Deactivation resistance of Pd/Au nanoparticle catalysts for water-phase hydrodechlorination

Article abstract of DOI:10.1016/j.jcat.2009.07.015

Palladium-decorated gold nanoparticles (Pd/Au NPs) have recently been shown to be highly efficient for trichloroethene hydrodechlorination, as a new approach in the treatment of groundwater contaminated with chlorinated solvents. Problematically, natural

Full text of DOI:10.1016/j.jcat.2009.07.015

Journal of Catalysis 267 (2009) 97–104
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Deactivation resistance of Pd/Au nanoparticle catalysts for water-phase
hydrodechlorination
Kimberly N. Heck a, Michael O. Nutt a, Pedro Alvarez b, Michael S. Wong a,c,d,
*
a Department of Chemical and Biomolecular Engineering, Rice University, 6100 S. Main Street, Houston, TX 77005, United States
b Department of Civil and Environmental Engineering, Rice University, 6100 S. Main Street, Houston, TX 77005, United States
c Department of Chemistry, Rice University, 6100 S. Main Street, Houston, TX 77005, United States
d Center of Biological and Environmental Nanotechnology, Rice University, 6100 S. Main Street, Houston, TX 77005, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Palladium-decorated gold nanoparticles (Pd/Au NPs) have recently been shown to be highly efficient for
trichloroethene hydrodechlorination, as a new approach in the treatment of groundwater contaminated
with chlorinated solvents. Problematically, natural groundwater can contain chloride and sulfide ions,
which are known poisons in Pd-based catalysis. In this study, the effects of chloride and sulfide on the
trichloroethene hydrodechlorination catalytic activity were examined for non-supported Pd/Au NPs
and Pd NPs, and alumina-supported Pd (Pd/Al2O3). Over the concentration range of 0–0.02 M NaCl, the
catalytic activity of Pd/Au NPs was unaffected, while the activities of both the Pd NPs and Pd/Al2O3 cat-
alysts dropped by ꢀ70%. Pd/Au NPs were found to be highly resistant to sulfide poisoning, deactivating
completely at a ratio of sulfide to surface Pd atom (S:Pdsurf) of at least 1, compared to Pd NPs deactivating
completely at a ratio of 0.5. Pd/Al2O3 retained activity at a ratio of 0.5, pointing to a beneficial role of the
Al2O3 support. Sulfide poisoning of Pd/Au NPs with different Pd surface coverages provided a way to
assess the nature of active sites. The gold component was found to enhance both Pd catalytic activity
and poisoning resistance for room-temperature, water-phase trichloroethene hydrodechlorination.
Ó 2009 Elsevier Inc. All rights reserved.
Received 6 May 2009
Revised 24 July 2009
Accepted 27 July 2009
Available online 9 September 2009
Keywords:
Palladium
Gold
Nanoparticles
Groundwater
Trichloroethene
Remediation
Sulfide
Chloride
Deactivation
1. Introduction
attenuation to vinyl chloride [4,5], the 4th most hazardous sub-
stance on the CERCLA list [4].
Chlorinated ethenes constitute a class of volatile organic com-
pounds considered to be some of the most harmful groundwater
contaminants. These compounds and their isomers are probable
human carcinogens, have been linked to liver damage, nervous sys-
tem damage, and lung damage, and, with extreme exposure, can
cause death in humans [1]. One of the most prevalent of these
compounds is trichloroethene (TCE) [2], which has also been found
at 832 of 1430 National Priority List Superfund sites at concentra-
tions far above the US EPA maximum containment level of 5 ppb
[3]. Still used as a metal degreaser and in textile manufacture,
TCE ranks 16th on the Comprehensive Environmental Response
and Compensation Liability Act of 2005 (CERCLA) priority list of
hazardous compounds [4]. It also shows up in groundwater as a re-
sult of biotic degradation of perchloroethene. TCE is particularly
threatening if left untreated, as it will degrade through natural
We recently reported on the high efficiency of palladium-deco-
rated gold nanoparticles (Pd/Au NPs) for the catalytic room-tem-
perature, water-phase hydrodechlorination (HDC) of TCE into
ethane (Scheme 1) [6,7]. We characterized Pd/Au NPs to have a
Pd-shell/Au-core structure using a combination of UV–vis spec-
troscopy, X-ray photoelectron spectroscopy, and reaction rate data
[7], with recent extended X-ray absorption spectroscopy results
showing that the Pd/Au NPs have a surface that is enriched in Pd
atoms, i.e., the NPs have a Au core with Pd on the surface [9] (Fang
et al., in preparation). The Pd/Au NP catalytic activity was directly
controlled by the Pd metal surface coverage of the 4-nm Au NPs. A
maximum rate (ꢀ1800 L/gPd/min) was found near 70% surface cov-
erage, nominally 100ꢁ higher than Pd/Al2O3 (ꢀ12 L/gPd/min). Ob-
served for some other catalytic reactions, Au promotes Pd metal
for TCE HDC catalytic activity possibly by creating a Pd–Au inter-
face as population of active sites, by inducing the formation of Pd
surface ensembles as the active sites, or by modifying the elec-
tronic structure of the Pd metal [8].
* Corresponding author. Address: Department of Chemical and Biomolecular
Engineering, Rice University, 6100 S. Main Street, Houston, TX 77005, United States.
Fax: +1 713 348 5478.
Toward their application to real groundwater systems, the long-
term durability of the Pd/Au NP catalysts needs to be addressed
with regard to catalyst deactivation. Groundwater typically
E-mail address: mswong@rice.edu (M.S. Wong).
0021-9517/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2009.07.015

98
K.N. Heck et al. / Journal of Catalysis 267 (2009) 97–104
of a H2PdCl4 solution (2.4 mM) combined with 69.6 mL of H2O [7].
A dark coffee-brown sol containing 4-nm Pd NPs (calculated parti-
cle concentration of 2.9 ꢁ 1015 NP/L) was the result. A Pd/Al2O3
(1 wt% Pd) catalyst was purchased from Sigma-Aldrich, and was
used as is.
Scheme 1. TCE HDC reaction.
contains a multitude of substances which could impact Pd/Au NP
catalytic performance [10–12]. Perhaps the most pertinent to ad-
dress are chloride (Clꢂ) and hydrosulfide (SHꢂ, ‘‘sulfide” for short)
ions, where the former can be found at concentrations ranging
from 1.0 to 1000 mg/L and the latter can be produced naturally
from native anaerobic bacteria. Both chemical species are known
to adsorb onto Pd surfaces [13–16] and to reduce Pd activity
[17–28]. The rapid poisoning of Pd-based catalysts during aque-
ous-phase hydrodechlorination by sulfide ions is a well-known
problem [18,29–31]. In this work, we studied the effects of sulfide
and chloride ions on TCE HDC reaction rates of Pd/Au NPs, observ-
ing that chloride had no effect and sulfide had a markedly less ef-
fect compared to Pd NPs and Pd/Al2O3. We performed catalytic
titration experiments using sulfide as the poison to gain insights
into the nature of the active sites for TCE HDC sites.
2.2. Catalyst testing
2.2.1. Reactor set-up
The catalytic testing of the NPs was conducted similar to our
previous studies [6,7]. Batch reactors were prepared by adding
168.6–172.5 mL of H2O (so as to keep the volume constant at
173 mL after the addition of the catalyst sol/suspension) and a stir
bar to a 250-mL, Teflon-sealed glass bottle. The bottles were sealed
with a cap outfitted with a rubber septum, and sparged with H2 gas
for 15 min to saturate the H2O and fill the headspace. TCE was then
added (3
approximately 43 ppm (43 mg/L, 325
concentration of 1200 ppm at 25 °C. Pentane (0.2
l
L, 99.5%, Aldrich), resulting in a liquid concentration of
M), far below the saturation
l, 99.7%, Burdick
l
l
& Jackson) was added as an internal standard for gas-chromatogra-
phy (GC) headspace analysis. The reactors were stirred rapidly for
at least 3 h or until the measured concentrations showed no
change over time, indicating total dissolution of the TCE and pen-
tane and the establishment of vapor–liquid equilibrium.
2. Experimental
2.1. Preparation of Pd/Au NPs and Pd NPs
2.2.2. Reaction analysis
The Pd/Au NPs were prepared as previously reported [7]. Briefly,
1 mL of a gold salt solution (1 wt% = 25 mM; prepared by adding
1 g of HAuCl4ꢃ3H2O (99.99%, Sigma-Aldrich) to 99 g of deionized
H2O) was added to 79 mL of deionized H2O. A second solution of
0.04 g of tannic acid, 0.05 g of trisodium citrate, and 0.0173 g of
K2CO3 in 20 mL of deionized water was also prepared. Both solu-
tions were heated to 60 °C, and then the tannic acid solution was
added rapidly to the gold salt solution under vigorous stirring.
The solution immediately turned reddish brown, indicating the
formation of Au NPs. The temperature of the solution was in-
creased until boiling, and allowed to boil for 25 min. Finally, the
solution (calculated particle concentration of 2.9 ꢁ 1015 NP/L)
was removed from the heat source and stored in a separate
container.
The catalyst was used directly without purification and intro-
duced into the reactor via syringe injection of an aliquot of sol
(containing Pd/Au or Pd NPs) or suspension (containing Pd/
Al2O3). For GC headspace analysis, 100-ll aliquot samples of the
headspace gas in the batch reactor were withdrawn with a gas-
tight syringe and injected into an Agilent Technologies 6890 GC
equipped with a flame ionization detector (FID) and a packed col-
umn (6-in ꢁ 1/8-in outer diameter) containing 60/80 Carbopack B/
1% SP-1000 (Supelco). Calibration curves were prepared for chlori-
nated ethenes, chlorinated ethanes, and ethane.
The pH of the reaction medium was monitored using pH paper
(Whatman Panpeha pH indicator strips, Sigma-Aldrich) before
injection of the catalyst and at the end of the reaction. At complete
conversion of TCE, the pH should drop to a value of 3.1 theoreti-
cally, due to HCl formation. For all catalyst samples tested, i.e.,
Pd/Au NPs, Pd NPs, and Pd/Al2O3, for both chloride and sulfide test-
ing studies, the initial reactor pH was ꢀ6.0 and the final pH was
ꢀ5.5. This slight pH decrease, as observed by others [18,32], indi-
cates that the reaction system has some buffering capacity that
limits pH change. The Pd and Pd/Au NP reaction systems contain
small amounts of citrate, carbonate, and tannic acid, which are buf-
fering ions. In the Pd/Al2O3 case, the alumina support could buffer
the solution around its point-of-zero charge (generally between pH
6.4 and 9.5) [33].
For the Pd coatings, a specified amount (15–60
H2PdCl4 solution (prepared from 0.0426 g of PdCl2 (99.99%, Sigma-
Aldrich) dissolved in 100 mL deionized water and 480 L 1 M HCl
lL) of a 2.4-mM
l
solution) was mixed with 2 mL of the as-synthesized Au NP sol.
The resulting solution was then bubbled with hydrogen gas
(99.999%, Matheson) for two min. After bubbling, the solution
was aged in a sealed vial overnight at room temperature prior to
use. Assuming complete reduction of Au and Pd ions into metal,
15, 30, and 60 lL of Pd salt solution yielded Pd/Au NPs containing
3.5, 6.8, and 12.8 wt% Pd, respectively. They, respectively, corre-
spond to 4-nm Au NPs with 15%, 30%, and 60% of monolayer
(ML) coverage by Pd metal atoms (Table 1). The particle size of
4 nm was derived from transmission electron microscopy analysis
as we reported previously [7].
2.2.3. Testing of chloride effect
Prior to charging with H2, TCE, pentane, and catalyst, solutions
of 1 M NaCl (prepared from 5.84 g 99.99% NaCl (Fisher) and
100 mL deionized water) were added to the reactor such that the
The Pd NPs were prepared using the same procedure for Au NP
synthesis, except for the Au salt solution was replaced by 10.4 mL
Table 1
Pd content of catalytic materials tested, and amounts charged to reactor.
Catalyst nanomaterials tested
Pd weight loading
(wt% Pd)
Estimated Pd dispersion
(%Pd surface atoms)
Calculated amount of surface Pd atoms,
charged to reactor (nmol)
Pd/Au NPs
15% ML
30% ML
60% ML
3.55
6.85
12.4
100
1
100
100
100
34.7
21a
72
72
72
89.2
493
Pd NPs
Pd/Al2O3
a
Taken from Ref. [32].

K.N. Heck et al. / Journal of Catalysis 267 (2009) 97–104
99
kobs is equivalent to ꢂ(dCTCE/dt)0/Ccat/CTCE,0, where CTCE,0 is the ini-
tial TCE concentration.
total reaction volume (accounting for the eventual added catalyst
sol/suspension) was 173 mL, giving final concentrations ranging
from 0 to 0.02 M chloride. The following catalyst aliquots were in-
jected: 2 mL of the 30%-Pd/Au NP sol; 1 mL of the Pd NP sol com-
bined with 1 mL of H2O; and 25 mg of Pd/Al2O3 suspended in 2 mL
of H2O.
As previously reported, Pd/Au NPs catalyze TCE HDC with a
first-order dependence on TCE concentration [6,7]. Fig. 1a shows
Pd/Au NPs with 30% Pd coverage exhibiting this first-order depen-
dence, in the absence and presence of additional chloride. The first-
order rate constants are listed out in Table 2. However, the mono-
metallic Pd NPs and Pd/Al2O3 catalysts did not show this behavior
(Fig. 1b and c). TCE conversions were linear with time, indicating
zero-order dependence on TCE concentration; conversions leveled
off above a conversion of ꢀ80%.
2.2.4. Testing of sulfide effect
Hereafter, the term ‘‘sulfide” refers to the hydrosulfide ion, the
dissociated form of hydrogen sulfide (H2S M H+ + SHꢂ, pKa = 7.05 at
20 °C [34]. A sulfide solution (0.01 M) was prepared by dissolving
0.0211 g Na2Sꢃ9H2O (98%, Sigma-Aldrich) in 8.78 mL deionized
water (Na2S + H2O M 2Na+ + SHꢂ + OHꢂ), and was combined with
catalyst sol/suspensions before injection.
For Pd/Au NP sols, 4 mL of 15%-Pd/Au NPs, 2 mL of 30%-Pd/Au
NPs, or 0.5 mL of 60%-Pd/Au NPs (Table 1) was treated with 0–
13 lL of the sulfide solution, to vary the sulfide:surface Pd atom
(S:Pdsurf) ratio from 0 to 1.8. Pd NP sol (1 mL) was diluted with
1 mL of H2O (so as to keep the final liquid volume in the reactor
constant at 173 mL), and treated with 0–9.0 lL of sulfide solution
to vary the sulfide:surface Pd atom (S:Pdsurf) ratio from 0 to 1. Sim-
According to the studies by Kopinke et al. [12], Reinhard and
coworkers [32], and us [6], Pd/Al2O3 catalysts were reported to be
first order in TCE. By running the reaction for a longer period of time
in this work, however, we observed clearly and reproducibly the
non-first-order dependence on TCE. We found that the reaction rates
for Pd NPs and Pd/Al2O3 catalysts were first order at low TCE concen-
trations (<1 ppm, close to the 1 ppm used by Kopinke et al. [12] and
by Reinhard and coworkers [32]) and non-first-order at high TCE
concentrations (29 ppm, as used in this study) [39]. An interesting
implicationis that the higher TCE concentrations lead to competitive
chemisorption of TCE on the active sites of Pd NPs and Pd/Al2O3 cat-
alysts, which does not happen with Pd/Au NPs.
ilarly, a Pd/Al2O3 suspension (25 mg powder in 2 mL of H2O) was
treated with 0–49 lL of sulfide solution. All suspensions were stir-
Since rate constants for catalysts exhibiting different rate laws
cannot be compared, we chose to compare initial reaction rates
and initial TOF values instead (Table 2). Whereas the TOF’s for
Pd/Au NPs were derived from initial reaction rates via the first-or-
der rate constants, the TOF’s for the pure Pd-based catalysts were
derived in a different manner. The initial reaction rate of the mono-
metallic Pd catalysts was characterized as
red briefly then left unstirred at room temperature for at least 3 h
prior to injection. The total reaction volume was 173 mL.
The concentrations of surface Pd atoms in the reactor after cat-
alyst injection were 0.42 lM, 0.51 lM, and 2.85 lM for the Pd/Au
NPs, Pd NPs, and Pd/Al2O3 catalysts, respectively. For Pd and Pd/
Au NPs [7], the magic cluster structural model was used [35–38]
and the complete reduction of the gold and palladium precursor
salts was assumed. We estimate that each of the nanoparticle sols
contains 2.91 ꢁ 1015 NPs/L. For the Pd NPs, we assume that the sur-
face atoms are contained only in the 7th shell, and hence the con-
ꢀ
ꢁ ꢂ
dCTCE
dt
ꢂ
Ccat ¼ ꢂr0TCE
0
centration of surface Pd in the sol is 88 lM. In the case of the
where ꢂr0TCE (units of mol-TCE/gPd/min) is the initial slope of the
first 3–4 data points of a TCE concentration–time profile divided
by Pd catalyst reactor charge amount. Accounting for exposed Pd
atoms, initial turnover frequency values were then calculated as
TOF ¼ ꢂr0TCE ꢁ Pd atomic weight ꢄ Pd dispersion/60, which is equiv-
alent to the TOF definition given earlier for Pd/Au NPs.
bimetallic Pd–Au NPs, if the added Pd is assumed to constitute a
partial 8th shell around the Au NP, we estimate that it covers
approximately 15–60% of the Au NP surface, resulting in a surface
Pd concentration of 18–72 lM. The amount of surface Pd on the
Pd/Al2O3 sample was estimated using the literature reported metal
dispersion of 21% [32], which indicates a surface concentration of
210 lg of surface Pd per gram of Pd/Al2O3.
3.2. Effect of chloride on the rate of HDC reaction
2.2.5. Testing the effect of adding Au NPs to Pd/Au NPs
To study the effects of adding Au NPs on sulfide deactivation of
Pd/Au NPs, 1–3 mL of the Au NP sol was mixed with 2 mL of the Pd/
As shown in Fig. 1 and Table 2, the activity of the Pd/Al2O3 and
Pd NPs fell drastically with increasing chloride concentration,
decreasing as much as 30% of its original activity at the highest
chloride concentration used. This chloride effect is seen in other
catalytic systems, such as both gas-phase oxidation [23–26] and
gas- and liquid-phase HDC [27,28] reactions. In gas-phase HDC,
HCl formed as a byproduct of the reaction and inhibited the activ-
ity by reversibly competing with the chlorinated reactant for active
sites [28,32]. In water-phase HDC, this species may play a similar
inhibitory role [27].
Au NPs prepared with 30
lL of H2PdCl4 (to get 30%-Pd/Au NPs) be-
fore adding 2.4 L of the sulfide solution.
l
3. Results and discussion
3.1. Determination of reaction rate constant
Reaction rate constants for the conversion of TCE by Pd-based
catalysts have generally been assumed to be pseudo first order in
TCE concentration, and zero order in H2 concentration (from very
high H2 concentrations) [6,7,32]:
In contrast to our observations with Pd/Al2O3 and Pd NPs, Lowry
and Reinhard reported that a chloride concentration of 0.028 M
had no effect on activity on the aqueous-phase HDC of TCE over
a Pd/Al2O3 catalyst [18]. This difference could be attributed to
the higher pH of 9.6 used by Lowry and Reinhard compared to
the lower pH used in our reaction system (i.e., initial pH of 6, low-
ering to 5.5 by the end of the reaction). The higher pH values led to
increased Pd/Al2O3-catalyzed TCE HDC rates, which could have
masked the chloride inhibitory effect we observed [18].
A reasonable explanation for the deactivation of the pure Pd
catalysts (Pd NPs and Pd/Al2O3) is the chemisorption of chloride
to the Pd surface. In an electrochemical study of halide adsorption
at Pd electrodes, Soriaga and coworkers showed that chloride can
ꢂdCTCE
¼ kmeasCTCE
dt
where the fitted first-order rate constant kmeas = kobs ꢁ Ccat, where
kobs (units of L/gPd/min) is the rate constant normalized by the
amount of catalyst present, and Ccat and CTCE are the concentrations
of the catalyst and TCE, respectively. Accounting for exposed Pd
atoms, initial turnover frequency values can be calculated as TOF = -
kobs ꢁ CTCE,0 ꢁ Pd atomic weight ꢄ Pd dispersion/60. It is noted that

100
K.N. Heck et al. / Journal of Catalysis 267 (2009) 97–104
a
1.0
0.8
0.6
0.4
0.2
0.0
b
1.0
0.8
0.6
0.4
0.2
0.0
[Cl] = 0 M
[Cl] = 0 M
[Cl] = 0.001 M
[Cl] = 0.005 M
[Cl] = 0.02 M
[Cl] = 0.001 M
[Cl] = 0.005 M
[Cl] = 0.02 M
0
20
40
60
80
100
0
20
40
Time (min)
60
80
Time (min)
1.0
0.8
0.6
0.4
0.2
0.0
0.60
0.50
0.40
0.30
0.20
0.10
0.00
c
d
(a) Pd/Au NPs (30% ML)
(b) Pd NPs
[Cl] = 0 M
[Cl] = 0.001 M
[Cl] = 0.005 M
[Cl] = 0.02 M
(c) Pd/Al2O3
0.000 0.005 0.010 0.015 0.020
0
20
40
Time (min)
60
80
Chloride Concentration (M)
Fig. 1. Fractional conversion of TCE vs. time for (a) Pd/Au NPs (30% ML), (b) Pd NPs, and (c) Pd/Al2O3 at various chloride concentrations. Points are experimental results. Solid
lines in panel (a) are first-order best fits, and dashed lines in panels (b) and (c) show initial slopes used to determine initial TCE HDC activity. (d) Initial turnover frequencies at
different chloride concentrations, with dashed lines drawn to guide the eye.
Table 2
Parameters and results from chloride poisoning studies for 30%-Pd/Au NPs, Pd NPs, and Pd/Al2O3. Reaction conditions: CTCE,0 = 221 lM, room temperature.
Initial rate ꢂr0TCE ꢁ 10ꢂ3
Catalyst
Reactor chloride concentration (mmol/L)
First-order rate constant kobs
(L/gPd/min)
Initial turnover frequency TOF
(mol-TCE/mol-Pdsurf/s)
(mol-TCE/gPd/min)
Pd/Au NPs (30% ML)
0
1
5
900
809
885
907
292.5
262.9
287.6
294.7
0.52
0.47
0.51
0.52
20
Pd NPs
0
1
5
–
–
–
–
28.1
23.5
13.3
9.9
0.050
0.042
0.024
0.018
20
Pd/Al2O3
0
1
5
–
–
–
–
42.7
30.0
25.8
18.3
0.076
0.052
0.045
0.031
20
rine atoms on Pd–Au catalytic surfaces in the closely related reac-
tion of 1,1-dichloroethylene HDC [41].
oxidatively adsorb to Pd(111) crystals under acidic conditions via
the reaction [40]
Whereas Pd-only catalysts deactivated significantly, the activity
of the Pd/Au NPs unexpectedly remained constant and unaffected
by Cl concentration at the same condition of near-neutral pH
(Fig. 1a and d), suggesting that the extent of chlorine surface spe-
cies formation on the Pd metal was lowered in the presence of Au.
This reduced binding affinity to chlorine atoms may be correlated
to the increased d-band electron density of Pd caused by the Au, as
previously observed through X-ray photoelectron spectroscopy [7].
HþðaqÞ þ ClꢂðaqÞ þ Pdð111Þ ! Pdð111Þ ꢂ Cl þ 1=2H2
The amount of Cl added to the reactors was far in excess of the
amount of surface Pd atoms, and so by way of the law of mass ac-
tion, the amount of chlorine surface species would also increase if
there was sufficient amount of hydronium ions H+(aq). Acid and
neutral pH conditions would favor chlorine surface species forma-
tion (leading to greater susceptibility to catalyst deactivation and
lower HDC rates) and basic pH would disfavor chlorine surface spe-
cies formation (leading to catalyst deactivation resistance and high-
er HDC rates). From in situ surface-enhanced Raman spectroscopic
investigations, we were recently able to detect surface-bound chlo-
3.3. Effect of sulfide on the rate of HDC reaction
Fig. 2 and Table 3 show the results for the sulfide poisoning
experiments on monometallic Pd and Pd/Au NP (30% ML) catalyst,

K.N. Heck et al. / Journal of Catalysis 267 (2009) 97–104
101
a
c
1.0
0.8
0.6
0.4
0.2
0.0
1.0
b
0.8
0.6
S:Pd = 0.0
S:Pd = 0.0
S:Pd = 0.1
S:Pd = 0.2
S:Pd = 0.3
0.4
0.2
0.0
S:Pd = 0.3
S:Pd = 0.6
S:Pd = 0.8
0
50
100
Time (min)
150
200
0
50
100
Time (min)
150
200
1.0
0.8
0.6
0.4
0.2
0.0
0.60
0.50
0.40
0.30
0.20
0.10
0.00
d
Pd/Au NPs (30% ML)
(a) Pd/Au NPs (30% ML)
Pd NPs
(b) Pd NPs
Pd/Al2O3
(c) Pd/Al2O3
S:Pd = 0.0
S:Pd = 0.1
S:Pd = 0.2
S:Pd = 0.3
S:Pd = 0.5
S:Pd = 1.0
0.00
0.25
0.50
S:Pdsurf ratio
0.75
1.00
0
50
100
Time (min)
Fig. 2. Fractional conversion of TCE vs. time for (a) Pd/Au NPs (30% ML), (b) Pd NPs, and (c) Pd/Al2O3 at various S:Pdsurf ratios. Points are experimental results. Solid lines in
panel (a) are first-order best fits, and dashed lines in panels (b) and (c) show initial slopes used to determine initial TCE HDC activity. (d) Initial turnover frequencies with
S:Pdsurf ratio, with dashed lines drawn to guide the eye.
Table 3
Parameters and results from sulfide poisoning experiments for 30%-Pd/Au NPs, Pd NPs, and Pd/Al2O3. Reaction conditions: CTCE,0 = 221 lM, room temperature.
Initial rate ꢂr0TCE ꢁ 10ꢂ3 (mol-
Catalyst
Sulfide added
g)
S:Pdsurf molar
ratio
First-order rate constant kobs (L/
gPd/min)
Initial turnover frequency TOF (mol-TCE/
mol-Pdsurf/s)
(
l
TCE/gPd/min)
Pd/Au NPs (30%
ML)
0
0.0
0.33
0.6
0.8
1.0
900.4
604.9
342
215
0
292.5
196.6
111.2
69.88
0
0.52
0.35
0.20
0.12
0
0.76
1.39
1.85
2.31
Pd NPs
0
0.0
0.1
0.2
0.3
0.5
–
–
–
–
–
28.1
10.4
5.47
2.88
0
0.050
0.018
0.010
0.005
0
0.29
0.57
0.85
1.43
Pd/Al2O3
0
0.0
0.1
0.2
0.3
0.5
1
–
–
–
–
–
–
42.7
37.4
39.8
40.9
40.6
14.7
0.076
0.066
0.071
0.072
0.072
0.026
1.58
3.16
4.74
7.90
15.8
in which sulfide-to-surface Pd atom ratios were considered. In the
absence of sulfide, the activity measured for the control catalysts
had initial TOF values of 0.076 and 0.050 mol-TCE/mol-Pdsurf/s
for the Pd/Al2O3 and Pd NP catalysts, respectively. The Pd/Au NP
catalyst had an initial TOF 0.52 mol-TCE/mol-Pdsurf/s, in agreement
with our earlier work (which used a higher initial TCE concentra-
been studied for its use as a gas-phase adsorbent of sulfide and
other sulfur containing compounds, where the adsorption is theo-
rized to occur either to strong Lewis acid sites of the Al2O3 or to
surface hydroxyl groups via hydrogen bonding.[42–45] While
there are no data available for the aqueous-phase adsorption of
sulfide, it is feasible that all of the sulfide could have been adsorbed
to the Al2O3. If each sulfide atom (the ionic radius of sulfide is
1.84 Å ? cross-sectional area of 10.63 Å2) populates an amount
of area on the support (BET surface area of Pd/Al2O3 = 155 m2/g
[32]) equal to its cross-sectional area, then 25 mg of catalyst would
have a monolayer adsorption sulfide capacity of 2.19 mmoles,
tion of 7 lL compared to 3 lL used in this study) [7].
TCE HDC activity of the Pd/Au NP and Pd NP catalysts monoton-
ically decreased with added sulfide, while added sulfide had little
effect on the Pd/Al2O3 catalyst until S:Pdsurf ratio of 1. The latter
observation could be due to the Al2O3 support material. Al2O3has

102
K.N. Heck et al. / Journal of Catalysis 267 (2009) 97–104
which is six orders of magnitude greater than the largest sulfide
amount added to the Pd/Al2O3 catalyst in our batch studies (i.e.,
0.49 lL–4.9 nmoles of sulfide).
the underlying Au uncovered [49,50]. It is well known that sulfide
compounds bind to Au surfaces [47,48,51,52]. Wierse et al. re-
ported that sulfide ions can adsorb weakly onto Au surfaces in
water up to surface saturation coverages of 0.5–0.6 S:Ausurf [52].
Adsorption onto any exposed Au would reduce the sulfide amount
available to bind to the presumed Pd-based active sites.
To test this idea, we combined varying amounts of gold-only
NPs to the Pd/Au NPs (30% ML) prior to the sulfide treatment step,
and determined the reaction rate constants of the resulting NP
mixtures. From Fig. 3, it can be seen that the rate constants did
not change when Au NPs were added, indicating that the sulfide
ions did not adsorb onto the Au NP surfaces. Otherwise, the rate
constants were expected to increase as increasing amounts of Au
NPs removed the sulfide from solution. The non-effect of additional
gold surface area indicated that the sulfide preferentially bonded
to Pd-based sites (i.e., Pd atoms or Pd–Au mixed sites) over pure
Au sites. According to Zhang et al., the Pd–S bond of thiol capped
Pd NPs is shorter than the Au–S bond in thiol capped Au NPs, sug-
gesting Pd–S bonding is stronger than Au–S bonding and therefore
more preferred [53].
The deactivation of Pd NPs by sulfide could be analyzed more
quantitatively. Complete deactivation observed at a S:Pdsurf ratio
of 0.5, suggesting that 2 Pd atoms are involved in the HDC reaction.
This ratio is close to those found from STM and EXAFS studies of
H2S on Pd surfaces, in which surface saturation corresponded to
a S:Pdsurf ratio of 0.43 (=3 S atoms chemisorbed on 7 Pd atoms
[14–16]) for Pd(111) and 0.50 for Pd(001) [13]. The rapid decrease
in TCE HDC reaction rate with sulfide content indicated preferen-
tially adsorption on Pd atoms of highest activity, which are often
correlated to atoms with the lowest coordination number [13,46].
Interestingly, the Pd/Au NP catalyst deactivated more slowly
than the Pd NPs and Pd/Al2O3. This material did not fully deactivate
until a S:Pdsurf ratio one of ꢀ1 was reached (Fig. 2d). The Au appar-
ently enhances the sulfide poisoning resistance of Pd for TCE HDC.
This beneficial presence of Au was seen in gas-phase hydrodesulfu-
rization thiophene using alloyed Pd–Au/Al2O3 catalysts by Venezia
et al. [47,48], for which it was concluded that Au prevented the
bulk formation of catalytically inactive Pd4S crystallites.
3.5. Effect of Pd/Au NP composition
3.4. Effect of adding Au NPs to Pd/Au NPs for TCE HDC
Sulfide deactivation experiments were carried out on Pd/Au NP
catalysts with varying Pd surface coverages (Fig. 4a). Besides 30%
ML coverage, 15% and 60% ML coverages were studied. All materi-
als exhibited lowered reaction rates with added sulfide, but the
deactivation characteristics differed. The 15% and 30% deactivated
with a linear dependence on S:Pdsurf ratio, and the much more ac-
tive 60%-ML Pd/Au NPs deactivated much more rapidly at low
S:Pdsurf ratios. Complete deactivation occurred at S:Pdsurf ratios of
1.8, 1.0, and 0.8 for 15%, 30%, and 60% ML coverages, respectively.
The 15% ML samples were poisoned completely with twice the
amount of sulfide per surface Pd atom, whereas the 30% and 60%
ML samples were poisoned completely at the roughly the same
sulfide level.
Analyzing the rate of deactivation change with S:Pdsurf ratio
provided additional insights into the sulfide poisoning effect
(Fig. 4b). For the 30% ML material, the rate of TOF change was con-
stant with increasing sulfide content (at a d(TOF)/d(S:Pdsurf) value
of ꢀꢂ0.57 sꢂ1), which likely indicated the NP catalysts contained
one population of active sites with equivalent catalytic activity.
For the 15% ML material, the rate of TOF change was also constant
at a lower value (with d(TOF)/d(S:Pdsurf) = ꢀꢂ0.15 sꢂ1), indicating
that there was one population of active sites with equivalent cata-
lytic activity and with less sensitivity to sulfides. This lower sus-
ceptibility to sulfide poisoning is consistent Pd/Au NPs (15% ML)
complete deactivating at a S:Pdsurf ratio that was nearly twice that
We considered explaining the enhanced deactivation resistance
the result of adsorption of sulfide onto the exposed Au surface of
Pd/Au NPs. Lambert and coworkers showed that Pd atoms form
two- and three-dimensional islands on Au(111) at room tempera-
ture at low (7%) and high (70%) Pd coverages, respectively, to leave
0.40
0.30
0.20
0.10
0.00
0
1
2
3
Additional Gold Sol (mL)
Fig. 3. Measured initial turnover frequencies for 30% Pd/Au NPs as a function of
additional Au NPs. The amount of sulfur S:Pdsurf ratio was set at 0.3. Line drawn to
guide the eye.
a
1.60
1.0
b
60% ML Pd/Au
Pd/Au NPs (60% ML)
0.0
1.20
0.80
0.40
0.00
30% ML Pd/Au
Pd/Au NPs (30% ML)
15% ML Pd/Au
-1.0
Pd/Au NPs (15% ML)
-2.0
-3.0
-4.0
60% ML Pd/Au
Pd/Au NPs (60% ML)
30% ML Pd/Au
Pd/Au NPs (30% ML)
15% ML Pd/Au
Pd/Au NPs (15% ML)
0.0
0.5
1.0
sSurf:Pd
1.5
2.0
0.0
0.5
1.0
1.5
2.0
S:Pdsurf ratio
S:Pd ratio
Fig. 4. (a) TOF values and (b) rates of TOF change with sulfur amount (calculated using 3-point numerical differentiation) for Pd/Au NPs with varying Pd surface coverages, as
a function of S:Pdsurf ratio. Lines are drawn to guide the eye.

K.N. Heck et al. / Journal of Catalysis 267 (2009) 97–104
103
of Pd/Au NPs (30% ML). This also led to Pd/Au NPs (15% ML) becom-
ing more active than Pd/Au NPs (30% ML) when S:Pdsurf ratios ex-
ceeded ꢀ0.60.
of 0.5, close to the reported surface saturation values for sulfide on
Pd surfaces. The Pd/Au catalysts completely deactivated only at
higher sulfide amounts, indicating a higher level of sulfide poison-
ing resistance. This improved sulfide resistance was not related to
the well-known ability of Au to bind to sulfides and thiol com-
pounds, but rather, to the formation of Pd–Au mixed metal active
sites. The Pd content of the Pd/Au NPs controlled the type of active
site populations on the particle surface. The NPs with a Pd coverage
of 60% ML were the most active but they deactivated most rapidly,
showing complete deactivation at S:Pdsurf = 1. The least active NPs
tested were those with a Pd coverage of 15% ML, which deactivated
least rapidly, with complete deactivation at S:Pdsurf of ꢀ2. These Pd/
Au NPs can be the basis for highly active and highly stable catalysts
for groundwater treatment applications.
The 60% ML material showed a more complex rate of TOF change
with sulfide content, in which d(TOF)/d(S:Pdsurf) = ꢀꢂ3.0 sꢂ1 at low
S:Pdsurf ratios and decreased to ꢀꢂ0.51 dsꢂ1 at high S:Pdsurf ratios.
This indicated the presence of more than one population of active
sites on the catalyst surface, in which the highly active sites deacti-
vated first in the presence of sulfides and the less active sites deac-
tivated more slowly. These less active sites of the Pd/Au NPs (60%
ML) could be equivalent to those of Pd/Au NPs (30% ML), based on
the approximately similar rates of TOF change. Furthermore, the
Pd/Au NPs (60% ML) did not contain the low-activity but more-
deactivation-resistant active sites found in Pd/Au NPs (15% ML).
Sautet and coworkers conducted both theoretical and experi-
mental studies on the dechlorination reaction of TCE over a model
(110) PdCu catalyst, consisting of alternating rows of Pd and Cu at
the surface [54,55]. It was shown that the carbon–carbon double
bond preferentially adsorbed on two adjacent Pd sites, while the
carbon–chloride bond cleavage was assisted by the Cu atoms, onto
which chloride atoms preferentially remained. These studies
underlined the importance of Pd and Cu adjacency in TCE dechlo-
rination, in which mixed metal surface atoms behave as catalyti-
cally active sites.
Acknowledgments
This work is supported by NSF (IGERT, DGE-0504425; CBEN,
EEC-0647452), the Welch Foundation (C-1676) and SABIC
Americas. We acknowledge helpful discussions with Mr. Y.-L.
Fang and Mr. R. J. Smith.
References
[1] N.C.f.E. Assessment, Sources, Emission, and Exposure for Trichloroethylene
(TCE) and Related Chemicals, US Environmental Protection Agency,
Washington, DC, 2001.
[2] J.S. Zogorski, J.M. Carter, T. Ivahnenko, W.W. Lapham, M.J. Moran, B.L. Rowe, P.J.
Squillace, P.L. Toccalino, The quality of our nation’s waters: Volatile organic
compounds in the nation’s ground water and drinking-water supply wells;
Circular 1292. US Geological Survey, Reston, VA, 2006.
[3] National Primary Drinking Water Regulations, United States Environmental
Protection Agency, 1995.
[4] ASTDR, Top 20 Hazardous Substances, 2005.
While it cannot be separated from ensemble (clustering of
atoms as active site), electronic (ligand), and geometric (e.g., strain,
interatomic distance changes) effects, this mixed metal site effect
could explain some of the poisoning trends of Pd/Au NPs with dif-
ferent Pd coverages. The 15%-ML material perhaps contains Pd me-
tal in the form of isolated atoms or small two-dimensional Pd
ensembles surrounded by Au atoms, which would not be found
in the NPs with higher Pd content. These Pd atoms or small ensem-
bles would have a large Pd–Au interface at which TCE HDC can oc-
cur and which sulfide can chemisorb and poison. The 30%-ML
material contains no isolated Pd atoms but contains larger Pd
ensembles such that its population of active sites is more active
and more sensitive to sulfide poisoning, with the 60%-ML material
containing even larger Pd ensembles. Baddeley et al. indicated that
Pd atoms assembled into three-dimensional ensembles (2 atom
layers thick) on a Au(111) surface at 70% monolayer coverage
[49,50], suggesting that Pd/Au NPs (60% ML) could also contain
3-D Pd ensembles. We thus associate the population of highly ac-
tive and sulfide-sensitive sites in this material to 3-D Pd ensembles
and the less active and less sulfide-sensitive sites to smaller 2-D
ensembles. The latter are found in the 30%-ML material but not
in the 15%-ML one. It is presumed that the Pd-shell/Au-core nano-
structure is retained and that these presumptive Pd ensembles do
not rearrange during the reaction. While our results do not confirm
these possibilities at this point, the technique of in situ extended X-
ray absorption spectroscopy as applied to the room-temperature
TCE HDC water-phase reaction catalyzed by ‘‘naked” nanoparticles
will be very helpful for verification.
[5] Y.J. An, D.H. Kampbell, J.W. Weaver, J.T. Wilson, S.W. Jeong, Environ. Pollut. 130
(2004) 325–335.
[6] M.O. Nutt, J.B. Hughes, M.S. Wong, Environ. Sci. Technol. 39 (2005) 1346–1353.
[7] M.O. Nutt, K.N. Heck, P. Alvarez, M.S. Wong, Appl. Catal. B 69 (2006) 115–125.
[8] B.G. Ponec V, Catalysis by Metals and Alloys, Elsevier, 1995.
[9] M.S. Wong, P.J.J. Alvarez, Y.-l. Fang, N. Akcin, M.O. Nutt, J.T. Miller, K.N. Heck, J.
Chem. Technol. Biotechnol. 84 (2009) 158–166.
[10] C. Schuth, N.A. Kummer, C. Weidenthaler, H. Schad, Appl. Catal. B-Environ. 52
(2004) 197–203.
[11] W.W. McNab, R. Ruiz, M. Reinhard, Environ. Sci. Technol. 34 (2000) 149–153.
[12] F.D. Kopinke, K. Mackenzie, R. Kohler, Appl. Catal. B 44 (2003) 15–24.
[13] D. Burgler, G. Tarrach, T. Schaub, R. Wiesendanger, H.J. Guntherodt, Phys. Rev.
B 47 (1993) 9963–9966.
[14] V.R. Dhanak, A.G. Shard, B.C.C. Cowie, A. Santoni, Surf. Sci. 410 (1998) 321–329.
[15] J.G. Forbes, A.J. Gellman, J.C. Dunphy, M. Salmeron, Surf. Sci. 279 (1992) 68–78.
[16] M.E. Grillo, C. Stampfl, W. Berndt, Surf. Sci. 317 (1994) 84–98.
[17] B. Coq, G. Ferrat, F. Figueras, J. Catal. 101 (1986) 434–445.
[18] G.V. Lowry, M. Reinhard, Environ. Sci. Technol. 34 (2000) 3217–3223.
[19] C.G. Schreier, M. Reinhard, Chemosphere 31 (1995) 3475–3487.
[20] M.N. Berube, B. Sung, M.A. Vannice, Appl. Catal. 31 (1987) 133–157.
[21] J. Sepulveda, N. Figoli, React. Kinet. Catal. Lett. 53 (1994) 155–160.
[22] P.C. Largentiere, N.S. Figoli, J. Chem. Technol. Biotechnol. 48 (1990) 361–368.
[23] D.O. Simone, T. Kennelly, N.L. Brungard, R.J. Farrauto, Appl. Catal. 70 (1991)
87–100.
[24] B. Miranda, E. Diaz, S. Ordonez, A. Vega, F.V. Diez, Appl. Catal. B 64 (2006) 262–
271.
[25] D. Roth, P. Gelin, M. Primet, E. Tena, Appl. Catal. A 203 (2000) 37–45.
[26] P. Marecot, A. Fakche, B. Kellali, G. Mabilon, M. Prigent, J. Barbier, Appl. Catal. 3
(1994) 283–294.
[27] C.H. Xia, H. Xu, W.Z. Wu, X.M. Liang, Catal. Commun. 5 (2004) 383–386.
[28] C.D. Thompson, R.M. Rioux, N. Chen, F.H. Ribeiro, J. Phys. Chem. B 104 (2000)
3067–3077.
4. Summary and conclusions
[29] N. Munakata, M. Reinhard, Appl. Catal. B 75 (2007) 1–10.
[30] D. Angeles-Wedler, K. Mackenzie, F.-D. Kopinke, Environ. Sci. Technol. 42
(2008) 5734–5739.
[31] M.G. Davie, H. Cheng, G.D. Hopkins, C.A. LeBron, M. Reinhard, Environ. Sci.
Technol. 42 (2008) 8908–8915.
[32] G.V. Lowry, M. Reinhard, Environ. Sci. Technol. 33 (1999) 1905–1910.
[33] J. Park, J.R. Regalbuto, J. Colloid Interface Sci. 175 (1995) 239–252.
[34] M.A.A. Schoonen, H.L. Barnes, Geochim. Cosmochim. Acta 52 (1988) 649–654.
[35] T. Teranishi, M. Miyake, Chem. Mater. 10 (1998) 594–600.
[36] L.N. Lewis, Chem. Rev. 93 (1993) 2693–2730.
This study analyzed the poisoning effects of Pd/Al2O3, Pd NP, and
Pd/Au NP catalysts for the aqueous-phase hydrodechlorination of
trichloroethene. Pd NPs and Pd/Al2O3 deactivated in the presence
of chloride due to Cl chemisorption onto Pd active sites. In contrast,
Pd/Au NPs did not deactivate, which was speculated to result from
the electronic effect of Au on Pd. Pd/Al2O3 catalyst showed little
deactivation behavior in the presence of sulfides due to their inter-
actions with the support; alumina can play an important role in
mitigating sulfide poisoning when operating with real groundwa-
ter. The Pd NP catalyst became completely inactive at a S:Pdsurf ratio
[37] J.M. Thomas, Pure Appl. Chem. 60 (1988) 1517–1528.
[38] A.L. Mackay, Acta Crystallogr. 15 (1962) 916.
[39] Y. Fang, K.N. Heck, P. Alvarez, M.S. Wong, in preparation.

104
K.N. Heck et al. / Journal of Catalysis 267 (2009) 97–104
[40] A. Carrasquillo, J.J. Jeng, R.J. Barriga, W.F. Temesghen, M.P. Soriaga, Inorg. Chim.
Acta 255 (1997) 249–254.
[41] K.N. Heck, B.G. Janesko, G.E. Scuseria, N.J. Halas, M.S. Wong, J. Am. Chem. Soc.
130 (2008) 16592–16600.
[42] R.W. Glass, R.A. Ross, J. Phys. Chem. 77 (1973) 2576–2578.
[43] A. Ionescu, A. Allouche, J.P. Aycard, M. Rajzmann, F. Hutschka, J. Phys. Chem. B
106 (2002) 9359–9366.
[47] A.M. Venezia, V. La Parola, V. Nicoli, G. Deganello, J. Catal. 212 (2002) 56–62.
[48] A.M. Venezia, V. La Parola, G. Deganello, B. Pawelec, J.L.G. Fierro, J. Catal. 215
(2003) 317–325.
[49] C.J. Baddeley, R.M. Ormerod, A.W. Stephenson, R.M. Lambert, J. Phys. Chem. 99
(1995) 5146–5151.
[50] A.W. Stephenson, C.J. Baddeley, M.S. Tikhov, R.M. Lambert, Surf. Sci. 398 (1998)
172–183.
[44] Y. Okamoto, M. Ohhara, A. Maezawa, T. Imanaka, S. Teranishi, J. Phys. Chem. 90
(1986) 2396–2407.
[45] O. Saur, T. Chevreau, J. Lamotte, J. Travert, J.C. Lavalley, J. Chem. Soc.-Faraday
Trans. I 77 (1981) 427–437.
[46] G. Somorjai, Introduction to Surface Chemistry and Catalysis, Wiley
Interscience, New York, 1994.
[51] R.L. McCarley, Y.T. Kim, A.J. Bard, J. Phys. Chem. 97 (1993) 211–215.
[52] D.G. Wierse, M.M. Lohrengel, J.W. Schultze, J. Electroanal. Chem. 92 (1978)
121–131.
[53] P. Zhang, T.K. Sham, Mater. Res. Soc. Symp. Proc. 738 (2003) G.13.14.1.
[54] Y. Jugnet, J.C. Bertolini, L. Barbosa, P. Sautet, Surf. Sci. 505 (2002) 153–162.
[55] S. Barbosa, J. Catal. 207 (2002) 127–138.