Sie sind auf Seite 1von 6

Applied Surface Science 357 (2015) 813

Contents lists available at ScienceDirect

Applied Surface Science


journal homepage: www.elsevier.com/locate/apsusc

Adhesion strength and nucleation thermodynamics of four metals


(Al, Cu, Ti, Zr) on AlN substrates
Yuan Tao, Genshui Ke, Yan Xie, Yigang Chen , Siqi Shi, Haibo Guo
Department of Electronic Information Materials, School of Materials Science and Engineering, Shanghai University, Shanghai 200444, China

a r t i c l e

i n f o

Article history:
Received 30 June 2015
Received in revised form 15 August 2015
Accepted 29 August 2015
Available online 1 September 2015
Keywords:
Aluminum nitride
Adhesion strength
Nucleation
Metallization
Density functional theory

a b s t r a c t
Devices based on AlN generally require adherent and strong interfaces between AlN and other materials,
whereas most metals are known to be nonwetting to AlN and form relatively weak interfaces with AlN.
In this study, we selected four representative metals (Al, Cu, Ti, and Zr) to study the adhesion strength
of the AlN/metal interfaces. Mathematical models were constructed between the adhesion strength and
enthalpy of formation of Almetal solid solutions, the surface energies of the metals, and the lattice
mismatch between the metals and AlN, based on thermodynamic parameters calculated using density
functional theory. It appears that the adhesion strength is mainly determined by the lattice mismatch, and
is in no linear correlation with either the Almetal solutions formation enthalpies or the metals surface
energies. We also investigated the nucleation thermodynamics of the four metals on AlN substrates. It
was found that Ti forms the strongest interface with AlN, and has the largest driving force for nucleation
on AlN substrates among the four metals.
2015 Elsevier B.V. All rights reserved.

1. Introduction
Aluminum nitride (AlN) is a ceramic compound with many technologically attractive properties, such as high thermal conductivity
(320 W m1 K1 ), excellent optical and dielectric properties, high
mechanical strength (greater than alumina and beryllia), strong
corrosion resistance, thermal and chemical stability, and a nontoxic nature, among others [1]. For these reasons, AlN is very
promising for or currently used in optical devices, surface-acoustic
wave devices, and high-power and high-temperature electronic
devices, and is now well recognized for its importance as a thermal
conductivity and insulation material in microelectronics, particularly in electronic packaging [25]. Most applications of AlN entail
mechanically stable and strong interfaces between AlN and other
materials, mostly metals and semiconductors. There is, however, a
known obstacle in this respect, namely poor adhesion of AlN with
most metals. Researchers found that only molten silicon at 1200 C
and aluminum at 850 C can wet AlN, and only titanium behaves as
an active additive to reduce contact angles on AlN [6,7]. Cu, one of
the mostly used metals in printed circuits, is known to form weak
interfaces with AlN. The interfaces have been of much concern to

Corresponding author.
E-mail address: yigangchen@shu.edu.cn (Y. Chen).
http://dx.doi.org/10.1016/j.apsusc.2015.08.243
0169-4332/ 2015 Elsevier B.V. All rights reserved.

researchers in either deposition [810] or metallization [1,6,11] of


AlN.
Besides the adhesion strength of AlN/metal interfaces, nucleation thermodynamics in lm growth is another factor to consider.
The reason is that an interface should rst be developed before
it can be used, and growth processes largely determine the interfaces structure, composition, and bonding. In the initial stage
of lm growth, particularly those that follow the VolmerWeber
lm growth mode where a lm grow by nucleation and coalescence of isolated islands, the formation of nuclei is dictated by
surface energies of the substrate and nuclei, interface energy of
substrate/nuclei, and volumetric Gibbs free energy change of the
nuclei. A similar set of parameters, i.e. the interface energy and surface energies, are also used to quantify mechanical strengths of the
metal/AlN interfaces. Therefore, the adhesion strength and nucleation thermodynamics are connected, even though they account for
very different aspects of an interface: the adhesion strength depicts
how strong the interface is, and the nucleation thermodynamics
depicts how easy the lm may grow on the substrate and develop
the interface.
Surface energies and interface energies are thermodynamically well dened, and are often important parameters for many
thermodynamic models [1214], but are difcult to measure
experimentally. Computations and simulations have become an
alternative and effective tool for obtaining these parameters. In
this work, we study the metal/AlN interface at the level of rst

Y. Tao et al. / Applied Surface Science 357 (2015) 813

principles. We choose four metals, i.e. Al, Cu, Ti, and Zr, based
on the following considerations. All the selected metals are close
packed, which allow for coherent interfaces, and previous studies
have shown that Al, Ti, and Zr are helpful to improve the interfacial
adhesion of AlN. Ti has been proposed for wetting AlN substrates for
high-temperature semiconductor devices, and was frequently used
as a mask of AlN in wet etching experiments [15,16]. Especially, Ti
behaves as an active additive, which reduces the contact angle of
molten metals on AlN to a value below 90 [6,7,17]. Zirconium is the
next element of the same group as Ti in the periodic table. Zr, when
added to Ag up to a few percent, leads to a considerable decrease
in contact angle of molten Ag on AlN from values much higher than
90 down to values close to 2030 [18]. Although Al is rarely used
in semiconductor devices, it is considered as a test case because of
its chemical afnity toward AlN. Al does not react with AlN, but can
be chemically adsorbed on AlNs surface and improve wetting of
other metals to the substrate [8,19]. Besides, coherent and sharp
interfaces could form when depositing AlN on pre-nitrided Al surfaces [8]. Cu is widely used in printed circuits, but is known to be
nonwetting to AlN [9,11,20].
2. The models and methods
The interfacial adhesion strength can be dened in several
different ways [21]. In thermodynamics the work of separation
(Wsep ) is conveniently used which denes the energy per unit area
required to separate the two parts of an interface and to form two
innitely separated free surfaces. A denition of Wsep could be
written as
Wsep =

1
(E1 + E2 Eint )
A

(1)

where Eint is the total energy of an interface system, E1 and E2


are the total energies of the two surface-containing systems that,
labeled respectively by subscripts 1 and 2, coalesce to form the
interface, and A is the interface area. Alternatively, interface energy
can also be used for characterizing an interfaces strength. An interface energy can be dened as the difference between the free energy
of the interface system, and the sum of free energies of its components in their reference bulk states [22], as

int

1
=
A

Eint

i Ni

(2)

where we assume the chemical system of the interface is composed


of one or more chemical systems, labeled by the subscript i, each
having Ni formulae and a chemical potential of i per formula.
The values of the work of separation and of the interface energy
are, of course, different because of the difference reference states
(zero-energy states), but both are capable of describing an interfaces adhesion strength. The different in the reference states can
be reconciled by surface energies of the two parts of an interface
( 1 and  2 ), and the work of separation (Wsep ) can be related to the
interface energy ( int ) by
Wsep = 1 + 2 int

Table 1
Interface registry of the AlN/metal interfaces and the lattice mismatch with respect
to the AlNs lattice parameters. The sign of the lattice mismatch depicts whether the
metals lattice constant is larger (+) or smaller () than AlNs.
Interface (AlN/metal)

Match (AlN//metal)

AlN/Al
AlN/Cu
AlN/Ti
AlN/Zr

( 7 7)R41//(3 3)R0

(3 3)R60//( 13 13)R46

(2 3 2 3)R30//( 13 13)R14
(1 1)R0//(1 1)R0

A: lattice
mismatch (%)
3.53
1.19
2.84
3.33

this case, chemical reservoirs with denite chemical potentials


can be used to supplement the nonstoichiometric system. In our
calculations, excess nitrogen atoms are from a reservoir of noninteracting N2 molecules, and excess metal (Al, Cu, Ti, and Zr) atoms
are from reservoirs of bulk metals in their thermodynamic equilibrium structures, i.e. face-center cubic for Al and Cu, and hexagonal
close-packed for Ti and Zr. All the chemical reservoirs are at the
state of temperature T = 0 K and pressure P = 0 Pa. We point out that
chemical reservoirs are unnecessary for the work of separation (Eq.
(1)).
To compute the energies in Eqs. (1) and (2), we have constructed
four sets of AlN/metal interfaces, namely AlN/Al, AlN/Cu, AlN/Ti,
and AlN/Zr. In this study, all the interfaces are parallel to the closepacked planes in AlN and the metals, i.e. (0 0 0 1) for AlN, Ti and
Zr, and (1 1 1) for Al and Cu. The AlN (0 0 0 1) surface is intrinsically
polar, and we have selected the Al-polar one, which is nonvolatile
and may be alloying with the selected metals.
Due to lattice mismatch between AlN and the metals, exceedingly large strains may result from direct joining of two (1 1) unit
cells. Therefore, appropriate rotation and resizing of the lattices
were applied to reduce the mismatch to below 5% with respect
to AlNs lattice parameters. Our scheme of the transformations is
shown in Table 1, in which A is a lattice mismatch dened as
A =

ametal aAlN
aAlN

(4)

Here ametal and aAIN are the lattice constants of a metal (after transformation, if any) and AlN. A positive A means the lattice constant
of the metal is larger than that of AlN, and the metal part is under
a compressive strain, otherwise the metal is under a tensile strain.
We pointed out that smaller lattice mismatches than those listed
in the table may be obtained, at the cost of larger computational
cells that require heavier computational loads.
When building the interface models, we varied the distance

between the two surfaces, or the interface width, from 1.0 to 7.0 A,
and calculated the total energies of the interfaces at each width.
is smaller than the
The lower bound of the interface width (1.0 A)
and when
typical bond lengths in Almetal solid solutions (2.5 A),
the energies converge to certain
the interface width is beyond 7 A,
constants. The calculated energies are used to t the universal binding energy relationship (UBER) [23], which may be expressed in the
following equation:

 dd 
0

(d) = 0 exp

(3)

where surface energies  1 and  2 must be dened with respect to


the same chemical reservoir as the interface energy  int (see below).
The denition in Eq. (2) is unequivocal for a unary system, such
as a grain boundary in a pure metal, because the unary system has a
denite thermodynamic reference state. For compounds, however,
both the interface system and the surface systems may be nonstoichiometric as compared to their reference states. For example,
a wurtzite AlN (0 0 0 1) surface may have a nitrogen-termination
or an aluminum-termination, corresponding respectively to an Nrich or Al-rich composition which differs from the bulk AlN. In

(5)

where (d) is the relative energy of the interface system with the
width d, and  0 , d0 , and l are interface-specic parameters. The
relative energy (d) is dened with respect to the state of innitely
separated surfaces, as in
(d) =

1
[E (d) Eint ()]
A int

(6)

where Eint (d) is the total energy of an interface with an interface


width of d. In our calculations, when d was larger than approxi Eint (d) approached a constant value for each set of
mately 56 A,
the interfaces, and could be taken as reference energy. One sees

10

Y. Tao et al. / Applied Surface Science 357 (2015) 813

Table 2
k-Point meshes used in the calculations and the resultant irreducible k-points in the
Brillouin zone.

Table 3
Fitted parameters of the UBER models of the four interfaces,  0 and d0 are the
minimum relative energy and the corresponding separation, respectively, in Fig. 1.

System

Mesh

Interface

d0 ()

 0 (J m2 )

l ()

Al, FCC
Cu, FCC
Ti, HCP
Zr, HCP
AlN, wurtzite
AlN/Al, interface
AlN/Cu, interface
AlN/Ti, interface
AlN/Zr, interface

23 23 23
19 19 19
13 13 9
13 13 9
995
441
331
331
551

364
364
105
105
36
24
3
5
5

AlN/Al
AlN/Cu
AlN/Ti
AlN/Zr

2.60
2.31
2.42
3.11

1.01
1.45
1.61
0.92

0.24
0.19
0.49
0.22

interface structures. The k-point meshes sizes and numbers of the


resultant irreducible k-points are listed in Table 2.
3. Results and discussions

from Eq. (5) that (d) reaches the minimum value of  0 at the equilibrium width d0 , and the rst derivative of (d) reaches maximum
at the width d0 + 1. According to Rose et al. [24], the cohesive energies or binding energies of metals have this universal form. This
phenomenon has been observed for adhesion [25] and cohesion
[26] of metals, as well as chemisorption on metal surfaces [27].
Hayes et al. [28] have even observed similar universal behavior of
energy-displacement curves for non-metallic systems (Al2 O3 and
Si).
The adhesion properties of the AlN/metal interfaces are calculated using density functional theory (DFT) [2931] implemented
in the program Vienna Ab Initio Simulation Package (VASP version 5.2) [32,33]. The potentials of nuclei and core electrons are
reproduced using the projector augmented wave method (PAW)
[34,35]. The PAW datasets are generated with the valence congurations of d10 s1 for Cu, d3 s1 for Ti, and d2 s2 for Zr, as shipped
in the VASPs potential database. Generalized gradient approximation (GGA) with the exchange-correlation functional in the form
of Perdew, Burke, and Ernzerhof (PBE) [36] is used to describe the
interactions in the electronic systems. We have applied the spindegenerated condition in our calculations as the material systems
in the present study are deemed nonmagnetic.
Our convergence tests showed a plane-wave cutoff of 500 eV
is sufcient to decrease the associated error in the calculated
energies to less than 3 meV/atom. We use a Gaussian distribution function for the electronic occupation with a width parameter
of 0.1 eV. Geometry optimizations are performed using the conjugate gradient algorithm until the residual forces are smaller than
0.05 eV/. All atoms in the AlN/metal interfaces are allowed to relax
except for boundary atoms whose z-coordinates are xed to attain
designed interface widths during the relaxations. The convergence
criteria for self-consistent electronic iterations are 105 eV. The kpoints are generated using the MonkhorstPack scheme [37] and
tested for the convergence criteria of 3 meV/atom for both bulk and

We built the AlN/metal interface structures and calculated the


relative energies as dened in Eq. (6). For each set of the interfaces,
were tted
the relative energies versus interface widths (17 A)
to the UBER model, and the tting results are shown in Fig. 1 and
Table 3. Each curve in Fig. 1 has an energy minimum that is minus
the work of separation of the corresponding interface. A smaller
negative minimum of the relative-energy curves in Fig. 1 thus
means a larger positive work of separation, and accordingly more
energy is required to break apart the interface structure. The work
of separation of AlN/Ti is about 1.61 J m2 , which is approximately
the same as the cleavage energy of pure Al along (1 1 1) plane, but
is only about half the cleavage energy of TiAl alloys [38,39]. Since
the chemical environment is very different in this interface system
from that in TiAl alloys, the difference in the work of separation
(or cleavage energy) in the two systems is unsurprising. Nevertheless, the comparison of the numbers is much more important
than the numerical values, and using the numbers listed in Table 3
we may sort the work of separation of the interfaces in the order
AlN/Ti > AlN/Cu > AlN/Al > AlN/Zr. By the connection between work
of separation and adhesion strength of an interface, the AlN/Ti interface has the largest adhesion strength among all the four interfaces,
and the AlN/Zr interface has the smallest adhesion strength.
To understand the connection between the adhesion strength
and other properties of the interfaces, we have constructed mathematical models. One may think that the work of separation is
related to the surface energies of constituent surfaces that form
the interface, because by denition a work of separation is the
energy difference between the two free surfaces and the interface.
As shown in Fig. 2, it appears that Wsep is not linearly dependent
on the surface energies of four metals. Higher order of polynomial
relationships between Wsep and the surface energies are probable,
but more data are required for a reliable data regression than in the
present study. The results disagree with a previous research that

Fig. 1. The interfaces relative energies versus interface widths of the four interfaces.
The markers are calculated values, and the curves are tted UBER models.

Fig. 2. The relationship between surface energy  and Wsep .

Y. Tao et al. / Applied Surface Science 357 (2015) 813

Fig. 3. The relationship between formation enthalpy of Almetal solid solutions and
Wsep .

showed Wsep was negatively correlated to the surface energies of


the constituent surfaces [13].
We conjure chemical afnities between Al and the four metals
contribute to the work of separation, based on the following considerations. The interfaces are formed between Al-terminated AlN
surface and close-packed planes of the four metals. When the interface width is small enough, chemical bonds should form among the
metal atoms and immediately neighboring Al atoms. The formation
enthalpies of Almetal solid solutions versus Wsep of AlN/metal
interfaces are given in Fig. 3. According to the results of linear
least-square regressions, we found the linear correlation is rather
weak since the tted coefcient a in y = ax + b is nearly zero (see
Table 4), where y represents Wsep , and x represents enthalpies of
formation. Thus, we conclude that there is no obvious linear relationship between Wsep and the formation enthalpy of Almetal
solid solutions.
The difference in the work of separation between the AlN/Ti and
AlN/Zr is, however, unexpectedly large. Ti and Zr belong to the IVB
group in the periodic table, and share many similarities in chemical
and physical properties. The two elements have the same crystal
structure (hexagonal close-packed or HCP), similar surface energies
(see Table 5), and very similar negative formation enthalpies with
Al (61 kJ mol1 for TiAl, and 83 kJ mol1 for ZrAl). It is thus anticipated that the work of separation should be close to each other,
contrary to the calculation results.
The major difference between the two metals is their lattice
mismatch with respect to the AlN lattice. We assume the lattice
mismatch of the interface would contribute to the difference of
Wsep , and have studied the correlation between Wsep and the
Table 4
The tted parameters of the linear models y = ax + b in Figs. 24.
x
H
A
 surf

a
2

7.1E4 J m kJ
0.11 J m2
0.22

mol

b (J m2 )

Fig. No.

1.28
1.3
1.49

Fig. 2
Fig. 3
Fig. 4

Fig. 4. The relationship between strain and Wsep .

corresponding strains, as shown in Fig. 4. We can see that Wsep is


qualitatively well approximated by a linear function of the strain.
We show the parameters of the linear function y = ax + b in Table 4.
We found that the larger values of Wsep are obtained when the
surface unit cell of the metal part is smaller than that of the AlN,
such that a high Wsep corresponds to the interface with the metal
part under a tensile strain. Since the strain in the metal is tensile,
a bond stretching in the metal layer induces an interfacial bond
strengthening and thus increases the interface cohesion [13]. In
general, from Fig. 4 it can be concluded that the Wsep is larger for
interfaces with the metals under tensile strains than those under
compressive strains.
As for the nucleation thermodynamics, interface energies are
energy barriers that must be overcome in a nucleation process,
and are more important than the work of separation. Fig. 5 is an
illustrative gure of the nucleation formation of metals on an AlN
substrate. The energy change in a nucleation process may be calculated by the equation
E = metal A + (AlN + int ) A + GV V

 surf (J m2 )

 int (J m2 )

Wsep (J m2 )

Al
Cu
Ti
Zr

0.81
1.25
0.97
1.40

1.93
1.96
2.40
2.41

1.01
1.45
1.61
0.92

(7)

where A and V are the surface area and volume of the nuclei,
respectively, A is the interface area of the wetting model,  metal
and  AIN are the surface energies of metal and AlN, respectively,
 int is the interface energy of metal/AlN interface, and GV is the
change in volumetric free energy. The energy change E can be
deemed as the driving force of the nucleation process. Fig. 5 is an
illustration of a metal nuclei on an AlN substrate, where the contact angle  depends on the surface energies of the metal and AlN,
and also interface energy of the AlN/metal interface, as depicted
in the Youngs equation. If we assume the nuclei shape is semispherical and the contact angle is 90 , which distinguish wetting
from nonwetting, we can calculate the areas A and A, the volume
V, and the energy change according to Eq. (7). Fig. 6 shows the driving force versus the nucleis size (characterized by diameter d). We
can see that the driving force decrease very quickly with diameter
for small nuclei, and then gradually stabilized for large nuclei. For
small nuclei, the surface-to-volume ratio is large, and the contribution from surfaces and interfaces is signicant compared to the

Table 5
The surface energies ( surf ) of metals, and interface energies ( int ) and work of
separation (Wsep ) of the AlN/metal interfaces. Note the Wsep is opposite of  0 in
Table 3.
Metal

11

Fig. 5. The schematics of metal nuclei on an AlN substrate.

12

Y. Tao et al. / Applied Surface Science 357 (2015) 813

Ti is the most competitive metal to form an adherent and strong


interface with AlN among the four metals.
Acknowledgements
This study is primarily supported by Shanghai Pujiang Program
under the grant No. 11PJ1403400 and National Natural Science
Foundation of China (NSFC) for Young Scholars under the grant No.
51302166, and Innovation Program of Shanghai Municipal Education Commission under the grant No. 14YZ012. The computations
were performed on the high-performance computing platform of
Shanghai University.
References

Fig. 6. Driving force as a function of nucleis size for the four metals with a common
wetting angle of 90 on the AlN substrate.

contribution from the bulk. When the nucleis size increases, the
surface-to-volume ratio decreases, and the driving force is mainly
from the bulks contribution (GV ), which is negative and constant
(invariant with nucleis size). From Fig. 6 we see that the driving
forces for the nucleation of Ti and Zr on AlN are close, and those
for Al and Cu are almost indistinctive. The driving forces of the four
metals are in the order Zr Ti < Cu (Al). Since a smaller negative
driving force corresponds to a larger tendency of nucleation, the
smaller values for Ti and Zr mean the two metals nucleate more
spontaneously than Al and Cu on AlN substrates.
4. Conclusions and remarks
In this paper, we used rst principles to study the interfaces of
four metals (Al, Cu, Ti, Zr) with AlN of Al-termination. The adhesion strength of the interfaces was characterized by the work of
separation. We found that larger work of separation is associated
with the AlN/Ti and AlN/Cu interfaces that have their metal parts
under tensile strains, and smaller work of separation is associated
with the AlN/Al and AlN/Zr interfaces that have their metal parts
under compressive strains. It appeared that the strain, or the lattice mismatch, mostly determines the interfacial adhesion strength.
Other properties, such as the metals surface energies, and chemical afnities with Al, were tested but were found to be in no linear
correlation with the work of separation. For the nucleation thermodynamics, we found that the nucleation of Zr and Ti should proceed
with larger energy decrease than that of Al and Cu. This means that
Zr and Ti should nucleate more easily than Al and Cu under same
conditions. The nucleation thermodynamics indicates that Zr and
Ti can form chemically adherent interfaces with AlN. Based on the
adhesion strength and nucleation thermodynamics, Ti is the best
metal that can form strong and adherent interface with AlN, thus
can be used for metallization of AlN, or as a buffer layer for enhancing adhesion and nucleation of otherwise nonwetting materials on
AlN substrates.
It is worth noting that the driving force of nucleation is for the
condensation of metal atoms in a gas state onto AlNs surface, and
form an interface between the nuclei and AlN. This is sharply different from the adhesion strength, which is for an interface that is
already formed. It seems the interfacial adhesion of AlN/Cu is strong
(close to AlN/Ti and better than AlN/Al), but the nucleation of Cu on
AlN is difcult (compared to Ti and Zr). The interfacial adhesion of
AlN/Al is relatively weak because of the lattice mismatch, which has
Al metal under compressive strain in the AlN/Al interface system.

[1] F. Miyashiro, N. Iwase, A. Tsuge, F. Ueno, M. Nakahashi, T. Takahashi, High


thermal conductivity aluminum nitride cermaic substrates and packages, IEEE
Trans. Compon. 13 (1990) 313319.
[2] A.J. Fischer, A.A. Allerman, M.H. Crawford, K.H.A. Bogart, S.R. Lee, R.J. Kaplar,
W.W. Chow, S.R. Kurtz, K.W. Fullmer, J.J. Figiel, Room-temperature direct
current operation of 290 nm light-emitting diodes with milliwatt power
levels, Appl. Phys. Lett. 84 (2004) 33943396.
[3] S. Rajan, P. Waltereit, C. Poblenz, S.J. Heikman, D.S. Green, J.S. Speck, U.K.
Mishra, Power performance of AlGaN-GaN HEMTs grown on SiC by
plasma-assisted MBE, IEEE Electron Dev. Lett. 25 (2004) 247249.
[4] Y. Taniyasu, M. Kasu, T. Makimoto, An aluminium nitride light-emitting diode
with a wavelength of 210 nanometres, Nature 441 (2006) 325328.
[5] C. Ozgit-Akgun, E. Goldenberg, A.K. Okyay, N. Biyikli, Hollow cathode
plasma-assisted atomic layer deposition of crystalline AlN, GaN and
AlxGa1-xN thin lms at low temperatures, J. Mater. Chem. C 2 (2014)
21232136.
[6] N.Y. Taranets, Y.V. Naidich, Wettability of aluminum nitride by molten metals,
Powder Metall. Met. Ceram. 35 (1996) 282285.
[7] S. Zhu, W. Wosinski, Joining of AlN ceramic to metals using sputtered Al or
Ti lm, J. Mater. Process. Technol. 109 (2001) 277282.
[8] W. Wang, W. Yang, Z. Liu, Y. Lin, S. Zhou, H. Qian, F. Gao, G. Li, Epitaxial
growth of high quality AlN lms on metallic aluminum substrates,
CrystEngComm 16 (2014) 4100.
[9] W. Wang, W. Yang, Z. Liu, Y. Lin, S. Zhou, H. Qian, F. Gao, H. Yang, G. Li,
Epitaxial growth of homogeneous single-crystalline AlN lms on
single-crystalline Cu(1 1 1) substrates, Appl. Surf. Sci. 294 (2014)
18.
[10] H. Yang, W. Wang, Z. Liu, W. Yang, G. Li, Epitaxial growth mechanism of
pulsed laser deposited AlN lms on Si(1 1 1) substrates, CrystEngComm 16
(2014) 3148.
[11] C.-Y. Su, C.T. Pan, M.-S. Lo, Microstructure and Mechanical Properties of
AlN/Cu Brazed Joints, J. Mater. Eng. Perform. 23 (2014) 32993304.
[12] H.B. Guo, Y. Qi, X. Li, Adhesion at diamond/metal interfaces: a density
functional theory study, J. Appl. Phys. 107 (2010) 033722.

[13] J.I. Beltrn, M.C. Munoz,


Ab initio study of decohesion properties in
oxide/metal systems, Phys. Rev. B 78 (2008) 245417.
[14] J.M. Howe, Bonding, structure, and properties of metal/ceramic interfaces:
part 1 chemical bonding, chemical reaction, and interfacial structure, Int.
Mater. Rev. 38 (1993) 233256.
[15] D. Zhuang, J.H. Edgar, Wet etching of GaN AlN SiC: a review, Mater. Sci. Eng. R
48 (2005) 146.
[16] W. Guo, J. Xie, C. Akouala, S. Mita, A. Rice, J. Tweedie, I. Bryan, R. Collazo, Z.
Sitar, Comparative study of etching high crystalline quality AlN and GaN, J.
Cryst. Growth 366 (2013) 2025.
[17] J. Li, P. Wei, Q. Huang, J. Chen, Z. Zhang, Mechanism of titanium deposition on
AlN surface by molten salt reaction, Mater. Lett. 57 (2003)
13691373.
[18] A. Koltsov, F. Hodaj, N. Eustathopoulos, A. Dezellus, P. Plaindoux, Wetting and
interfacial reactivity in AgZr/sintered AlN system, Scr. Mater. 4 (2003)
351357.
[19] T. Klnk, S. Klnk, S. Baba, . ztrk, S. Danisman, S. Savas, The effect of
alumina and aluminium nitride coating by reactive magnetron sputtering on
the resin bond strength to zirconia core, J. Adv. Prosthodont. 5 (2013)
382387.
[20] G.R. Prin, T. Bafe, M. Jeymond, N. Eustathopoulos, Contact angles and
spreading kinetics of Al and AlCu alloys sintered AlN, Mater. Sci. Eng. A
Struct. Mater. Prop. Microstruct. Process. 298 (2001) 3443.
[21] C. Stamp, A.J. Freeman, Structure and stability of transition metal nitride
interfaces from rst-principles: AlN/VN, AlN/TiN, and VN/TiN, Appl. Surf. Sci.
258 (2012) 56385645.
[22] D.J. Siegel, L.G. Hector, J.B. Adams, Ab initio study of Al-ceramic interfacial
adhesion, Phys. Rev. B 67 (2003) 092105.
[23] A. Banerjea, J.R. Smith, Origins of the universal binding-energy relation, Phys.
Rev. B 37 (1988) 66326645.
[24] J.H. Rose, J.R. Smith, J. Ferrante, Universal features of bonding in metals, Phys.
Rev. B 28 (1983) 18351845.

Y. Tao et al. / Applied Surface Science 357 (2015) 813


[25] J. Ferrante, J.R. Smith, Theory of metallic adhesion, Phys. Rev. B 19 (1979)
39113920.
[26] J.H. Rose, J.R. Smith, F. Guinea, J. Ferrante, Universal features of the equation
of state of metals, Phys. Rev. B 29 (1984) 29632969.
[27] J.R. Smith, J. Ferrante, J.H. Rose, Universal binding-energy relation in
chemisorption, Phys. Rev. B 25 (1982) 14191422.
[28] R.L. Hayes, M. Ortiz, E.A. Carter, Universal binding-energy relation for crystals
that accounts for surface relaxation, Phys. Rev. B 69 (2004) 172104.
[29] J.P. Perdew, R.G. Parr, M. Levy, J.L. Balduz, Density-functional theory for
fractional particle number: derivative discontinuities of the energy, Phys. Rev.
Lett. 49 (1982) 16911694.
[30] P. Hohenberg, Inhomogeneous electron gas, Phys. Rev. 136 (1964) B864.
[31] W. Kohn, L.J. Sham, Self-consistent equations including exchange and
correlation effects, Phys. Rev. B 140 (1965) A1133.
[32] G. Kresse, J. Hafner, Ab initio molecular-dynamics simulation of the
liquid-metal-amorphous-semiconductor transition in germanium, Phys. Rev.
B 49 (1994) 1425114269.

13

[33] G. Kresse, J. Furthmller, Efciency of ab-initio total energy calculations for


metals and semiconductors using a plane-wave basis set, Comput. Mater. Sci.
6 (1996) 1550.
[34] P.E. Blchl, Projector augmented-wave method, Phys. Rev. B 50 (1994)
1795317979.
[35] G. Kresse, D. Joubert, From ultrasoft pseudopotentials to the projector
augmented-wave method, Phys. Rev. B 59 (1999) 17581775.
[36] J.P. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approximation
made simple, Phys. Rev. Lett. 77 (1996) 38653868.
[37] H.J. Monkhorst, J.D. Pack, Special points for Brillouin-zone integrations, Phys.
Rev. B 13 (1976) 51885192.
[38] L. Wang, J.X. Shang, F.H. Wang, Y. Zhang, A. Chroneos, Unexpected
relationship between interlayer distances and surface/cleavage energies in
-TiAl: density functional study, J. Phys. Condens. Matter 23 (2011) 265009.
[39] P. Lazar, R. Podloucky, Cleavage fracture of a crystal: density functional theory
calculations based on a model which includes structural relaxations, Phys.
Rev. B 78 (2008) 104114.

Das könnte Ihnen auch gefallen