ArticlePDF Available

Electronic structure investigation of Ti3AlC2, Ti3SiC2, and Ti3GeC2 by soft-X-ray emission spectroscopy



The electronic structures of epitaxially grown films of Ti(3)AlC(2), Ti(3)SiC(2), and Ti(3)GeC(2) have been investigated by bulk-sensitive soft x-ray emission spectroscopy. The measured high-resolution Ti L, C K, Al L, Si L, and Ge M emission spectra are compared with ab initio density-functional theory including core-to-valence dipole matrix elements. A qualitative agreement between experiment and theory is obtained. A weak covalent Ti-Al bond is manifested by a pronounced shoulder in the Ti L emission of Ti(3)AlC(2). As Al is replaced with Si or Ge, the shoulder disappears. For the buried Al and Si layers, strongly hybridized spectral shapes are detected in Ti(3)AlC(2) and Ti(3)SiC(2), respectively. As a result of relaxation of the crystal structure and the increased charge-transfer from Ti to C, the Ti-C bonding is strengthened. The differences between the electronic structures are discussed in relation to the bonding in the nanolaminates and the corresponding change of materials properties.
Electronic structure investigation of Ti3AlC2,Ti
3SiC2, and Ti3GeC2by soft x-ray
emission spectroscopy
M. Magnuson,1J.-P. Palmquist,2M. Mattesini,1,4 S. Li,1R. Ahuja,1O. Eriksson,1J. Emmerlich,3O. Wilhelmsson,2
P. Eklund,3H. Högberg,3L. Hultman,3and U. Jansson2
1Department of Physics, Uppsala University, P.O. Box 530, S-75121 Uppsala, Sweden
2Department of Materials Chemistry, The Ångström Laboratory, Uppsala University, P.O. Box 538, SE-75121 Uppsala, Sweden
3Department of Physics, IFM, Thin Film Physics Division, Linköping University, SE-58183 Linköping, Sweden
4Departamento de Física de la Materia Condensada, Universidad Autónoma de Madrid, E-28049, Spain
Received 20 December 2004; revised manuscript received 7 June 2005; published 1 December 2005
The electronic structures of epitaxially grown films of Ti3AlC2,Ti
3SiC2, and Ti3GeC2have been investi-
gated by bulk-sensitive soft x-ray emission spectroscopy. The measured high-resolution Ti L,CK,AlL,SiL,
and Ge Memission spectra are compared with ab initio density-functional theory including core-to-valence
dipole matrix elements. A qualitative agreement between experiment and theory is obtained. A weak covalent
Ti-Al bond is manifested by a pronounced shoulder in the Ti Lemission of Ti3AlC2. As Al is replaced with Si
or Ge, the shoulder disappears. For the buried Al and Si layers, strongly hybridized spectral shapes are detected
in Ti3AlC2and Ti3SiC2, respectively. As a result of relaxation of the crystal structure and the increased
charge-transfer from Ti to C, the Ti-C bonding is strengthened. The differences between the electronic struc-
tures are discussed in relation to the bonding in the nanolaminates and the corresponding change of materials
DOI: 10.1103/PhysRevB.72.245101 PACS numbers: 78.70.En, 71.15.Mb, 71.20.b
Recently, the interest in nanolaminated ternary Mn+1AXn
denoted 211, 312, and 413, where n=1, 2, and 3, respec-
tivelycarbides and nitrides, so-called MAX phases, has
grown significantly. Here, Mis an early transition metal, Ais
apelement, usually belonging to the groups IIIA and IVA,
and Xis either carbon or nitrogen.1–4 The large interest is due
to the fact that these layered materials exhibit a unique com-
bination of metallic and ceramic properties,5including high
strength and stiffness at high temperatures, resistance to oxi-
dation and thermal shock, as well as high electrical and ther-
mal conductivity. These unusual macroscopic properties are
closely related to the electronic and structural properties of
the constituent atomic layers on the nanoscale.
This family of compounds has a hexagonal structure with
nearly close-packed layers of the Melements interleaved
with square-planar slabs of pure Aelements, where the X
atoms fill the octahedral sites between the Matoms. How-
ever, the Aelements are also located at the center of trigonal
prisms that are larger than the octahedral sites and thus better
accommodate the Aatoms. Among the 312 phases, there are
three carbides which belong to this family, namely, Ti3AlC2,
Ti3SiC2, and Ti3GeC2. Figure 1 shows the crystal structure of
the 312-phases.
Up to now, the most studied MAX material is Ti3SiC2,
which consists of hexagonal layers stacked in the repeated
sequence of Si-TiII-C-TiI-C-TiII, where the unit cell consists
of two formula units. Ti atoms can occupy either of two
different sites; TiII which has both C and Si neighbors and
TiIwhich only has C neighbors. Most of the research on
Ti3SiC2has incorporated processing and mechanical proper-
ties of sintered bulk compounds.1,4,5 However, in many tech-
nological applications where, e.g., high melting points, cor-
rosion resistance, electrical and thermal conductivity as well
as low-friction gliding properties are required, high-quality
thin film coatings are more useful than bulk materials. Al-
though MAX phases and related compounds have been stud-
ied previously, a full knowledge of why these materials ob-
tain certain properties has not yet been obtained. One reason
for why a complete understanding is lacking lies in the dif-
ficulties in obtaining accurate electronic structure measure-
ments of internal atomic layers.
Previous experimental investigations of the electronic
structure of Ti3SiC2include valence-band x-ray photoemis-
FIG. 1. Color onlineThe hexagonal crystal structure of the
Ti3AC2phases. The Ti atoms have two different sites, denoted TiI
and TiII. Every fourth layer in the 312-phases is interleaved with
layers of the pure A-group element.
PHYSICAL REVIEW B 72, 245101 2005
1098-0121/2005/7224/2451019/$23.00 ©2005 The American Physical Society245101-1
sion XPS.6–8 However, XPS is a surface-sensitive method
which is not element specific. Theoretically, it has been
shown by ab initio band structure calculations that there
should be significant differences between the partial density-
of-states DOSof Ti and C when Si is exchanged for an-
other Aelement.9Thus, if it would be possible to perform
bulk-sensitive and element-specific electronic structure mea-
surements and compare them to ab initio band structure cal-
culations for a series of compounds with different Aele-
ments, one could obtain the most reliable information about
the differences in the internal electronic structure of the bur-
ied layers.
In this paper, we investigate and compare the electronic
structures of Ti3AlC2,Ti
3SiC2, and Ti3GeC2with each other,
using the bulk-sensitive and element-specific soft x-ray emis-
sion SXEspectroscopy method with selective excitation
energies around the Ti 2p,C1s,Al2p,Si2p, and Ge 3p
thresholds. The SXE technique is more bulk sensitive than
electron-based techniques such as XPS and x-ray absorption
spectroscopy XAS. Due to the involvement of both valence
and core levels, each kind of atomic element can be probed
separately by tuning the excitation energy to the appropriate
core edge. The SXE spectroscopy follows the dipole selec-
tion rule and conserves the charge neutrality of the probed
system. This makes it possible to extract both elemental and
chemical near ground-state information of the electronic
structure. The SXE spectra are interpreted in terms of partial
valence band DOS weighted by the transition matrix ele-
ments. The main objective of the present investigation is to
systematically study the nanolaminated internal electronic
structures and the influence of hybridization among the con-
stituent atomic planes in the Ti3AC2materials using the com-
bination of x-ray emission spectroscopy and density-
functional calculations. By comparing the partial electronic
structures of the three 312 systems, important information
about the bonding is achieved, creating a basis for the under-
standing of the unusual materials properties.
A. X-ray emission measurements
The SXE measurements were performed at the undulator
beamline I511-3 at MAX II MAX-lab National Laboratory,
Lund University, Sweden, comprising a 49-pole undulator
and a modified SX-700 plane grating monochromator.10 XAS
spectra at the Ti 2pandC1sedges were recorded in total
electron yield mode by measuring the sample drain current
as a function of the photon energy of the incident monochro-
matized synchrotron radiation. The XAS spectra were nor-
malized to the photocurrent from a clean gold mesh intro-
duced into the synchrotron radiation beam in order to correct
for intensity variations of the incident x-ray beam. During
the XAS measurements at the Ti 2pandC1sedges, the
resolution of the beamline monochromator was about 0.1 eV.
The SXE spectra were recorded with a high-resolution
Rowland-mount grazing-incidence grating spectrometer11
with a two-dimensional detector. The Ti Land C Kx-ray
emission spectra were recorded using a spherical grating
with 1200 lines/ mm of 5 m radius in the first order of dif-
fraction. The Al L,SiL, and Ge Mspectra were recorded
using a grating with 300 lines/mm, 3 m radius in the first
order of diffraction. During the SXE measurements at the Ti
2p,C1s,Al2p,Si2p, and Ge 3pedges, the resolutions of
the beamline monochromator were 1.6, 1.0, 0.3, 0.2, and 0.4
eV, respectively. The SXE spectra were recorded with spec-
trometer resolutions 0.7, 0.2, 0.2, 0.2, and 0.2 eV, respec-
tively. All the measurements were performed with a base
pressure lower than 510−9 Torr. In order to minimize self-
absorption effects,12 the angle of incidence was 30° from the
surface plane during the emission measurements. The x-ray
photons were detected parallel to the polarization vector of
the incoming beam in order to minimize elastic scattering.
B. Deposition of Ti3AC2(A=Al,Si,Ge) films
The MAX-phase thin films were deposited by dc magne-
tron sputtering base pressure 510−10 Torrfrom el-
emental targets of Ti and C, and Al, Si or Ge in an argon
discharge pressure of 4 mTorr.
-Al2O3000lwas used as
substrate. However, to promote a high quality growth of the
MAX phases, a 200 Å thick seed layer of TiC0.7111was
initially deposited. The Ti3AlC2film was 5000 Å thick,
while the Ti3SiC2and Ti3GeC2films were 2000 Å thick. For
further details on the synthesis process, the reader is referred
to Refs. 13–15.
diffractograms of the deposited films are
shown in Fig. 2. As observed, the peaks originate from
Ti3AC2000ltogether with peaks from the TiC111seed-
layer and the
-Al2O30006substrate. In Fig. 2a, small
contributions of Ti3Al and Ti2AlC can also be observed, and
in Fig. 2b, a small peak from Ti5Si3Cxall marked with
arrows. The low intensities of the small additional peaks
show that the films mainly consist of single-phase MAX
phases. Furthermore, the fact that the diffractograms only
FIG. 2. X-ray diffraction pattern from Ti3AlC2,Ti
3SiC2, and
Ti3GeC2thin films. Sdenotes the contribution from the substrate.
The arrows in aindicate small contributions of Ti2AlC0002
leftand Ti3Al right. The arrow in bcorresponds to Ti5Si3Cx.
MAGNUSON et al. PHYSICAL REVIEW B 72, 245101 2005
show Ti3AC2of 000l-type suggests highly textured or ep-
itaxial films. X-ray pole figures verified that the growth in-
deed was epitaxial, and determined the relation to
Ti3AC2000l/ /TiC111//Al
2O3000lwith an in-plane ori-
entation of Ti3AC2210/ /TiC110//Al
2O3210. The epi-
taxial growth behavior has also been documented by trans-
mission electron microscopy TEM.16–20 XPS-analysis
depth profiles of the deposited films within the present study
using a PHI Quantum instrument, showed after 60 seconds
of Ar sputtering a constant composition without any con-
tamination species. From the diffractograms in Fig. 2, the
values of the c-axis were determined to be 18.59 Å, 17.66 Å,
and 17.90 Å for the Ti3AlC2,Ti
3SiC2, and Ti3GeC2MAX
phases, respectively.
A. Calculation of the x-ray emission spectra
The x-ray emission spectra were computed within the
single-particle transition model by using the full potential
linearized augmented plane wave FPLAPWmethod.21 Ex-
change and correlation effects were taken into account
through the generalized gradient approximation GGAas
parametrized by Perdew, Burke, and Ernzerhof.22 A plane
wave cutoff, corresponding to RMTKmax= 8, was used for
all the investigated phases. The charge density and potentials
were expanded up to =12 inside the atomic spheres, and the
total energy was converged with respect to the Brillouin zone
The emission intensity Icfor a hole created in the cth core
shell can be written in cgs units as23
where FcErepresents the spectral distribution given by23–25
In the above equation
cM is the core wave function, q
wave vector of the incident photon,
the polarization ten-
sor, and Ek
jare the energy and the wave function of
the jth valence band at vector k
. The energy of the emitted
photon is represented by
=EEc, where Ecis the core
energy level. The emission spectra were computed by using
the electric-dipole approximation which means that only the
transitions between the core states with orbital angular mo-
mentum to the ±1 components of the valence bands were
considered. The core-hole lifetimes used in the calculations
were 0.73 eV, 0.27 eV, 0.45 eV, 0.5 eV, and 2.0 eV, for the Ti
2p,C1s,Si2p,Al2p, and Ge 3pedges, respectively. A
direct comparison of the calculated spectra with the mea-
sured data was finally achieved by including the instrumental
broadening in the form of Gaussian functions corresponding
to the experimental resolutions see experimental section
II A. The final state lifetime broadening was accounted for
by a convolution with an energy-dependent Lorentzian func-
tion with a broadening increasing linearly with the distance
from the Fermi level EFaccording to the function a+bE
EF, where the constant ais in units of eV and bis
dimensionless.26 For Ti, C, Al, and Si awas 0.01 eV and b
was 0.05. For Ge, a variable lifetime broadening of the va-
lence band was not feasible since the expanded energy re-
gion also contains the 3dcore levels.
B. Balanced crystal orbital overlap population (BCOOP)
In order to study the chemical bonding of the Ti3AlC2,
Ti3SiC2, and Ti3GeC2compounds, we also calculated the
BCOOP function by using the full potential linear muffin-tin
orbital FPLMTOmethod.27 In these calculations, the muf-
fin tins are kept as large as possible without overlapping one
another, so that the muffin tins fill about 66% of the total
volume. To ensure a well-converged basis set, a double basis
with a total of four different
2values is used. For Ti, we
include the 4s,4p, and 3das valence states. To reduce the
core leakage at the sphere boundary, we also treat the 3sand
3pcore states as semicore states. For Al and Si, 3s,3p, and
3dare taken as valence states. For Ge, we used semicore 3d
and valence 4s,4p, and 4dbasis functions. The resulting
basis forms a single, fully hybridizing basis set. This ap-
proach has previously proven to give a well-converged
basis.28 For the sampling of the irreducible wedge of the
Brillouin zone, we use a special-k-point method29 and the
number of kpoints we used is 216 for the self-consistent
total energy calculation. In order to speed up the conver-
gence, a Gaussian broadening of width 20 mRy is associated
with each calculated eigenvalue.
A. Ti Lx-ray emission spectra
Figure 3 topshows Ti L2,3 SXE spectra of Ti3AlC2,
Ti3SiC2, and Ti3GeC2excited at 458 eV, 459.8 eV, 463.5 eV,
and 477 eV photon energies, corresponding to the 2p3/2 and
2p1/2 absorption maxima vertical bars in the XAS spectra
and nonresonant excitation. The XAS measurements have
been made to identify the resonant excitation energies for the
SXE measurements. For comparison of the SXE spectral
profiles, the measured spectra are normalized to unity and are
plotted on a common photon energy scale and relative to the
EFusing the 2p1/2 core-level XPS binding energy of 460.8
eV in Ti3SiC2.8The main L3and L2emission lines are ob-
served at −8.7 eV and −2.5 eV on the relative energy scale.
Contrary to pure Ti, a peak structure is also observed at
−16.2 eV. This is the result of Ti-C hybridization. As ob-
served, the Ti L2,3 SXE spectra appear rather delocalized
wide bandswhich makes electronic structure calculations
suitable for the interpretation of the spectra. Calculated spec-
tra for the two Ti sites TiIand TiII, their sum and difference
are also shown in the lower part of Fig. 3. Although SXE is
a site-selective spectroscopy, separate contributions from the
two crystallographically different Ti sites are not experimen-
tally resolved due to their negligible energy difference. For
resonant excitation at the 2p1/2 absorption maximum 463.5
eV, the L2emission resonates and the L3/L2ratio is close to
2:1. For nonresonant, continuum excitation far above the 2p
thresholds 477 eV, the L3/L2ratio is 6:1. In general, non-
resonant L3/L2emission ratios in 3dtransition metals have
been observed to be higher than the statistical weight of
atomic levels 2:1due to the opening of the Coster-Kronig
process at the 2p1/2 threshold. The Coster-Kronig decay from
the 2p1/2 core-level leads to a shorter lifetime and a larger
Lorentzian width for the 2p1/2 core state than for the 2p3/2
state. For a clear comparison between the experimental and
calculated spectral structures, the nonresonant emission spec-
tra excited at 477 eV are suitable since they resemble the
occupied electronic states, and resonant phenomena are
avoided. In the fitting procedure to the 477 eV SXE spectra,
we employed the experimental values for the L3/L2ratio of
6:1 and the L2,3 peak splitting of 6.2 eV which is larger than
our calculated ab initio spin-orbit splitting of 5.7 eV. Note
that the Ti 2ppeak splitting obtained from XAS, XPS Ref.
30and SXE Ref. 31spectra are not exactly the same due
to screening and different number of electrons in the final
states. The spectral weight at the EFis significantly higher
for the calculated TiII atoms which are directly bonded to the
Aatoms than for the TiIatoms. This suggests that the TiII
atoms contribute more to the conductivity than the TiIatoms.
The TiII-TiIdifference spectra are a measure of the difference
in the electronic structure and hence, the bonding strength
between the 3dstates of the TiII and TiIatoms and the s,p,
dstates of the Aelement. The largest difference in the
TiII-TiIbonding is observed for the Ti3AlC2system. The
most significant difference between the studied systems is
the pronounced shoulder observed both in experiment and
calculationsin Ti3AlC2indicated by the arrow at the top in
Fig. 3which is not observed in the other systems. This
shoulder has a splitting from the main line of 1.5 eV. From
the calculated band structure, the nature of this shoulder is
related to a series of flat bands not shownwith Ti 3dchar-
acter which are shifted towards the EF.
B. C Kx-ray emission spectra
Figure 4 topshows experimental C KSXE spectra of
3SiC2, and Ti3GeC2excited at 284.8 eV and 310
eV photon energies. The XAS measurements have been
FIG. 3. Top, experimental Ti L2,3 SXE spectra of Ti3AlC2,
Ti3SiC2, and Ti3GeC2excited at 458, 459.8, 463.5, and 477 eV,
indicated by the vertical bars in the XAS spectra. The XAS spectra
are aligned to the 2p1/2 threshold and the 2p3/2,1/2 peak splitting is
indicated. Bottom, fitted spectra of the two different Ti sites, TiIand
TiII using the experimental L2,3 peak splitting of 6.2 eV and the
L3/L2ratio of 6:1 in comparison to the 477 eV experimental spec-
tra. Note that the fitted spectra for Ti3SiC2and Ti3GeC2are almost
identical. The sum and difference of the TiII and TiIcontributions
are shown in the middle. The arrow indicates the shoulder of the
Ti3AlC2system and the EFis indicated by the vertical dotted line.
FIG. 4. Top, experimental C KSXE spectra excited at 284.8 and
310 eV. The resonant excitation energy of 284.8 eV is indicated by
the vertical bar in the XAS spectra. Bottom, calculated spectra. The
vertical dotted line indicates the EF.
MAGNUSON et al. PHYSICAL REVIEW B 72, 245101 2005
made to identify the resonant excitation energy vertical bar
for the SXE measurements. The experimental spectra are
plotted both on a photon energy scale and relative to the EF
using the C 1score-level XPS binding energy of 281.83 eV
in Ti3SiC2.8Calculated spectra are shown at the bottom of
Fig. 4. The main peak at −2.6 eV has shoulders on both the
low- and high-energy sides at −4.2 eV and −2 eV. The
agreement between the experimental and calculated spectra
is good although the low-energy shoulder at −4.2 eV is more
pronounced in the experiment. The main peak and the shoul-
ders correspond to the occupied C 2porbitals hybridized
with the Ti 3dand Aspdbonding and antibonding orbitals of
the valence bands. As observed both in the experimental and
calculated spectra, the high-energy shoulder is less pro-
nounced in Ti3AlC2in comparison to Ti3SiC2and Ti3GeC2
which have similar spectral shapes. It should also be noted
that the chemical shift between Ti3AlC2and the other two
studied compounds is very small both in the experiment and
in the theoretical work. Experimentally, the upward chemical
shifts for Ti3AlC2and Ti3SiC2are both 0.05 eV while the
theory predicts a somewhat larger upward shift of +0.24 eV
for Ti3AlC2, in contrast to observations.
C. Al Lx-ray emission spectra
Figure 5 topshows an experimental Al L2,3 SXE spec-
trum of Ti3AlC2measured nonresonantly at 120 eV photon
energy. The experimental spectrum is plotted on a photon
energy scale and relative to the EFusing the Al 2p1/2 core-
level XPS binding energy of 71.9 eV for Ti3AlC2.7A calcu-
lated spectrum with the L2,3 spin-orbit splitting of 0.438 eV
and the L3/L2ratio of 2:1 is shown at the bottom. Comparing
the experimental and calculated spectra, it is clear that the
main peak at −3.9 eV of the SXE spectrum is dominated by
3sfinal states. The partly populated 3dstates form the broad
peak structure close to the Fermi level and participate in the
Ti-Al bonding in Ti3AlC2.Al3pstates dominate in the upper
part of the valence band but do not directly contribute to the
L2,3 spectral shape since they are dipole forbidden. The con-
tribution of the Al 3pstates can be probed using SXE at the
Al K-edge. For the Al L2,3 SXE spectrum, the valence-to-
core matrix elements are found to play an important role to
the spectral shape. In contrast to Al L2,3 SXE spectra of pure
Al, which have a sharp and dominating peak structure within
1 eV below EF, the Al L2,3 SXE spectrum of Ti3AlC2has a
strongly modified spectral weight towards lower energy. A
similar modification of the Al L2,3 SXE spectral shape has
also been observed in the metal aluminides.32 Comparing the
spectral shape to the aluminides, the appearance of the broad
low-energy shoulder around −6 eV in the Al L2,3 SXE spec-
trum of Ti3AlC2can be attributed to the formation of hybrid-
ized Al 3sstates produced by the overlap of the Ti 3dorbit-
als. This interpretation is supported by our first-principle
D. Si Lx-ray emission spectra
Figure 6 topshows an experimental Si L2,3 SXE spec-
trum of Ti3SiC2measured nonresonantly at 120 eV photon
energy. The experimental spectrum is plotted on a photon
energy scale and relative to the EFusing the Si 2p1/2 core-
level XPS binding energy of 99.52 eV for Ti3SiC2.8A calcu-
FIG. 5. Top, experimental Al L2,3 SXE spectrum of Ti3AlC2.A
calculated spectrum is shown below. The vertical dotted line indi-
cates the EF.
FIG. 6. Top, experimental Si L2,3 SXE spectrum of Ti3SiC2.A
calculated spectrum is shown below. The vertical dotted line indi-
cates the EF.
lated spectrum with the L2,3 spin-orbit splitting of 0.646 eV
and the L3/L2ratio of 2:1 is shown at the bottom. In the Si
spectrum, the 3s-character is concentrated to the bottom of
the valence band with the main peak at −7 eV. The shift of
the main peak towards lower energy is related to the extra
valence electron in Si in comparison to Al. As in the case of
Al, the partly populated Si 3dstates in the upper part of the
valence band also participate in the Ti-Si bonding in Ti3SiC2.
The Si 3p-states in the upper part of the valence band do not
directly contribute to the L2,3 spectral shape since they are
dipole forbidden, but can be probed using SXE at the Si
K-edge. The valence-to-core matrix elements are found to
play an important role to the spectral shape of the Si L2,3
SXE spectrum. Notably, the observed spectral shape of the Si
L2,3 SXE spectrum of Ti3SiC2appears rather different from
single crystal bulk Si which has a pronounced double
structure.33,34 On the other hand, the spectral shape of the Si
L2,3 SXE spectrum of Ti3SiC2is similar to what has been
observed for the metal silicides.35,36 Comparing the Si L2,3
SXE spectrum of Ti3SiC2to the silicides, the low-energy
shoulder on the main peak around −8.5 eV can be attributed
to the formation of hybridized Si 3sstates produced by the
overlap of the Ti 3dstates, which is also supported by our
E. Ge Mx-ray emission spectra
Figure 7 topshows experimental Ge M2,3 SXE spectra
of Ti3GeC2measured nonresonantly at 150 eV 27 eV above
the 3p3/2 absorption maximumand 165 eV photon energies.
The experimental spectra are plotted on a photon energy
scale and relative to the EFusing the Ge 3p1/2 core-level
XPS binding energy of 125.5 eV for Ti3GeC2.37 A calculated
spectrum is shown at the bottom, both for the 4sd valence
band full curveand the 3dcore levels dashed curve.It
should be noted that the measured Ge M2,3 emission is two
orders of magnitude weaker than the Al and Si L2,3 emission
which makes the measurements more demanding. The shal-
low 3dcore levels shown at the left consists of two peaks
from the Ge M4,5 M3and M4M23d3p3/2,1/2transi-
tions with energies of −32.7 and −29.1 eV relative to the EF.
The observed Ge M2,3 spin-orbit splitting of 3.6 eV Ref. 38
is somewhat smaller than the calculated value of 4.3 eV and
the calculated 3dcore levels dashed curveare also closer to
the EFby 3.9 eV. The measured M3/M2intensity ratio of
about 6/5:1 differs from the statistical ratio of 2:1. The in-
tensity of the 3dcore levels was divided by a factor of 10 to
match the experimental data. In addition to the calculated
DOS, the overlap by the matrix elements play an important
role for the intensity of the shallow 3dcore levels. However,
the general agreement between the measured and calculated
spectral shapes is good.
The 4sd valence band within 10 eV from the EFconsists
of a double structure with M3and M2emission peaks ob-
served at −12 eV and −8 eV, respectively. This double struc-
ture is significantly different from the three-peak structure
observed in the M2,3 emission of bulk Ge.39 In bulk Ge, the
4sd DOS of the M3and M2emission bands are not only
separated by the spin-orbit splitting of 3.6 eV, but are also
split up into two subbands, separated by 3.5 eV as in the
case of single crystal bulk Si. The result is a triple structure
in the valence band of bulk Ge where the upper and lower
emission peaks are solely due to M3and M2emissions, re-
spectively, while the main middle peak is a superposition of
both M3and M2contributions. On the contrary, the Ge 4sd
DOS in Ti3GeC2consists of a single peak structure with
more spectral weight towards the EF. A shift of peak struc-
tures towards the EFin comparison to bulk Ge has been
observed in Ge Kemission probing the 4pvalence statesof
the metal germanides.40 The Ge 4pstates in the upper part of
the valence band are not directly observed in the M2,3 emis-
sion spectra since they are dipole forbidden but are indirectly
reflected due to the mixing of sand pstates in the solid.
Finally, we note that at 150 eV excitation energy, a weak
energy loss structure is superimposed onto the valence band.
This weak loss structure, tracking 30 eV below the excitation
energy, originates from localized dd excitations from the
shallow 3dcore level. At the excitation energy of 165 eV 42
eV above the 3p3/2 absorption maximumthe dispersing loss
structure does not affect the spectral profile of the valence
FIG. 7. Top, experimental Ge M2,3 SXE spectra of Ti3GeC2
excited at 150 and 165 eV. A calculated spectrum is shown below.
For comparison, the calculated spectrum has been divided by a
factor of 10 for the 3dcore level.
MAGNUSON et al. PHYSICAL REVIEW B 72, 245101 2005
F. Chemical bonding
By relaxing the cell parameters of the Ti3AC2phases A
=Al, Si, and Ge, it was possible to calculate the equilibrium
caxis. They were determined to be 18.64 Å, 17.68 Å, and
17.84 Å for Ti3AlC2,Ti
3SiC2, and Ti3GeC2, respectively.
These values are in excellent agreement with the experimen-
tal values in Sec. II B. In order to analyze the chemical bond-
ing, we show in Fig. 8 the calculated BCOOP of the three
Ti3AC2systems.41,42 This analysis provides more informa-
tion about the bonding and is the partial DOS weighted by
the balanced overlap population. The BCOOP is a function
which is positive for bonding states and is negative for anti-
bonding states. The strength of the covalent bonding can be
determined by summing up the area under the BCOOP
curve. The energy position of the peaks also gives an indi-
cation of the strength of the covalent bonding. First, compar-
ing the areas under the BCOOP curves and the distances of
the main peaks around −3 eV from the EF, it is clear that the
Ti 3d-C 2pbonds are generally much stronger than the Ti
3d-Aelement pbonds. Additional bonding occurs around
−10 eV with an important Ti 3d-C 2soverlap, corresponding
to −16.2 eV when the L2,3 peak splitting is taken into ac-
count see Fig. 3. The areas under the TiII-C curves are
larger than for the TiI-C curves, which indicates stronger
bonds. This implies that the TiII atoms lose some bond
strength to the nearest neighbor Aatoms, which to some
degree is compensated with a stronger TiII-C bond. This ob-
servation implies that the Ti-C bonds in principle can be
stronger in the Ti3AC2phases than in TiC. This can also be
observed in our calculated value for the Ti-C bond length
which is 2.164 Å for TiC, in good agreement with the ex-
perimental value of 2.165 Å.4For comparison, the calculated
shorter TiII-C bond lengths for Ti3AlC2,Ti
3SiC2, and
Ti3GeC2are 2.086 Å, 2.097 Å, and 2.108 Å, respectively,
indicating stronger bonds. Second, comparing the BCOOP
curves between the three systems show that the TiI,II-C
BCOOP curves of Ti3AlC2are the most intense and are
somewhat shifted towards the EF. Third, the TiII3d-Al 3p
BCOOP peak is located at about −1 eV below EFindicated
by the arrow, while the TiII3d-Si 3pand TiII3d-Ge 4p
BCOOP peaks are located around −2 eV below the EF. This
is an indication that the TiII3d-Al 3pbond is weaker than the
TiII3d-Si 3pand TiII3d-Ge 4pbonds which has also been
confirmed by the differences in bond lengths.9Our calculated
TiII-A bond lengths for Ti3AlC2,Ti
3SiC2, and Ti3GeC2are
2.885 Å, 2.694 Å, and 2.763 Å, respectively. Fourth, the
calculated C 2p-A soverlaps not shownhave significantly
different shapes in comparison to the other overlaps which is
an indication that this bond has a weaker covalent character.
For the Al L2,3 SXE spectrum of Ti3AlC2, presented in
Sec. IV C, the BCOOP calculations confirm that the broad
low energy shoulder at −6 eV can be attributed to bonding
Al 3sorbitals hybridized with Ti 3dandC2porbitals. The
main peak at −3.9 eV corresponds to hybridization with
bonding C 2porbitals and the shoulder at −3 eV is due to
both bonding Ti 3dandC2porbitals. The broad structure
around −1 eV is due to bonding Ti 3dorbitals and antibond-
ingC2porbitals. In the case of the Si L2,3 SXE spectrum of
Ti3SiC2presented in Sec. IV D, the shoulder at −8.5 eV is
attributed to hybridization with bonding Ti 3dorbitals and
antibonding C 2porbitals. The peak structure at −3.5 eV
corresponds to hybridization with bonding Ti 3dorbitals and
antibonding C 2porbitals while the wiggles between −1 eV
and −3 eV are related to bonding Ti 3dandC2porbitals of
varying strength.
The present results provide understanding of the elec-
tronic structure and hence the bonding in the Ti3AC2phases
A=Al,Si,Geas a basis for an understanding of their un-
usual material properties. Our SXE investigation confirms
that the Ti 3d-Apbonding is different in Ti3AlC2in com-
parison to Ti3SiC2and Ti3GeC2. The calculations show that
the double peak in the Ti L2,3 SXE spectrum mainly originate
from the TiII atoms. As a consequence of increasing the c
axis of the hexagonal 312 unit cell to its equilibrium length
i.e., increasing the cdistance in the zdirection in Fig. 1and
the simultaneous symmetry breaking of the A-element mirror
plane in the trigonal prisms, the TiII3d-Apbond is shortened
and the TiI3d-C pbond is lengthened. In the case of Ti3SiC2
FIG. 8. Color onlineCalculated balanced crystal overlap popu-
lation BCOOPfor Ti3AlC2,Ti
3SiC2, and Ti3GeC2. The arrow
indicates the energy shift of the TiII 3d-Al 3porbital overlap.
and Ti3GeC2, the splitting of the double-structure stemming
from the TiII atoms in the calculated Ti 3dDOS is consider-
ably smaller 共⬃0.5 eVthan for Ti3AlC21.5 eV. As a con-
sequence, no double-peak structure is resolved and observed
in the sum of the TiIand TiII peak contributions for Ti3SiC2
and Ti3GeC2. The significantly larger peak splitting in
Ti3AlC2is related to the difference in bonding character as
shown by the BCOOP analysis in Fig. 8. This implies that
the TiII3d-Al 3pbond is less covalent than the TiII3d-Si 3p
and TiII3d-Ge 4pbonds. Our results confirm the general
trend in the Ti-Abond strength where the Ti-Si and Ti-Ge
bonds are stronger than the Ti-Al bond.9As the valence elec-
tron population of the Aelement is increased by replacing
the partly filled valence band of a IIIA element by the iso-
electronic valence band of a IVA element, e.g., by replacing
Al to Si or Ge, more charge is placed into the Ti-Abond.
This is directly reflected in a stronger bond and implies a
change in the elastic properties. Ti3SiC2and Ti3GeC2are
therefore stiffer and have significantly larger Young’s modu-
lus and electrical conductivity than Ti3AlC2. However, all
MAX phases are generally softer and have lower Young’s
modulus and higher electrical conductivity than the parent
compound TiC. This makes crystallographically oriented
MAX-phase films particularly useful as coating material in
applications where, e.g., high temperatures, electrical and
thermal conductivity, and low-friction gliding and/or sliding
properties are required. One example of an application where
MAX phase materials have been shown to be useful is com-
ponents such as electrical switches and contacts.
An interesting result is that the TiII-C bond is stronger in
the MAX phases than in TiC, as suggested by the differences
in bond lengths discussed in Sec. IV F. This may seem con-
tradictory but can be understood from the following argu-
ments. If the Aatoms in the MAX phases are replaced by C,
the result is a highly twinned Ti3C3structure see Fig. 1.
The result of detwinning is monocarbide TiC. Preliminary
calculations not shownsuggest that a replacement of all A
elements by C in the 312-crystal structure results in a sub-
stantially smaller charge-transfer from Ti to C compared to
the MAX phases. Equivalently, inserting Aelement planes
into a highly twinned TiC structure implies a substantial
charge transfer from Ti to C and a strengthening of the TiII-C
bond. Here, the weaker TiII-A bond in one direction is com-
pensated by the strengthening of the TiII-C bond in another
direction. The BCOOP analysis confirms that the covalent
Ti 3d-C 2pbond is generally much stronger than the
Ti 3d-Apbond in the MAX phases. The same type of obser-
vation has been made when point vacancies are introduced
into monocarbide, TiC.43 The insertion of vacancies into TiC
resembles that Aplanes are introduced and implies a
strengthening of the Ti-C bond. This is the major reason for
the nanolaminated character in the MAX phases with delami-
nation by the Ti-Abond breaking as an important step in the
deformation mechanism.4It is also known that a larger num-
ber of the same kind of bonds in a system reduces the
strength of each individual bond.41,44 The influence of differ-
ent Aelements on the bond strength in the MAX phases will
be subject to further investigation.
We have made a systematic electronic structure investiga-
tion of Ti3AlC2,Ti
3SiC2, and Ti3GeC2using the combination
of soft x-ray emission spectroscopy and calculations. For all
the emission processes, excellent agreement between experi-
ment and theory is generally observed. However, differences
in the Ti and Ge peak splittings and energy positions are
observed. This provides an important test of the usefulness of
theoretical calculations. It also shows that the interpretation
of the emission spectra of these materials is best understood
as delocalized band states and that excitonic effects are of
minor importance.
The analysis of the electronic structures provides in-
creased understanding of the hybridization and chemical
bonding of the materials. We find that the Ti-Al bonding in
Ti3AlC2has less covalent character than the Ti-Si bond of
Ti3SiC2and Ti-Ge bond of Ti3GeC2. Emission spectra of Si
in Ti3SiC2,AlinTi
3AlC2, and Ge in Ti3GeC2appear very
different from the pure Al, Si, and Ge elements indicating
strong hybridization between the Aatoms with Ti and C.
Changes in the Ti-Abond strength is shown to have direct
implications on the Ti-C bonding. Therefore, the Aelement
substitution and corresponding tuning of the valence electron
population from the unfilled states of Al to the isoelectronic
states in Si and Ge implies that the unusual material proper-
ties of these nanolaminated carbide systems can be custom-
made by the choice of Aelement.
The authors would like to thank the staff at MAX-lab for
experimental support. This work was supported by the Swed-
ish Research Council, the Göran Gustafsson Foundation, the
Swedish Strategic Research Foundation SSFMaterials Re-
search Programs on Low-Temperature Thin Film Synthesis
and the Swedish Agency for Innovation Systems VIN-
NOVAProject on Industrialization of MAX Phase Coatings.
1V. H. Nowotny, Prog. Solid State Chem. 5,271971.
2W. Jeitschko and H. Nowotny, Monatsch. Chem. 98, 329 1967.
3H. Wolfsgruber, H. Nowotny, and F. Benesovsky, Monatsch.
Chem. 98, 2403 1967.
4M. W. Barsoum, Prog. Solid State Chem. 28, 201 2000.
5M. W. Barsoum and T. El-Raghy, Am. Sci. 89, 334 2001.
6N. I. Medvedeva, D. L. Novikov, A. L. Ivanovsky, M. V. Kuz-
netsov, and A. J. Freeman, Phys. Rev. B 58, 16042 1998.
7S. Myhra, J. A. A. Crossley, and M. W. Barsoum, J. Phys. Chem.
Solids 62,8112001.
8S. E. Stoltz, H. I. Starnberg, and M. W. Barsoum, J. Phys. Chem.
Solids 64, 2321 2003.
9Y. Zhou, Z. Sun, X. Wang, and S. Chen, J. Phys.: Condens.
Matter 13, 10001 2001.
MAGNUSON et al. PHYSICAL REVIEW B 72, 245101 2005
10 R. Denecke, P. Vaterlein, M. Bassler, N. Wassdahl, S. Butorin, A.
Nilsson, J.-E. Rubensson, J. Nordgren, N. Mårtensson, and R.
Nyholm, J. Electron Spectrosc. Relat. Phenom. 101-103, 971
11 J. Nordgren and R. Nyholm, Nucl. Instrum. Methods Phys. Res. A
246, 242 1986; J. Nordgren, G. Bray, S. Cramm, R. Nyholm,
J.-E. Rubensson, and N. Wassdahl, Rev. Sci. Instrum. 60, 1690
12 S. Eisebitt, T. Böske, J.-E. Rubensson, and W. Eberhardt, Phys.
Rev. B 47, 14103 1993.
13 H. Högberg, L. Hultman, J. Emmerlich, T. Joelsson, P. Eklund, J.
M. Molina-Aldareguia, J.-P. Palmquist, O. Wilhelmsson, and U.
Jansson, Surf. Coat. Technol. 193,62005.
14 J.-P. Palmquist, O. Wilhelmsson, T. Nyberg, and U. Jansson,
Appl. Phys. Lett. 85, 1066 2004.
15 J. Emmerlich, H. Högberg, Sz. Sasvári, P. O. Å. Persson, L. Hult-
man, J.-P. Palmquist, U. Jansson, J. M. Molina-Aldareguia, and
Zs. Czigány, J. Appl. Phys. 96, 4817 2004.
16 H. Högberg, P. Eklund, J. Emmerlich, J. Birch, and L. Hultman, J.
Mater. Res. 20, 779 2005.
17 J. M. Molina-Aldareguia, J. Emmerlich, J.-P. Palmquist, U. Jans-
son, and L. Hultman, Scr. Mater. 49, 155 2003.
18 T. Seppänen, J.-P. Palmquist, P. O. Å. Persson, J. Emmerlich, J.
Molina, J. Birch, U. Jansson, P. Isberg, and L. Hultman,
SCANDEM Tampere, Finland, 2002.
19 J.-P. Palmquist, S. Li, P. O. Å. Persson, J. Emmerlich, O. Wil-
helmsson, H. Högberg, M. I. Katsnelson, B. Johansson, R.
Ahuja, O. Eriksson, L. Hultman, and U. Jansson, Phys. Rev. B
70, 165401 2004.
20 J.-P. Palmquist, U. Jansson, T. Seppanen, P. O. Å. Persson, J.
Birch, L. Hultman, and P. Isberg, Appl. Phys. Lett. 81, 835
21 P. Blaha, K. Schwarz, G. K. H. Madsen, D. Kvasnicka, and J.
Luitz, WIEN2K, An Augmented Plane Wave
Local Orbitals
Program for Calculating Crystal Properties Karlheinz
Schwarz, Techn. Universität Wien, Austria, 2001.
22 J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77,
3865 1996.
23 J. E. Müller and J. W. Wilkins, Phys. Rev. B 29, 4331 1984.
24 G. Breit and H. A. Bethe, Phys. Rev. 93, 888 1954.
25 U. Fano and J. W. Cooper, Rev. Mod. Phys. 40, 441 1968.
26 A. Santoni and F. J. Himpsel, Phys. Rev. B 43, 1305 1991.
27 J. M. Wills, O. Eriksson, and M. Alouani, “Full potential LMTO
total energy and force calculations,” in Electronic Structure and
Physical Properties of Solids Springer, Berlin, 2000, p. 148.
28 J. M. Wills and B. R. Cooper, Phys. Rev. B 36, 3809 1987;39,
4945 1989.
29 D. J. Chadi and M. L. Cohen, Phys. Rev. B 8, 5747 1973.
30 J. C. Fuggle and N. Mårtensson, J. Electron Spectrosc. Relat.
Phenom. 21, 275 1980.
31 J. A. Bearden, Rev. Mod. Phys. 39,781967.
32 K. Ichikawa, J. Phys. Soc. Jpn. 37, 377 1974.
33 J. E. Rubensson, D. Mueller, R. Shuker, D. L. Ederer, C. H.
Zhang, J. Jia, and T. A. Callcott, Phys. Rev. Lett. 64, 1047
34 P. O. Nilsson, J. Kanski, J. V. Thordson, T. G. Andersson, J.
Nordgren, J. Guo, and M. Magnuson, Phys. Rev. B 52, R8643
35 J. J. Jia, T. A. Callcott, W. L. O’Brien, Q. Y. Dong, J. E. Rubens-
son, D. R. Mueller, D. L. Ederer, and J. E. Rowe, Phys. Rev. B
43, 4863 1991.
36 E. Z. Kurmaev and G. Wiech, J. Non-Cryst. Solids 70, 187
37 Due to screening and charge-transfer from Ge to C, there is a 0.3
eV low-energy shift of the Ge 3p3/2 and 3p1/2 XPS binding
energies in Ti3GeC2121.4 and 125.5 eV, respectivelyin com-
parison to single-crystal bulk Ge 121.7 and 125.8 eV, respec-
38 The 3.6 eV splitting in the measured emission spectra is the dif-
ference between the 3p3/2,1/2 and 3d5/2,3/2 spin-orbit splittings
observed in XPS, i.e., 4.1–0.5 eV.
39 P. O. Nilsson, S. Mankefors, J. Guo, J. Nordgren, D. Debowska-
Nilsson, W. X. Ni, and G. V. Hansson, Phys. Rev. B 64, 115306
40 I. V. Kavich, B. N. Yatsyk, and A. B. Nazarchuk, Phys. Met.
Metallogr. 5, 1094 1985.
41 R. Hoffmann, Solids and Surfaces—A Chemist’s View of Bonding
in Extended Structures Wiley-VCH, New York, 1988.
42 A. Grechnev, R. Ahuja, and O. Eriksson, J. Phys.: Condens.
Matter 15, 7741 2003.
43 H. W. Hugosson, P. Korzhavyi, U. Jansson, B. Johansson, and O.
Eriksson, Phys. Rev. B 63, 165116 2001.
44 L. Pauling, The Nature of the Chemical Bond Cornell University
Press, Ithaca, New York, 1960.
... In the case of polycrystalline Ti 3 SiC 2 , thermopower was found negligible over a wide temperature range [15], a fact later explained by a compensation of electron and holelike contributions (although those contributions take place out of or in plane, respectively, and are thus not compensated in single crystals, polycrystals average them over all random directions, thus leading to a net zero thermopower). For single crystals, a substantial thermopower anisotropy was predicted [16][17][18], and later confirmed by in-plane measurements performed on single-crystalline thin films [19]. A prominent role is played by highly anisotropic intertwined pockets centered on the H point of the hexagonal Brillouin zone (BZ). ...
... For a many-band metallic system like Ti 3 SiC 2 or other MAX phases, the determination of FS and near Fermi level band structure is necessary to be able to rigorously derive the conductivity and other Onsager coefficients [21,22,24]. The FS and band structure of Ti 3 SiC 2 are featured in several ab initio studies [16][17][18][25][26][27] but little work has been done to experimentally measure them, mainly due to the lack of availability of bulk single crystal required for measuring either de Haas-van Alphen oscillations or angle-resolved photoemission spectroscopy (ARPES). ...
... DFT calculations of Ti 3 SiC 2 electronic structure are not new [16][17][18][25][26][27], but we had to repeat some because, among other things, (i) the best fit to the experimentally determined FS requires slightly shifting the Fermi energy value, (ii) comparing theory and experiment requires computations along uncommon directions or points, and (iii) precise FS fitting also requires the use of large grids. Assessing the electronic structure of a given solid through numerical calculations has become highly useful when interpreting the results of photoemission experiments. ...
An in-depth study of the Fermi surface and electronic structure of the ternary nanolamellar carbide Ti3SiC2 is reported. The outputs of angle-resolved photoemission spectroscopy (ARPES) measured on single crystals and density functional theory calculations are systematically compared. The band structure is found to be dominated by Ti d orbitals near the Fermi level. Band inversions near K points are responsible for the intricate shape of the Ti3SiC2 pockets. Previous theory found that those pockets were giving opposite contributions leading to zero thermoelectric power in polycrystalline samples. Here we directly visualize them by ARPES, describe their precise shape, and examine the way they should behave regarding thermoelectric power. The presence of linear band crossings and the effect of spin-orbit coupling on the near Fermi level electronic structure are discussed, as well as potential consequences for magnetotransport and spin Hall effect.
... The individual bonding strength can be realized by comparing the peak area under the −pCOHP curves of different bonds and from the peak energy positions. The most intense peaks for Ti(2a)-C and Ti(4f)-C bonds at approximately −3 eV and the corresponding areas under the peaks imply that the Ti(4f)-C bonds are slightly stronger than the Ti(2a)-C ones, because Ti(4f) atoms lose some bond strength by transferring charge to the neighboring Al atoms, which is compensated by strengthening the Ti(4f)-C bonds to some extent [51]. Furthermore, the peak nearest the Fermi level corresponds to the Ti(4f)-Al bond, and its area is smaller than that of the Ti-C bonds, indicating that the Ti(4f)-Al bond is much weaker than the Ti-C ones. ...
Full-text available
In the MAX phase, the nucleation and motion of basal dislocations dominate the plastic deformation at room temperature. Therefore, information on the dislocation core properties and energy barrier for dislocation glide is essential to understand their plasticity. Dislocations in MAX phases are mainly 〈a〉-type basal edge dislocations (b=a/3[112̄0]) and can glide in the basal plane (typical slip) or form kink bands depending on the load direction and crystal orientation owing to the strong plastic anisotropy. In this study, we determined the core structure, Peierls barrier, and corresponding stress of a/3(0001)[112̄0] edge dislocations in Ti3AlC2 MAX phase by multiple computational approaches using the semidiscrete variational Peierls–Nabarro method, density functional theory, and classical bond-order-potential-based molecular statics calculations. The a/3[112̄0] basal edge dislocation was dissociated into two Shockley partial dislocations with a 2–2.5b-wide stacking fault in between. These partial dislocations glide between the Ti(4f) and Al atomic layer. For straight dislocation motion, the Peierls barrier was single-humped and the corresponding Peierls stress was estimated to be approximately 182 MPa.
... Remarkably, these structures show excellent Ohmic contact with SiC which was even retained after 1000 hrs of aging at 600 C (Fig. 3d), making MAX useful for numerous high temperature electronic device applications. Ti3GeC2 films by magnetron sputtering on Al2O3 (000l) substrate and investigated their electronic structures by soft x-ray emission spectroscopy [62]. Following the earlier reports, before the deposition of 312 phases, they also grew 20 nm TiC0.7 (111) the buffer layer to promote high quality growth. ...
Full-text available
Layered nanolaminate ternary carbides, nitrides and carbonitrides with general formula Mn+1AXn or MAX (n = 1, 2, or 3, M is an early transition metal, A is mostly group 13 or 14 element, and X is C and/or N) has revolutionized the world of nanomaterials, due to the coexistence of both ceramic and metallic nature, giving rise to exceptional mechanical, thermal, electrical, chemical properties and wide range of applications. Although several solid-state bulk synthesis methods have been developed to produce a variety of MAX phases, however, for certain applications, the growth of MAX phases, especially in its high-quality epitaxial thin films form is of increasing interest. Here, we summarize the progress made thus far in epitaxial growth and property evaluation of MAX phase thin films grown by various deposition techniques. We also address the important future research directions to be made in terms of thin-film growth. Overall, in the future, high-quality single-phase epitaxial thin film growth and engineering of chemically diverse MAX phases may open up interesting new avenues for next-generation technology.
... [5][6][7]14 Despite being an important parameter, there are only a few theoretical/experimental studies reported on this topic, 13,15−21 and a fundamental understanding of the interaction between the M n+1 X n -layers and the A-layers is not yet established. Calculations of partial density of states indicate bonding between Ti 3d and Al 3p in Ti 3 AlC 2 , see, e.g., Zhou et al. 13 The strength and nature of this bonding have generally been suggested to be weak and even of weak covalent character; 19,20 however, conclusive evidence thereof remains to be presented. ...
Full-text available
The inherently nanolaminated Ti3AlC2 is one of the most studied MAX-phase materials. MAX-phases consists of two-dimensional Mn+1Xn-layers (e.g., T3C2-layers) with strong internal covalent bonds separated by weakly interacting A-layers (e.g., Al-layers), where the repetitive stacking of the Mn+1Xn-layers and the A-layers suggests being the foundation for the unusual but attractive material properties of the MAX-phases. Although being an important parameter, the nature of the bonding between the Mn+1Xn-layers and the A-layers has not yet been established in detail. The X-ray photoelectron spectroscopy data presented in this paper suggest that the weak interaction between the Ti3C2-layers and the Al-layers in Ti3AlC2 is through electrostatic attraction facilitated by a charge redistribution of the delocalized electrons from the Ti3C2-layers to the Al-layers. This charge redistribution is of the same size and direction as between Ti atoms and Al atoms in TiAl alloy. This finding opens up a pathway to predict and improve MAX-phase materials properties through A-layer alloying, as well as to predict new and practically feasible MXene compounds.
... Par rapport aux autres composés, le réseau cristallin de la phase Ti3SiC2 se déforme peu sous irradiation, avec notamment une augmentation de 1% du paramètre de maille c pour une irradiation de 0,1 dpa. Cette plus forte tolérance aux radiations est liée à la présence des liaisons Ti-Si qui sont plus fortes que les liaisons Ti-Al [79]. Les effets de l'irradiation sur Ti3SiC2 et Ti3AlC2, présentés en Figure I-5, montrent la formation de microfissures après irradiation à 121°C. ...
Full-text available
This study aims to improve oxidation resistance of nuclear fuel claddings in accident conditions. In this context, Cr-Al-C and Cr2AlC coatings deposition and their behavior were studied. Firstly, we investigated the influence of HiPIMS process parameters on the properties of the plasma and the deposited films. Despite more intense ionic fluxes due to the HiPIMS process, coatings do not crystallize without an additional energy supply. Partially crystallized Cr2AlC thin films were obtained by a 500°C annealing of as-deposited Cr-Al-C coatings. This two-step process is a viable solution to protect nuclear claddings with Cr2AlC coating while maintaining the metallurgical properties of the zirconium-based substrates. Secondly, the assessment of the oxidation resistance of as-deposited and annealed coatings revealed significant protective effect against rapid oxidation under dry and wet air at high temperatures (up to 1200°C) owing to the formation of a continuous oxide layer. During the first stages of oxidation, this layer is made of α-Al2O3 and Cr2O3 for as-deposited coating while only α-Al2O3 is present for the annealed one. Because of Al depletion, coatings later deteriorate and form a residual and porous intermediate chromium carbide (Cr7C3) layer which further fully oxidizes. It was shown that the inward diffusion of Al with Zr also accelerates the coating deterioration. To improve the oxidation resistance of these coatings, multilayered architectures were developed. Adding a molybdenum interlayer as diffusion barrier globally decreased the oxidation resistance of the coating. In contrast, topping Cr-Al-C and Cr2AlC with a Cr layer improved oxidation behavior over single-layer coatings.
... As a typical MAX phase, Ti 3 GeC 2 has a theoretical density of 5.55 g/cm 3 and a hexagonal crystal structure [8,9]. Ti 3 GeC 2 exhibits moderate hardness (2 ~ 6 GPa), high Young's modulus (~ 197 GPa), excellent machinability, high yield strength, and signi cant plasticity at 1300 ℃ [7,10]. ...
Full-text available
Ti 3 GeC 2 compound has excellent electrical and mechanical properties, which can be used as a potential reinforcement for Ag-based electrical contact materials. However, few reports have been addressed on the synthesis of its powders. The present study, therefore, presents a new route to fabricate high-purity Ti 3 GeC 2 powders from Ti/Ge/TiC by pressureless sintering in vacuum. The influence of sintering temperature and Ge content on the powder purity was studied and the optimized conditions are found to be 1550 ℃ and Ti:Ge:TiC=1:1.1:2. In addition, the reaction path of Ti/Ge/TiC was revealed to be three steps: First, liquid Ge starts to react with Ti to form Ti 5 Ge 3 compounds at around 1140 ℃; Then Ti 5 Ge 3 consumes the liquid Ge and TiC to form Ti 2 GeC at around 1200 ℃; Finally, Ti 2 GeC transfers into Ti 3 GeC 2 with the consumption of residual TiC. The present study lays the foundation for the subsequent applications of Ti 3 GeC 2 powders in high performance composites.
... Feng et al., also used magnetron sputtering to grow the Ti2AlC phase films [22]. 3 Magnuson et al. investigated the electronic structure of the dc magnetron sputtered Ti3XC2 (X = Al, Si and Ge) films by soft x-ray emission spectroscopy [23]. Mauchamp et al., grew epitaxial (000l) Ti2AlC phase thin films by dc magnetron sputtering and showed the anisotropic nature of the electronic transport [24]. ...
Full-text available
Recently, nanolaminated ternary carbides have attracted immense interest due to the concomitant presence of both ceramic and metallic properties. Here, we grow nanolaminate Ti3AlC2 thin films by pulsed laser deposition on c-axis-oriented sapphire substrates and, surprisingly, the films are found to be highly oriented along the (103) axis normal to the film plane, rather than the (000l) orientation. Multiple characterization techniques are employed to explore the structural and chemical quality of these films, the electrical and optical properties, and the device functionalities. The 80-nm thick Ti3AlC2 film is highly conducting at room temperature (resistivity of 50 micro ohm-cm), and a very-low-temperature coefficient of resistivity. The ultrathin (2 nm) Ti3AlC2 film has fairly good optical transparency and high conductivity at room temperature (sheet resistance of 735 ohm). Scanning tunneling microscopy reveals the metallic characteristics (with finite density of states at the Fermi level) at room temperature. The metal-semiconductor junction of the p-type Ti3AlC2 film and n-Si show the expected rectification (diode) characteristics, in contrast to the ohmic contact behavior in the case of Ti3AlC2 on p-Si. A triboelectric-nanogenerator-based touch-sensing device, comprising of the Ti3AlC2 film, shows a very impressive peak-to-peak open-circuit output voltage of 80 V. These observations reveal that pulsed laser deposited Ti3AlC2 thin films have excellent potential for applications in multiple domains, such as bottom electrodes, resistors for high-precision measurements, Schottky diodes, ohmic contacts, fairly transparent ultrathin conductors, and next-generation biomechanical touch sensors for energy harvesting.
Full-text available
Layered nanolaminate ternary carbides, nitrides and carbonitrides with general formula Mn+1 AXn or MAX (n = 1, 2, or 3, M is an early transition metal, A is mostly group 13 or 14 element, and X is C and/or N) has revolutionized the world of nanomaterials, due to the coexistence of both ceramic and metallic nature, giving rise to exceptional mechanical, thermal, electrical, chemical properties and wide range of applications. Although several solid-state bulk synthesis methods have been developed to produce a variety of MAX phases, however, for certain applications, the growth of MAX phases, especially in its high-quality epitaxial thin films form is of increasing interest. Here, we summarize the progress made thus far in epitaxial growth and property evaluation of MAX phase thin films grown by various deposition techniques. We also address the important future research directions to be made in terms of thin-film growth. Overall, in the future, high-quality single-phase epitaxial thin film growth and engineering of chemically diverse MAX phases may open up interesting new avenues for next-generation technology.
Für die Herstellung von Leistungshalbleiterbauelementen sind reproduzierbare, niederohmige Kontakte elementar. Hierzu ist es notwendig, dass die zugrundeliegenden Bildungsmechanismen verstanden sind und darauf aufbauend das elektrische Verhalten der Kontakte modelliert werden kann. Im Rahmen dieser Arbeit wurden Ti/Al-basierte ohmsche Kontakte auf p-dotiertem 4H-SiC realisiert, anschließend elektrisch charakterisiert und strukturell analysiert und aufbauend auf den Ergebnissen ein numerisches Simulationsmodell entwickelt. Zur Untersuchung des Einflusses von Prozessvariationen auf das ohmsche Verhalten wurden unterschiedliche Probensätze hergestellt. Auf jedem Probensatz wurden Transferlängenmethode (engl. Transfer Length Method) (TLM) Strukturen realisiert, welche es erlauben, den hergestellten ohmschen Kontakt und die darunterliegende Halbleiterschicht elektrisch zu charakterisieren. Hierdurch konnte der Kontaktwiderstand R_C bzw. der spezifische Kontaktwiderstand rho_C als Kennwert für den ohmschen Kontakt und der Schichtwiderstand R_sh als Kennwert für die zugehörige Halbleiterschicht ermittelt werden. Die elektrische Charakterisierung der hergestellten Proben zeigte, dass alle Proben ohmsches Verhalten aufwiesen. Weiterhin zeigten die Ergebnisse, dass der Schichtwiderstand des Halbleitergebiets der hergestellten Proben signifikant höher war, als aus der Theorie zu erwarten gewesen wäre. Diese Abweichung konnte mit der Kompensation der freien Löcher durch vorhandene Elektronen bzw. durch implantationsbedingte Kompensationszentren erklärt werden. Zur Beschreibung der Kompensation wurde in dem, im Rahmen dieser Arbeit entwickelten, Simulationsmodell der Kompensationsgrad f_comp verwendet. Bezüglich des spezifischen Kontaktwiderstands unterschieden sich die hergestellten Probensätze jedoch deutlich, abhängig davon, ob der abgeschiedene Ti/Al-Metallstapel ein umgebendes Passivierungsoxid berührte. Bei den Prozessvarianten mit Abstand zwischen dem abgeschiedenen Ti/Al-Metallstapel und dem umgebenden Passivierungsoxid wurden, ausgehend von den nominellen Kontaktflächen, effektive, spezifische Kontaktwiderstände von 0,5 mOhmcm² bei Raumtemperatur erreicht. Dieser Wert entspricht den, aus der Literatur bekannten, spezifischen Kontaktwiderständen für niederohmige Kontakte. Die Prozessvarianten ohne Abstand wiesen, ausgehend von den nominellen Kontaktflächen, einen effektiven, spezifischen Kontaktwiderstand von 700 mOhmcm² bei Raumtemperatur auf. Dieser sehr große Unterschied konnte durch strukturelle Analysen des ohmschen Kontaktbereichs auf eine Reaktion des Al mit dem umgebenden SiO2 zurückgeführt werden. Hierdurch wurde die Al-Konzentration in dem Ti/Al-Metallstapel signifikant verringert, was zu einem unvollständigen Aufwachsen von Ti3SiC2 an der 4H-SiC-Oberfläche und somit zu einer unvollständigen ohmschen Kontaktbildung führte. Aus dieser unvollständigen Kontaktbildung resultiert eine effektive Reduktion der ohmschen Kontaktfläche und somit eine Erhöhung des Kontaktwiderstands. Die Reduktion der Kontaktfläche wurde mit Hilfe des Flächenfaktors f_area im numerischen Simulationsmodell beschrieben. Weiterhin konnten aus dem temperaturabhängigen Verhalten des spezifischen Kontaktwiderstands die Schottkybarrierenhöhen phi_B aller Probensätze ermittelt werden. Diese zeigten einen Anstieg der Schottkybarrierenhöhe mit steigender Al-Konzentration an der 4H-SiC-Oberfläche. Vergleicht man diese Ergebnisse mit aus der Literatur bekannten Werten, zeigt sich übereinstimmendes Verhalten. Dieses Verhalten widerspricht allerdings dem, aus den physikalischen Grundlagen bekannten Zusammenhang zwischen Schottkybarrierenhöhe und Dotierstoffkonzentration, da die Schottkybarrierenhöhe aufgrund des Schottky-Effekts mit steigender Dotierung fallen sollte. Dieser Widerspruch wurde mit der Hypothese, dass sich die Al-Konzentration an der 4H-SiC-Oberfläche während der ohmschen Kontaktbildung erhöht, im Rahmen dieser Arbeit erstmals erklärt. Da die Erhöhung der Al-Konzentration durch die Eindiffusion beziehungsweise Ansammlung von Al in die 4H-SiC-Oberfläche entstanden ist, wurde zur Beschreibung dieser Erhöhung eine Gaußsche Normalverteilung mit der Dosis der eindiffundierten Al-Konzentration N_Al,Dosis und der zugehörigen Diffusionslänge L_Al,Diff verwendet. Unter Verwendung der Parameter f_comp, f_area, N_Al,Dosis und L_Al,Diff wurde ein numerisches Simulationsmodell in Sentaurus TCAD implementiert. Mit dessen Hilfe konnte das elektrische Verhalten der hergestellten Probensätze temperaturabhängig mit sehr guter Genauigkeit simuliert werden. Zur Verifizierung des numerischen Simulationsmodells wurden an der 4H-SiC-Oberfläche Sekundärionen-Massenspektrometrie (SIMS) Analysen mit einer Tiefenauflösung im Nanometerbereich durchgeführt. Diese Analysen konnten einen signifikanten Anstieg der Al-Konzentration an der 4H-SiC-Oberfläche in den ersten Nanometern nachweisen und somit die im Rahmen dieser Arbeit aufgestellte Hypothese bestätigen. Somit konnte in dieser Arbeit gezeigt werden, dass sich die Al-Konzentration an der 4HSiC-Oberfläche während der ohmschen Kontaktbildung signifikant erhöht und diese Erhöhung entscheidend ist für die Bildung von ohmschen Kontakten auf p-dotiertem 4H-SiC bei der Verwendung eines Ti/Al-basierten Metallstapels.
Full-text available
The MAX phase materials such as layered ternary carbides that simultaneously exhibit characteristics of metallic and ceramic materials have received substantial interest in recent years. Here, we present a systematic investigation of the electronic, structural stabilities, and elastic properties of Ti3(Al1-nSin)C2 (n = 0,1) MAX phase materials using the ab initio method via a plane-wave pseudopotential approach within generalized-gradient-approximations. The computed electronic band structures and projected density of states show that both Ti3SiC2 and Ti3AlC2 are metallic materials with a high density of states at the Fermi level emanating mainly from Ti-3d. Using the calculated elastic constants, the mechanical stability of the compounds was confirmed following the Born stability criteria for hexagonal structures. The Cauchy pressure and the Pugh’s ratio values establish the brittle nature of the Ti3SiC2 and Ti3AlC2 MAX phase materials. Due to their intriguing physical properties, these materials are expected to be suitable for applications such as thermal shock refractories and electrical contact coatings.
Full-text available
The electronic structure of the silicocarbide Ti3SiC2 has been determined by the full-potential linear-muffin-tin-orbital (FLMTO) method. The spectra of the core-electron levels and valence bands of Ti3SiC2 have been obtained by x-ray photoelectron spectroscopy (XPS) and compared with the results of FLMTO calculations and x-ray-emission spectroscopy (XES) data. Using XPS data of the inner electron levels (Ti 2p3/2, Si 2p, and O 1s) and the results of band calculations, the nature of chemical bonding in the silicocarbide was analyzed. The high plasticity of Ti3SiC2 is explained by a weak interaction between the layers comprising Ti6C octahedra and plane nets composed of silicon atoms. The electronic and cohesive energy properties of the nonstoichiometric Ti3SiC and hypothetical Ti3SiC2-based solid solutions (SS's), namely Ti3SiCN and Ti3SiCO, were simulated by the FLMTO method. An analysis of the cohesive properties shows probable destabilization of the hexagonal structure of Ti3SiC2 in the presence of C vacancies and oxygen impurities. By contrast, the partial substitution of N for C (Ti3SiCxN1-x SS's) should lead to an increase in the cohesive properties of the crystal.
Full-text available
We have grown single-crystal thin films of Ti2GeC and Ti3GeC2 and a new phase Ti4GeC3, as well as two new intergrown MAX-structures, Ti5Ge2C3 and Ti7Ge2C5. Epitaxial films were grown on Al2O3(0001) substrates at 1000 °C using direct current magnetron sputtering. X-ray diffraction shows that Ti–Ge–C MAX-phases require higher deposition temperatures in a narrower window than their Ti–Si–C correspondences do, while there are similarities in phase distribution. Nanoindentation reveals a Young’s modulus of 300 GPa, lower than that of Ti3SiC2. Four-point probe measurements yield resistivity values of 50–200 μΩcm. The lowest value is obtained for phase-pure Ti3GeC2(0001) films.
The Al L2,3 emission spectra of Al-Mn, Al-Co, Al-Ni and Al-Cu alloys with various composition of transition metals have been measured with a grazing incidence spectrometer. The results show that the Fermi edges of the Al L2,3 emission spectra of the alloys do not markedly shift with the change of phase, and that the energy levels arising from the 3d states of the transition metal and contributing to the formation of the conduction band of the alloy locate at a few electron volts below the Fermi edge in any phases of the aluminum-transition metal alloys.
The Mn+1AXn layered carbide/nitride-derived phases, where M is an early transition metal, A is an A-group element and X is N or C, have an unusual combination of mechanical, electrical and thermal properties. The surface and crystal-chemistries of two members with n=2 and 3 have been investigated by X-ray photoelectron spectroscopy. The results show that the constituent species are characterized by low binding energies, sometimes exceptionally so. The energies are 281.0 281.5 and 396.9 eV, respectively, for C 1s and N 1s, both of which are at, or below, the lowest values measured for carbides and nitrides. Similarly the Ti 2p energies, in the range 454.0 454.7 eV, are comparable to that of Ti metal and TiC, while the energy of the Al 2p, 72.0 eV, is lower by ca. 0.8 eV than that for Al metal. The signature of the XTi6 octahedral units in the stacking sequence is reminiscent of the corresponding units in TiN, and it is found that a decrease in Ti 2p binding energy is correlated with decrease in average X-Ti bond length. The low binding energies of the A-type species, Al and Si, in planar coordination suggest that binding and thus screening may be derived from out-of-plane interactions. Results relevant to oxidation arising from exposure to air have been obtained. Yes Yes
Commentary on errors in an earlier article on the nature of the chemical bond. Keywords (Audience): First-Year Undergraduate / General
Ultrathin Ge wetting layers, buried in Si(100), have been investigated by soft x-ray emission spectroscopy. With the assistance of ab initio density functional theory calculations the electronic structure of the layers could be established. In particular Si bulk states, splitted and resonating in the Ge layers, were identified.
A survey of some more recent results on the structural chemistry of compounds between transition elements and IVb group elements (carbon, silicon, germanium and tin) will be presented. There are essentially two large classes of compounds to be discussed, characterized by uniform structural principles, namely transition element carbides and related phases on the one hand and defect disilicide structure compounds and derivatives on the other.Starting with the problem of carbon ordering in transition element carbides and hydrogen containing carbides which reveal the significance of the octahedral [T6C]-group, numerous complex carbides of the general formula TxMyCz (T = transition element, M = another transition or B-group element) can be explained by means of a few common structural features. Perovskite carbides of formula T3MC, corresponding to the filled up Cu3Au-type or the filled up U3Si-type structures, β-Mn carbides of formula T3M2C, corresponding to the filled up β-Mn-type structure and K-carbides, related to the Mn3Al9Si-type structure are characterized by linking of the [T6C]-groups by corners. H-phase carbides of formula T2MC and carbides having Ti3SiC2-type structure exhibit linking of the [T6C]-groups by edges. A similar mode of linking also occurs for carbides with V3AsC-type or the filled Re3B-type structures, although in some cases such as VCr2C2 the trigonal prismatic [T6C]-group intervenes. Finally, the η-carbides having filled Ti2Ni-type and carbides of formula T5M3C with filled up Mn5Si3-type structure can be regarded as built up by linking of the octahedral [T6C]-groups by faces. The geometrical factor within the carbides is strongly supported by the short T-C-distances in the structural element [T6C], thus the formation and architecture of complex carbides may be understood from a topo-chemical point of view, for example, Ti2GeC (H-phase carbide) consists of the sum of TiC (octahedral group) and TiGe (trigonal prism).The second class of compounds, which are derived from the TiSi2-type structure, also belong to an uniform geometrical principle; however, some influence of the electronic concentration on the defect of the B-group element (Si,Ge) and the cell parameter will be recognized. The peculiar structural principle can be described by a partial lattice of the transition metal atom corresponding almost perfectly to that of the Ti-atoms of the TiSi2-type while the second partial lattice (Si,Ge or Sn) according to the defect of these atoms is expanding in one direction of the generating (110) plane. As a consequence of the mutual interference of T- and B-group element atoms a helicoidal structural element of the respective Si, Ge, etc., atoms results. Thus, the arrangement is characterized by a typical subcell and occasionally by extremely long c-axis. That also means, fairly complex compositions occur such as Mn11Si19, Mo13Ge23, V17Ge31 or Rh39 (Ga0.5Ge0.5)58. The problem of pseudo-homogenous domains of compounds arise in as far as within fairly small regions of composition a split according to different multiple of subcell will be observed, such as Mn11Si19(MnSi1.727); Mn26Si45(MnSi1.730); Mn15Si25(MnSi1.733) and Mn27Si47(MnSi1.741). A similar change in the multiple of subcells that means independent phases, takes place by substituting either the transition element by another or by substituting the B-group element by another B-group element such as Cr37Ge59As4 which corresponds to the Rh10Ga17-type while Cr11Ge19 is isotypic with Mn11Si19. In general, lowering of the overall electron concentration diminishes the defect of the B-group element in the compound.