively. Thick and thin arrows denote reactions investi-
gated and not investigated in the present study, respectively. Molecules observed in
the present study are enclosed in thick squares.
20 K. Although they speculated on the production of H2O2 and H2O
based on their experimental results, they were unable to verify the
production of these molecules since no analysis of the products
was performed. To investigate the formation mechanism of water
molecules in interstellar molecular clouds, we focus on reactions
(5)–(7) and the subsequent reactions (2) and (3), and performed
H/D addition experiments to solid O2 at 10 K.
2. Experimental
Experiments were performed using the Apparatus for SUrface
Reaction in Astrophysics (ASURA) system described previously
[14,15]. Briefly, solid O2 with an 8 monolayer (ML) thickness was
produced by vapor deposition on an aluminum substrate at 10 K
at a deposition rate of 25 pm sꢀ1 in an ultrahigh vacuum chamber
at a pressure of about 10ꢀ8 Pa. Atomic H and D were produced by
a microwave-induced plasma in a Pyrex tube and transferred via a
series of PTFE and aluminum tubes to the target. The dissociation
fraction was about 20% at least. The atomic beams could be cooled
to 20 K in the aluminum tube that was connected to an He refriger-
ator. The flux of atoms was controlled by varying the temperature of
the aluminum tube and measured by the method of Hidaka et al.
[16]. The fluxes of both the H and D atoms were 2 ꢁ 1014 cmꢀ2 sꢀ1
in the present experiments. Infrared absorption spectra of the sam-
ple solid during irradiation by atoms were measured in situ by Fou-
Fig. 2. Spectral change in solid O2 due to (a) cold H exposure and (b) D exposure.
The spectra were obtained by subtracting the initial absorption spectra of non-
irradiated samples from the spectra of H(D)-irradiated samples. Lines indicate ne-
wly formed molecules. Spikes with arrows indicate noise caused by vibration of the
He refrigerator.
production of D2O2 and D2O in D-exposure experiments, confirm-
ing that the H2O produced in the present experiments is not a con-
taminant. The production of H2O2(D2O2) and H2O(D2O) was also
indicated in TPD spectra obtained after H(D)-exposure.
rier transform infrared spectroscopy with a resolution of 4 cmꢀ1
.
After irradiation, TPD spectra of the sample were obtained using a
quadrupole mass spectrometer at a heating rate of 4 K minꢀ1
.
The amount of products are calculated and plotted as a function
of exposure time in Fig. 3. These were obtained from the band
areas in the spectra and the reported and estimated integrated
band strengths. The integrated band strengths used are
3. Results and discussion
Fig. 2a shows the change in a typical absorption spectra on irra-
diation by 70 K-hydrogen atoms at 10 K. This figure shows the
absorbance variations from the initial spectrum of the sample, so
that bands appearing above the baseline indicate an increase in
absorbance. H2O2 appears immediately with bands at 3250, 2830,
1.2 ꢁ 10ꢀ17 and 2.1 ꢁ 10ꢀ17 molecule cmꢀ1 for H2O (1655 cmꢀ1
)
[17] and H2O2 (1347 cmꢀ1) [18], respectively. Since no integrated
band strengths for D2O at 1200 cmꢀ1 and D2O2 at 1033 cmꢀ1 have
been reported, we estimated these in the following manner. Ber-
gren et al. [19] measured the optical constants of OH(OD)-stretch-
ing modes in amorphous H2O and D2O at 3200 cmꢀ1 and
2400 cmꢀ1, respectively. Assuming that the ratio of integrated
band strengths for H2O and D2O in OH(OD)-stretching modes are
the same as those in OH(OD)-bending modes, we estimated the
integrated band strength of OD-bending modes for D2O
(1200 cmꢀ1) to be 8 ꢁ 10ꢀ18 molecule cmꢀ1. No data is available
for D2O2, not even optical constants. So we assumed that the ratio
of integrated band strengths for H2O/D2O = 0.67 is the same as that
for H2O2/D2O2, and estimated the integrated band strength of OD-
bending modes for D2O2 (1033 cmꢀ1) to be 1.5 ꢁ 10ꢀ17 molecule
and 1405 cmꢀ1, then H2O follows at 3432, 1650, and 820 cmꢀ1
.
No product was observed when solid O2 was exposed to an H2
beam at 10 K, or when H atoms were sprayed onto the Al substrate
without solid O2, demonstrating that the formation of H2O2 and
H2O in the present experiments is due to H atom irradiation onto
O2. No intermediate radicals (e.g., HO2) were observed, suggesting
that the reaction rate of (5) is much slower than that of (6). In this
experiment, the H flux is sufficiently large so that the reaction (5)
immediately proceeds to produce HO2. The formation of H2O2 and
H2O were also observed when solid O2 was exposed to an H beam
at temperatures between 15 K and 20 K. Fig. 2b clearly reveals the

N. Miyauchi et al. / Chemical Physics Letters 456 (2008) 27–30
29
mation of OH by reaction (7), there are two possible reaction routes
for further H2O formation: reactions (2) and (3). This is because the
H-beams used in the present experiments do not consist of pure H
atoms, but rather are a mixture of H and H2. Reaction (2) has no
activation barrier, while the activation energy of reaction (3) is
5.16 kcal/mol [7]. It seems probable that H2O is formed by reaction
(2), although we currently have no direct evidence for this. Fig. 3b
shows that D2O2 and D2O are produced very rapidly and efficiently
in the D-exposure experiments, although the formation of D2O is
slightly slower than in H-exposure experiments.
In the present experiments, it is impossible to directly measure
the decrease in O2, since O2 is infrared inactive. Thus, it is reason-
able to assume that the summation of 0.5H2O and H2O2 is equal to
the reduction of O2. The O2 loss becomes saturated after long expo-
sure. This type of saturation was also found previously for the suc-
cessive hydrogenation of CO at 8–15 K by Watanabe et al. [14,21].
They concluded from the dependence of the saturation value on
the sample thickness that the saturation is related not to the bal-
ance between the forward and reverse reactions, but rather to
the diffusion length of H in solid CO and CO–H2O mixed ice. This
suggests that in the present experiments the reaction occurs only
at the surface. The structure of the molecular layer may thus be
onion-like, with H2O on the top, H2O2 in the middle, and O2 at
the bottom. The amount of H2O2 is greater than that of D2O2, sug-
gesting that the diffusion length of H into O2 is longer than that of
D into O2. On the other hand, the amount of H2O is the same as that
of D2O, suggesting that the water formation reaction occurs only at
surface. The small amount of observed H2O compared to H2O2 may
be attributed to desorption of products caused by the release of the
heat of reaction.
To discuss the reaction mechanism based on the isotope effect,
we next derive effective reaction rate constants by fitting the data
in Fig. 3. To simplify the problem, we assume that reactions (5)–(7),
(2) and (3) can be simplified to the following consecutive reaction:
k1
k2
O2 ! H2O2 ! 2H2O;
ð8Þ
where ki is the reaction rate constant of the ith reaction. Since, it is
very difficult to measure the surface density of H atoms, nH, in the
present study, we were not able to obtain ki directly. Instead, we de-
fine effective rate constants: k0Hi ¼ kinH. Since nH is considered to be
time-independent, these reactions are assumed to be pseudo-first-
order reactions.
Furthermore, we assume that the products have a layered struc-
ture, as mentioned above. In the deepest layer, no reaction of O2
with H atoms occurs. In the middle layer that has a thickness of
a, the reaction from O2 to H2O2 occurs, but that from H2O2 to
H2O does not occur. In the uppermost layer with a thickness of b,
the entire reaction (8) occurs. The rate equations for the number
densities ni of O2, H2O2 and H2O are given by
Fig. 3. Variation in column densities of products (a) for H2O2 and H2O with H
exposure and (b) for D2O2 and D2O with D exposure. Each data point represents
the average of three measurements, and error bars represent the statistical errors.
The upper abscissa indicates the fluence of H(D) atoms. The solid lines are fitted
to the curves defined by Eqs. (12)–(14). The amount of O2 reduction is assumed to be
equal to the sum of the increase in H2O2 (D2O2) and 1/2H2O (1/2D2O). Fitting curves
of H2O and D2O shown are two times larger than those obtained from Eq. (14).
cmꢀ1. This result is consistent with the spectral features of OH and
OD-bending modes of amorphous H2O2 and D2O2 measured by
Lannon et al. [20], which show that the ratio of peak heights for
H2O2/D2O2 is 0.75.
dn0
dt
dn1
dt
dn2
dt
¼ ꢀk0H1n0;
ð9Þ
ð10Þ
ð11Þ
¼ k0H1n0 ꢀ k0 n1;
Fig. 3a clearly shows that the formation of H2O2 and H2O is very
rapid and efficient; H2O2 and H2O are observed even after exposure
for 5 s. When H atoms are irradiated onto solid CO at 10 K [14], the
production of H2CO and CH3OH are observed after 30 s and 1 min,
respectively. The present experimental results confirm that reac-
tions (5)–(7), which were initially proposed based on theoretical
considerations [8,6], proceed at 10 K. Since reaction (5) essentially
has no barrier [9], it is natural that this reaction proceeds very rap-
idly. While reaction (7) has an activation energy of 3.6–4.3 kcal/
mol [10], it has been suggested that this reaction does not proceed
by an Arrhenius-type reaction at 10 K, but rather proceeds by a
tunneling reaction even at 10 K. As described later, isotope effect
(H or D) on reaction (7) has been observed, strongly supporting
that reaction (7) proceeds by the tunneling reaction. After the for-
H2
¼ 2k0H2n1:
The solutions of Eqs. (9)–(11) for the initial conditions of
n0ðt ¼ 0Þ ¼ n00, and n1(0) = n2(0) = 0 are given by
0
0
n0
H1t
H1t
¼ beꢀk þ aeꢀk
;
ð12Þ
ð13Þ
n00
ꢀ
ꢁ
ꢀ
ꢁ
0
0
0
H1t
n1
n00
k0H1
H2t
H1t
¼ b
eꢀk ꢀ eꢀk
þ a 1 ꢀ eꢀk
;
k0H1 ꢀ k0H2
"
#
ꢀ
ꢁ
0
0
n2=2
n00
1
0
0
H1t
H2t
¼ b 1 þ
kH2eꢀk ꢀ kH1eꢀk
;
ð14Þ
k0H1 ꢀ k0H2

30
N. Miyauchi et al. / Chemical Physics Letters 456 (2008) 27–30
very quickly with H to form H2O2 and H2O within 104ꢀ5 years as
shown in Fig. 3a.
Table 1
Effective reaction rate constants (minꢀ1
)
Recent astronomical observations have revealed that the abun-
dances of some deuterated molecules are up to four orders of mag-
nitude greater than the cosmic D/H ratio of 1.5 ꢁ 10ꢀ5. In the case
of water, a HDO/H2O ratio of 0.03 has been measured in a solar-
type protostar [25]. Many theoretical models, including pure gas
phase models [26] and gas-grain models [27], have been proposed.
However, only a few experimental studies have been performed to
verify the latter models [16,28]. In the case of formaldehyde and
methanol, the successive addition of D to CO (reaction (17)) is
not favorable for producing deuterated formaldehyde and metha-
nol [16]. Instead, H–D substitution in formaldehyde and methanol
are necessary to achieve the observed abundances of deuterated
spiecies [28]. However, Nagaoka et al. [28] found experimentally
that no deuteration of H2O occurs by H–D substitution under D
exposure at 10–20 K, even for fluences of up to 5 ꢁ 1018 cmꢀ2. Con-
sidering the ratios of k0H1=kD0 1 ¼ 1 and kH0 2=kD0 2 ¼ 8 obtained in the
present experiments and the D/H atom ratio of 0.1 or less expected
in molecular clouds [29], deuterium addition to O2 is favorable for
producing the observed amount of HDO; the observed HDO/H2O
ratio of 0.03 can be achieved in a time scale between 104 and 105
years. Although the discussion is indefinite at present, the experi-
mental results presented in this study provide a basis for discuss-
ing H/D fractionation.
Temperature (K)
Reaction
O
2 + H
H2O2 + H
O2 + D
D2O2 + D
CO + H
CO + D
k0H1
k0H2
k0D1
k0D2
k0H3
k0D3
10
15
12.8
3.9
12.0
0.49
0.14a
0.41c
0.014b
0.033c
a, b, c: obtained with the H(D) flux of 1ꢁ1014 cmꢀ2 sꢀ1 [22].
a
For pure CO [15].
For pure CO [16].
For 0.8 ML CO on amorphous H2O [16].
b
c
where n0 þ n1 þ n2=2 ¼ n00. In the case of D exposure, the same reac-
tion as that for H exposure is assumed to occur, namely,
k0D1
k0D2
O2 ! D2O2 ! D2O;
ð15Þ
where k0Di ¼ kinD, and nD is the surface density of D atoms. Effective
rate constants, k0Hi and kD0 i, obtained by fitting the data are shown in
Table 1 together with those for the first steps of successive hydro-
genation (deuteration) of CO [15,16]:
0
CO k!H3 HCO ! H2CO ! CH3O ! CH3OH;
ð16Þ
ð17Þ
0
CO k!D3 DCO ! D2CO ! DH3O ! DH3OD:
Using the relation of nH/nD = 1 [16], the ratios of the reaction
rate constants are estimated: k0H1=kD0 1 ¼ 1 and kH0 2=kD0 2 ¼ 8. The for-
mer result states that there is no difference in rate constants be-
tween O2 + H and O2 + D at 10 K, and it is consistent with the fact
that reaction (5) has essentially no barrier [9]. The latter result
for the ratio of reaction rate constants between H2O2 + H and
D2O2 + D ðk0H2=kD0 2 ¼ 8Þ is reasonable for a tunneling reaction with
an activation energy of 3.6–4.3 kcal/mol [10]. In the case of
CO + H(D), the ratio of reaction rate constants, k0H3=kD0 3, lies be-
tween 10 (from Table 1) and 13 [16] with an activation energy of
about 4 kcal/mol. In any case, these data are very useful for esti-
mating the barrier height and width of a potential, and potential
energy surface as discussed by Hidaka et al. [16]. As already men-
tioned qualitatively, the rate constants of (8) and (15) are one to
two orders of magnitude greater than those of (16) and (17).
Acknowledgements
This work was partly supported by a Grant-in-Aid for Scientific
Research from the Japan Society for the Promotion of Science and
the Ministry of Education, Science, Sports, and Culture of Japan.
References
[1] B. Schmitt, C. de Bergh, M. Festou (Eds.), Solar System Ices, Kluwer, Dordrecht,
1998.
[2] J. Chela-Flores, F. Raulin (Eds.), Chemical Evolution: Physics of the Origin and
Evolution of Life, Kluwer, Dordrecht, 1996.
[3] L.B. d’Hendecourt, L.J. Allamandola, J.M. Greenberg, Astron. Astrophys. 152
(1985) 130.
[4] T.I. Hasegawa, E. Herbst, C.M. Leung, Astrophys. J. Suppl. 82 (1992) 167.
[5] A.G.G.M. Tielens, The Physics and Chemistry of the Interstellar Medium,
Cambridge University Press, Cambridge, 2005.
[6] H.M. Cuppen, E. Herbst, Astrophys. J. 668 (2007) 294.
[7] H.I. Schiff, in: B.M. McCormac (Ed.), Physics and Chemistry of Upper
Atmospheres, Reidel, Dordrecht, 1973, p. 85.
[8] A.G.G.M. Tielens, W. Hagen, Astron. Astrophys. 114 (1982) 245.
[9] S.P. Walch, C.M. Rohlfing, C.F. Melius, C.W. Bauschlicher Jr., J. Chem. Phys. 88
(1988) 6273.
[10] H. Koussa, M. Bahri, N. Jaidane, Z. Ben Lakhdar, J. Mol. Struct.—Theochem 770
(2006) 149.
[11] K. Hiraoka et al., Astrophys. J. 498 (1998) 710.
4. Astrophysical implications
Our results have several implications for ice in space. Assuming
a hydrogen number density of 1 cmꢀ3, the H fluences in a 10 K
molecular cloud will be 1.3 ꢁ 1016 and 1.3 ꢁ 1017 cm ꢀ 2 over 104
and 105 years, respectively. In the present experiment, these flu-
ences approximately correspond to exposure times of 1 and
10 min. Considering that the lifetime of molecular clouds is of
the order of 106 years, it is reasonable to say that water formation
by reactions (5)–(7) occurs very quickly in molecular clouds. Based
on computer simulations of surface reaction networks, Tielens and
Hagen [8] and Cuppen and Herbst [6] conjectured that reactions
(4)–(7) occur in cold molecular clouds in which the ratio H/H2 is
low. The present results strongly support their conclusion experi-
mentally, although the contribution of reactions (1)–(3) is still
unclear.
Solid O2 is expected to occur on the grain surface with H2O and
CO ice in interstellar molecular clouds and to be observable by
infrared telescopes since the 6.44 lm O@O stretching mode is ob-
servable when O2 is surrounded by other molecules [23] and be-
cause the 4.67 lm feature of CO is altered when O2 is mixed with
CO [24]. However, no positive detection has been reported so far
[24]. Present experiments clearly explain why no O2 has been ob-
served: if an O2 molecule is formed on grain surface it will react
[12] F. Dulieu et al., in: S. Diana, J.L. Lemaire, (Eds.), Abstracts of Molecules in Space
& Laboratory, Paris, 2007, p. 124.
[13] R. Klein, M.D. Scheer, J. Chem. Phys. 31 (1959) 278.
[14] N. Watanabe, A. Nagaoka, T. Shiraki, A. Kouchi, Astrophys. J. 616 (2004) 638.
[15] N. Watanabe et al., Planet. Space Sci. 54 (2006) 1107.
[16] H. Hidaka, A. Kouchi, N. Watanabe, J. Chem. Phys. 126 (2007) 204707.
[17] P.A. Gerakins, W.A. Schutte, J.M. Greenberg, E.F. van Dishoeck, Astron.
Astrophys. 296 (1995) 810.
[18] M.L. Loeffler, B.D. Teolis, R.A. Baragiola, J. Chem. Phys. 124 (2006) 104702.
[19] M.S. Bergren, D. Schuh, M.G. Sceats, S.A. Rice, J. Chem. Phys. 69 (1978) 3477.
[20] J.A. Lannon, F.D. Verderame, R.W. Anderson Jr., J. Chem. Phys. 54 (1971) 2212.
[21] N. Watanabe, T. Shiraki, A. Kouchi, Astrophys. J. 588 (2003) L121.
[22] N. Watanabe et al., Astrophys. J. 668 (2001) 1001.
[23] P. Ehrenfreund, R. Breukers, L. dfHendecourt, J.M. Greenberg, Astron.
Astrophys. 260 (1992) 431.
[24] B. Vandenbussche et al., Astron. Astrophys. 346 (1999) L57.
[25] B. Parise et al., Astron. Astrophys. 431 (2005) 547.
[26] H. Roberts, T.J. Miller, Astron. Astrophys. 361 (2000) 388.
[27] A.G.G.M. Tielens, Astron. Astrophys. 119 (1983) 177.
[28] A. Nagaoka, N. Watanabe, A. Kouchi, Astrophys. J. 29 (2005) L29.
[29] H. Roberts, E. Herbst, T.J. Miller, Mon. Not. R. Astron. Soc. 340 (2002) 283.