The roles of chloride ions in the direct formation of H2O 2 from H2 and O2 over a Pd/SiO2 catalyst in a H2SO4/ethanol system

Article abstract of DOI:10.1016/j.jcat.2006.02.003

Hydrogen peroxide was produced directly from H₂ and O ₂ over a Pd/SiO₂ catalyst in a slurry of ethanol that had been acidified with H₂SO₄. All reactions were carried out at atmospheric pressure and at 10 °C. The effects of chloride ions on the reactions involving the formation of H₂O₂, as well as the formation of water via the combustion of H₂ and the reduction or decomposition of H₂O₂, were investigated by adding small amounts of HCl to the system. In the absence of chloride ions, the net rate of H₂O₂ formation was very small, because the combustion reaction was dominant. At a Cl^- concentration of 4×10-4 M, the formation of H₂O₂ proceeded at a reasonable rate, with a selectivity of ca. 45% based on H₂ conversion. During the early stages of the reaction, hydrogen was consumed both in the formation of H ₂O₂ and in the combustion reaction, but as more peroxide was formed, the reduction of H₂O₂ became a significant reaction. The positive effect of Cl^- on the surface Pd appears to result from the inhibition of OO bond breaking in both dioxygen and hydrogen peroxide. The former effect limits the combustion reaction, and the latter limits the decomposition and reduction reactions. Chloride ions also inhibit the dissolution of palladium in H₂SO₄; in the presence of small amounts of Cl^-, the loss of Pd was generally <20% of the total amount in the catalyst.

Full text of DOI:10.1016/j.jcat.2006.02.003

Journal of Catalysis 239 (2006) 237–243
www.elsevier.com/locate/jcat
The roles of chloride ions in the direct formation of H O from H and O
2
2
2
2
over a Pd/SiO catalyst in a H SO /ethanol system
2
2
4
∗
Qingsheng Liu, Jack H. Lunsford
Department of Chemistry, Texas A&M University, College Station, TX 77842-3012, USA
Received 3 November 2005; revised 6 February 2006; accepted 8 February 2006
Abstract
Hydrogen peroxide was produced directly from H and O over a Pd/SiO catalyst in a slurry of ethanol that had been acidified with H SO .
2
2
2
2
4
◦
All reactions were carried out at atmospheric pressure and at 10 C. The effects of chloride ions on the reactions involving the formation of H O ,
2
2
as well as the formation of water via the combustion of H and the reduction or decomposition of H O , were investigated by adding small
2
2 2
amounts of HCl to the system. In the absence of chloride ions, the net rate of H O formation was very small, because the combustion reaction
2
2
−
−4
was dominant. At a Cl concentration of 4 × 10 M, the formation of H O proceeded at a reasonable rate, with a selectivity of ca. 45% based
2
2
on H conversion. During the early stages of the reaction, hydrogen was consumed both in the formation of H O and in the combustion reaction,
2
2 2
−
but as more peroxide was formed, the reduction of H O became a significant reaction. The positive effect of Cl on the surface Pd appears to
2
2
result from the inhibition of O–O bond breaking in both dioxygen and hydrogen peroxide. The former effect limits the combustion reaction, and
the latter limits the decomposition and reduction reactions. Chloride ions also inhibit the dissolution of palladium in H SO ; in the presence of
2
4
−
small amounts of Cl , the loss of Pd was generally <20% of the total amount in the catalyst.
2006 Elsevier Inc. All rights reserved.
©
Keywords: Hydrogen peroxide; Palladium/silica catalyst; Chloride promotion
1
. Introduction
catalyst [10], is to increase the net rate of H2O2 formation and
the selectivity for the conversion of H2 to H2O2.
The four reactions that should be considered are given in
Scheme 1. Only reaction I results in the formation of H2O2; the
other three reactions result in the formation of H2O and thus
Hydrogen peroxide is currently produced by the Riedl–
Pfleiderer process, which involves the sequential hydrogenation
and oxidation of an alkyl anthraquinone [1]. The direct synthe-
sis of hydrogen peroxide from H2 and O2 over supported palla-
dium, however, has attracted recent attention because the direct
process may yield fewer side products and may be less ex-
pensive. The early work of Pospelova et al. [2,3] demonstrated
that an inorganic acid, such as HCN or HCl, was essential for
achieving large yields of peroxide in the direct synthesis. More
recent studies have focused on the positive effects of chloride
and bromide ions [4–12], but because of the complexity of the
reaction network, the specific roles of these halide ions have not
been completely elucidated. The effects of adding these ions,
either directly to the reaction slurry or indirectly through the
−
are undesirable. Pospelova et al. [2] suggested that Cl slows
down the decomposition of H2O2, whereas, Hutchings et al. [9]
−
concluded that Br suppresses the hydrogen combustion (reac-
tion II). Burch and Ellis [7] pointed out that halide ions may
inhibit the dissociation of dioxygen either by blocking adja-
cent sites on the palladium surface or by modifying the electron
density on the metal surface, and thus affect its reactivity. We
−
previously demonstrated that relatively large amounts of Br
*
Corresponding author.
Scheme 1. Reactions involved in the direct formation of H O .
E-mail address: lunsford@mail.chem.tamu.edu (J.H. Lunsford).
2 2
0021-9517/$ – see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2006.02.003

238
Q. Liu, J.H. Lunsford / Journal of Catalysis 239 (2006) 237–243
3
(
e.g., 1 × 10− M) resulted in a decrease in the net formation
rate, presumably by extensively blocking the surface palladium
and decreasing the rate of reaction I [12]. The effect of Cl is
−
less, even at much larger concentrations [12].
The research described here provides more detailed infor-
mation on the relative importance of reactions II–IV in limiting
the net rate of H O formation, the selectivity, and the maxi-
2
2
mum concentration of H2O2 that can be achieved for a range of
−
Cl concentrations while maintaining essentially the same pro-
ton concentration. Ethanol was chosen as the solvent because
the solubilities of H and O in this liquid phase were larger
2
2
than in water.
2
. Experimental
The catalyst used in this study was 5 wt% Pd/SiO pre-
Fig. 1. Direct formation of H2O2 in H2SO4/ethanol system with (2) 4 ×
−
4
−
−
−4
−5
10
M Cl introduced initially, (") no Cl present initially and 4 × 10
M
M
−
−
Cl introduced after 3 h, and (Q) no Cl present initially and 2 × 10
−
2
Cl introduced after 1 h: (A) H O concentration, (B) conversion of H , and
2
2
2
2−
pared by the incipient wetness method using aqueous PdCl4
(C) selectivity for H2O2.
and Cab-O-Sil M5 silica, which is a fumed silica with a sur-
2
−1
2−
face area of about 200 m g . The PdCl4 was formed by
dissolving PdCl2 in concentrated HCl, followed by dilution
with distilled water. After the solution was added to the sil-
At appropriate times after the reaction had been initiated by
the addition of catalyst, the gas phase was analyzed and the
conversion of H determined. In addition, an aliquot of the
2
◦
ica, the material was dried in air at 120 C overnight, heated
in flowing O2/He (20/40 mL min ) at 400 C for 3 h, cooled
in flowing He to 300 C, and finally reduced in flowing H2/He
20/40 mL min ) for 2 h. Previous XPS results confirmed that
liquid phase (ca. 0.2 g) was removed and analyzed for H O us-
2
2
−
1
◦
4 2 4
ing a colorimetric method with a TiOSO /H SO reagent [13].
From the net rate of H O formation and the rate of H con-
2 2 2
version, the selectivity for H O could be determined using the
2 2
◦
−1
(
the supported palladium was present as Pd(0) both after the re-
duction step and after the catalytic reaction in ethanol [12].
The reactions were typically carried out at atmospheric pres-
following equation:
Rate of H2O2 formation (mol/min)
Rate of H2 conversion (mol/min)
SH O =
× 100.
(1)
2
2
◦
sure and at 10 C in glass reactors described previously [8]. One
of these reactors was open to the atmosphere, whereas the other
reactor was closed and had a vent that exited to a gas chromato-
graph (HP 5890 with a 5A molecular sieve column operating
3
. Results
3
.1. Formation of H2O2
◦
at 30 C), so the composition of the gas could be analyzed for
H2 and O2. To improve the accuracy of the analysis, a mixture
of H2/N2 containing 10% N2 was used. The gases were intro-
duced at a total flow rate of 50 mL min via a glass frit (Ace,
The formation of H2O2 as a function of time at two differ-
−
ent chloride concentrations is depicted in Fig. 1A. At a Cl
−
1
−4
concentration of 4 × 10 M, the net formation of H2O2 in-
4
–8 µm) at the bottom of the reactor.
creased in a nearly linear manner for a period of 5 h to
a level of 1.78 wt%. This corresponds to a rate of formation
The liquid phase consisted of 50 or 60 mL of ethanol acidi-
−
1
−1
−1 −1
of 34 mmolH O g min (102 molH O kg h ). In a sep-
cat
fied with sulfuric acid to 0.12 M. The lesser amount of ethanol
was used for the decomposition reactions; the larger amount,
for the formation reactions. The desired chloride concentration
was attained by adding appropriate amounts of aqueous HCl.
The addition of both acids introduced small amounts of water
to the system.
2
2
Pd
2
2
arate experiment, the reactor was thoroughly cleansed of any
−
residual Cl , and the reaction was carried out in the usual man-
−
ner for 3 h. In the absence of added Cl there was almost no net
formation of H2O2 (<0.008 wt%), but when Cl was added to
−
−
4
a level of 4×10 M, the net rate of H2O2 formation was about
The O /H ratio of 4:1 used in these experiments is in the
2
2
the same as that observed in the initial experiment. In a pre-
−
explosive regime; therefore, appropriate precautions should be
taken. In particular, the dry catalyst should not be added to
O /H gas mixture. Hence, the protocol for adding the cata-
vious paper [14], we stated that even in the absence of Cl ,
H O was produced with the H SO /ethanol system; however,
2
2
2
4
2
2
it now appears that the reactor had not been thoroughly cleaned,
−
lyst was as follows: (a) The ethanol was exposed to the O2/H2
gas mixture via the frit for 30 min; (b) 5 mL of the liquid phase
was taken out of the reactor, and a slurry was made with 42 or
and the residual Cl was sufficient to promote the net reaction.
−
Only very small amounts of Cl are necessary to initiate the net
formation of H O , as illustrated by the experiment in which
2
2
−
5
−
5
0 mg of catalyst; and (c) this slurry was added to the remain-
the slurry was made only 2 × 10 M in Cl . In this case the
ing solution in the reactor. The amount of catalyst depended on
the amount of ethanol; that is, 42 mg of catalyst was used with
net H O formation rate was initially about the same as that
2
2
observed with the much larger chloride ion concentration, but
−
5
0 mL of ethanol, and 50 mg of catalyst was used with 60 mL
after 2 h at this smaller Cl concentration, the rate decreased
of ethanol.
markedly.

Q. Liu, J.H. Lunsford / Journal of Catalysis 239 (2006) 237–243
239
Pd dissolved, the molar equivalent would be 3.9 × 10− M. It
was also shown that the relatively large particles of palladium
black formed during the catalytic reaction were not oxidized to
4
2+
Pd under these conditions. These larger unoxidized particles
were observed only after the reaction had been carried out at
−
−3
the largest Cl concentration, namely 4 × 10 M.
2+
Here it should be noted that the Pd is in a dynamic state
with palladium remaining on the catalyst and deposited on the
frit, as well as any that may be present in the slurry as palladium
black. Thus, the amount of Pd2 in solution is expected to be
+
somewhat less than the total amount lost from the catalyst. The
2
+
results of Fig. 2A clearly indicate that the concentration of Pd
−
4
reached a maximum value of 0.30×10 M after about 45 min,
and then decreased abruptly. This maximum was <10% of the
total amount of Pd in the system. After the abrupt decrease, the
−
4
concentration increased slowly to 0.29×10 M after 420 min.
2+
During the early stage of the reaction, the dissolved Pd appar-
ently was in a metastable state, and after about 45 min into the
−
4
−
−6
−4
Fig. 2. (A) Net formation of H O at (2) 4×10 M Cl and (") 4×10
M
M
2
2
0
−
2+
reaction, more than half of it was reduced to a form of Pd that
Cl ; (a) concentration of Pd in solution as a function of time for 4×10
−
Cl ; (B) Net decrease in the concentration of H O in the presence of H and
is not readily reoxidized. Thereafter, part of this palladium was
2
2
2
−
4
−
2+
O
(O /H = 4) with 4 × 10 M Cl . The initial concentration of H O in
slowly oxidized to Pd . It is important to note that during the
2
2
2
2 2
(
B) was 4.6 wt%.
2+
first 45 min, while the Pd concentration was first increasing
and then decreasing, the net rate of H2O2 formation remained
constant.
In a set of two experiments carried out for 4 h and 7 h under
the same conditions, it was determined that the amount of Pd
It is also instructive to consider the effect of chloride con-
centration on the H2 conversion level (Fig. 1B). Surprisingly, in
−
the absence of Cl , when there was almost no net formation of
−
4
H2O2, the conversion of H2 was the greatest, about 55%. When
remaining on the surface was the equivalent of 3.3 × 10 and
−
−
4
Cl was added to the system, H2 conversion decreased propor-
2
.9 × 10 M, respectively. Thus, the amount of Pd lost from
−
tionally to the amount of Cl introduced. In the presence of
the surface was about 20%, and at least half of this loss occurred
relatively early in the reaction. In fact, only 10% of the loss
occurred between the fourth and seventh hours of the reaction.
Thus, it seems unlikely that loss of palladium from the surface
can account for the decrease in the net rate of H O formation
−
Cl , H2 conversion generally increased with time, which had
a negative effect on selectivity, as shown in Fig. 1C. For ex-
−
−4
ample, at a Cl level of 4 × 10 M, the selectivity decreased
−
−5
from 50 to 42% in 5 h. At a Cl level of 2 × 10 M, the de-
2
2
2+
crease in selectivity was much larger. Obviously, the selectivity
after 5 h (Fig. 2A). Moreover, the marked changes in the Pd
−
was nearly zero in the absence of added Cl .
The longer-term net formation of H2O2 at two different Cl
concentrations is shown in Fig. 2A. Even at a Cl concentration
concentration after 45 min of reaction were not reflected in the
H O formation rate.
−
2
2
−
As a blank experiment, the loss of palladium was followed in
−4
−
of 4 × 10 M, the net rate began to decrease after 5 h, and at
the usual H SO /ethanol system but with no Cl present. Unex-
2
4
−
6
2+
a concentration of 4 × 10 M, the rate decreased after 1 h, and
the H2O2 concentration reached a maximum at ca. 3 h. At 4 ×
pectedly, the Pd concentration reached a maximum value of
−
4
0.88 ×10 M after 50 min, which is considerably greater than
−
3
−
−4
−
1
0
M Cl (results not shown), the net rate of H2O2 formation
that observed when the slurry was 4×10 M in Cl . But after
−
4
−
2+
was somewhat greater than at 4 × 10 M Cl for the first 3 h,
but at longer reaction times, the rate became smaller. After 5 h,
the H2O2 concentration was 2.0 wt%.
370 min of reaction, the Pd concentration in the absence of
−
4
chloride ions had decreased to only 0.05 × 10 M. After 7 h
under reaction conditions, the Pd remaining on the SiO was
2
−
4
In an attempt to understand the origin of this loss in activ-
ity with time, the loss Pd from the catalyst and the amount
the equivalent of 2.6 × 10 M. Therefore, one long-term role
of the chloride is to reduce the loss of Pd in the sulfuric acid
solution somewhat. This inhibition was also observed for other
chloride concentrations.
of Pd2 in solution were determined for the system contain-
+
−4
−
ing 4 × 10 M Cl . The latter is also shown in Fig. 2A. The
2+
concentration of Pd in solution was determined using the
method of Onishi [15]. The amount of palladium remaining
on the surface was determined after different reaction times
by first centrifuging the slurry, thereby removing the catalyst
and any other solid material from the slurry. Then an aque-
ous 0.6 M HCl solution was added to the solid residue. After
allowing the new slurry to stand overnight in air, the liquid
3.2. Reduction and decomposition of H O2
2
Because the loss of palladium cannot fully account for the
decrease in the net rate of H O2 formation after ca. 5 h,
2
other explanations must be sought. One explanation is that
a steady state is reached between the true rate of H O forma-
2
2
phase was analyzed for Pd2 . The method was confirmed by
+
tion and its rate of reduction and/or decomposition. Evidence
analyzing a fresh Pd/SiO2 catalyst. Assuming that all of this
for such a steady state is found in the net rate of H2O2 reduc-

240
Q. Liu, J.H. Lunsford / Journal of Catalysis 239 (2006) 237–243
tion/decomposition depicted in Fig. 2B. This experiment was
reactor; the results are shown in Fig. 4. The first-order rate con-
stants are also given in Table 1. The largest rate constant was
observed with the N /H mixture, which suggests the role of
−
carried out in the standard system with O2/H2 = 4 and a Cl
−4
concentration of 4 × 10 M; hydrogen peroxide was added to
the initial slurry to attain a concentration of 4.6 wt%. Over 7 h,
the concentration of H2O2 decreased to 2.2 wt%, comparable
to the 2.0 wt% formed in the conventional system to which no
H2O2 was added (Fig. 2A). A similar experiment was carried
2
2
H in the loss of H O ; however, as shown below, the reduction
2
2
2
−
reaction is not the dominant pathway in the absence of Cl .
Oxygen, in contrast, had a stabilizing effect on the H O .
2
−
2
Although these experiments show the effects of Cl ion con-
centration and gas composition on the stability of H2O2, they do
not provide information on the more fundamental question con-
cerning the relative importance of reduction and decomposition
reactions. To explore these reactions, we took advantage of the
fact that the reduction reaction consumes H2 while the decom-
position reaction releases O2. Using the closed reactor, a set of
experiments was conducted in which a N2/H2 (4:1) mixture was
passed through the standard H2SO4/ethanol solution that con-
tained different concentrations of chloride ions. At a specific
time, H2O2 was introduced to a level of 1.7 wt%, after which
the H2 and O2 concentrations were determined as functions of
−
−5
out at a Cl concentration of 2 × 10 M. Hydrogen peroxide
was added to achieve an initial concentration of 1.7 wt%. In
the presence of O2 and H2, the H2O2 concentration decreased
in a nearly first-order manner to 0.4 wt% over 4 h, in reason-
able agreement with the steady-state concentration of 0.6 wt%
shown in Fig. 1A. Thus, a nearly common steady-state level of
H2O2 was achieved by both the net loss and the net formation
reactions.
To further explore the net decomposition rate of H2O2 at dif-
ferent chloride ion concentrations, the reaction was allowed to
−
proceed for 1 h with O2/H2 = 4 at a particular Cl concentra-
tion, and then ca. 1.7 wt% H2O2 was added to the system. The
concentration of H2O2 was followed as a function of time; the
−
−4
time. At a Cl concentration of 4 × 10 M, the conversion
of H2 was nearly 50% during much of the decomposition re-
action (Fig. 5), and O2 was detected at only the background
level. Hence, only the reduction reaction occurred at this chlo-
−
results are given in Fig. 3. In the absence of any added Cl , the
net loss of H2O2 was rapid (Fig. 3A), and after 90 min, the con-
centration approached the detection limit. The results are shown
−
as a first-order plot in Fig. 3B for this case and for several Cl
Table 1
−
concentrations. Clearly, the Cl ion concentration had a sig-
Rate constants for the decomposition of H O
2
2
nificant effect on the net loss of H2O2, with the most marked
−
[
(
Cl ]
M)
Gases
composition
Decomposition rate
−
6
−5
−3
−
change between 4×10 and 2×10 M. At 4×10 M Cl ,
there was a net increase in H2O2 concentration. Although the
net change in H2O2 concentration was not strictly first order,
the results are given as first-order rate constants in Table 1 for
−
1 a
constants (min
)
−
−
2
2
0
0
0
0
N /H (4/1)
O /H (4/1)
2 2
He only
O only
2
N2/H2 (4/1)
O /H (4/1)
2 2
6.5 × 10
4.2 × 10
2.8 × 10
2.1 × 10
8.6 × 10
3.5 × 10
7.0 × 10
3.0 × 10
4.1 × 10
−5.6 × 10
2
2
−2
−
−
2
purposes of comparison. Over the range of Cl investigated, the
−
−
−
−
−
−
4
6
5
5
4
3
−3
−2
−3
−3
−4
−4
rate constants differed by more than two orders of magnitude.
The decrease in H2O2 concentration in the absence of added
4 × 10
4
× 10
−
2 × 10
O2/H2 (4/1)
O /H (4/1)
Cl was determined with different gasses passing through the
4
4
4
× 10
× 10
× 10
2
2
O /H (4/1)
2
2
O /H (4/1)
2
2
a
When O and H are present these are rate constants for the net loss of
2
2
H O .
2
2
Fig. 3. Decomposition of H O under O /H = 4; 1.7 wt% H O was added
2
2
2
2
2 2
6
0 min after the reaction had begun: (A) H O concentration with respect to
2 2
−
−
time in the absence of Cl and (B) ln[H O ] versus time in the presence of
Fig. 4. Decomposition of H2O2 in the absence of Cl ions with different gases:
2
2
−
−6
−
−5
−
−5
M
(
") 0 M Cl , (1) 4 × 10 M Cl , (Q) 2 × 10 M Cl , (2) 4 × 10
(") N2/H2 = 4, (2) O2/H2 = 4, (a) He and (Q) O2, as well as (!) N2/H2 = 4
−
−4
−
−3
−
−4
−
Cl , (a) 4 × 10 M Cl and (P) 4 × 10 M Cl .
in the presence of 4 × 10 M Cl .

Q. Liu, J.H. Lunsford / Journal of Catalysis 239 (2006) 237–243
241
Fig. 5. The conversion of H (2) and the concentration of H O (") as a func-
2
2 2
−
−4
tion of time in flowing N /H = 4. The Cl concentration was 4 × 10 M.
2
2
Fig. 6. (A) The conversion of H and (B) the ratio of decomposition-to-
2
reduction (D/R) as a function of time in flowing N /H = 4: (") in the ab-
2
2
−
−5
−
ride level. In a parallel reaction, the decrease in the amount of
H2O2 was followed; these results are also shown in Fig. 5. For
these conditions, the loss of H2O2 was nearly zero order un-
til almost all of the peroxide had been reduced; nevertheless,
for comparison purposes, the results are plotted in a first-order
reaction format in Fig. 4, and the rate constant is given in Ta-
ble 1. Near the end of the reaction, H2 conversion decreased
because the residual amount of H2O2 became small. Because
H2 conversion remained essentially constant throughout most
of the reaction, even as the H2O2 concentration decreased sig-
nificantly, hydrogen transport may have been a limiting factor.
With smaller chloride ion concentrations, the decomposition
reaction became significant, as indicated by the presence of O2
as a product. From the amounts of O2 and H2 in the effluent
from the reactor and the stoichiometry of the decomposition
and reduction reactions, a ratio of decomposition to reduction
sence of Cl and (a) in 1 × 10 M Cl .
4
. Discussion
4
.1. Analysis of the kinetic data
Interesting conclusions concerning the relative rates of the
reactions described in Scheme 1 can be deduced from the ki-
−4
−
netic data. We first consider the system with 4 × 10 M Cl .
The experiments described above, which were carried out in
the absence of added O2, confirmed that the decomposition of
H2O2 (reaction IV) is minimal at this chloride concentration
and that most of the H2O2 is lost via reduction (reaction III).
Under normal reaction conditions, however, the extent to which
H2 is consumed by reactions I and II has not been determined,
but one can qualitatively evaluate the significance of these reac-
tions by considering the selectivity for H2O2 formation and the
extent to which reaction III might occur. During the early stages
of the reaction, the H2O2 concentration would be small; there-
fore, the loss of peroxide by reduction would approach zero.
The observed selectivity of about 50% suggests that about half
of the H2 reacts to form H2O2 and half reacts to form H2O.
From Fig. 1A, one can determine that the net rate of H2O2 for-
(
D/R) can be calculated. Fig. 6 depicts the conversion of H2 and
−
−5
the D/R ratio for systems containing no Cl and 1 × 10
M
−
−5
−
Cl . With 1 × 10 M Cl , H2 conversion was 42% during
most of the reaction, which means that the extent of the hy-
drogen consumption decreased only slightly with a 40-fold de-
−
crease in the Cl concentration. But the D/R ratio increased
−4
−
−5
−
from 0 with 1 ×10 M Cl to ca. 2.4 with 1 × 10 M Cl .
The latter D/R ratio was attained soon after the introduction of
H2O2 to the system, but as the H2O2 was decomposed/reduced,
the D/R ratio decreased in a continuous manner. A D/R ratio of
−
5
−1
mation is 8.4 × 10 molmin , but this is also the true rate
of peroxide formation, because secondary reactions are negli-
gible. The rate of H2 consumption would be twice this amount;
−
ca. 2 suggests that at the intermediate Cl concentration, H2O2
−
5
−1
is formed under normal reaction conditions (i.e., H2 + O2), but
that is, 16.8 × 10 molmin . From the results of Fig. 5, it
is evident that the reduction of H2O2 also can be significant,
but again, its contribution to the overall reaction depends on
the amount of peroxide that has accumulated. In the absence
of O2, the rate of reduction, based on H2 consumption, was
it is decomposed according to reaction IV.
When no chloride was added to the system, the initial D/R
ratio was 14 (Fig. 6B), which indicates that decomposition
mainly occurred. Again, as the H2O2 was consumed, the D/R
ratio decreased. As shown in Fig. 6A, H2 conversion first in-
creased to a maximum of 35% and then decreased. The smaller
initial conversion of H2 may result from the formation of O2
bubbles on the surface, which inhibits the diffusion of H2 to the
Pd particles. The decrease resulted from the nearly complete
depletion of H2O2.
−
4
−1
1.0 ×10 molmin for H2O2 concentrations > ca. 0.8 wt%,
but during the actual formation of H2O2, the rate of reduction
would be much less than this, because reactions I and II would
also compete for the available H2. The purpose of this analysis
is to show that the reduction of the peroxide will become sig-
nificant only at the higher H2O2 concentrations. The decrease

242
Q. Liu, J.H. Lunsford / Journal of Catalysis 239 (2006) 237–243
in the net rate of formation after 4 h (Fig. 2A) probably results
from the contribution of this reduction reaction.
can affect such an ensemble, or there are long-range electronic
effects that modify the catalytic activity of the palladium.
Whether by site blocking or through an electronic effect, the
−
In the absence of added Cl ions, the contribution of reac-
−
tion II appears to be quite different. The net formation of H2O2
is very small (Fig. 1A), which conceivably could result from the
very rapid decomposition or reduction of H2O2, however the
rates of either or both of these reactions would have to be much
greater than those determined from the results of Fig. 4. The H2
major role of Cl appears to be that of limiting the combustion
reaction (reaction II), probably by keeping the O–O bond intact
as O2 is adsorbed. If the combustion reaction can be limited,
H2O2 will be formed, but it is subject to both decomposition
and reduction. The decomposition reaction also involves the
−
−
conversion decreased by only a modest amount when Cl was
breaking of an O–O bond, and the effect of Cl on this reaction
added to the system (Fig. 1B), suggesting that the reduction re-
action was not responsible for the small net formation of H O .
may be similar to its effect on the combustion reaction. Previ-
ously, a mechanism for the dissociation of H2O2 over gold via
OH radicals was proposed [16]. The mechanism for the reduc-
2
2
·
Moreover, the rate of H2 conversion in the absence of chloride
−4
−1
ions was ca. 2.3 × 10 molmin (Fig. 5), which is only 28%
more than the true formation rate and clearly not sufficient to
account for the fact that there was almost no net formation of
H O . As shown in Fig. 6, the decomposition reaction became
tion of H2O2 is not known, but also may involve the formation
·
of OH radicals, followed by the reaction of these radicals with
−
surface H atoms. Thus, the major positive role of Cl appears to
be the inhibition of O–O bond breaking in both O2 and H2O2.
2
2
less significant as the H2O2 concentration approached zero. It
seems reasonable, therefore, that the inability to form H2O2 in
the absence of chloride ions is a result of the rapid combustion
of H2 (reaction II), which is in contrast to the case in which the
5. Conclusion
Chloride ions, even in small concentrations, have a strong
positive effect on the net formation of hydrogen peroxide. Their
main role is inhibiting the combustion of hydrogen; however,
−
−4
Cl concentration was 4 × 10 M.
−
At intermediate Cl concentrations, reactions II, III, and IV
all probably contribute to the inefficient utilization of H2. For
low concentrations of H2O2, D/R < 1; however, at higher con-
centrations of peroxide, decomposition is relatively more im-
portant (Fig. 6). But again the decomposition/reduction reac-
tions do not become significant in the net synthesis reaction
until an appreciable concentration of H O is attained.
they also limit the reduction and decomposition of H O . In
2 2
−
the absence of Cl ions, the dominant reaction is the combus-
−
tion of H2, but on addition of Cl ions, the direct formation of
H2O2 becomes more favorable. Moreover, as chloride is added
to the system, the reduction of H2O2 becomes favored over
the decomposition of the peroxide. The fundamental role of
2
2
−
Cl appears to be the inhibition of O–O bond breaking dur-
ing the adsorption of dioxygen and hydrogen peroxide. This
may be achieved by blocking Pd ensembles on the surface or
by electronic effects. Chloride ions have a secondary role in
4
.2. Interpretation of the results
It is obvious from the preceding sections that chloride ions
influence the side reactions and sequential reactions that limit
the net rate of H O formation, the selectivity for H O , and
2+
controlling the dissolution of the supported palladium as Pd
−
(
or PdCl42 ) in the liquid phase.
2
2
2
2
the maximum concentration that can be attained in the semi-
batch system. In previous work attention was given to the role
Acknowledgment
−
−
of Cl (or Br ) in blocking sites for the dissociation of O ,
2
The authors gratefully acknowledge financial support from
DuPont.
thereby inhibiting the combustion reaction [7,9]. With the new
information concerning the importance of the different possi-
ble reactions, it now seems that Cl may play several roles,
−
References
depending on the concentration of halide and the amount of
peroxide in the system. Before considering the mechanism by
−
[1] W.T. Hess, in: J.I. Kroschwitz, M. Howe-Grant (Eds.), Kirk–Othmer En-
cyclopedia of Chemical Technology, vol. 13, fourth ed., Wiley, New York,
which Cl operates, one should note the concentrations of the
Pd, particularly the surface concentration, and the concentra-
1995, p. 961.
−
tion of Cl ions. Based on the size of the Pd particles reported
[2] T.A. Pospelova, N.I. Kobozev, E.N. Eremin, Russ. J. Phys. Chem.
(Trans.) 35 (1961) 143.
in an earlier study [12], the Pd dispersion has been estimated to
be about 20%; that is, about 20% of the Pd atoms are on the sur-
face. The molar equivalent of surface Pd in our system would be
[3] T.A. Pospelova, N.I. Kobozev, Russ. J. Phys. Chem. (Trans.) 35 (1961)
584.
[4] L.W. Gosser, US Patent 4,889,705 (1989), to DuPont.
[5] K.T. Chuang, B. Zhou, US Patent 5,338,531 (1994).
[6] J. Wanngard, EP 0816286 (1998), to Eka Chemicals AB.
−
5
about 8×10 M. The results of Fig. 1A demonstrate, however,
−5
−
that even 2×10 M Cl leads to the formation of a significant
−
6
−
amount of H2O2, and a slurry that is 4×10 M in Cl also ex-
[7] R. Burch, P.R. Ellis, Appl. Catal. B: Environ. 42 (2003) 203.
8] S. Chinta, J.H. Lunsford, J. Catal. 225 (2004) 249.
[
hibits activity for the net formation of H2O2 (Fig. 2). Moreover,
−
[9] P. Landon, P.J. Collier, A.F. Carley, D. Chadwick, A.J. Papworth, A. Bur-
rows, C.J. Kiely, G.J. Hutchings, Phys. Chem. Chem. Phys. 5 (2003) 1917.
10] V.R. Choudhary, C. Samanta, A.G. Gaikwad, Chem. Commun. (2004)
it is unlikely that all of this Cl is adsorbed on the Pd; some of it
is complexed as PdCl42 . This comparison of Pd and Cl con-
centrations suggests that either large ensembles of Pd atoms are
required for the combustion reaction, and a single chloride ion
−
−
[
2054.
[11] S. Chinta, J.H. Lunsford, J. Catal. 226 (2004) 471.

Q. Liu, J.H. Lunsford / Journal of Catalysis 239 (2006) 237–243
243
[12] Y.-F. Han, J.H. Lunsford, Catal. Lett. 99 (2005) 13.
[15] H. Onishi, Photometric Determination of Traces of Metals, Wiley, New
[13] I.R. Cohen, T.C. Purcell, A.P. Altshuller, Environ. Sci. Technol. 1 (1967)
York, 1989.
2
47.
[16] M. Suh, P.S. Bagus, S. Pak, M.P. Rosynek, J.H. Lunsford, J. Phys. Chem.
B 104 (2000) 2736.
[
14] Y.-F. Han, J.H. Lunsford, J. Catal. 230 (2005) 322.