X-ray diffraction study of gold nitride films: Observation of a solid solution phase

Abstract

The structure of nitride containing gold films produced by reactive ion sputtering and nitrogen plasma etching is investigated using x-ray photoelectron spectroscopy and x-ray diffraction. It is found that gold nitride is a solid solution of nitrogen atoms dissolved in a fcc gold matrix. Differences between the strain and lattice parameters of gold and gold nitride films were observed and are explained by interstitial nitrogen present in the latter.

Full text

Page 1

X-ray diffraction study of gold nitride films: Observation of a solid solution
phase
L. Alves,1,a?T. P. A. Hase,2M. R. C. Hunt,3A. C. Brieva,1and L. Šiller1,a?
1School of Chemical Engineering and Advanced Materials, University of Newcastle upon Tyne,
Newcastle upon Tyne NE1 7RU, United Kingdom
2Department of Physics, University of Warwick, Coventry CV4 7AL, United Kingdom
3Department of Physics, University of Durham, Durham DH1 3LE, United Kingdom
?Received 22 August 2008; accepted 25 October 2008; published online 10 December 2008?
The structure of nitride containing gold films produced by reactive ion sputtering and nitrogen
plasma etching is investigated using x-ray photoelectron spectroscopy and x-ray diffraction. It is
found that gold nitride is a solid solution of nitrogen atoms dissolved in a fcc gold matrix.
Differences between the strain and lattice parameters of gold and gold nitride films were observed
and are explained by interstitial nitrogen present in the latter. © 2008 American Institute of Physics.
?DOI: 10.1063/1.3040717?
I. INTRODUCTION
A large amount of gold is used by industry to produce
contacts and interconnects1and minute amounts of alloying
elements can be included to improve certain mechanical and
electrical properties.2There has been continuous research to
produce still harder conductive gold coatings with corre-
spondingly improved lifetimes. Metal nitrides are often used
in technological applications because of their desirable prop-
erties such as enhanced hardness, high melting point, chemi-
cal stability, and magnetic and electrical behaviors. As such
there has been a concerted effort to produce gold nitride for
the past 30 years. However, this compound was only recently
observed, by Šiller et al.3Gold nitride?s? AuNxcan be
formed on single crystals as well as over large area polycrys-
talline surfaces using a variety of techniques: These include
nitrogen ion irradiation of gold crystals in ultrahigh vacuum,3
nitrogen reactive ion sputtering ?RIS? of gold under high
vacuum conditions, nitrogen radio frequency ?rf? plasma
treatment of gold films,4reactive pulsed laser deposition in a
nitrogen containing atmosphere,5and through the pulsed arc
technique.6Conducting gold films containing AuNxshow an
increased hardness when compared with gold films produced
under similar conditions making them ideal for use in coat-
ing and interconnect technologies.4Unequivocal evidence
for the presence of gold nitride species has been obtained
using x-ray photoelectron spectroscopy ?XPS?, where two
distinct chemical states of nitrogen were observed in N 1s
spectra, which are associated with two nitride states that have
different thermal stabilities.7Theoretical studies7–9have pre-
dicted several possible structures for this stable solid phase
of gold nitride, such as triclinic Au3N,7anti-ReO3–Au3N,7
AuN2?fluorite?,8as well as zinc blende and rocksalt.9How-
ever, the concentrations of nitrogen incorporated into these
films7suggest that the formation of a solid solution phase is
also a possibility. A detailed knowledge of the long range
order and structure of possible nitride phases is very impor-
tant for the further optimization of gold nitride films. In this
paper, we present a detailed structural characterization using
XPS and x-ray diffraction ?XRD? to compare pure and ni-
trided gold films produced by different processes.
II. EXPERIMENTAL
Polycrystalline gold nitride films were prepared using
nitrogen RIS, under high vacuum, on standard 100 mm ?100?
silicon wafers using an Edwards AUTO 500 magnetron sput-
tering system. One of the films was further treated by etching
with a nitrogen rf plasma. For comparative purposes, pure
gold films were also prepared using conventional sputtering
in an Ar atmosphere. The Ar pressure was similar to that of
nitrogen during the nitride deposition. The film thicknesses
were determined using ellipsometry and were similar for all
samples ??2 ?m?.
XPS measurements were made at the National Centre for
Electron Spectroscopy and Surface Analysis ?NCESS?,
Daresbury, UK, using an ESCA 300 spectrometer with
monochromated Al K? radiation ?h?=1486.6 eV? with a
resolution of ?0.40 eV. All measurements were performed
in a normal emission geometry at room temperature with a
base vacuum of 2?10−9mbar.
XRD measurements were made at room temperature us-
ing the 11-axis Huber diffractometer on the XMaS beamline
?BM28? at the European Synchrotron Radiation Facility
?ESRF? in Grenoble, France. A monochromatic beam of 11
keV was produced from a Si ?111? monochromator with a
resolution of ?E/E=2?10−4. This energy was chosen such
that the incident energy was below the L3edge of gold,
thereby minimizing the fluorescent background from the
sample. The data were recorded using a vertical scattering
geometry with the diffractometer setup in a high resolution,
double axis configuration where the incident and scattered
beams were defined by narrow slits. The instrumental reso-
lution under this configuration was determined from the
width of the ?400? diffracted beam from a Si ?100? wafer and
was 0.038°?0.001° in 2?.
a?Authors to whom correspondence should be addressed. Electronic ad-
dresses: luis.alves@ncl.ac.uk and lidija.siller@ncl.ac.uk.
JOURNAL OF APPLIED PHYSICS 104, 113527 ?2008?
0021-8979/2008/104?11?/113527/5/$23.00© 2008 American Institute of Physics
104, 113527-1
Downloaded 23 Aug 2010 to 129.234.188.75. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

Page 2

III. RESULTS AND DISCUSSION
The presence of nitrogen within the gold nitride films
was confirmed using XPS. A typical N 1s XPS spectrum
from a gold nitride film used in this study is shown in Fig. 1.
The spectrum shows two broad peaks centered at 398 and
403 eV. On the basis of our previous work on gold nitride
films produced under similar conditions,7,10we assign peak A
as arising from nitrogen bonded directly to gold. Peak B
comes from nitrogen molecules trapped in subsurface
bubbles. The current instrument resolution does not allow us
to resolve peak A into separate peaks associated with the two
gold nitride states, which were observed previously.7The
XPS cross section for the nitrogen atoms is the same for
molecular nitrogen and nitrogen bonded to gold. Thus, from
the integrated intensities of the two peaks we estimate that
the ratio of the molecular nitrogen to that in a nitride envi-
ronment to be 60:40. Due to the noise in the data, we esti-
mate that this ratio is only precise to ?30%. By using the
transmission function of the spectrometer and literature val-
ues for photoionization cross sections and asymmetry
parameters,11it is also possible to use the XPS spectra to find
the concentration of nitrogen. In these samples the concen-
tration of nitrogen incorporated in the nitride phase ?peak A?
is 0.4%?0.2% ?expressed as an atomic ratio N:Au?. A simi-
lar concentration was recorded for the sample subjected to
further rf treatment.
Parallel beam XRD data were recorded from the three
different sample types. Detector scans recorded with an inci-
dent angle of 4° are shown in Fig. 2?a?. The polycrystalline
nature of all the samples is clearly demonstrated since all the
expected peaks from the Au fcc structure are present. A com-
parison between the pure gold films and those subjected to a
nitride treatment shows no change in the long range order.
The relative intensity changes of the diffraction peaks are
attributed to different textures in the films, discussed below.
In all cases, the diffraction data only show peaks associated
with the fcc structure and none that could be attributed to any
of the theoretically proposed AuNxstructures.7–9Moreover,
there is also no evidence of peak splitting or any systematic
peak shifts in the 2? scans which would otherwise indicate
an in-plane distortion, such as the ones that have been ob-
served when nitrides are formed.12
High quality diffraction data were recorded in the vicin-
ity of each reflection and fitted to a Pearson VII function
combined with a linear background in order to extract the
position and full width at half maximum ?Fig. 2?b??. On in-
corporation of nitrogen the peaks shift to slightly smaller
angles and broaden. The broadening is considerably larger in
the RIS sample when compared to the RIS+plasma treated
sample.
The lattice parameter for each film was obtained by plot-
ting diffraction order, ?h2+k2+l2? versus sin2??B?, where ?B
is the Bragg angle. The fitted value of the lattice parameter
for the pure gold film is aAu=4.0784?0.0006 Å in excellent
agreement with the fully relaxed bulk value of 4.0786?2? Å.
Incorporation of nitrogen causes the lattice to expand
slightly, the lattice parameters being aRIS=4.083?0.001 Å
and aRIS+plasma=4.0818?0.0004 Å for the RIS and RIS plus
rf plasma samples, respectively. Both nitrogen treatments
thus result in a similar lattice expansion, ?0.12?0.02?% for
the RIS treatment and ?0.09?0.01?% for the RIS plus rf
plasma treatment.
Despite being polycrystalline in nature, the pure gold
film showed strong ?111? texture ?Fig. 3?a??. Rocking curves
recorded at the ?111? reflection were clearly centered on the
Bragg angle ?Fig. 3?. The mosaic of the gold film, measured
from the width of the rocking curves, was determined by
fitting a Gaussian to be 7.5°?0.1°. Such a mosaic distribu-
tion is similar to that observed in other metallic films pro-
duced by sputtering. Scanning the planes through the scatter-
0
10
20
391393395397399401
Energy (eV)
403405
Intensity (arb. Units)
A
B
FIG. 1. N 1s photoemission spectrum obtained from a gold nitride film.
Peak A is associated with gold nitride while peak B is associated with
molecular nitrogen trapped beneath the surface.
(220)
(311)
(222)
(400)
(331)
(420)
(422)
(511)
(111)
(200)
FIG. 2. ?a? Parallel beam XRD data for the three sample types ?sequential
data offset for clarity?. ?b? Normalized scans in the vicinity of the ?511? Au
peak for the three samples. Experimental data are shown as points and fits to
a Pearson VII function with a constant background are shown as lines.
113527-2Alves et al.J. Appl. Phys. 104, 113527 ?2008?
Downloaded 23 Aug 2010 to 129.234.188.75. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

Page 3

ing plane ??? measured the tilt which was similarly broad,
5.91°?0.03°. All other reflections gave rocking curves that
were typical of randomly orientated grains. On incorporating
nitrogen into the samples the degree of ?111? texturing re-
duced considerably ?Fig. 3?b??. While the ?111? planes were
still distributed around the surface the rocking curves broad-
ened to 14.7°?0.1° for the RIS sample and became even
more randomly orientated for the RIS+RF treatment. We did
not observe any change in the other reflections on incorpo-
ration of nitrogen. Accompanying the change in mosaic was
a corresponding broadening of the tilt ??? distribution.
Coupled with the changes in the texture of the films on
addition of nitrogen is a clear increase in the width of the
diffraction peaks probed normal to the diffracting planes
?Fig. 2?b??. This increase in width can either be associated
with a decrease in grain size or an increase in microstrain ?,
within a grain arising from a distribution of lattice constants.
The peak broadening from finite particle size D is given by
the well known Scherrer equation:
?size?2?? =
??
D cos??B?,
?1?
where ? is a constant depending on the line shape profile and
?Bis the Bragg angle. In a cubic material, the broadening
associated with microstrain has a different dependence on the
scattering angle:
?strain?2?? = 2? tan??B?.
?2?
The total width of a diffraction peak is a combination of the
two terms and depends on the line shape of the diffraction
peak.13The fitted line shapes in Fig. 2?b? have shape param-
eters which closely resemble a Cauchy profile and the total
breadth is then a linear sum of the size and strain terms. As
the measured widths are an order of magnitude greater than
the instrumental broadening, a plot of the width multiplied
by the cosine of the Bragg angle against the sine of the
Bragg angle allows the two sample dependent broadening
terms to be separated and measured:
?hkl?2??cos??B
hkl? =k?
D+ 2? sin??B
hkl?.
?3?
Such a Williamson–Hall ?WH? analysis for the pure gold
film is shown in Fig. 4.
In the pure gold film the textured ?111? and ?222? peaks
were significantly sharper in 2? when recorded in a Debye–
Scherrer ??/2?? geometry and were therefore excluded from
the WH analysis. In Fig. 4 there is only a very weak positive
linear regression of the untextured peaks. Within error there
is no evidence of any microstrains in the relaxed gold film
?=?0.1?0.2?%. This film is probably composed of two
grain size distributions, larger textured ?111? oriented grains
arranged parallel to the surface with the smaller grains being
randomly orientated. Assuming a constant of ?=1 in the
Scherrer equation gives grain sizes of 0.080?0.001 ?m for
the ?111? grains and 0.026?0.003 ?m for the random
grains.
A clear difference between the gold and the nitride films
is observed when comparing Figs. 4 and 5. The reduction in
the texture on incorporation of nitrogen, which was deduced
from the widths of the rocking curve, is also indicated in the
WH plots ?Fig. 5? as there is no longer any discernable dif-
ference between the ?111? peaks and the others.
A clear positive trend appears in Fig. 5 indicating that
microstrains have been introduced through the addition of
nitrogen. The different slopes show that the RIS+plasma
treatment results in a reduced strain state within the grains
when compared to the RIS treatment alone. The spread of the
data points around the best fit linear regression line, most
notable in the more highly strained RIS sample, could be due
to the assumption that the elastic constants are equal in all
directions, i.e., that ?d/d=?a/a=?b/b=?c/c. This may be
associated with the abrupt change in texture observed in the
0
100
200
300
400
500
600
0510152025
Sample angle (degree)
Intensity (arb. Units)
a
0
0.5
1
1.5
2
2.5
-149 141924 29
Sample angle (degree)
Intensity (arb. Units)
b
FIG. 3. Rocking curves obtained from ?-scans of the Au ?111? planes from
?a? a pure gold film and ?b? a gold nitride film.
0
0.002
0.004
0.006
0.008
0.01
0.2 0.40.6 0.8
sin(?B)
?(2?)cos(?)
Powder peaks
Textured peaks
FIG. 4. WH plot for the pure gold film. The textured ?111? planes are
considerably larger than the randomly orientated grains which show little
evidence of microstrain.
113527-3Alves et al.J. Appl. Phys. 104, 113527 ?2008?
Downloaded 23 Aug 2010 to 129.234.188.75. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

Page 4

nitride films. In principle the nature of the dislocations and
their density can be obtained from the strain broadening
term14but we do not have enough data to make any further
comment on this. The microstrain values ? which could only
reasonably obtained for the nitride films ?Fig. 5? are
?1.1?0.2?% for the RIS film ?Fig. 5?a?? and ?0.45?0.05?%
for the RIS plus rf plasma ?Fig. 5?b??. The grain size could
only be determined with any degree of accuracy using Eq.
?3? for the RIS+rf sample and is 0.06?0.02 ?m, larger than
the random orientated grains in the pure gold film.
IV. DISCUSSION
A comparison between the XPS and XRD data show that
the increase in the value of the lattice parameter and the
microstrains within grains on incorporation of nitrogen could
involve two sources. First, the presence of molecular nitro-
gen in the form of bubbles will produce a local strain field
around the bubble and second nitrogen incorporated into the
Au matrix will induce crystallographic changes. The lack of
any peaks not associated with fcc Au is explained either by
the presence of interstitial nitrogen forming a solid solution
with gold as the solvent or that the nitrogen exists as substi-
tutional defects on the Au fcc sites. The structural results
reported here are typically found in alloys with small inter-
stitial elements, such as hydrogen in various metallic
elements15and alloys.16Boron and carbon are also found in
interstitial sites in a range of metals including palladium,
platinum, iridium, and cobalt.17
In an attempt to separate the structural effects associated
with the two different nitrogen environments, we have per-
formed diffraction anomalous fine structure measurements.18
The scattered intensity was recorded as the incident energy
was scanned through the Au L3edge while maintaining the
sample at the ?111? diffraction condition. The fluorescence
signal was recorded on an energy sensitive detector placed
perpendicular to the scattering plane which minimizes the
contribution from diffusely scattered radiation from the inci-
dent beam. X-ray absorption fine structure ?XAFS? oscilla-
tions were observed on the high energy side of the edge ?Fig.
6?. These oscillations arise from the interference of the emit-
ted photoelectrons with the surrounding atoms and can be
used to investigate the local environment of the absorbing
element. As the technique is element specific, the oscillations
are only sensitive to the local environment ?within 1 nm? of
the Au atoms. As the absorption process occurs uniformly in
the bulk of the film, these experiments will be more sensitive
to changes in the local environment associated with the gold-
nitrogen bonds and not to the strain fields coming from the
molecular nitrogen bubbles.
The fringes were extracted by subtracting a third order
polynomial background correction from the data above the
edge ?L3=11.919 keV?. Given the very small changes ob-
served in the bulk lattice parameter from the XRD, it is not
surprising that it is difficult to identify any changes in the
periodicity of the fringes seen in Fig. 6. However, it appears
that the interatomic distances surrounding the Au atoms are
not greatly affected by the incorporation of nitrogen. The
amplitude of the fringes is inversely proportional to the size
of the distribution in radial distances and is therefore closely
related to the microstrain measured in XRD. The damping of
the fringes seen in Fig. 6 is greatest for the RIS sample
which is the sample with the largest microstrain deduced
from the XRD data. This would suggest that the strain pro-
files we observe in the XRD most likely arise from the ef-
fects of the nitride and are not from the molecular bubbles.
V. CONCLUSION
Gold nitride films prepared by RIS and, in a separate
case, treated by rf plasma were investigated with XPS and
XRD and compared with pure gold films produced under
0
0.005
0.01
0.015
0.02
0.20.40.6 0.8
Powder Peaks
Textured Peaks
b
?(2?)cos(?)
sin(?B)
0
0.005
0.01
0.015
0.02
0.20.40.60.8
sin(B)
?(2?)cos(?)
a
Powder Peaks
Textured Peaks
?
FIG. 5. WH plot from samples treated by ?a? RIS and ?b? RIS+plasma. The
reduction in the slope in ?b? indicates a reduced microstrain relative to ?a?.
-40
-30
-20
-10
0
10
20
30
40
11.91212.1
Energy (KeV)
12.212.312.412.5
RIS
RIS + RF Plasma
Pure Gold
Intensity (arb. Units)
FIG. 6. XAFS-like oscillations close to the L3edge of Au for the three
sample types.
113527-4Alves et al.J. Appl. Phys. 104, 113527 ?2008?
Downloaded 23 Aug 2010 to 129.234.188.75. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

Page 5

similar conditions. No diffraction peaks associated with any
of the distinct gold nitride phases theoretically predicted
could be observed. The fcc lattice of the nitride films was
found to be dilated ?by about 0.1%? with respect to pure gold
on incorporation of the nitrogen. The nitride films were also
found to display significant dispersion in the lattice param-
eter most likely associated with the disorder arising from
nitrogen incorporation, while the pure gold sample was
strain-free to within the experimental precision. These struc-
tural changes upon nitride formation, coupled with the low
nitrogen concentration deduced from XPS, indicate that gold
nitrides produced by ion sputtering and plasma etching tech-
niques are solid solutions with the nitrogen residing in inter-
stitial sites in the fcc gold lattice.
ACKNOWLEDGMENTS
We wish to thank Paul R. Coxon for careful reading of
the manuscript. This work was performed on the EPSRC-
funded XMaS beamline at the ESRF. We are grateful to the
beamline team of S. D. Brown, D. F. Paul, P. Normile, L.
Bouchenoire, and P. Thompson for their invaluable assis-
tance, and to S. Beaufoy and J. Kervin for additional support.
We acknowledge EPSRC for access to NCESS and financial
support under Grant No. EP/C006011/1.
1Y. Okinaka, Gold Bull. 33, 117 ?2000?.
2W. S. Rapson and T. Groenewald, Gold Usage ?Academic, London, 1978?.
3L. Šiller, M. R. C. Hunt, J. W. Brown, J.-M. Coquel, and P. Rudolf, Surf.
Sci. 513, 78 ?2002?.
4L. Šiller, N. Peltekis, S. Krishnamurthy, Y. Chao, S. J. Bull, and M. R. C.
Hunt, Appl. Phys. Lett. 86, 221912 ?2005?.
5A. P. Caricato, M. Fernàndez, G. Leggieri, A. Luches, M. Martino, F.
Romano, T. Tunno, D. Valerini, A. Verdyan, Y. M. Soifer, J. Azoulay, and
L. Meda, Appl. Surf. Sci. 253, 8037 ?2007?.
6A. Devia, H. A. Castillo, V. J. Benavides, Y. C. Arango, and J. H.
Quintero, Mater. Charact. 59, 105 ?2008?.
7S. Krishnamurthy, M. Montalti, M. G. Wardle, M. J. Shaw, P. R. Briddon,
K. Svensson, M. R. C. Hunt, and L. Šiller, Phys. Rev. B 70, 045414
?2004?.
8R. Yu and X. F. Zhang, Phys. Rev. B 72, 054103 ?2005?.
9M. B. Kanoun and S. Goumri-Said, Phys. Lett. A 362, 73 ?2007?.
10Yu. V. Butenko, L. Alves, A. C. Brieva, J. Yang, S. Krishnamurthy, and L.
Šiller, Chem. Phys. Lett. 430, 89 ?2006?.
11J. J. Yeh and I. Lindau, At. Data Nucl. Data Tables 32, 1 ?1985?.
12A. Munitz, G. Kimmel, J. C. Rawers, and R. J. Fields, Nanostruct. Mater.
8, 867 ?1997?.
13G. K. Williamson and W. H. Hall, Acta Metall. 1, 22 ?1953?.
14T. Ungár, A. Révésez, and A. Borbély, J. Appl. Crystallogr. 31, 554
?1998?.
15Hydrogen in Metals I and II, edited by G. Alefeld and J. Völkl ?Springer-
Verlag, Berlin, 1978?.
16B. Baranowski, S. Majchrzak, and T. B. Flanagan, J. Phys. F: Met. Phys.
1, 258 ?1971?.
17M. Beck, M. Ellner, and E. J. Mittemeijer, Acta Mater. 49, 985 ?2001?.
18Resonant Anomalous X-Ray Scattering: Theory and Applications, edited
by G. Materlik, C. J. Sparks, and K. Fischer ?North-Holland, Amsterdam,
1994?, pp. 389–420.
113527-5 Alves et al.J. Appl. Phys. 104, 113527 ?2008?
Downloaded 23 Aug 2010 to 129.234.188.75. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions