Surface and bulk structural response of Pt black upon its hydrogen treatment and catalytic reaction with n-hexane

Article abstract of DOI:10.1039/b007837i

Pt black has been studied by X-ray and ultraviolet photoelectron spectroscopies (XPS, UPS) and XRD after reduction, after presintering and after subjection to various in situ treatments. All samples contained carbon and oxygen impurities. Sintering decreased the abundance of C. Samples introduced from the air contained a large amount of oxygen whose amount dropped markedly upon keeping the Pt in UHV for several hours. XPS and UPS revealed considerable amounts of nondissociatively chemisorbed CO after this treatment. An in situ hydrogen treatment at 603 K decreased the O content further but increased the concentration of surface carbon. The Pt 4f regions showed some oxidized Pt both before and after sintering. Pt reached an almost clean metallic state after UHV treatment. This state seemed to remain unchanged after further manipulations. Hardly any electronic interaction could thus be observed between Pt and its main impurity: C, which was present mostly as graphite and C_xH_y polymer after treatment with n-hexane plus hydrogen. n-Hexane alone produced mostly "disordered" surface carbon species. The intensity of higher-order X-ray reflections of Pt was suppressed upon sintering. This anisotropy was reversed after in situ H₂ treatments, inducing recrystallization with preferential formation of higher Miller-index planes, (220) and (311). These reflections were again suppressed after exposure to n-hexane. Thus, adsorbate-induced solid-state rearrangement occurred as a result of the interplay of surface and subsurface impurities. The catalytic reactions of n-hexane over Pt black subjected to different pretreatments were different: abundant (220) and (311) reflections promoting isomerisation, C₅-cyclisation and also hydrogenolysis. Carbon accumulation decreased the "intrinsic" activity of the Pt fraction detected by XPS. Selectivities, in turn, were governed by the crystallite structure as well as by the presence of composite platinum-carbon sites. "Disordered" surface carbon species poisoned all skeletal reactions and favoured dehydrogenation to hexenes.

Full text of DOI:10.1039/b007837i

View Article Online / Journal Homepage / Table of Contents for this issue
Surface and bulk structural response of Pt black upon its hydrogen
treatment and catalytic reaction with n-hexane
Zoltan Paal,ab Ute Wild,a Attila Wootsch,b Josef Find¤a and Robert Schlogla
  Ž
a Fritz-Haber-Institut der MPG, Faradayweg 4È6, D-14195 Berlin, Germany
b Institute of Isotope & Surface Chemistry, CRC, Hungarian Academy of Sciences, P. O. Box 77,
Budapest, H-1525 Hungary. E-mail: paal=iserv.iki.kfki.hu
Received 27th September 2000, Accepted 4th April 2001
First published as an Advance Article on the web 8th May 2001
Pt black has been studied by X-ray and ultraviolet photoelectron spectroscopies (XPS, UPS) and XRD after
reduction, after presintering and after subjection to various in situ treatments. All samples contained carbon
and oxygen impurities. Sintering decreased the abundance of C. Samples introduced from the air contained a
large amount of oxygen whose amount dropped markedly upon keeping the Pt in UHV for several hours. XPS
and UPS revealed considerable amounts of nondissociatively chemisorbed CO after this treatment. An in situ
hydrogen treatment at 603 K decreased the O content further but increased the concentration of surface
carbon. The Pt 4f regions showed some oxidized Pt both before and after sintering. Pt reached an almost clean
metallic state after UHV treatment. This state seemed to remain unchanged after further manipulations.
Hardly any electronic interaction could thus be observed between Pt and its main impurity: C, which was
present mostly as graphite and C H polymer after treatment with n-hexane plus hydrogen. n-Hexane alone
x y
produced mostly ““disorderedÏÏ surface carbon species. The intensity of higher-order X-ray reÑections of Pt was
suppressed upon sintering. This anisotropy was reversed after in situ H treatments, inducing recrystallization
2
with preferential formation of higher Miller-index planes, (220) and (311). These reÑections were again
suppressed after exposure to n-hexane. Thus, adsorbate-induced solid-state rearrangement occurred as a result
of the interplay of surface and subsurface impurities. The catalytic reactions of n-hexane over Pt black
subjected to di†erent pretreatments were di†erent: abundant (220) and (311) reÑections promoting
isomerisation, C -cyclisation and also hydrogenolysis. Carbon accumulation decreased the ““intrinsicÏÏ activity
5
of the Pt fraction detected by XPS. Selectivities, in turn, were governed by the crystallite structure as well as by
the presence of composite platinumÈcarbon sites. ““DisorderedÏÏ surface carbon species poisoned all skeletal
reactions and favoured dehydrogenation to hexenes.
photoelectron spectroscopic and X-ray di†raction studies9,10
Introduction
conÐrmed the restructuring of Pt black upon its exposure to
O , H and n-hexane/H mixtures. Hydrogen treatment of Pt
The ““Ñexible surfaceÏÏ concept1,2 of Somorjai underlines the
importance of ““rough surfacesÏÏ in catalysis as opposed to
close-packed, low Miller-index faces. Reconstruction induced
by chemisorbed entitiesÈincluding reactants of catalytic
processesÈcan also produce ““roughÏÏ surfaces.3 O , H or
2
2
2
black at 600 K resulted in ““retentionÏÏ of hydrogenÈ
occupying partly subsurface positionsÈas shown by radio-
tracer studies.11 UPS results indicated that H at 600 K
penetrated into the subsurface layers and forced O to the
2
2
CO, each adsorbed at 1 bar pressure, generated adsorbate-
speciÐc surface reconstruction of the Pt (110) single crystal
plane.4 Thermal restructuring of Pt particles required at least
surface. This O reacted with hydrogen when the sample was
ads
left at 600 K overnight and produced minor but stable surface
species like OH and H O.12,13
2
1073 K.5
Regeneration with O and H is a well-established process
2
2
High-surface Pt black that had been prepared by aqueous
for the restoration of the activity of dispersed Pt. Even a poly-
reduction of H PtCl with formaldehyde underwent consider-
able sintering during its very Ðrst contact with H between
D480 and D680 K.6 No sintering occurred in vacuum,
crystalline foil can be puriÐed more efficiently by this
2
6
““chemicalÏÏ method than by ion sputtering alone.12 Such
2
treatments of Pt black bring about a stable and reproducible
catalyst activity and selectivity in n-hexane model reac-
tions.8,13,14 The ““roughÏÏ surfaces of unsupported Pt
agglomerates7,8 ““do chemistryÏÏ in the sense of SomorjaiÏs
concept.1,2 We present here an integrated approach combin-
ing XRD to study structural changes and photoelectron spec-
troscopy to monitor sample composition and the valence
states of the components in Pt black after its treatment in
various atmospheres. Test runs with n-hexane/hydrogen mix-
tures after analogous treatments complete the picture. The
results of the physical and catalytic measurements correlated
well with each other. These experiments represent two sides of
helium or air under the same conditions.6 Hydrogen may
have removed the majority of surface impurities and pro-
moted coalescence of particles puriÐed in this way.6 This
sintering process resulted in agglomerates of rounded, larger
crystallites.7 When the solution of H PtCl was reduced with
2
6
hydrazine, somewhat larger, polyhedral crystallites were
formed. They also became rounded upon their Ðrst hydrogen
treatment without considerable crystallite growth.8 Recent
¤
Present address: Lehrstuhl II fu
Ž
Ž
r Technische Chemie, TU
Munchen, D-85747 Garching, Germany.
2148
Phys. Chem. Chem. Phys., 2001, 3, 2148È2155
This journal is ( The Owner Societies 2001
DOI: 10.1039/b007837i

View Article Online
energy (PE) \ 12 eV. A Mg-Ka anode was used for XPS
(
PE \ 48 eV). The peaks were evaluated after satellite subtrac-
tion and Shirley background subtraction. The binding energy
BE) was calibrated to the Au 4f 7/2 line (BE \ 84.0 eV).
(
Atomic compositions were determined from the peak areas
with literature sensitivity factors18 using the homogeneous
composition model. Di†erence spectra were obtained either by
direct subtraction or after normalizing the intensities at an
appropriate BE value.19
X-Ray di†raction
XRD measurements were carried out on samples taken from
the spectrometer at room temperature. A STOE double-circle
transmission mode di†ractometer was used in DebyeÈScherrer
geometry, equipped with a Ge(111) primary monochromator
and a position-sensitive detector, using Cu-Ka radiation,
j \ 154.06 pm. XRD measurements were recorded for
samples 1, 2, 4, 6 and 7, interim states between in situ treat-
ments (3 and 5 in our case) could not be measured. The inten-
sities of individual reÑections were normalized to that of (111).
Fig. 1 X-Ray di†ractograms of Pt black in di†erent states. Inten-
sities have been normalized to the strongest, (111), line. The thick ver-
tical lines show the relative intensities according to the JPCD Ðle.
Note the anisotropic recrystallisation after treatments.
the ““pressure gapÏÏ15 between catalytic and surface chemical
studies; its bridging being the hope of most scientists in this
Ðeld.
Catalytic reactions
Experimental
The catalytic behavior of Pt black was tested by reacting a
standard mixture of n-hexane (nH) and hydrogen at
Catalysts and treatments
p(nH) : p(H ) \ 13 : 160 mbar. A closed-loop reactor was
2
used. Sampling took place after 5, 20, 35 and 50 min reaction
Pt black was reduced from an aqueous solution of H PtCl
time. The catalyst pretreatments in the photoelectron spectro-
meter were simulated as closely as possible. In one run the
reactor volume was kept in vacuum at 603 K for several hours
and the hydrogen-rich reaction mixture was introduced
directly afterwards (analogous to preparation of 3). Catalytic
2
6
by D10% hydrazine at 313 K8 and washed with bidistilled
water (sample 1). Its speciÐc surface was D10 m2 g~1. This
preparation was presintered in H Ñow at 473 K for 1 h,6
2
(
sample 2). Its speciÐc surface decreased to 2.6 m2 g~1. Photo-
electron spectra of samples 1 and 2 were measured after
storing them in air for a few months. Electron spectroscopy
was repeated after having kept the ““as-receivedÏÏ sample 2 in
UHV for D60 h (sample 3). Exposure to 270 mbar H fol-
lowed in the preparation chamber of the electron spectrometer
runs were also carried out after the Ðrst H exposure (as after
2
preparation of 4). These were compared with results corre-
sponding to the preparation of 6 (sample regenerated several
times, with well-reproducible activity and selectivity). The
catalytic properties were also checked after two exposures to
2
(
sample 4). Another Pt black sample from the same batch was
n-hexane: with or without H , as in the preparation of 5 and
2
exposed to a mixture of n-hexane (nH) and hydrogen
7
, respectively. Turnover frequencies were calculated using
(
nH/H \ 13 : 80 mbar; sample 5) and was regenerated subse-
“
“nominalÏÏ Pt values obtained from hydrogenÈoxygen titra-
tion results, after treating the presintered catalysts with hydro-
2
s
quently by 27 mbar O , evacuation and 270 mbar H (sample
6
5
2
2
). Sample 7 was obtained after regeneration and exposure to
gen (corresponding to sample 4).
3 mbar nH alone. Measurements on samples 6 and 7 were
preceded by a few hydrocarbon exposures and subsequent
regenerations. All treatments to produce samples 4–7 lasted
for 20 min at 603 K.
Results
Structure of di†erent samples
XPS and UPS
Fig. 1 shows the high-angle sections of X-ray di†ractograms
obtained on Ðve catalyst samples. The X-ray di†raction
pattern of the unsintered Pt (1) showed a slight line broaden-
ing (crystallite size D20 nm, cf. the value of 23 nm reported
for an analogous preparation8). Sintering increased crystallite
size only slightly (D25 nm), as opposed to the behaviour of Pt
reduced by formaldehyde.6 Results on those samples have
been reported recently.10 The relative intensities of individual
reÑections of the unsintered sample (sample 1) were close to
literature values for bulk Pt.20 The X-ray di†ractogram of
presintered Pt (2) showed a marked decrease in the intensities
of higher-order reÑections. This agrees with earlier electron
micrographs revealing that this process changed the poly-
hedral contours of the crystallites to a rounded shape rather
than promoting particle growth. The loss of speciÐc surface
must have been due to agglomeration of the crystallites.8 The
Details of the methods of electron spectroscopy have been
reported earlier.16,17 Dry powder samples were placed on a
stainless steel sample holder evacuated and treated in the
preparation chamber of a Leybold LHS 12 MCD instrument.
UPS spectra were excited by He II radiation (40.8 eV), pass
Table 1 Surface composition of samples in various stages of their
treatment
Composition (atom%)
Sample
O 1s
C 1s
Pt 4f
1
2
3
4
5
6
7
a
15
24
4.5
1
3
2
18
12
11
15
27
23
40
67
64
84.5
84
70
75
59.5
Ðrst H treatment (sample 4) increased the relative intensities
2
of the (311) and (220) reÑections. This anisotropy was
enhanced when the regenerative treatment followed an expo-
0.5
sure to nH/H at 603 K. (sample 6). When the Pt black was
2
exposed to 53 mbar n-hexane without hydrogen (sample 7),
a Disregarding the D7% Cl and D4% N impurity from reduction of
rearrangement in the opposite direction took place, the
higher-order reÑections were suppressed again. Repeated
H PtCl by hydrazine, disappearing completely after Ðrst evacuation.
2
6
Phys. Chem. Chem. Phys., 2001, 3, 2148È2155
2149

View Article Online
Fig. 2 O 1s region of Pt black in di†erent states. The Pt 4f 7/2 inten-
sities were normalized and the same factor used for the O 1s inten-
sities.
Fig. 4 Pt 4f region Pt black samples in di†erent states, together with
their di†erence spectra. Intensities were normalized at the peak posi-
tion of the Pt 4f 7/2 line.
exposure to n-hexane at di†erent temperatures produced
storage in air. Subsequent treatments reduced the percentage
of oxygen still further.
To compare the individual XPS regions of samples 1 to 4,
the intensities of the Pt 4f 7/2 peaks were normalized to the
reÑections analogous to sample 7.9 The lattice constants
(
(
0.3920 to 03924 nm) were very close to the bulk Pt state
0.3923 nm). Very small lattice strain values (up to 0.08%)
were observed. They showed no correlation with the crystal
anisotropy.
peak height measured after H treatment (sample 4). The same
2
normalization factor was applied thereafter to the intensities
of the other regions. The major component of the O 1s region
of untreated Pt samples (Fig. 2) was at D530 eV correspond-
ing to oxygen bound directly to the metal, accumulated
during lengthy storage in air. Its chemical state is not clear: it
may be adsorbed oxygen orÈas observed21 with a highly dis-
Photoelectron spectroscopy after di†erent treatments
Initial treatments of Pt black. Table 1 shows the composi-
tions of samples 1 to 7 determined from XPS line intensities.
Considerable oxygen and carbon contributions appeared in
both untreated samples. Evacuation (sample 3) removed from
sample 2 most of the oxygen that had accumulated during
perse Pt/SiO Èa nearly monomolecular surface layer of PtO.
2
Its amount was larger in sample 2 than in sample 1. Leaving
the sample in UHV for D60 h (sample 3) removed D85% of
the oxygen, as estimated from peak areas (Fig. 2). In addition
to surface OH/H O, a component in the O 1s region at
2
BE B 532.5 eV appeared. This can be assigned to chemisorbed
CO.12 Hydrogen treatment (sample 4) completed the oxygen
removal. Carbon was the major impurity in all cases. It
appeared12 mainly as graphitic C, poly-C H and oxidized C
x y
(
Fig. 3). Narrower peaks appeared upon UHV and hydrogen
Fig. 3 C 1s region of Pt black in di†erent states. The Pt 4f 7/2 inten-
sities were normalized and the same factor used for the C 1s inten-
sities.
Fig. 5 He II UP spectrum of the unsintered (1), sintered (2) and H -
2
treated (4) Pt black samples together with their di†erence spectra
(intensity normalization at 2.45 eV).13
2150
Phys. Chem. Chem. Phys., 2001, 3, 2148È2155

View Article Online
Fig. 6 Pt 4f region of Pt black samples subjected to H treatment for
2
the Ðrst time (4), after exposure to n-hexane/H (5) and after a sub-
2
intensity normalization.
sequent regeneration (6). All di†erence spectra were calculated after
Fig. 8 C 1s region of Pt black samples 4, 6, 7 and their di†erence
spectra calculated as measured.
(
BE B 284.3 eV) being the main component. The Pt 4f 7/2
treatments (3 and 4) pointing to advancing graphitisation. The
marked component at BE B 286 eV in sample 3 (Fig. 3) agrees
well with that reported by Belton and Schmieg for adsorbed
CO.22 The C 1s intensity increased after contacting the
peaks were similar in both samples 1 and 2 with maxima
shifted to higher BE values than that corresponding to Pt0.
The Pt 4f peak did not change much upon presintering. The
negative peak in the di†erence spectra 4 [ 1 and 4 [ 2 with
vacuum-treated Pt with H (from sample 3 to 4) graphitic C
2
*
BE ca. ]0.8 eV corresponds to ““PtO ÏÏ.23 This conÐrmed
ads
that adsorbed oxygen was the main impurity in the untreated
samples.24 Even UHV treatment was sufficient to remove
most of the oxygen. These experimental data are not sufficient
to attribute any deÐnite structure to this oxygen species,
which may not be identical with the atomically chemisorbed
O that would desorb from Pt single crystal surfaces at higher
temperatures.25 The di†erence spectrum 4 [ 3 appeared as a
nearly straight horizontal line (Fig. 4) indicating that hydro-
gen treatment did not purify Pt further.
The He II UP spectra of untreated catalysts (Fig. 5) showed
rather low Fermi-edge intensities. The normalized di†erence
spectrum 1 [ 2 was rather featureless, except for a broad
intensity excess in the unsintered Pt between 2.5 and 4.5 eV.
An increased intensity in this region was attributed to
restructuring of a Pt (100) single crystal surface.26 This would
be in agreement with the loss of some irregularities of the
freshly reduced crystallites upon sintering in H 6h8 as shown
2
also by XRD (Fig. 1). The Fermi-edge intensities conÐrmed
the considerable puriÐcation of Pt after UHV and H2 treat-
ments (sample 4). Nondissociatively adsorbed CO appeared in
spectrum 3 (not shown). Its origin cannot be established
without doubt, hence it will not be discussed further. The
small amount of other adsorbed species at BE B 5 and D7È8
eV may be tentatively assigned to various surface oxygenates,
such as O, OH or H O.12,13,27
2
Exposure to n-hexane and regeneration. Contacting Pt black
with n-hexane (nH) plus H modeled what might happen to
2
Fig. 7 C 1s region of Pt black samples 4, 5, 6. Treatments, see Fig. 6.
Di†erence spectra were calculated after intensity normalization.
the catalyst during alkane reaction in a closed system. The Pt
4f region in sample 5 was identical to that observed after the
Phys. Chem. Chem. Phys., 2001, 3, 2148È2155
2151

View Article Online
Carbon intensities were markedly di†erent in samples 4, 5,
and 7 (Table 1). The C 1s spectra of samples 5 and 6 (Fig. 7)
6
were similar to those reported earlier.9,10 The normalized dif-
ference spectra between the carbonized and regenerated cata-
lysts showed that O /H treatmentÈremoving relatively little
2
2
carbon (Table 1)Èdecreased the amount of graphitic C com-
ponent. At the same time the contribution due to ““atomic CÏÏ
in surface or subsurface positions increased. The distribution
of C 1s components in sample 4 (Ðrst contact with H ) was
2
more similar to that in sample 5 (exposed to nH/H ) than to
2
that in sample 6. This means that there was some graphite
present after the Ðrst contact of Pt with H (Fig. 3) as opposed
2
to more ““Pt/CÏÏ after repeated regenerations (cf. the di†erence
spectra 5 [ 4 in Fig. 7). Exposure to n-hexane without H
2
(
sample 7) not only produced more carbon but also its BE
appeared at a lower value. Di†erence spectra in Fig. 8 calcu-
lated without intensity normalization conÐrm the presence of
Fig. 9 He II UP spectra of sintered Pt black samples 5, 6 and 7.
““atomic CÏÏ in the regenerated sample (5 [ 6). A high amount
of C 1s component at BE B 284.1 eV appeared in sample 7.
The related He II UP spectra (Fig. 9) indicate rather
marked Fermi-edges after each treatment, pointing to the
presence of metallic Pt (as in Fig. 6). The peak doublet at 9.1
and 11.5 eV indicated the presence of adsorbed CO in samples
5 and 6.29 The increased intensity in the region 5.5È10 eV in
spectrum 7 conÐrms the presence of a high amount of carbon
after exposure to 53 mbar nH.
Ðrst contact with H (Fig. 6, spectrum 5 [ 4 calculated
2
without intensity normalization). The Pt 4f doublet indicated
a predominant Pt0 state, hence there was hardly any chemical
interaction between Pt and its major impurity, C.28 The di†er-
ence spectrum 6 [ 4 (Fig. 6) indicated some PtO in sample
ads
4
, this excess being less in sample 7 (spectrum 7 [ 4), in agree-
ment with the surface Pt composition.
Table 2 n-Hexane transformation (13 mbar) in 160 mbar hydrogen on various catalystsa
Reaction time/min
5
20
35
50
First run after evacuation (sample 3):
TOF/h~1
23
30
18
27
20
5
12
29
19
27
20
5
9
29
19
26
21
5
7
30
19
26
21
4
Selectivity (%): \C
6
Isomersb
MCPc
Benzene
Hexenes
Run after Ðrst H treatment (sample 4):
2
TOF/h~1
29
25
19
29
19
8
13
25
22
30
19
4
9.5
7.5
Selectivity (%): \C
24
23
29
19
5
24
23
29
19
5
6
Isomersb
MCPc
Benzene
Hexenes
After exposure to n-hexane/H at 603 K (sample 5):
2
TOF/h~1
12
37
16
16
24
7
5.5
4
37
17
17
26
3
3
38
16
17
26
3
Selectivity (%): \C
37
17
17
26
3
6
Isomersb
MCPc
Benzene
Hexenes
Regenerated, H /O 603 K (sample 6):d
2
2
TOF/h~1
20 ^ 0.82
40 ^ 1.5
18 ^ 0.9
16 ^ 0.8
24 ^ 0.7
2 ^ 0.5
10
39
19
16
24
2
7
39
19
15
25
2
5.5
Selectivity (%): \C
40
19
14
26
1
6
Isomersb
MCPc
Benzene
Hexenes
After exposure to n-hexane without H at 603 K (sample 7):
2
TOF/h~1
2
25
8
10
22
35
1.2
1
34
11
14
26
15
0.8
Selectivity (%): \C
32
11
13
24
20
35
11
14
26
14
6
Isomersb
MCPc
Benzene
Hexenes
a Treatments analogous to those performed in the photoelectron spectrometer. Catalyst mass: 16 mg for 3 and 4; 50 mg for 5, 6, 7. Typical
conversions were 3 to 15%; somewhat higher for 6 and lower for 7. b Sum of methylpentanes. c Methylcyclopentane. d The TOF and selectivity
values at 5 min have been averaged from 21 independent runs after the same regeneration carried out after various test reactions with nH ] H
2
mixtures on the same charge of catalyst. Values for longer times have been taken from one long run.
2152
Phys. Chem. Chem. Phys., 2001, 3, 2148È2155

View Article Online
Catalytic results
“SkeletalÏÏ reactions of n-hexane in mixtures with hydrogen at
tion of (graphitic or disordered30,31) surface carbon entities
that were, likely, not thick enough to suppress the Pt signal
but deactivated the active Pt ensembles in their vicinity. In
spite of the known difficulties of transforming directly XPS
data for real surface compositions, their correlation with
activity data is remarkable. As for the activities of individual
reactions, n-hexane pretreatment decreased Ðrst processes
involving formation and/or rupture of CÈC bonds. All activ-
ities dropped to a disproportionately high degree for sample 7.
However, the dehydrogenation activity to hexenes even
increased and remained constant in both deactivated samples
“
D600 K involve aromatization, C -cyclisation and isomer-
5
isation as well as hydrogenolysis.8,13,14 Table 2 summarises
the catalytic results. The turnover frequencies (TOF) for the
three catalysts treated in vacuum or in hydrogen prior to the
run (samples 3, 4 and 6) were rather close to each other. The
table contains average values for the 5 min run on the
“
“regeneratedÏÏ sample 6, indicating the good reproducibility of
the catalytic properties of this state.
The selectivities (S) were also well-reproducible after stan-
dard regeneration (Table 2). The di†erences after various
pretreatments were rather pronounced. Sample 6 produced
most fragments. The most active catalysts after vacuum and
(
6 and 7) in agreement with the suggestion that single-atom Pt
atoms (or ““PtCÏÏ ensembles) can be the appropriate active
sites.32,33
Ðrst H treatments (samples 3 and 4) produced most isomers
2
Discussion
(
S D 20%) and methylcyclopentane (MCP) up to S D 30%.
The fragmentation selectivity was highest with the ““steady-
stateÏÏ regenerated Pt (sample 6). The benzene selectivity was
rather constant (20È25%), that of hexenes remained low (2 to
Surface spectroscopy and catalytic measurements represent
two sides of the ““material and pressure gapÏÏ, as pointed out
by Bonzel.15 Nevertheless, the catalytic properties and active
sites on single crystals, metal Ðlms and unsupported metal
powders can essentially be discussed in an analogous way.34
The necessity of leaving ““the comfortable world of single
crystalsÏÏ has been pointed out recently.35 Unsupported Pt
black may be a unique model catalyst, its study being comple-
mentary to rather than competing with single crystal investi-
gations. (i) Its particles are certainly more ““ÑexibleÏÏ than that
of macroscopic single crystals. (ii) The size of individual crys-
tallites is large enough to carry out XRD but small enough
that this bulk method can also monitor the reconstruction of
a few surface layers.10 (iii) It exposes di†erent crystal planes
like catalysts of practical importance. (iv) The absence of
carrier facilitates electron spectroscopy. (v) Interactions
between very small metal particles and the support, arising
from chemical bonds present on the metal/support interface36
are also eliminatedÈremember that the catalytic behaviour of
5
%), but its standard deviation was largest. Deactivation
during a test run caused an approximately fourfold decrease
in the initial TOF value during 50 min reaction time. The
selectivities were, in turn, constant during this period (Table
2).
Deactivation with nH/H (sample 5) increased the surface
2
carbon concentration and also decreased the TOF by about
50% but caused hardly any selectivity changes as compared to
sample 6. The highly deactivated sample 7 was, however, in a
class of its own. Its carbon percentage was about twice that of
the regenerated sample 6 (Table 1) but its activity was lower
by an order of magnitude. It produced initially mainly
hexenes (up to 35%), at the expense of C saturated products
and fragments, but with hardly any change in the benzene
selectivity. The selectivity changes during longer runs indi-
cated a slow reactivation of skeletal processes, as shown also
by the TOF value at the end of run.
The number of surface Pt atoms used for calculating the
TOF values in Table 2 was measured in a state corresponding
to sample 4. XPS indicated 84% Pt on its surface. Table 3
contains corrected TOF values of other samples by consider-
ing the ratio of their actual Pt percentages measured by XPS
to the value observed for sample 4. Sample 3 showed the same
percentage of Pt, so no correction was necessary. It was less
active than sample 4. In all other cases the percentage of Pt
shown by XPS was lower, therefore, fewer Pt atoms were
responsible for the transformation measured and thus their
corrected TOF values were higher than those in Table 2. The
overall corrected activity of samples 5 and 7 also proved to be
lower with this calculation method, indicating that the likely
number of ““workingÏÏ active sites was lower than estimated
from the XPS Pt signal. This lower catalytic activity (sample 5
and especially sample 7) was concomitant with the accumula-
6
0
.2% Pt/Al O (Pt B 1 nm) could not be modeled by single
2 3
crystal faces, in contrast to catalysts with 2 to 10% Pt
loading.37
The (rounded) crystallites of our Pt black expose a great
variety of surface microfacets.38 It has been pointed out
already by Taylor39 that the reactant itself may ““selectÏÏ the
appropriate active sites. Surface reconstruction may be a sig-
niÐcant step in creating those sites, in agreement with the
“
“Ñexible surfaceÏÏ concept.1,2 It may occur largely as a result
of the e†ect of chemisorbed species.4 All these may agree with
the earlier idea (postulated then on a rather intuitive basis40)
that the catalyst, the reactant and other components present
create together the active ““catalytic systemÏÏ.
Rather marked changes occurred on the surface of the ““as-
receivedÏÏ sample 2 during its Ðrst treatment in the spectro-
meter (samples 3 and 4). In spite of various amounts of
impurities present, a reasonably clean Pt0 state appeared after
these treatments (Fig. 4). Surface O entities were easily
removed. The residual O appeared mainly as adsorbed OH
Table 3 Normalized turnover frequencies (TOF) of various n-
hexane reactionsa
and H O (Fig. 2), being, likely, the reaction products between
2
hydrogen and oxygen.12,13
Sample
Overall
\C
Iso
MCP
Benzene
nH/
Although the e†ect of O may be more conspicuous with
large single crystal surfaces,4 the most important component
inducing reconstruction of unsupported catalyst particles was
found to be hydrogen.6 Exposing a 6(111) ] (100) stepped Pt
6
“
3
4
“FreshÏÏ catalysts
23
29
6.9
7.2
4.2
5.5
6.2
8.5
4.6
5.5
1.1
2.3
single crystal surface to H caused a, reversible, doubling of
“
“UsedÏÏ catalysts
2
step height and terrace width.41 A surface H : Pt ratio as low
6
5
7
22.5
14.5
2.8
9
5.4
0.7
4
2.3
0.22
3.6
2.3
0.28
5.4
3.5
0.6
0.5
1
1
as 1 : 1 was sufficient to induce reconstruction of Pt(100).42 A
highly enhanced mobility of [PtÈH] complexes was reported
as compared to single Pt atoms.43 The XRD of ““as-reducedÏÏ
Pt black (1) compared with that measured upon Ðrst sintering
a The TOF values of Table 2 normalized to the Pt% measured in
sample 4 (analogous to the pretreatment for H ÈO titration) on the
basis of the Pt percentages of Table 1. TOF for individual reactions
were calculated from the initial selectivities of Table 2.
2
2
in
H
(2) conÐrms the considerable hydrogen-induced
2
restructuring.6,8,10 Higher-index reÑections: (220), (311) and
(222) were favoured after H treatments in the electron
2
Phys. Chem. Chem. Phys., 2001, 3, 2148È2155
2153

View Article Online
spectrometer (samples 4 and 6). Contacting Pt with hexane
reverse reconstruction which also
appeared after long storage in air. The less pronounced
amount of disordered carbon (BE B 284.1 eV) was most dele-
terious to all skeletal reactions. Dehydrogenation activity
increased in carbonized catalysts. In addition, the treatment
with n-hexane in the absence of hydrogen (sample 7) resulted
not only in the accumulation of much carbon but also in sup-
pression of the (311) and (222) reÑections. Thus, the activity of
Pt black is governed not only by its purity but also by its
(
sample 7) caused a
anisotropy after the Ðrst contact of Pt with H (sample 4 vs.
2
sample 6) indicates that movement of H and C atoms (present
in ample amount after nH treatments) between di†erent posi-
tions can be important in crystallite reconstruction. These
atoms were reported to be more stable on the Pt surface
restructuring under the e†ect of H .
2
above 250 K than CH entities.44
x
The reconstruction demonstrated by XRD after H treat-
Acknowledgements
2
ment may have been related to penetration of hydrogen atoms
Z.P. thanks the Max-Planck-Gesellschaft for inviting him as a
guest scientist to the Fritz-Haber-Institut, Berlin. The catalytic
measurements in Budapest were supported by the Hungarian
National Science Foundation, Grant No T 025599.
into subsurface layers.11,13 These might obviously have mobi-
lized other subsurface entities. One of the possible explana-
tions for the increased C 1s intensity after H treatment (Fig.
2
3)
could be their agglomeration into three-dimensional
(
possibly graphite-like) carbon promoted by hydrogen.45
References
Indeed, the Pt signal in ion scattering spectroscopy of a
carbon-covered Pt black increased by about 6 times, at the
same time, the percentage of Pt calculated from the Pt 4f XPS
signal only increased from 56 to 70%.16 This points to the
presence of thin C deposits permitting the underlying Pt to
appear in XPS. This reconstruction of surface carbon may be
the reason why sample 4 was more active than the apparently
1
2
3
G. A. Somorjai, Catal. L ett., 1992, 12, 17.
G. A. Somorjai, J. Mol. Catal. A, 1996, 107, 39.
G. A. Somorjai, in Introduction to Surface Chemistry and Cataly-
sis, Wiley, New York, 1994.
4
5
6
B. J. McIntyre, M. Salmeron and G. A. Somorjai, J. V ac. Sci.
T echnol. A, 1993, 11, 1964.
S. Nishiyama, S. Matsuura, H. Morita, S. Tsuruya and M. Masai,
Appl. Catal., 1985, 15, 185.
““cleanerÏÏ sample 3. The hydrogen-induced migration of sub-
T. Baird, Z. Paa
T rans. 1, 1973, 69, 50; T. Baird, Z. Paa
Chem. Soc., Faraday T rans. 1, 1973, 69, 1237.
Z. Paal, H. Zimmer, J. R. Gunter, R. Schlogl and M. Muhler, J.
Catal., 1989, 119, 146.
Z. Paal, Zh. Zhan, E. Fu
9.
J. Find, Z. Paa
9.
J. Find, Z. Paa
Stud. Surf. Sci. Catal., 2000, 130, 2291.
11 Z. Paal and S. J. Thomson, J. Catal., 1973, 30, 96.
Z. Paal, S. Schlogl and G. Ertl, J. Chem. Soc., Faraday T rans.,
992, 88, 1179.
13 Z. Paal and Z. Zhan, L angmuir, 1997, 13, 3752.
Z. Paal, X. L. Xu, J. Paal-Lukacs, W. Vogel, M. Muhler and S.
Schlogl, J. Catal., 1995, 152, 252.
15 H. P. Bonzel, Surf. Sci., 1977, 68, 236.
16 Z. Paal, K. Matusek and M. Muhler, Appl. Catal. A, 1997, 149,
13.
Z. Paal and S. Schlo

l and S. J. Thomson, J. Chem. Soc., Faraday
surface C atoms to the surface, however, cannot be excluded.
Exposure to n-hexane resulted in accumulation of surface
C, present, in part, as three-dimensional overlayers:12,46

l and S. J. Thomson, J.
7
8
9
0

Ž
Ž
graphite as well as poly-C H . The BE values for carbon
x y

Ž
lo
Ž
p and B. Tesche, J. Catal., 1995, 156,
gl and U. Wild, Catal. L ett., 2000, 65,
l, H. Sauer, R. Schlogl, U. Wild and A. Wootsch,
species likely present on the Pt surface have been summarized
recently.30 More graphite was formed from nH/H (sample 5)
than with n-hexane alone (sample 7). Nongraphitic carbon
1
2
l, R. Schlo
 Ž
1
(
BE B 284.1 eV) was present in the latter case. Counter claims
1

Ž
can be found concerning its structureÈwe may identify them
with the ““disordered carbon overlayerÏÏ47 or with the ““several
hydrocarbon fragments intermediate between ethylidyne and

1
2

Ž
1
a dehydrogenated C speciesÏÏ.48 This particular BE can arise
from several, single and/or multiple bonds between C and Pt,
2

1
4



more numerous than in the case of graphite and di†erent from
that of ““Pt/atomic CÏÏ entities. The latter was more abundant
after H2 treatment. Its presence can point to (a) hydrogen
forcing some C atoms from subsurface layers to the surface
and/or (b) splitting up of a fraction of polymerized surface
entities. This latter process may be important in regeneration
and can lead to carbon removal as methane.49
Ž

1
1
1
7
8

Ž
gl, Surf. Interface Anal., 1992, 19, 524.
Practical Surface Analysis, ed. D. Briggs and M. P. Seah, Wiley,
Chichester, 1990, Appendix 6, vol. 1, p. 635.
1
9
U. Wild, N. Pfa
1997, 357, 420.
20 JPCD Ðle (No. 4802).
nder and R. Schlogl, FreseniusÏ J. Anal. Chem.,
Ž Ž
Skeletal reactions of n-hexane reportedly require various
active sites. They are ensembles of threefold symmetry for
2
2
2
1
2
3
V. Gnutzmann and W. Vogel, J. Phys. Chem., 1990, 94, 4991.
D. N. Belton and S. J. Schmieg, Surf. Sci., 1990, 233, 131.
K. S. Kim, N. Winograd and R. E. Davis, J. Am. Chem. Soc.,
benzene formation,50 two neighbouring atoms for C ring
5
closure and opening51 while larger ensembles may possibly be
active for fragmentation.52 The importance of ““ledge
1
971, 93, 6296.
Z. Paal, K. Matusek and M. Muhler, Appl. Catal. A, 1997, 149,
113.
structuresÏÏ (or Ðve-coordinated ““B sitesÏÏ) as the active sites
2
4

5
of C -cyclisation, isomerisation and single hydrogenolysis was
5
also pointed out.37 The selectivity of these four main reactions
25 D. M. Collins and W. E. Spicer, Surf. Sci., 1977, 69, 85; P. R.
Norton, K. Griffiths and P. E. Bindner, Surf. Sci., 1985, 138, 125.
was roughly equal on samples 3 to 6, the most active sample 4
2
6
H. P. Bonzel, C. R. Helms and S. Kelemen, Phys. Rev. L ett., 1975,
35, 1237.
showing the highest selectivity for C saturated products,
6
especially MCP (Table 2). The promoting e†ect of hydrogen
2
2
7
8
Z. Paal, R. Schlogl and G. Ertl, Catal. L ett., 1992, 12, 331.
 Ž
in the gas phase7,8,53 was explained in terms of much surface
hydrogen favouring their not too deeply dehydrogenated
intermediates.54 In addition, hydrogen treatment enhanced
the (220) and (311) reÑections (e.g. in sample 6). This can also
contribute to the high isomerisation and hydrogenolysis selec-
R. Sundararajan, G. Pet°, E. Koltay and L. Guczi, Appl. Surf.
Sci., 1995, 90, 165.
29 M. Kiskinova, G. Pirug and H. P. Bonzel, Surf. Sci., 1983, 133,
3
21.
N. M. Rodriguez, P. E. Anderson, U. Wild, R. Schlo
Wootsch and Z. Paal, J. Catal., 2001, 197, 365.
31 B. Torok, A. Molnar, I. Palinko and M. Barto
45, 295; A. Fasi, J. T. Kiss, B. Torok and I. Palinko
A, 2000, 200, 189.
3
0
Ž
gl, A.

tivities,55 since the so-called ““B -sitesÏÏ, with (311) or (110)
structure, reportedly favoured the related skeletal reactions:
5
Ž
Ž
Š




k, J. Catal., 1994,
, Appl. Catal.
1

Ž
Ž


isomerization and single hydrogenolysis.37 The almost con-
stant benzene selectivity may point to the stability of three-
atom ensembles50 during reconstruction. They may possibly
be in step or kink conÐguration.3,55
Surface (hydro)carbonaceous adsorbates ensured optimum
conditions for nondegradative reactions.56 This would explain
the high and reproducible activity of sample 6, even in the
presence of 20È25% ““invisible carbonÏÏ.57 The pronounced
32 V. Ponec, Adv. Catal., 1983, 32, 149.
3
3
Z. Paa
l, H. Groeneweg and J. Paal-Lukacs, J. Chem. Soc.,
  
Faraday T rans., 1990, 86, 3159.
3
3
4
5
S. C. Davis and K. J. Klabunde, Chem. Rev., 1982, 82, 153.
M. W. Roberts, Catal. L ett., 2000, 67(1), citing G. A. Somorjai, on
p. 50.
36 J. T. Miller, B. C. Meyers, F. S. Modica, G. S. Lane, M. Vaar-
kamp and D. C. Koningsberger, J. Catal., 1993, 143, 395.
2154
Phys. Chem. Chem. Phys., 2001, 3, 2148È2155

View Article Online
3
7
8
F. Garin, S. Aeiyach, P. Le
3
2

gare
23; A. Dauscher, F. Garin and G. Maire, J. Catal., 1987, 105,
33.

and G. Maire, J. Catal., 1982, 77,
46 Z. Hlavathy and P. Tetenyi, Surf. Sci., 1998, 410, 39.
 
47 S. M. Davis and G. A. Somorjai, J. Catal., 1980, 65, 78.
48 N. Freyer, G. Pirug and H. P. Bonzel, Surf. Sci., 1983, 126, 487,
citation from p. 490.
3
G. Ehrlich, Adv. Catal., 1969, 14, 255; a structure is depicted on
p. 314.
49 K. Matusek and Z. Paa

l, React. Kinet. Catal. L ett., 1999, 67, 241;
3
4
9
0
H. S. Taylor, Proc. R. Soc. L ondon, Ser. A, 1925, 108, 105.
K. Matusek, A. Wootsch, H. Zimmer and Z. Paa

l, Appl. Catal. A,
P. Te
974, 83, 37.
G. Maire, F. Bernhardt, P. Le

te

nyi, L. Guczi and Z. Paa

l, Acta Chim. Acad. Sci. Hung.,
2000, 191, 141.
1
50 P. Biloen, J. N. Helle, H. Verbeek, F. M. Dautzenberg and
4
1

gare

and G. Lindauer, in Proc. 7th
W. M. H. Sachtler, J. Catal., 1980, 63, 112.
  
51 H. Zimmer, M. Dobrovolszky, P. Tetenyi and Z. Paal, J. Phys.
International V acuum Congress & 3rd International Conference on
Solid Surfaces, V ienna, 1977, p. 861.
Chem., 1986, 90, 4758.
52 J. R. Anderson, Adv. Catal., 1973, 23, 1.
53 Z. Paal, in Hydrogen E†ects in Catalysis, ed. Z. Paa
Menon, Dekker, New York, 1988, p. 449.
54 Z. Paal, G. Szekely and P. Tetenyi, J. Catal., 1979, 58, 108.
55 G. A. Somorjai and F. Zaera, J. Phys. Chem., 1982, 86, 3070.
56 A. Sarkany, J. Chem. Soc., Faraday T rans. 1, 1988, 84, 2267; A.
Sarkany, J. Chem. Soc., Faraday T rans. 1, 1989, 85, 1523.
57 P. G. Menon, J. Mol. Catal., 1990, 59, 207.
4
4
2
3
B. Klotzer and E. Bechtold, Surf. Sci., 1993, 295, 374.
Ž
S. Horch, H. T. Lorensen, S. Helveg, E. Laegsgaard, I. Stenst-
gaard, K. W. Jacobsen, J. K. Norskov and F. Besenbacher,
Nature, 1999, 398, 134.


l and P. G.




4
4
5
M. Wolf, O. Deutschmann, F. Behrendt and J. Warnatz, Catal.
L ett., 1999, 61, 15.
W. T. Owens, N. M. Rodriguez and R. T. K. Baker, J. Phys.
Chem., 1992, 96, 5048.


4


Phys. Chem. Chem. Phys., 2001, 3, 2148È2155
2155