ISSN 10637826, Semiconductors, 2010, Vol. 44, No. 9, pp. 1192–1198. © Pleiades Publishing, Ltd., 2010.

Original Russian Text © R.I. Romanov, V.V. Zuev, V.Yu. Fominskii, M.V. Demin, V.V. Grigoriev, 2010, published in Fizika i Tekhnika Poluprovodnikov, 2010, Vol. 44, No. 9,
pp. 1229–1235.

SEMICONDUCTOR STRUCTURES,
INTERFACES, AND SURFACES

Electrical Properties of ThinFilm Structures Formed

by Pulsed Laser Deposition of Au, Ag, Cu, Pd, Pt, W, Zr Metals
on n6HSiC Crystal
R. I. Romanov, V. V. Zuev, V. Yu. Fominskii^, M. V. Demin, and V. V. Grigoriev
MEPhI National Research Nuclear University, Kashirskoe sh. 31, Moscow, 115409 Russia
^email: [email protected]
Submitted February 16, 2010; accepted for publication February 24, 2010

Abstract—Diode structures with ideality factors of 1.28–2.14 and potential barriers from 0.58 to 0.62 eV on
the semiconductor side were formed by pulsed laser deposition of Au, Ag, Cu, Pd, Pt, W, and Zr metal films
on n6HSiC crystal without epitaxial layer preparation. A high density of surface acceptor and donor states
was formed at the metal–semiconductor interface during deposition of the laserinduced atomic flux, which
violated the correlation between the potential barrier height and metal work function. The barrier heights
determined from characteristic currents and capacitance measurements were in quite good agreement. For
the used lowresistance semiconductor and contact elements, the sizes of majority carrier (electron) deple
tion regions were determined as 26–60 nm.
DOI: 10.1134/S1063782610090162

1. INTRODUCTION sponding functional characteristics of structures

(devices). For example, some processes are applicable
Silicon carbide (SiC) attracts particular attention to fabrication of highvoltage Schottky diodes [2],
due to its unique electrical characteristics, which while other processes are applicable to production of
make it possible to develop devices functioning under photodetectors and gas sensors [5].
difficult conditions (at elevated temperatures, voltage
and electric fields higher than those for Si, and others). To extend the concepts on the effect of various fac
For example, it is required to fabricate metal–semi tors on important parameters of thinfilm structures
conductor contacts under practical conditions of on silicon carbide, systematic experimental studies are
technological processes. In this case, the formation of required directed to the investigation of the depen
defects and intermediate layers at the boundary of dence of electrical properties of the contact region on
materials is inevitable. In turn, these defects and layers the physicochemical nature of a metal, conditions of
can affect both electrical characteristics and the func its deposition and annealing. In [6], using the example
tional shape of macroscopically observed dependences of the Pt/SiC thinfilm structure, it was found that its
(e.g., current–voltage characteristics of grown struc electrical properties depend on the energy parameters
tures). To control current flow processes, it is suffi of the laserinduced atomic flux. The present study is
cient to slightly vary concentrations of electrically devoted to a comparative analysis of electrical proper
active defects or impurities (~1013–1015 cm–3) with ties of thinfilm structures on 6HSiC, formed by
corresponding electron binding energies relative to the pulsed laser deposition of various metals (Au, Ag, Cu,
conduction band, which have an effect on the concen Pd, W, Zr). The structures were grown using high
tration and lifetime of equilibrium and nonequilib speed vapor–plasma flux induced in vacuum by pulsed
rium charge carriers [1]. The problem arises of select laser irradiation of puremetal targets. According to
ing technological processes for attainment of required the data of [6], the upper boundary of the energy spec
physical properties of surface and bulk defects to attain trum of atoms and ions incident on the substrate under
specified electrical characteristics of structures. The these conditions can extend to a few hundred elec
problem is solved by performing experimental studies tronvolts. Therefore, the electrical properties of the
of electrical characteristics for different contact fabri contact region can depend on both the physicochem
cation approaches, i.e., purely thermal methods; ical nature of a deposited metal and the defect forma
implantation with or without thermal drivein diffu tion mechanism in the crystal surface layer. A compar
sion; magnetron, ion, or laser deposition followed by ative analysis of the electrical properties of thinfilm
annealing; or combinations of these methods [2–4]. It structures based on various metals allowed us to deter
is also clear that the choice of these or those techno mine the role of different factors in the formation of
logical processes will be different for achieving corre the metal–silicon carbide contact region.

1192
 ELECTRICAL PROPERTIES OF THINFILM STRUCTURES FORMED 1193

2. EXPERIMENTAL I, A
Experiments were performed using 6HSiC with 10−2 (a) 2
out an epitaxial layer supplied by Semiconductor 6 5
1
Crystals Co. According to the technical certificate, its 10−3 4
3
resistivity is ρ = 0.03–0.2 Ω cm. This value, at the −4
10
average mobility μn = 500 cm2/(V s) [7], corresponds
to the carrier concentration of 6.2 × 1016–4.2 × 10−5
1017 cm–3. At the effective density of states Nc ≈
1019 cm–3 at 300 K, it corresponds to the electron 10−6
Fermi level–conduction band distance of 0.154–
0.1 eV. The wafer thickness was 350 μm; the shear 10−7
plane ((0001) Si face) was polished on one side and
ground on the other side. 10−8
A Ni film was preliminarily grown on the ground
(rear) surface of the SiC substrate by pulsed laser dep I, A 0 0.1 0.2 0.3 0.4 0.5 U, V
osition; then it was annealed in vacuum at a tempera 10−2 2
ture of ~1173 K for 15 min. Due to high surface (b) 1
recombination of carriers on the ground surface and 10−3
physicochemical properties of nickel [4], such treat
ment provided the formation of an electrical contact 10−4
characterized by sufficient ohmic properties with
respect to current flow processes involving metals 10−5 6
deposited on the polished (front) wafer surface. The
10−6 5
area of the rear Ni contact on wafers was 1 cm2.
Pulsed laser deposition of metals (Au, Ag, Cu, W, 10−7 3
Zr, Pd) performed through a circular aperture 2 mm in
diameter, tightly pressed against the polished wafer 10−8 4
surface. Wafers were preliminarily rinsed in ethanol
and distilled water. The deposition conditions for 10−3 10−2 10−1 U, V
metal films were chosen considering the results of [6],
where Pt/SiC diode structures were grown and stud Fig. 1. Experimentally measured I–V characteristics under
ied. Puremetal targets (99.9%) were irradiated with forward bias for various metal/SiC structures, plotted on
(a) semilog and (b) log–log scales: (1) Au, (2) W, (3) Ag,
nanosecond pulses of an Nd:YAG laser. The radiation (4) Pd, (5) Zr, and (6) Cu.
pulse energy was ~50 mJ, the energy density on the
target surface in the focal spot was ~10 J/cm2. The
pulse repetition rate was 25 Hz. The laserinduced with U can be observed as usual. To provide efficient
vapor–plasma flux was deposited under vacuum con measurements of the I–V characteristics and wafer
ditions on a SiC substrate placed normally to the flux capacitance, the structures were placed under a tung
direction at a distance of 5 cm from the target. The sten needle 50 μm in diameter the weight of which
deposited layer thickness was ~100 nm. The vacuum with additional loads provided a sufficient force of
chamber was pumped out by a turbomolecular pump structure pressing onto the second (rear) contact, i.e.,
to a residual gas pressure of ~10–4 Pa. a tungsten or steel polished plate connected with
Current–voltage (I–V) characteristics were mea another power supply pole. The load resistors con
sured mostly under steadystate conditions. To test the nected in series to the structure could be chosen from
effect of Joule heating, measurements with short the range of 102–105 Ω. Measurements were per
pulses (~10 μs) and large duty ratios (~104 s) were used formed at room temperature. The conductivity type
in some cases. The minimum measured currents could be determined from the thermopower sign dur
under steadystate conditions and pulsed currents ing tungsten needle heating; the contact photovoltage
were ~10–9 and ~10–7 A. sign upon exposure to a blue lightemitting diode or an
The capacitance was measured by the Lissajous fig incandescent lamp yielded information on the band
ure parameters, when the amplitude of the reactive bending at the metal–semiconductor interface.
capacitance component of the voltage signal can be
easily measured using a twocoordinate oscilloscope
3. EXPERIMENTAL RESULTS
(S183) mode. Such a method allowed capacitance
AND DISCUSSION
determination in a wide frequency range from 103 to
2 × 105 Hz in the region of linear rise with frequency at Figures 1 and 2 show the I–V characteristics mea
applied voltage amplitudes of 5, 10, and 15 mV. As an sured under steadystate conditions for the grown
external dc voltage U is applied, capacitance variation metal/SiC structures. The dependences of the current

SEMICONDUCTORS Vol. 44 No. 9 2010

1194 ROMANOV et al.

I, A (a) eU
1 I ( U ) = I 0 exp  . (2)
−5 2 ηkT
10
5 In the logI–U scale, these dependences can be plotted
3 as straight lines with the slope 0.434e/(ηkT) and the
10−6 value of logI0 cut on the vertical axis. In this case, the
4 ideality factors η can be determined. The measured
electrical parameters of the grown metal–SiC struc
10−7
tures are listed in the table along with data on the
Pt/n6HSiC thinfilm structure studied by the
10−8 authors and in part published in [6].
In the analysis of the results obtained, it was
assumed that, the closer the factor η to unity, the
0 0.5 1.0 1.5 closer the fabricated metal–semiconductor barrier to
I, A
U, V the ideal Schottky barrier in the degree of sharpness of
10−4 (b)
the physical boundary between metal and semicon
ductor in the degree of the effect of generation–
10−5 recombination processes in the space charge depletion
2
5 region on the current transmission and in the degree of
1 the effect of leakage currents flowing in parallel to the
10−6 3
main flow. Therefore, by comparing the values of η, we
can conclude that the best conditions of the formation
10−7 4
of a barrier close to an ideal one were observed for
tungsten (η = 1.28) and gold (η = 1.48); are slightly
worse for copper and zirconium (η = 1.8); and much
10−8 worse for silver (η = 2.1), platinum (η = 1.9), and pal
ladium (η = 2.14). These results suggest that the
10−2 10−1 100 U, V sharpest metal–semiconductor interface was probably
formed when depositing the laserinduced W and Au
Fig. 2. Experimentally measured I–V characteristics under atom fluxes and/or the lowest density of surface
reverse bias for various metal/SiC structures, plotted on (a) charged levels was formed than in the case of other
semilog and (b) log–log scales: (1) Au, (2) W, (3) Ag, (4) metals. It is worth noting that W and Au differ appre
Pd, and (5) Zr. ciably in chemical activity with respect to substrate
elements and residual gas molecules. Tungsten can
I on the applied voltage U in the forward direction can potentially be involved in the formation of chemical
be described in the following functional form: bonds with substrate carbon and interact with residual
gas, forming carboxide compounds. For gold, such
I ( U ) = I 0 ⎛ exp  – 1⎞ ,
eU processes are of low efficiency.
(1)
⎝ ηkT ⎠ The saturation current density j0 can be expressed
where e is the elementary charge, k is the Boltzmann in terms of the effective Richardson constant A and
constant, and T is the temperature. At eV Ⰷ ηkT, this temperature and height of the potential barrier Φ on
dependence takes the form the metal side as the following functional dependence

Electrical parameters of metal/n6HSiC structures grown by pulsed laser deposition

Work C(0), C(0)/S, Ub(0.5C), U k, I0, Uk = Φ – μn,
Metal function, d, nm η Φ, eV
eV pF pF/cm2 V eV 10–7 A eV

Au 5.1 8 × 103 2.54 × 105 40 1.6–1.9 0.53–0.63 1.48 0.8 0.72 0.62
Ag 4.3 1.2 × 104 3.8 × 105 26 1.7–1.9 0.57–0.63 2.08 0.8 0.72 0.62
Cu 4.4 9× 103 2.9 × 105 34 1.2–1.5 0.4–0.5 1.8 35 0.63 0.53
Pt 5.32 8 × 103 2.54 × 105 40 2.3 0.73–0.77 1.9 0.6 0.73 0.63
Pd 4.8 1.1 × 104 3.5 × 105 28 2.3 0.77 2.14 0.6 0.73 0.63
W 4.54 1.1 × 104 3.5 × 105 28 1.9 0.63 1.28 2 0.7 0.6
Zr 3.9 5.3 × 103 1.7 × 105 60 2.6 0.87 1.8 7 0.66 0.56

SEMICONDUCTORS Vol. 44 No. 9 2010

 ELECTRICAL PROPERTIES OF THINFILM STRUCTURES FORMED 1195

Φ neutral regions was a small fraction of the applied volt

j 0 = AT exp ⎛ – ⎞ .
2
(3) age (~1%).
⎝ kT⎠
For all elements, the I–V reverse portions did not
The effective Richardson constant is given by A = flatten out, i.e., were not described by dependence (1)
ek2m/ប4 (ប is Planck’s divided by 2π); at the effective with I0 = const at voltages U Ⰷ kTη/e. At low voltages
electron mass m = 0.7m0 (m0 is the free electron mass), (U < kTη/e), forward and reverse portions coincided
it is 100 A/(cm K)2. and were linear according to (1) at I0 = const (Figs. 1b
After determining I0 from the data of Fig. 1, we and 2b). From this fact, it followed that current flow
determined the saturation current density j0 = I0/S, through the depletion layer at low voltages is caused by
where S is the metalcontact area and S ≈ π × 10–2 cm2. the mechanism of abovebarrier electron motion,
Taking into account expression (3), Φ was determined which is analytically described by expression (1) with
by the formula I0 = const. At large reverse biases, the I–V character
2 2 istics were linear, sublinear, and then superlinear
Φ = kT ln AT = kT ln AT
S. (4) (Figs. 2a, 2b), i.e., abovebarrier electron transport
j0 I0 from metal to semiconductor was added by an electron
The values of Φ calculated for various metals are flux caused by other mechanisms. The generation
listed in the table along with the experimental values of mechanism in the space charge layer seems hardly
currents I0 and potential barrier heights Uk on the probable due to the high bandgap energy (~3 eV).
semiconductor side given by Uk = Φ – μn. For definite Thermal carrier generation at T = 300 K cannot
ness, the Fermi energies μn are taken as 0.1 eV. A com induce an appreciable current even under conditions
parative analysis of Φ and Uk for various laserdepos of complete separation of heatinduced electron–hole
ited metals showed closeness of their values. This can pairs by an external field. Generation via localized
be caused by a large number surface states, estimated states in the band gap is limited by the process via the
˜ s ≈ 1014 cm–2 eV–1 for the case where the differ levels farthest from allowed bands; in a widegap
as N material, this generation remains low even for levels in
ence between work functions of elements has a insig the midgap. Potential barrier lowering due to mirror
nificant effect on the barrier height, which is mainly image forces, which should lead to the dependences
controlled by electron drift from the semiconductor to logI0 ∝ (Uk – U)0.25 [9], is also not discernible in
these surface levels [8]. The work functions for the experimental I–V characteristics under reverse bias.
used metals are also listed in the table. The surface The absence of this effect can indicate both the degree
states of such a density could, most likely, be formed of surface “roughness” after film deposition and
when highspeed atoms and ions are introduced with masking of this phenomenon by another one with a
the formation of defects capable of trapping electrons, stronger effect on the current.
i.e., acceptortype defects.
Under used conditions, tunneling transfer of elec
We note that the rear contact functioned in the trons from surface states to the semiconductor con
reverse direction in studying the barriers in the forward duction band seems most probable. In this case, under
direction. Judging from good functional dependences conditions of steadystate current, electrons, which
of the current on voltage (2) in the forward direction, have left surface states, were substituted with electrons
an appreciable voltage drop in the Ni/SiC rear contact from metal. The total current was composed of above
region in the current range under study occurred only barrier and tunneling currents. The increase in the
at comparatively high currents, at the I–V characteris tunneling current could be caused by a decrease in the
tic “tail.” Furthermore, such a functional dependence thickness for tunneling from surface levels under
suggests that the contribution of other processes in this reverse bias [9]. Impact ionization from lowenergy
region, except for those controlled by carrier passage levels cannot also be excluded, since fields at U = 1 V
through the barrier, is insignificant. It can be assumed and a layer thickness of ~50 nm can be as high as ~2 ×
that laser deposition of metals occurred under such 105 V/cm.
physical conditions for forward currents that, even in
the absence of guard rings, the effect of parallel chan The capacitances measured at zero bias C(0) and
nels (mechanisms) of current flow was insignificant. capacitances per unit area are listed in the table. Using
We note that this statement is valid for the combina the parallelplate capacitor formula C = εS/4πd, we
tion of the deposition area size and SiC wafer thick estimated the spacecharge region thickness d at
ness used in this study. The material resistivity is also ε/4π ≈ 1 (see table). For all used metals, the thick
important; a small voltage drop should be at the elec nesses were within comparable physical limits from
trically neutral structure part connected in series with 26 to 59.2 nm. This fact can also be explained within
the depletion region. The resistance of this part did not the model of surface states controlling the space
exceed 0.2 Ω. For comparison, e.g., the resistance of a charge region size (thickness). The capacitance
structure with a gold film at zero bias was ~3 × 105 Ω. decreased with the reverse bias voltage; i.e., the thick
The voltage drop at neutral regions was ~2 × 10–4 V at ness increased by the law d ∝ (Uk + U)1/2. At such
a current of 10–3 A (Fig. 1a); i.e., the voltage drop at dependence d(U), the capacitance decreased by half at

SEMICONDUCTORS Vol. 44 No. 9 2010

1196 ROMANOV et al.

U = 3Uk. The table lists the experimental voltages by other methods for fabricating Schottky barriers [3,
Ub(0.5C(0)) at which the capacitance C was half the 4] promoted relatively high currents through the main
value of C(0); from these voltages, the values of Uk = channel, which exceed possible shunting currents. In
Ub(0.5C(0))/3 were determined. We can see that the the same context, it should be noted that perturbed
values of Uk calculated by this method are identical to regions of the crystal can become relatively highresis
those found by analyzing currents under forward bias. tivity ones and play the role similar to that of guard
The thickness d was estimated by the depletionlayer rings [3].
ε ( Uk + U ) It is necessary to consider the problem of the possi
formula d = 
 [9] as 44.2 nm at Uk = ble effect of thin layers of “impurities” formed unin
2πen 0 tentionally at the metal–SiC interface on the charac
0.62 eV, U = 0, ε = 12, and n0 = 4.2 × 1017 cm–3. This teristics of fabricated structures. The mechanism of
value is in agreement with experimental ones (see their formation can be associated with interaction of
table). the incident flux of excited atoms and ions with the
substrate surface containing impurity atoms and mol
The time between collisions of electrons with such ecules (main components are oxygen, carbon, and
scattering centers as phonons, impurities, and elec hydrogen). These molecules can be on the initial crys
trons was estimated by the formula τ = m*μn/e as 2.8 × tal surface and can be adsorbed from the residual
10–13 s at the mobility μn ≈ 800 V/(cm2 s) and the effec atmosphere in the deposition chamber. If a thin
tive mass m* = m0 = 9.1 × 10–28 g. At the thermal (<5 nm) insulator layer (e.g., SiOx) arises between
velocity Vt ≈ 107 cm/s at room temperature, the mean metal and semiconductor n6HSiC, a fraction Ui of
free path is l = τVt ≈ 28 nm, which is comparable to the the external voltage can be at the insulator and another
measured thickness of the depletion layer. This vali fraction Ud can be at the space charge layer in the
dates the application of the diode theory to the analy semiconductor, so that U = Ui + Ud at Ud = U/η and
sis of I–V characteristics when determining I0 [10]. Ui = (1 – 1/η)U [4]. In this case, the surface potential
The question why I–V characteristics under for at the insulator–semiconductor interface changes by
ward bias are close to model (ideal) ones even in the Ui relative to metal; thus, electrons tunneling from
absence of guard rings is noteworthy. This may be due metal to semiconductor through insulator should
to several causes. It is known that the guard ring should overcome the barrier ϕ = ϕ0 + eUi. This will change the
prevent shunting of current flow in the main channel heat flux of electrons from metal to semiconductor.
controlling the I–V characteristic shape [9]. Shunting The electron flux crossing the insulator–semiconduc
can occur over the surface through surface channels tor interface is given by
and due to a nonuniform distribution of a strength
ened electric field in a material around metal elec I MS = DI 0 exp ( – eU i /kT )
(5)
trodes. The experimental data show that barrier layers ≡ DI 0 exp ( – ( eU/kT ) ( 1 – 1/η ) ),
were formed due to surface states for all studied met
als. This layer encloses the metal electrode on all sides, where D is the insulator transmissivity for the tunnel
preventing leakage to the surface. It is possible that the ing transfer, close to unity at small thicknesses [8].
wide gap of silicon carbide played a certain role. This Hence, the flux from semiconductor to metal is writ
circumstance caused a relatively low rate of intrinsic ten as
carrier generation at T ≈ 300 K and, hence, small leak
age currents due to separation of produced electron– I SM = DI 0 exp ( eU d /kT ) ≡ DI 0 exp ( eU/ηkT ), (6)
hole pairs arriving at the depletion region with high where I0 is the equilibrium heat flux expressed by for
electric field. The electric field strength initiating mula (3). Accordingly, the total flux through the inter
breakdown processes was not reached in measure face is given by
ments of I–V characteristics. For SiC, it is ~2.2 ×
106 V/cm (a blue glow in the crystal under the needle I = DI 0 [ exp ( eU/ ( ηkT ) )
was observed only at a voltage of ≈65 V, when a nega (7)
tive voltage pulse was applied to the 50μmdiameter – exp ( – ( eU/kT ) ( 1 – 1/η ) )],
tungsten needle directly pressed onto the clean sur This is the expression for the current, which is valid
face). It should also be taken into account that the under above physical conditions and replaces expres
deposited metal electrode was much larger than the sion (1), which ignores a change in the flux from metal
barrier layer thickness and the plate thickness; there due to a change in the potential barrier height [9]. At
fore, currents from side regions were significantly high voltages (U Ⰷ kT/e) in the forward direction, we
lower than the current from the central contact region can obtain previous expression (2) from (7), from
controlling the shape of the I–V characteristic. A large which we can determine η and DI0 using a corre
fraction of field (hence, current) lines passed along the sponding graphic dependence. At low voltages, we
shortest geometrical way between contacts, rather obtain from (7)
than from perimeter. Furthermore, relatively low
potential barriers in comparison with those obtained I ≈ DI 0 ( eU/kT ). (8)

SEMICONDUCTORS Vol. 44 No. 9 2010

 ELECTRICAL PROPERTIES OF THINFILM STRUCTURES FORMED 1197

If the flux from metal were unchanged, the expression ity factor was obtained for platinum deposited on an
I ≈ DI 0 ( eU/ ( ηkT ) ). (9) 4HSiC epitaxial layer [4] and for Al and Mg deposited
from a heated tungsten boat on a carefully chemically
would be valid. prepared 6HSiC epitaxial layer 2–4 μm thick [11].
Since, in physical meaning, η > 1, an analysis of the The effective thickness of the depletion layer at the
experimental data in the linear I–V portion allows us, interface between laserdeposited metal and silicon
using the I/V ratio (conductance), to draw a conclu carbide was in the range of 26–59.2 nm. The potential
sion concerning the presence of an insulator layer, tak barrier height on the semiconductor side, calculated
ing into account DI0 and η determined at high volt by the results of current measurements under forward
ages. For example, for zirconium on n6HSiC, DI0 = bias was ~0.53–0.62 eV which is in quite good agree
(6–7) × 10–7 A and η = 1.8 for the portion of high volt ment with the results based on capacitance measure
ages; for low voltages, DI0 = 5 × 10–7 A. Thus, for zir ments (0.53–0.87 eV). The height of potential barriers
conium, the experimental data are closer to condition was independent of the work function of a deposited
(8), rather than to (9), and show the possibility of the element, which could be caused by the formation of
formation of an intermediate insulating layer. The donor and acceptortype surface states with high
problem of mechanisms and efficiency of the forma densities (~1014 cm–2 eV–1) at the metal–SiC inter
tion of thin insulating layers during metal deposition face. The effect associated with the formation of a thin
on silicon carbide requires additional study. It may be intermediate layer of insulator (SiOx) at the metal–
noted that zirconium oxide belongs to “highk” SiC interface cannot be excluded.
dielectrics characterized by high permittivity. How
ever, η for Zr is appreciably lower than that for other It was found that, the closer the ideality factor to
elements not susceptible to oxidation (Pd, Pt, Ag). unity, the closer the conditions at the metal–SiC
If the coefficient η is constant in the definition interface to ideal ones corresponding to the diode
domain of the I–V characteristic, the rate of charge model of abovebarrier motion of electrons at a abrupt
change at the surface states at the insulator–semicon phase interface without changing the barrier at the
ductor interface is constant with voltage change. This interface due to a change in the bound charge at sur
in turn necessarily requires unoccupied surface states face states. It is obvious that the degree of the interface
for additional to equilibrium electron localization at sharpness and the characteristics of surface states (the
them under electric current. Under reverse bias (U < electron density and binding energies) are controlled
0), both processes described by expression (7) are by the kinetic energy and electronic and charge states
involved in the current formation. For high voltages of incident atoms. Therefore, the ideality factor is an
(eU(1 – 1/η) Ⰷ kT), a linear dependence should be attribute of differences of these atomic parameters,
observed in the “logI–U” coordinates, from which we which manifest themselves during interaction with the
can determine ηr different in the general case from ηf in n6HSiC semiconductor lattice. Later on, varying
the forward direction; however, DI0 cut on the U = 0 physicochemical characteristics of deposited atoms,
axis should be the same as for the forward direction. in particular, the degree of electron excitation of
The linearity is observed in the experimental atoms, the effect of this excitation on features of
dependences, but the values of DI0 are not identical formed metal–semiconductor transition regions can
(Fig. 2b). This indicates more complex processes of be traced by changes in the ideality factor from the
current flow under reverse bias than is described by viewpoint of forming active electronic states of
(7). Also it is only at low voltages that the forward and defects. Thus, the problem of controlling the proper
reverse portions of the I–V characteristic coincide and ties of obtained interfacial defects which control the
obey relation (8) with identical values of DI0 (Figs. 1b, values and functional form of structure characteristics
2b). The functional dependences I(U) under reverse can be solved.
bias are similar for different deposited metals, which is
indicative of identical mechanisms of the current flow
whose complex nature should be clarified. We note REFERENCES
that the currents are significantly lower under reverse
bias than under forward bias, and the contribution of 1. A. D. Kiryukhin, V. V. Grigor’ev, A. V. Zuev, and
shunting channels can appear with higher probability. V. V. Zuev, Fiz. Tekh. Poluprovodn. 42, 271 (2008)
[Semiconductors 42, 266 (2008)].
2. I. V. Grekhov, P. A. Ivanov, I. D. Il’inskaya, O. I. Kon’kov,
4. CONCLUSIONS A. S. Potapov, and T. P. Samsonov, Fiz. Tekh. Polupro
Structures with diode characteristics were formed vodn. 42, 211 (2008) [Semiconductors 42, 211 (2008)].
during deposition of laserinduced vapor–plasma
atomic fluxes of various metals (Au, Ag, Cu, Pt, Pd, 3. R. Weiss, L. Frey, and H. Ryssel, Appl. Surf. Sci. 184,
413 (2001).
W, Zr) on 6HSiC. The ideality factors of these struc
tures under forward bias depended on the metal type 4. M. Soshacki, A. Kolendo, J. Szmidt, and A. Werlowy,
and varied in the range of 1.28–2.14. The same ideal Solid State Electron. 49, 585 (2005).

SEMICONDUCTORS Vol. 44 No. 9 2010

1198 ROMANOV et al.

5. C. I. Muntele, D. Lla, E. K. Williams, D. B. Poker, 9. A. Milnes and D. Feucht, Heterojunctions and Metal
D. K. Hensley, D. J. Larkin, and L. Muntele, Mater. Semiconductor Junctions (Academic Press, New York,
Sci. Forum, 338–342, 1443 (2000). 1972; Mir, Moscow, 1975).
6. V. Yu. Fominskiі, R. I. Romanov, A. G. Gnedovets, 10. V. L. BonchBruevich and S. G. Kalashnikov, Physics of
V. V. Zuev, and M. V. Demin, Fiz. Tekh. Poluprovodn. Semiconductors (Nauka, Moscow, 1977) [in Russian].
44, 556 (2010) [Semiconductors 44, 537 (2010)].
7. T. T. Mnatsakanov, L. I. Pomortseva, and S. N. Yurkov, 11. J. R. Waldrop, R. W. Grant, Y. C. Wong, and R. F. Davis,
Fiz. Tekh. Poluprovodn. 35, 406 (2001) [Semiconduc J. Appl. Phys. 72, 4757 (1992).
tors 35, 394 (2001)].
8. G. E. Pikus, Principles of the Theory of Semiconductor
Devices (Nauka, Moscow, 1965), p. 104 [in Russian]. Translated by A. Kazantsev

SEMICONDUCTORS Vol. 44 No. 9 2010