Surface Science 600 (2006) 1771–1779

www.elsevier.com/locate/susc

New interpretations of XPS spectra of nickel metal and oxides

a,1
Andrew P. Grosvenor , Mark C. Biesinger a, Roger St.C. Smart a,b,*
, N. Stewart McIntyre a

a
Surface Science Western, University of Western Ontario, London, Ont., Canada N6A 5B7
b
Applied Centre for Structural and Synchrotron Studies (ACeSSS),2 University of South Australia, Mawson Lakes SA 5095, Australia

Received 5 May 2005; accepted for publication 27 January 2006

Available online 17 February 2006

Abstract

A current interpretation of XPS spectra of Ni metal assumes that the main 6 eV satellite is due to a two hole c3d94s2 (c is a core hole)
final state effect. We report REELS observation in AES at low voltages of losses (plasmons and inter-band transitions) corresponding to
the satellite structures in Ni metal 2p spectra. The satellite near 6 eV is attributed to a predominant surface plasmon loss. A current inter-
pretation of Ni 2p spectra of oxides and other compounds is based on charge transfer assignments of the main peak at 854.6 eV and the
broad satellite centred at around 861 eV to the cd9L and the unscreened cd8 final-state configurations, respectively (L is a ligand hole).
Multiplet splittings have been shown to be necessary for assignment of Fe 2p and Cr 2p spectral profiles and chemical states. The assign-
ments of Ni 2p states are re-examined with intra-atomic multiplet envelopes applied to Ni(OH)2, NiOOH and NiO spectra. It is shown
that the free ion multiplet envelopes for Ni2+ and Ni3+ simulate the main line and satellite structures for Ni(OH)2 and NiOOH. Fitting
the NiO Ni 2p spectral profile is not as straightforward as the hydroxide and oxyhydroxide. It may involve contributions from inter-
atomic, non-local electronic coupling and screening effects with multiplet structures significantly different from the free ions as found
for MnO. A scheme for fitting these spectra using multiplet envelopes is proposed.
 2006 Elsevier B.V. All rights reserved.

Keywords: Nickel; Metals; Oxides; X-ray photoelectron spectroscopy

1. Introduction ical states of nickel in systems like these is potentially of

considerable value in understanding and control of corro-
X-ray photoelectron spectroscopy (XPS) has been exten- sion and dissolution. In principle, the oxidation states of
sively used in studies of the surface chemistry of metals, al- nickel in the surface region can be determined from the
loys, oxides and hydroxides with particular reference to binding energies (BE) and their chemical shifts in XPS
oxidation, corrosion and chemical attack including acid spectra. Previous research has supported assignment of
dissolution (e.g. [1–3]). Nickel has assumed some increased BEs of 852.6, 854.6 and 856.1 eV to Ni 2p3/2 XPS spectra
prominence with its recent use in high temperature corro- for Ni0, Ni2+ and Ni3+, respectively in oxidized Ni, oxides,
sion-resistant alloys and their reaction products (e.g. [4]). hydroxides and oxyhydroxides (e.g. [3,5,6]). This assign-
The assignment and quantitation of different surface chem- ment can be compromised by the complex main line split-
ting due to multiplet contributions in oxides and by
satellite structures at higher BEs particularly in the most
*
Corresponding author. Address: Applied Centre for Structural and intense Ni 2p spectra where broad maxima and extended
Synchrotron Studies (ACeSSS), University of South Australia, Mawson satellite structures are often measured.
Lakes SA 5095, Australia. Tel.: +61 883023353; fax: +61 883023683. The literature on the electronic structure, theoretical
E-mail address: [email protected] (R.St.C. Smart). models and energy loss spectroscopy (ELS) of Ni metal
1
Now at: Gunning/Lemieux Chemistry Centre, University of Alberta,
Edmonton, Alta., Canada T6G 2G2.
and NiO has recently been comprehensively reviewed by
2
Applied Centre for Structural and Synchrotron Studies, Division of Hagelin-Weaver et al. [7,8]. They have also extensively
Information Technology, Engineering and Environment. examined alternative assignments of ELS features based

0039-6028/$ - see front matter  2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2006.01.041
1772 A.P. Grosvenor et al. / Surface Science 600 (2006) 1771–1779

on both theoretical predictions and experimental inter- and intra-atomic electronic interactions without any
behaviour. It is useful to relate this information to the core ad hoc empirical parameters. This work demonstrates that
level XPS assignments of features in nickel metal and coupling of many-electron angular momenta (AM) and re-
oxides. coupling of high spin, open d-shell electrons, going beyond
Beginning with nickel metal ELS, they suggest that more exchange coupling of open p-shell with fixed d-shell cou-
intense features at 6.0 and 9.5 eV are due to surface and pling, is essential for simulation of the complex Mn2+ 2p
bulk plasmon losses respectively with weaker intra- and in- structure. It also shows that it is difficult to assign a single
ter-band transitions at 3.7 and 7.1 eV. In XPS spectra, the 2p BE (or chemical shift) to each ionic state of Mn or, by
well-known satellite 6 eV above the main emission line implication, other transition metal ion species. A multi-
for Ni 2p3/2 (at 852.6 eV) has previously been assigned component spectral envelope is necessary for each ion.
(e.g. [9]) to a two hole c3d94s2 (c is a core hole) final state Hagelin-Weaver et al. [7,8] have also examined energy
effect because optical and ELS experiments at that time had losses in NiO spectra again extensively correlating assign-
failed to show any structure near 6 eV and that plasmon ments of ELS features based on both theoretical predic-
losses could not therefore explain this satellite. We have tions and experimental behaviour. They conclude that
re-examined these measurements and conclusions to relate there are more major losses at 4.3, 7.4 and 18.6 eV due to
recent ELS evidence to the interpretation of XPS core level intra-and inter-band transitions with a bulk plasmon loss
spectra. We report that both reflection electron ELS at 22.0 eV.
(REELS) and Ni 2p3/2 spectra do show energy losses In practice, a simpler fit to the overall multiplet and sa-
corresponding to surface and bulk plasmons with evidence tellite envelope for each oxidation state of nickel may be a
also for weaker loss features in satellite structures of the requirement for quantitation. This view is supported by
Ni 2p3/2 spectra likely to be associated with final state other recent research. The central importance of intra-
effects as suggested by Hagelin-Weaver et al. [7,8]. atomic multiplet structures in fitting 2p spectra from Fe
In recent literature, the assignment of Ni 2p3/2 spectra of and Cr oxides has already been demonstrated [13,14]. A
oxides and other compounds is based on charge transfer practical scheme for the fitting procedure has been de-
assignments of the main peak (at 854.6 eV in NiO) and scribed for each ionic state based on the Gupta and Sen
the broad satellite (centred at 861 eV in NiO) to the (GS) calculations [15,16]. It is recognized that these multi-
cd9L and the unscreened cd8 final-state configurations plet components do not include inter-atomic AM coupling
respectively (L is a ligand hole) (e.g. [5,9]). The same model contributions or relativistic effects. Nevertheless, these free
has been previously accepted to explain main line and satel- ion, intra-atomic multiplet calculations do include the most
lite profiles in 2p spectra of Cu and Mn oxides [9]. Quan- important intra-atomic (including spin–orbit) coupling and
tum chemical cluster model calculations for NiO6 [10] have successfully simulated the spectral envelopes neces-
confirm the lowest final state with cd9L character. A satel- sary for interpretation of the Fe2+, Fe3+ [13] and Cr2+,
lite arising from the overlap of the frozen ground state Cr3+, Cr5+ states [14]. They have also provided a relatively
(with the core electron removed) and the unscreened final simple, practical means of resolving the different oxidation
state of mainly cd8 character is shifted 9.5 eV from the states of these metals.
main peak. A second satellite with predominantly cd10L2 In this paper, we examine the extension of these multi-
character is found shifted 6.0 eV to higher BE. This simu- plet fitting envelopes to Ni 2p spectra from model com-
lation produces a single peak for the main emission. Add- pounds of both Ni2+, in Ni(OH)2 and NiO, and Ni3+ in
ing the six extra (frozen) NiO6 clusters to give a Ni7O36 c-NiOOH and b-NiOOH.
cluster [10] now produces an additional peak on the high
BE side of the main line due to a screening process by an 2. Experimental
electron that does not come from the oxygen ligands in
the central NiO6 cluster but from a neighbouring NiO6 2.1. Materials
unit. These calculations have produced the extra peak, nor-
mally attributed to defects (e.g. Ni3+), as intrinsic to the Nickel metal was sourced from Puratronic (Alfa Aesar,
bulk NiO although the authors note that its intensity is Mass, USA with 99.995% purity. It was prepared for
strongly affected by defects. Extension of these simulations REELS examination in the Auger microscope by first
[11] using multiple clusters and ligand charge transfer has cleaning the surface using an Ar ion sputter beam and
been shown to reproduce the complex multiplet structure examining the surface oxide concentration, based on the
in the Ni 2p spectrum. However, quantum chemical simu- O KLL signal, for extinction. At this point no remaining
lation [12] of 2p spectra for MnO, with complex satellite C could be detected on the surface.
structure similar to NiO, has shown that these features XPS spectra from two sources of Ni(OH)2 were re-
can be produced without invoking charge transfer. In par- corded. Nickel hydroxide powder (61% Ni) was purchased
ticular, the simulation produces a split main peak and two from Alfa Aesar (Mass, USA). a-Ni(OH)2 was supplied by
satellites at higher BE of similar form to the Mn 2p spectra Inco Limited. It was characterized by X-ray diffraction
(and to many Ni 2p spectra (e.g. [3,5])). This ab initio cal- from the (0 0 3) peak areas compared with the (0 0 1) peak
culation includes relativistic electron correlation from both area of b-Ni(OH)2 and found to be 92% a-phase.
 A.P. Grosvenor et al. / Surface Science 600 (2006) 1771–1779 1773

c-NiOOH, also supplied by Inco Limited, was made by

oxidation of the a-Ni(OH)2 It was characterized by X-ray (a)
diffraction from the (0 0 3) peak areas compared with the
(0 0 1) peak area of b-NiOOH and found to be 98% c-
phase.
b-NiOOH was made by the oxidation of Ni(NO3)2 solu-
tion with bromine in aqueous KOH keeping the tempera-
ture below 20 C in an ice bath. A black powder of
mixed Ni(II)/Ni(III) was obtained. It is known that b-
NiOOH readily reduces to Ni3O2(OH)4 in ambient condi-
tions [17]. It was characterized by FTIR as a KBr pellet
giving spectra corresponding to NiOOH in the data base.
Nickel oxide powder with 99.998% purity was obtained
from Puratronic (Alfa Aesar, Mass, USA). 10 8 6 4 2 0
Energy (eV)
2.2. Instrumental analysis

The nickel reflection electron energy loss spectroscopy (b)

(REELS) spectra were recorded in a PHI 660 Auger
electron spectrometer. Beam energies between 200 and
3000 eV were examined.
All oxide samples were analysed in a Kratos Axis Ultra
XPS. A monochromatic Al Ka X-ray source was used for
all samples, along with pressures in the analysis chamber
of 10 6–10 7 Pa. The resolution function of the instrument
has been found to have a width of 0.35 eV using the silver
Fermi edge. To control charging of the samples, the charge
neutralizer filament was used during all experiments. The
Ni metal sample was sputter cleaned under vacuum
(4 kV, 15 mA) to remove any oxides that had formed on 10 8 6 4 2 0
the surface and then annealed at 600 C so as to remove Energy (eV)
any point defects that may have been formed during sput- Fig. 1. Gupta and Sen multiplet envelopes (Table 1) for (a) Ni(II) and
tering. The Ni oxide, oxyhydroxide and hydroxide samples (b) Ni(III) free ions.
were all prepared and transferred into the XPS under Ar to
produce pristine surfaces for analysis.
The conditions used for all of the survey scans were as fol- GS-calculated envelopes for the Ni 2p3/2 multiplets of
lows: energy range 1100–0 eV, pass energy 160 eV, step size Ni2+ and Ni3+ are shown in Fig. 1. The calculated envelope
0.7 eV, sweep time 180 s and X-ray spot size 700 · 400 lm. has been fit, as in previous fits to Fe and Cr envelopes
For the high-resolution spectra an energy range of 40– [10,11] using the minimum number of components (6) to
20 eV was used, depending on the peak being examined, reproduce the envelope with minimal residuals as evident
with a pass energy of 10 eV and a step size of 0.05 eV. in Fig. 1. Gaussian (70%)—Lorentzian (30%) profiles were
All spectra were first analysed using the CasaXPS soft- used for each component. The full set of relative peak posi-
ware [18]. Peak shifts due to any apparent charging were tions, FWHM and % contributions for each of Ni2+ and
normalized with the C 1s peak set to 284.8 eV. After calibra- Ni3+ is given in Table 1. The FWHM range used (1.1–
tion, the background from each spectrum was subtracted 1.5 eV) was chosen based on the use of a similar range by
using a Shirley-type background to remove most of the previous authors [20,21]. As with the Fe and Cr GS multi-
extrinsic loss structure [19]. All survey scans were analysed plet fits [13,14], it is necessary to include an additional min-
to determine the stoichiometry of the compound by using or contribution in the region between the main line
the appropriate sensitivity factors and to determine the multiplets and the related satellites. From the work of
amount of adventitious carbon and contaminants present. Hagelin-Weaver et al. [7,8], it appears likely that this broad
band may arise from the combined losses at 4.3 eV, 6.0
2.3. Analysis of multiplet splitting using nickel free and 7.4 eV due to intra-and inter-band transitions allowed
ion states by breakdown of dipole selection rules in polycrystalline
samples with delocalized electrons.
The Ni 2p3/2 envelope from compounds containing a In both Ni2+ and Ni3+ cases, it is clear that it is difficult
high-spin Ni cation was fit using peaks corresponding to assign a single BE, even as a main peak centre of gravity
to the GS multiplets and shake-up-related satellites. The (CG), to these chemical states. Nevertheless, it appears that
1774 A.P. Grosvenor et al. / Surface Science 600 (2006) 1771–1779

Table 1 peaks (FWHM 2.5 eV) with BE near 3.7 eV (minor) and
Relative peak positions, FWHM and % contributions for Ni2+ and Ni3+ 6.0 eV (major) above the main line at 852.6 eV.
GS envelopes
Fig. 3 gives the REELS spectra recorded at 0.2 keV
Species Peak label E (eV) FWHM (eV) % Contribution beam voltage for clean nickel. This voltage is represented
Ni2+ 1 0.0 1.1 21.9 because it gave the best resolution of loss features for sur-
2 1.0 1.3 29.9 face sensitivity on the PHI 660 instrument. At higher ener-
3 2.4 1.3 36.9
Satellite 1 4.8 1.1 5.8
gies, the elastic peak overlapped the lower energy loss
Satellite 2 6.0 1.3 4.9 peaks. The 5.5 and 7.0 eV loss peaks increased in intensity
Satellite 3 8.5 1.1 0.6
Ni3+ 1 0.0 1.2 18.7 6e+5
2 1.0 1.3 16.4
3 1.9 1.3 13.6
5e+5
5
4 2.8 1.2 38.8
Satellite 1 6.3 1.4 8.0
Satellite 2 7.8 1.5 4.5 4e+5

Intensity
3e+5
the Ni3+ state has a predominance of intensity at higher BE
in the main line. This is consistent with the observation of 2e+5
Ni3+ growth in the K-doped oxidized Ni(1 1 0) experiments
of Carley et al. [5] and the quantum calculations for defect 1e+5
structures [10] where intensity increases in a multi-peak (a)
envelope near 856.1 eV. 0
The experimental spectra were fit using components and 0 50 100 150 200
intensities (peak areas) similar to those in Fig. 1 and Table Kinetic Energy (eV)
1 except that the FWHM of the main peaks were allowed
1.2e+5
to vary (up to 1.5 eV) to optimize the fit. A single broad, d
weak peak (FWHM 3.2–3.4 eV) was also added at higher e c
BE (shifted 12 eV) to complete the fits. This may be
1.2e+5
due to truncation error in background subtraction or to
b
other extended losses beyond 8.5 eV [7,8]. The fits were
Intensity

adequate for Ni(OH)2 and NiOOH compounds. After the 1.1e+5

2p3/2 envelope was fitted using the above method, the main
peak CG was determined using the GS multiplets only.
a
1.1e+5
3. Results
(b)

3.1. Nickel metal spectra 1.0e+5

196 198 200 202
Fig. 2 shows the Ni 2p3/2 XPS spectrum from clean nickel Kinetic Energy (eV)
metal with the two obvious satellite intensities fit by broad
2000 e
20000

18000 0 d
A
16000 -2000 c
dy/dx

14000
-4000
b
CPS

12000
-6000
10000 a
A+6.0eV
8000 A+3.7eV -8000
(c)
6000
-10000
4000 196 198 200 202
Kinetic Energy (eV)
864 862 860 858 856 854 852 850 848
Binding Energy (eV) Fig. 3. REELS Ni metal spectra at 0.2 kV beam voltage. (a) Elastic peak
with loss structure, (b) expanded region of loss structure below elastic
Fig. 2. Ni 2p3/2 spectrum from clean nickel metal. peak, (c) first derivative spectrum of loss structure.
 A.P. Grosvenor et al. / Surface Science 600 (2006) 1771–1779 1775

and sharpness at lower voltages (100–200 eV), an observa- Fig. 5 shows the Ni 2p3/2 spectrum from c-NiOOH
tion also made in the ELS work of Hagelin-Weaver et al. which is well fitted with the GS Ni3+ multiplet set with
[7,8] for their 6 eV peak, suggesting a surface rather than the additional broad surface peak at 858.1 eV. This fit does
bulk process. Fig. 3a shows the main elastic peak at not require any contributions from any other Ni states
204.5 eV with obvious loss structure at lower kinetic ener- including Ni2+. The FWHM of the four main line contri-
gies. Fig. 3b is the region close to the main peak. Analysis butions are 1.4, 1.5, 1.4 and 1.4 eV, respectively with
of the extrinsic losses at 200 eV using the Tougaard increasing BE.
QUASES-XS-REELS program 2.1 [22] produces a very Fig. 6 shows the Ni 2p3/2 spectrum from b-NiOOH This
broad, continuous envelope across 200–190 eV. The model- spectrum is closely similar to that reported previously by
ing of the extrinsic losses does not seem to have been done Moroney et al. [6] for b-NiOOH after some dehydration
in previous REELS interpretation making identification of at 200 C This oxyhydroxide is a mixed Ni2+/Ni3+ phase
the positions of loss maxima less certain. Even with this as noted above. The OH :O2 ratio from O 1s spectra
correction, the attributions of specific loss peak energies was 2:1 confirming the decomposition to Ni3(OH)4O2. It
in this and other work (see Table 1 in Ref. [7]) are diffi- is well fit with the combination of GS Ni2+ and Ni3+ mul-
cult to deconvolute. Broader peaks may contain more tiplet sets in 1:2 ratio again with the additional broad loss
than one origin. Recognising this limitation, superimposed peak at 858.1 eV. It is important to emphasise here that we
on this broad extrinsic loss peak we appear to distinguish are not using 12 independent components in this fit but two
loss peaks near 3.5 eV (a), 4.5 eV (b), 5.5 eV (c), 7.0 eV sets of 6 components in which only the relative proportion
(d) and 8.5 eV (e). The first derivative spectrum in Fig. 3c
shows the positions of these peaks more clearly. Peaks (c)
and (d) are the most prominent corresponding to losses
across the 6 eV satellite in the Ni 2p3/2 spectrum from clean
nickel metal. Similar REELS loss structures are observed 4500
for nickel metal, with decreasing intensity relative to the
main reflection, up to 1000 eV beam energies. It is noted 4000
that no loss peaks were observed for a NiO(1 0 0) single
crystal face at 500 eV beam energy.
CPS

3500

3.2. Nickel hydroxide and oxyhydroxide spectra

3000
The XPS spectra from the two nickel hydroxide sources
gave identical multiplet fits without changing any of the 2500
parameters in Table 1. Fig. 4 shows the Ni 2p3/2 spectrum
from Ni(OH)2 which is well fit with the GS Ni2+ multiplet
870 865 860 855 850
set with the additional broad loss peak at 857.9 eV. This fit
Binding Energy (eV)
does not require any contributions from any other Ni states
(see below). The FWHM of the main line contributions are Fig. 5. Ni 2p3/2 spectrum of c-NiOOH fitted with Ni(III) multiplet
1.2, 1.4 and 1.5 eV, respectively with increasing BE. envelope.

2800
9000
2600
8000
2400
7000
2200
CPS
CPS

6000
2000

5000 1800

4000 1600

1400
3000

870 865 860 855 850 870 865 860 855 850
Binding Energy (eV) Binding Energy (eV)

Fig. 4. Ni 2p3/2 spectrum from Ni(OH)2 fitted with Ni(II) multiplet Fig. 6. Ni 2p3/2 spectrum of b-NiOOH fitted with Ni(II) and Ni(III)
envelope. multiplet envelopes.
1776 A.P. Grosvenor et al. / Surface Science 600 (2006) 1771–1779

loss peak near 5.5 eV may be the surface plasmon assigned

4000
by Hagelin-Weaver et al. [7,8] at 6.0 eV in their work. They
assign a peak near 7.0 eV to a Ni 3d (low spin) to Ni 4p
3500 (low spin) transition in surface atoms. The weaker 8.5 eV
peak may be due to a multiple or combination of these loss
peaks or to a bulk plasmon assigned by Hagelin-Weaver
CPS

3000
et al. [7] at 9.0 eV.
In the comprehensive summary and interpretation of
2500
ELS and optical (with dielectric function) data on Ni metal
by Hagelin-Weaver et al. [7], convincing evidence for a sur-
2000 face plasmon loss near 6 eV is presented although, as they
note, the reported peak position varies up to 8.3 eV in more
than 10 reports. At lower primary beam energies
870 865 860 855 850
(<200 eV), the peak centre shifts towards lower energy loss
Binding Energy (eV)
values closer to 6 eV while at higher beam energies, the
Fig. 7. NiO fitted with Ni(II) multiplet structure allowing variation of the intensity of a higher energy loss peak increases and shifts
multiplet spacings. towards values above 9 eV. These observations are consis-
tent with our REELS measurements. Other papers have
described trends in measurements of EELS for different
of the two sets is allowed to vary giving the Ni2+/Ni3+ transition metals consistent with surface plasmon losses.
ratio. The early paper by Robins and Swan [25] suggested that
the peaks at loss energies less than 10 eV may be due to
3.3. NiO spectra plasmons but their beam excitation energy was high and
the peaks very broad. Ingram et al. [26] extended this work
The Ni 2p3/2 spectra from polycrystalline NiO are clo- confirming the assignment to a surface plasmon near
sely similar to those reported previously for NiO powders 6.0 eV. Kemeny and Shevchik [27] examined the XP spec-
prepared by sintering above 1000 C [3]. This spectrum tra of Ni in a Ni90Zn10 alloy and suggested that, if the
cannot be satisfactorily fit with the unaltered Ni2+ multi- Ni satellite is from plasmon loss, it should also be found
plets in Fig. 1a or with FWHM variation alone. A satisfac- in the Zn 2p3/2 spectrum. They conclude that it is not but
tory fit can only be achieved by allowing variation in the do not discuss a small broad satellite above 6 eV separation
BE positions of the multiplet contributions to the main energy. A satellite peak is present in Zn metal with similar
peak. An illustration of one such fit is given in Fig. 7. Here, loss energy to that found in the Ni90Zn10 alloy albeit with
the spacings of the three peaks making up the main peak lower intensity than satellites in Ni or Co. It is also noted
are altered from 0.95 and 1.4 eV to 0.5 and 1.6 eV, respec- that a two-hole final state effect should result in an asym-
tively. As with Fe and Cr 2p spectra, there is also a broad metric satellite profile due to Doniach-Sunjic processes
peak associated with other intrinsic losses at a higher bind- [28]. The profile in Fig. 2 is, in our view, extended but
ing energy than the main peak multiplets. In all of these not asymmetric.
cases, this contribution is necessary to model the valley It remains possible that the 3.5 eV peak is also a surface
between the main peak and the satellite structures and plasmon rather than a shake-up peak. This interpretation
is consistent with the inter-band losses attributed by has been followed by Grosvenor et al. [29] in their paper
Hagelin-Weaver et al. [7,8]. on XP spectra of Co metal and phosphides. They also mea-
sured REELS peaks at 3.0 and 5.0 eV from Co metal. Their
4. Discussion evidence suggests that both are surface plasmons differing
only in the depth at which the plasmon oscillation is occur-
4.1. Nickel metal spectra ring, i.e. the higher energy loss is from deeper layers but not
true bulk.
The REELS results in Fig. 3 show that a variety of loss Overall, the combination of these losses in Ni 2p XPS
processes can be measured in the clean nickel metal surface. spectra gives rise to the broad peaks fit well with the 3.7
Comparing our loss energies with the careful analysis of and 6.0 eV components of Fig. 2. These components are
alternative assignments by Hagelin-Weaver et al. [7,8], not taken to represent specific processes but are likely, on
the first loss near 3.5 eV is likely to be associated with the basis of the REELS evidence, to represent both plas-
the Ni 3d (high spin) to Ni 4s (low spin) transition in sur- mon and some shake-up losses from the primary Ni 2p
face atoms. ELS peaks in the 4.2–4.8 eV range have been emission. In fact, it will be clear that it is possible to fit this
attributed to either inter-band transitions (e.g. [23]) or sur- extended profile with the set of 5 bands, corresponding to
face plasmons [24] and there is convincing evidence for sur- the 3.5, 4.5, 5.5, 7.0 and 8.0 eV loss peaks in Fig. 3. We
face loss processes at 4.0–4.5 eV in optical and dielectric have not done this here because it is inherently simpler
function data (summarized in Table 2, Ref. [7]). The larger and more practical to use the two peak fit shown in Fig. 2.
 A.P. Grosvenor et al. / Surface Science 600 (2006) 1771–1779 1777

This suggests that the current explanation of satellite evant literature [7–12] on alternative explanations of the
structure in Ni XP spectra should be re-examined. This the- origin of the complex main line and satellite structures
ory compares initial and final states attributing the main illustrates the continuing discussion on whether these fea-
line to c3d104s and the satellite to c3d94s2 after photoioni- tures can be explained by extended multiplet interactions
zation. The argument (and BE estimates), summarized in or require ligand charge transfer to simulate the overall
Hüfner [9], is that the additional Coulomb attraction due emission envelope. The necessity to vary the positions of
to the core hole can pull the conduction band below the the GS AM multiplet contributions can potentially be ex-
Fermi level so that an additional charge is put into the plained in two ways.
3d orbitals, i.e. 3d10 for the ground state (main line). The In the quantum chemical simulation of Ni 2p XPS line
higher BE state is said to arise from the 3d band not being shapes for NiO by van Veenendaal and Sawatsky [10],
completely filled and screening is produced instead within interactions of the central Ni atom (core hole c) with either
the wide 4s band. This posits a two-hole state, core hole neighbouring NiO6 octahedra or with oxygen-deficient de-
and 3d hole, in an excited state 6 eV above the ground fects broadened the main line and the satellite. In these
state. If this is the origin of the satellite, these peaks should multiple cluster calculations, the lowest final state still
not be found in REELS data since no photoionization has has predominant cd9L character with charge transfer from
occurred. The observation of the REELS loss peaks makes the adjacent oxygen atoms. Additions to the satellite struc-
it more likely that this is the true origin of the Ni metal ture arise from intersite charge transfer screening and in-
satellite structures. This was acknowledged by Hüfner [9] creased width of the cd10L2 states due to oxygen banding
to be the most obvious explanation but it was accepted that effects. The strong influence of the non-local screening ef-
no loss functions had previously been observed for nickel fects, with contributions from dn+1dn 1 charge transfer
metal at that time. states and associated multiplet structure, on the energies
and line shapes of core-level photoemissions has been dem-
4.2. Nickel hydroxide and oxyhydroxides onstrated in the calculations of Atanasov and Reinen [11].
Hence, the multiplet and satellite envelopes in Ni 2p spec-
The fitting of these Ni 2p spectra from Ni(OH)2, tra from NiO are unlikely to be fit with the free ion GS
c-NiOOH and b-NiOOH with relatively simple multiplet components.
envelopes for Ni2+, Ni3+ and their combination (respec- Bagus et al. [12] have also shown convincingly for MnO
tively) suggests that inter-atomic AM interactions are not that a complete fit to the 2p spectra can be obtained using
dominant in these structures. The intra-atomic GS multi- all inter-atomic wave function mixing and coupling/re-
plet interactions are sufficient to adequately model the coupling AM contributions. These contributions are not
complex envelopes of the Ni 2p spectra in these cases with- included in the free ion GS intra-atomic calculations.
out invoking charge transfer or other final state effects. When this is done, the complex spectra can be interpreted
without explicit ligand charge transfer as in the current
4.3. Nickel oxide interpretation of the main line as the cd9L screened
emission. In this theory, however, it is noted that the O
Modelling this oxide with GS multiplet states is obvi- 2p band is overlapped with the Ni 3d band in NiO UV pho-
ously more complex than the hydroxide and oxyhydroxide toemission. This suggests that inter-atomic wave function
compounds. The summary in the Introduction of some rel- mixing is favoured in the structure contributing to the

858.5

858 NiF2
857.5
Ni 2p 3/2 BE (eV)

857
NiCl 2
856.5

856
NiBr 2
855.5

855 Ni(OH)2
NiO
854.5
NiI2
854
2.4 2.9 3.4 3.9
Average Ligand Electronegativity

Fig. 8. Change of core Ni 2p3/2 BE (CG values) with electronegativity of the ligand.
1778 A.P. Grosvenor et al. / Surface Science 600 (2006) 1771–1779

Energy Separation Between Core Line and

5.8

First Satellite Peak (eV)

5.6

5.4

5.2

4.8
2.4 2.9 3.4 3.9
Average Ligand Electronegativity

Fig. 9. BE separation between Ni 2p3/2 core line and first satellite peak with electronegativity of the ligand.

AM contributions in the Bagus et al. treatment. The origin metal satellite structures rather than the two-hole final state
of the emissions from the main line and satellites may only theory. This final state effect would not be observed in
be resolved by further theoretical modelling of the NiO REELS spectra without photoemission.
structure using all inter-atomic wave function mixing and Ni 2p spectra from Ni(OH)2, c-NiOOH and b-NiOOH
coupling/re-coupling AM contributions. are adequately fitted with relatively simple Gupta and
Consideration of the position of the BEs resulting from Sen intra-atomic free ion multiplet envelopes for Ni2+,
the electronegativity effects of neighbouring ligands is also Ni3+ and their combination respectively implying limited
informative. The CG values calculated from the curve fits inter-atomic angular momenta electronic interactions.
for Ni(OH)2, NiOOH and NiO spectra are 855.3, 855.8 Modelling NiO with GS multiplet states is more complex
and 854.7 eV, respectively. An examination of the BE shifts than the hydroxide and oxyhydroxide compounds necessi-
(taken from the CGs) and main line/satellite separation tating multiplet contributions different from the free ion
with ligand electronegativity [30] is instructive (Figs. 8 GS Ni2+ envelope for adequate fitting. A more complete
and 9). These figures contain data from the samples mea- analysis may require all inter-atomic wave function mixing
sured here and the nickel halides (fluoride, chloride, bro- and coupling/re-coupling AM contributions not included
mide and iodide to be discussed in a subsequent paper). in the free ion GS intra-atomic calculations. The evidence
Fig. 9 shows a reasonable progression of satellite displace- suggests that assigning nickel states from single BE values
ment with increasing ligand electronegativity from NiI2 of Ni 2p3/2 spectra alone may carry significant uncertainty
(2.5) through NiO (3.5) to NiF2 (4.0). Conversely, the value and may require fitting with more complete multiplet enve-
for the NiO BE is aberrant in Fig. 8 compared with the lopes before quantitative analysis.
other compounds. A possible reason for this might be that
the BE of the Ni 2p3/2 peak does not represent the Made-
Acknowledgements
lung effect of the electronegativity of the ligands alone.
The effect of electronegativity alone should produce a BE
We thank Dr. Sridath Ramamurthy for his expert assis-
near 856 eV. In fact, the NiO result appears to represent
tance with the Auger REELS measurements and Brad
the strong effect of the oxide ligands on both BE and mul-
Kobe (Surface Science Western) for continuing discussions
tiplet splitting. It is likely therefore that inter-atomic wave
on nickel XPS spectra. The a-Ni(OH)2 and c-NiOOH were
function mixing, as proposed by Atanasov and Reinen [11]
kindly supplied by Inco Limited, Mississauga, Ontario,
and Bagus et al. [12], shifts the true BE of the Ni 2p spectra
Canada with thanks to Dr. Feng Zou. The considerable
in NiO.
assistance of the reviewers with additional literature has
improved the paper.
5. Conclusions

REELS results at low beam energies have shown that References

surface plasmons and inter-band transitions can be mea-
sured in the clean nickel metal surface. The larger loss [1] M.W. Roberts, Adv. Catal. 29 (1980) 55.
[2] N.S. McIntyre, Corrosion and surface analysis, in: J.C. Riviere, S.
peaks near 5.5 and 7.0 eV can be attributed to plasmon
Myhra (Eds.), Handbook of Surface and Interface Analysis, vol. 643,
losses with corresponding less-intense losses near 3.5 and Marcel Dekker, New York, 1998.
4.5 eV, respectively. The observation of the REELS loss [3] M.W. Roberts, R.St.C. Smart, J. Chem. Soc. Faraday I 80 (1984)
peaks makes it likely that this is the true origin of the Ni 2957.
 A.P. Grosvenor et al. / Surface Science 600 (2006) 1771–1779 1779

[4] N.S. McIntyre, R.D. Davidson, T.L. Walzak, A.M. Brennenstuhl, [16] R.P. Gupta, S.K. Sen, Phys. Rev. B 12 (1975) 15, GS II.
F. Gonzalez, S. Corrazza, Corros. Sci. 37 (1995) 1059. [17] A.F. Cotton, G. Wilkinson, Advanced Inorganic Chemistry, second
[5] A.F. Carley, S.D. Jackson, J.N. O’Shea, M.W. Roberts, Surf. Sci. ed., Interscience Wiley, New York, London, Sydney, 1966, Section
440 (1999) L868–L874. 29-G-5.
[6] L.M. Moroney, R.St.C. Smart, M.W. Roberts, J. Chem. Soc. [18] http://www.casaxps.com.
Faraday Trans. I 79 (1983) 1769–1778. [19] B.S. Norgren, M.A.J. Somers, J.H.W. de Wit, Surf. Interface Anal.
[7] H.A.E. Hagelin-Weaver, J.F. Weaver, G.B. Hoflund, G.N. Salaita, 21 (1994) 378.
J. Electron, Spec. Rel. Phenom. 134 (2004) 139–171. [20] N.S. McIntyre, D.G. Zetaruk, Anal. Chem. 49 (1977) 1521.
[8] H.A.E. Hagelin-Weaver, J.F. Weaver, G.B. Hoflund, G.N. Salaita, [21] J.R. Mycroft, H.W. Nesbitt, A.R. Pratt, Geochim. Cosmochim. Acta
J. Alloys, Compounds 389 (2005) 34–41. 59 (1995) 721.
[9] S. Hufner, Photoelectron spectroscopy, Solid State Science Series, [22] QUASES-XS-REELS Version 2.1, QUASES Tougaard Inc. (Odense,
vol. 82, Springer-Verlag, 1995 (Chapter 3 and references therein). Denmark). Available from: <www.quases.com>.
[10] M.A. van Veenendaal, G.A. Sawatsky, Phys. Rev. Lett. 70 (1993) [23] E. Sickafus, F. Steinrisser, Phys. Rev. B 6 (1972) 3714.
2459. [24] H.B. Koldoziej, B. Rozenfeld, Acta Phys. Pol. A 48 (1975) 765.
[11] M. Atanasov, D. Reinen, J. Electron, Spectrosc. Relat. Phenom. 86 [25] J.L. Robins, J.B. Swan, Proc. Phys. Soc. 76 (1960) 857.
(1997) 185. [26] J.C. Ingram, K.W. Nebesny, J.E. Pemberton, Appl. Surf. Sci. 45
[12] P.S. Bagus, R. Broer, W.A. de Jong, W.C. Nieuwpoort, F. Parmi- (1990) 247.
ginai, L. Sangaletti, Phys. Rev. Lett. 84 (2000) 2259. [27] P.C. Kemeny, N.J. Shevchik, Solid State Comm. 17 (1975) 255.
[13] A.P. Grosvenor, B.A. Kobe, M.C. Biesinger, N.S. McIntyre, Surf. [28] S. Doniach, M. Sunjic, J. Phys. C 3 (1970) 285.
Interface Anal. 36 (2004) 1564. [29] A.P. Grosvenor, S.D. Wik, R.G. Cavell, A. Mar, Inorg. Chem. 44
[14] M.C. Biesinger, C. Brown, J.R. Mycroft, R.D. Davidson, N.S. (2005) 8988–8998.
McIntyre, Surf. Interface Anal. 36 (2004) 1550. [30] L. Pauling, The Nature of the Chemical Bond, third ed., Cornell
[15] R.P. Gupta, S.K. Sen, Phys. Rev. B 10 (1974) 71. University Press, 1960, p. 93.