Sunteți pe pagina 1din 10

Available online at www.sciencedirect.

com

Thin Solid Films 516 (2008) 1755 – 1764


www.elsevier.com/locate/tsf

Transparent electronics: Schottky barrier and heterojunction considerations


J.F. Wager
School of Electrical Engineering and Computer Science, Oregon State University, Corvallis, OR 97331-5501, United States
Received 23 June 2006; received in revised form 14 June 2007; accepted 21 June 2007
Available online 4 July 2007

Abstract

Transparent electronics employs wide band gap semi-conductors which are transparent in the visible portion of the electromagnetic spectrum
for the fabrication of electronic devices and circuits. Current and future transparent electronics applications require the use of wide band gap oxide
semi-conductor interfaces as contacts and rectifiers, as well as for passivation and barrier-shaping layers. Modern Schottky barrier and
heterojunction theory can be applied to the assessment of such interfaces, and is reviewed for this purpose from a charge transfer, energy band
diagram perspective. Ideal interface formation theory is envisaged as originating from Fermi level mediated charge transfer giving rise to a
macroscopic interfacial dipole, while non-ideal theory involves charge neutrality level mediated charge transfer giving rise to a microscopic
interfacial dipole. This interface formation theory is applied to the problem of indium tin oxide (ITO) – zinc oxide and ITO – tin oxide interfaces,
confirming their utility as injecting source-drain contacts in transparent thin-film transistors.
© 2007 Elsevier B.V. All rights reserved.

PACS: 73.40.Lq; 73.40.Ns; 73.61.Le; 81.05.Hd


Keywords: Transparent electronics; Transparent thin-film transistor; Schottky barrier; Heterojunction; Oxide electronics; Interface

1. Introduction The next two sections of this paper constitute a summary of


modern Schottky barrier and heterojunction theory [13–15]
Transparent electronics is an emerging technology which from a charge transfer, energy band diagram perspective. Ideal
employs wide band gap semi-conductors that are transparent in and non-ideal theory, for both Schottky barriers and hetero-
the visible portion of the electromagnetic spectrum for the junctions, are envisaged as originating from, respectively, Fermi
fabrication of electronic devices and circuits [1–12]. Current level mediated charge transfer giving rise to a macroscopic
state-of-the-art transparent thin-film transistors (TTFTs) utilize interfacial dipole, and from charge neutrality level mediated
high-low carrier concentration oxide semi-conductor interfaces charge transfer giving rise to a microscopic interfacial dipole.
for injecting source and drain contacts [2–12]. It is not clear This standpoint facilitates recognition of similarities between
whether these contacts are best modeled via Schottky barrier Schottky barrier and heterojunction formation, which is useful
or heterojunction theory. However, it is evident that a more when scrutinizing transparent electronics interface physics
fundamental understanding of wide band gap oxide semi- issues. n-type Schottky barrier formation and n–n isotype
conductor interfacial physics is highly desirable for these and heterojunctions, in which both semi-conductors are n-type, are
for future transparent electronics applications. Thus, the goals of emphasized in the following discussion, as these topics are
the work described herein are to first provide an overview of perceived to be of primary current interest to transparent
modern Schottky barrier and heterojunction theory, and to then electronics.
apply this theory to the problem of indium tin oxide (ITO) –
zinc oxide (ZnO) and ITO – tin oxide (SnO2) interfaces in order 2. Schottky barrier theory
to assess their applicability for use as TTFT injecting source and
drain contacts. First, consider the formation of a metal–semi-conductor
junction from the perspective of ideal Schottky barrier theory, as
illustrated by the energy band diagrams shown in Fig. 1 for the
E-mail address: jfw@ece.orst.edu. specific case of an n-type semi-conductor. Fig. 1a depicts
0040-6090/$ - see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.tsf.2007.06.164
1756 J.F. Wager / Thin Solid Films 516 (2008) 1755–1764

negative dipole, as indicated by + and − signs and the lower left-


going arrow, and a concomitant negative electric field, ξ, also
shown in Fig. 1a.
Fig. 1b shows an energy band diagram for the Schottky
barrier formed after the metal and semi-conductor come into
intimate contact for this ideal model case. The effect of charge
transfer is evident from the positive curvature of the semi-
conductor conduction band near the interface; positive
curvature indicates that the semi-conductor space charge is
positive. Additional parameters in Fig. 1b that arise as a
consequence of charge transfer during metal–semi-conductor
interface formation are the n-type semi-conductor Schottky
barrier height, ϕBn, the local vacuum level, ELVAC, and the
built-in potential, VBI. ΦBn is the potential barrier that a metal
electron at the Fermi level needs to overcome in order to be able
to transit into the semi-conductor. Similarly, VBI is the potential
barrier which an electron at the bottom of the conduction band
in the bulk of the semi-conductor needs to surmount in order to
transit into the metal. Finally, an important feature of this ideal
model interface is the fact that the local vacuum level is
continuous across the interface.
Fig. 1. Energy band diagrams for (a) an isolated metal and an isolated n-type Next, consider the formation of a metal–semi-conductor
semi-conductor (valence band not shown) with Φ M N Φ S, and (b) the junction from the perspective of non-ideal Schottky barrier
corresponding Schottky barrier band structure for an ideal interface (i.e., charge theory. The two non-ideal energy band diagram situations
exchange is exclusively Fermi level mediated, as indicated by the arrow shown
in (a)). Ideal charge exchange results in the formation of a macroscopic negative
corresponding to the ideal Schottky barrier theory of Fig. 1 are
dipole, as indicated by the arrow labeled ξ in (a). indicated in Figs. 2 and 3.
Initially, consider Fig. 2. The non-ideal case shown in Fig. 2a
is identical to the ideal case indicated in Fig. 1a except for the
energy band diagrams for an isolated metal and an isolated n-
type semi-conductor (valence band not shown) with a relative
metal to semi-conductor work function ΦM N ΦS. Note that ΦM,
ΦS, and the semi-conductor electron affinity, χS, are all defined
as potentials which are referenced to the vacuum level, EVAC.
When the metal and semi-conductor of Fig. 1a are brought
into intimate contact, charge transfer occurs across the interface,
setting up a space charge layer in the semi-conductor. This
interfacial space charge situation can be construed as compris-
ing an extended or macroscopic dipole in which negative,
electronic charge is transferred from the semi-conductor to the
metal–semi-conductor interface. This negative interfacial
charge sheet is balanced by an equal density of positive charge
in the extended space charge region in the semi-conductor. This
charge transfer-induced space charge distribution constitutes a
dipole in the sense that if the positive space charge is assumed to
be localized at its average, charge-centroid position, the sheet of
negative interfacial charge and the positive charge-centroid
sheet define an extended or macroscopic dipole with a spatial
dimension on the order of the space charge region dimension,
i.e., ∼ 0.01–10 μm.
The direction of charge transfer, in the context of this ideal
Schottky barrier model, is determined by the relative Fermi
level positions of the metal and semi-conductor. For the ΦM N ΦS Fig. 2. Energy band diagrams for (a) an isolated metal and an isolated n-type
case shown in Fig. 1a, the metal Fermi level energy is lower semi-conductor (valence band not shown) with ΦM N ΦS and with ΦM b ΦCNL,
and (b) the corresponding Schottky barrier band structure for a non-ideal
than that of the semi-conductor. Thus, as these isolated materials
interface (i.e., charge exchange is Fermi level mediated (upper arrow in (a)) and
come into intimate contact, electrons flow from the semi- also occurs due to misalignment of the metal Fermi level and the semi-conductor
conductor to the metal, as indicated by the upper left-pointing charge neutrality level (bottom arrow in (a)). Non-ideal charge exchange results
arrow in Fig. 1a. This gives rise an extended, macroscopic in the formation of a microscopic positive dipole, as indicated in (b).
J.F. Wager / Thin Solid Films 516 (2008) 1755–1764 1757

instead of a positive direction as indicated in Fig. 2. The positive


or negative nature of the microscopic dipole depends upon the
relative alignment of the metal Fermi level and the semi-
conductor charge neutrality level when these materials are
isolated from one another.
The ideal model complement to Fig. 1, in which ΦM b ΦS, is
given in Fig. 4. Fermi level mediated charge transfer occurs in
opposite directions for Figs. 1 and 4. Fig. 4 is characterized by a
positive macroscopic dipole and a negative Schottky barrier
height, ϕBn. A negative Schottky barrier height establishes this
contact as Ohmic, meaning that current flow occurs readily,
independent of which voltage polarity is applied to the metal,
since there is no appreciable barrier to the flow of charge across
the metal–semi-conductor interface.
Fig. 5 illustrates one of the non-ideal Schottky barrier
situations corresponding to the ideal theory case of Fig. 4. As
shown in Fig. 5a, the isolated semi-conductor charge neutrality
level is at a lower energy than the metal Fermi level, so that non-
ideal charge transfer gives rise to a positive microscopic dipole.
As evident from Fig. 5b, this positive dipole decreases the
Fig. 3. Energy band diagrams for (a) an isolated metal and an isolated n-type
semi-conductor (valence band not shown) with ΦM N ΦS and with ΦM N ΦCNL, magnitude of the negative Schottky barrier height, resulting in
and (b) the corresponding Schottky barrier band structure for a non-ideal less interfacial semi-conductor band bending and a smaller
interface (i.e., charge exchange is Fermi level mediated (upper arrow in (a)) and built-in potential (dashed lines) compared to the ideal case
also occurs due to misalignment of the metal Fermi level and the semi-conductor (solid lines). Comparing Figs. 5 and 2, note that a positive
charge neutrality level (bottom arrow in (a)). Non-ideal charge exchange results
microscopic dipole decreases the magnitude of the Schottky
in the formation of a microscopic negative dipole, as indicated in (b).
barrier when the barrier is negative (in fact, it can even invert the
polarity of ϕBn from negative to positive, as illustrated in
inclusion of a charge neutrality level, ECNL, at the semi- Fig. 5b), but increases the magnitude of a positive Schottky
conductor surface. Physically, the charge neutrality level
corresponds to the branch point within the semi-conductor
band gap, above (below) which an interface state is predom-
inantly conduction (valence) band derived, or acceptor-like
(donor-like) [13,14]. From an interface formation perspective,
ECNL is an energy established by the nature of the bulk semi-
conductor band structure and whose relative alignment estab-
lishes the direction of non-ideal charge transfer during interface
formation. Thus, in addition to the ideal Fermi level mediated
charge transfer discussed with respect to Fig. 1a, a second, non-
ideal charge transfer occurs due to misalignment of the metal
Fermi level and the semi-conductor charge neutrality level, as
indicated by the bottom, right-going arrow shown in Fig. 2a.
This non-ideal charge exchange results in the formation of a
positive microscopic dipole, of atomic dimensions, i.e., ∼ 0.2–
0.3 nm, as indicated in Fig. 2b, and a corresponding increase in
ϕBn and VBI (dashed lines), compared to their corresponding
ideal model values (solid lines) [16]. Note that the interfacial
dipole makes the local vacuum level discontinuous at the
interface for this non-ideal model.
The non-ideal Schottky barrier case shown in Fig. 3 is
similar to that illustrated in Fig. 2 except that non-ideal charge
exchange results in the formation of a microscopic negative
dipole, as indicated in Fig. 3b, and a corresponding decrease in
ϕBn and VBI (dashed lines), compared to their corresponding Fig. 4. Energy band diagrams for (a) an isolated metal and an isolated n-type
semi-conductor (valence band not shown) with Φ M b Φ S, and (b) the
ideal model values (solid lines). Again note that this interfacial
corresponding Schottky barrier band structure for an ideal interface (i.e., charge
microscopic dipole makes the local vacuum level discontinuous exchange is exclusively Fermi level mediated, as indicated by the arrow shown
at the interface for this non-ideal model, but that the dipole is in (a)). Ideal charge exchange results in the formation of a macroscopic positive
directed in a negative direction for the case shown in Fig. 3, dipole, as indicated by the arrow labeled ξ in (a).
1758 J.F. Wager / Thin Solid Films 516 (2008) 1755–1764

built-in potential of an n-type semi-conductor for the non-ideal


Schottky barrier model is given by
nSB
VBI ¼ UM  US þ DSB : ð4Þ

Several comments are warranted with regard to the non-ideal


Schottky barrier model quantitative formulation given by Eqs.
(1)–(4). First, the ideal Schottky barrier model obtains if the
microscopic dipole correction term, ΔSB, is equal to zero. Thus,
this non-ideal model formulation includes the ideal model as a
limit when one non-ideal model parameter, ΔSB, goes to zero.
Second, assessment of Eq. (2) reveals that non-ideal charge
transfer depends upon misalignment of the semi-conductor charge
neutrality level and the metal Fermi level, i.e., (ΦCNL − ΦM), and
also upon (1 − S), which physically is related to the dynamic
electronic screening properties of the semi-conductor, as reflected
by the magnitude of its high frequency relative dielectric constant.
In the no screening limit, the ideal model again obtains, i.e.,
ϵ∞ →1, so that S → 1, ΔSB →0, ϕBn →ΦM −χS, and VBI nSB
→ΦM −
ΦS. In the perfect screening limit, the Fermi level pinning
condition results, i.e., ϵ∞ → ∞, so that S → 0, ΔSB → ΦCNL − ΦM,
Fig. 5. Energy band diagrams for (a) an isolated metal and an isolated n-type ϕBn → ΦCNL − χS, and VBI nSB
→ ΦCNL − ΦS. (1 − S) represents the
semi-conductor (valence band not shown) with ΦM b ΦS and with ΦM b ΦCNL, fraction of the metal Fermi level — semi-conductor charge
and (b) the corresponding Schottky barrier band structure for a non-ideal
neutrality level misalignment that contributes to the microscopic
interface (i.e., charge exchange is Fermi level mediated (upper arrow in (a)) and
also occurs due to misalignment of the metal Fermi level and the semi-conductor interfacial dipole after semi-conductor screening is taken into
charge neutrality level, as indicated by the bottom arrow shown in (a)). Non- account [15]. Third, the sign and magnitude of ΔSB directly
ideal charge exchange results in the formation of a microscopic positive dipole, establishes the sign and magnitude of the local vacuum level
as indicated in (b). discontinuity, and also modifies ϕBn and VBI compared to their
ideal model values, as illustrated by the energy band diagrams
nSB
shown in Figs. 2, 3, and 5. Fourth, Eq. (4) indicates that VBI can
be a positive or a negative quantity, depending on the relative
barrier (Fig. 2). The other non-ideal Schottky barrier situation
magnitudes of ΦM and ΦS, and on the sign and magnitude of ΔSB.
corresponding to the ideal theory case of Fig. 4 in which ΦM N nSB
Note that if VBI = + for the ideal model case, i.e., if ΦM N ΦS and
ΦCNL and ΦM b ΦS for an n-type semi-conductor is not sketched,
ΔSB = 0, then ELVAC occurs on the semi-conductor side of the
since it is unlikely that any practical metal–semi-conductor nSB
junction (e.g., Figs. 1–3), whereas if VBI = − for the ideal model
combination will satisfy these conditions.
case, i.e., ΦM b ΦS and ΔSB = 0, then ELVAC occurs on the metal
A convenient quantitative formulation for the barrier height
side of the junction (e.g., Figs. 4 and 5).
of an n-type semi-conductor for the non-ideal Schottky barrier
A corresponding quantitative formulation for the barrier
model is given by
height of a p-type semi-conductor for the non-ideal Schottky
Bn ¼ UM  vS þ DSB ; ð1Þ barrier model may be obtained by recognizing that ϕBn +
ϕBp = EG / q [17], where EG is the semi-conductor band gap, thus
where ΔSB is the Schottky barrier microscopic dipole correction yielding
due to non-ideal charge transfer mediated by misalignment of
the metal Fermi level and the semi-conductor charge neutrality Bp ¼ IPS  UM  DSB ; ð5Þ
level [15], and which is equal to
where IPS is the semi-conductor ionization potential (equal to a
DSB ¼ ð1  SÞðUCNL  UM Þ; ð2Þ sum of the semi-conductor electron affinity and band gap). A
corresponding quantitative formulation for the built-in potential
where S is the interface parameter, empirically prescribed by
of a p-type semi-conductor for the non-ideal Schottky barrier
Mönch [13] as
model is given by
1
S¼ ; ð3Þ pSB
VBI ¼ US  UM  DSB : ð6Þ
1 þ 0:1ðϵl  1Þ2
where ϵ∞ is the semi-conductor high frequency relative dielectric Note the similarity of Eqs. (5) and (6) compared to Eqs. (1)
constant. The high frequency, rather than the low frequency and (4).
dielectric constant is used since the microscopic dipole is of In summary, Schottky barrier trends are qualitatively and
atomic dimensions such that dynamic rather than static screening quantitatively treated in terms of ideal (Fermi level mediated)
is operative. A corresponding quantitative formulation for the and non-ideal (charge neutrality level mediated) charge transfer,
J.F. Wager / Thin Solid Films 516 (2008) 1755–1764 1759

giving rise to macroscopic and microscopic dipoles for the ideal


and non-ideal model cases, respectively.

3. Heterojunction theory

A discussion of heterojunction theory from a charge transfer,


energy band diagram perspective is now provided which
benefits from the previous Schottky barrier theory explication.
Fig. 6 is the heterojunction analog of Fig. 1, illustrating
energy band diagrams for abrupt heterojunction formation
according to ideal theory. Fig. 6a shows energy band diagrams
for isolated n-type semi-conductors (valence bands not shown),
labeled 1 and 2, with work functions Φ2 b Φ1. Fig. 6b displays
an energy band diagram after the two semi-conductors are in
intimate contact. As in the Schottky barrier situation shown in
Fig. 1, a macroscopic negative dipole is formed as a
consequence of Fermi level mediated electronic transfer, from
semi-conductor 2 to semi-conductor 1 for the heterojunction
case indicated in Fig. 6. However, ideal heterojunction
formation, as shown in Fig. 6b, differs from ideal Schottky
Fig. 7. Energy band diagrams for (a) two isolated n-type semi-conductors,
barrier formation, as indicated in Fig. 6b, in two primary ways. (valence bands not shown) with Φ2 b Φ1, χ2 b χ1, and ΦCNL2 N ΦCNL1, and (b) the
First, negative charge transferred to semi-conductor 1 is corresponding heterojunction band structure for a non-ideal interface (i.e.,
accommodated in an extended interfacial accumulation region, charge exchange is Fermi level mediated (upper arrow in (a)) and also occurs
instead of an interfacial sheet, as is the case for Schottky barrier due to misalignment of the semi-conductor charge neutrality levels, as indicated
formation. The negative curvature of the semi-conductor 1 by the bottom arrow shown in (a)). Non-ideal charge exchange results in the
formation of a microscopic positive dipole, as indicated in (b).
conduction band near the interface confirms this charge to be
negative. Second, a positive conduction band discontinuity,
ΔEC, is the heterojunction equivalent of a positive n-type semi- Fig. 7 is the heterojunction complement to the Schottky
conductor Schottky barrier height, ϕBn. Finally, note that the barrier situation shown in Fig. 2, illustrating energy band
local vacuum level is continuous across the interface for this diagrams for heterojunction formation according to non-ideal
ideal model situation. theory for the ΦCNL2 N ΦCNL1 case. Fig. 7a shows that non-ideal
electronic charge transfer occurs from semi-conductor 1 to
semi-conductor 2, leading to the formation of a microscopic
positive dipole, as given in Fig. 7b. One slight difference
between the mechanism of heterojunction and Schottky barrier
non-ideal charge transfer involves its dependence upon
misalignment of semi-conductor charge neutrality levels for
the heterojunction case, but on the misalignment of the metal
Fermi level and the semi-conductor charge neutrality level for
the Schottky barrier case. The most important aspect of Fig. 7b
is that the presence of this positive interfacial dipole leads to an
increase in the conduction band discontinuity, ΔEC. This results
in a widening of the interfacial space charge regions and an
increase in VBI, a portion of which, in general, is partitioned
1
across each semi-conductor, leading to the notation VBI and
2
VBI, as indicated in Fig. 7.
Fig. 8 is the heterojunction match to the Schottky barrier case
shown in Fig. 3, illustrating energy band diagrams for
heterojunction formation according to non-ideal theory for the
case ΦCNL2 b ΦCNL1. Fig. 8a shows that non-ideal electronic
charge transfer occurs from semi-conductor 2 to semi-conductor
1, leading to the formation of a microscopic negative dipole, as
Fig. 6. Energy band diagrams for (a) two isolated n-type semi-conductors, indicated in Fig. 8b. In contrast to the positive dipole case
(valence bands not shown) with Φ2 b Φ1 and χ2 b χ1, and (b) the corresponding
shown in Fig. 7b, the negative dipole case indicated in Fig. 8b
heterojunction band structure for an ideal interface (i.e., charge exchange is
Fermi level mediated, as indicated by the arrow shown in (a)). Ideal charge leads to a decrease in the conduction band discontinuity (in fact
exchange results in the formation of a macroscopic negative dipole, as indicated ΔEC ≈ 0 for the case illustrated in Fig. 8b) and a narrowing of
by the arrow labeled ξ in (a). the interfacial space charge regions. Also, in contrast to the
1760 J.F. Wager / Thin Solid Films 516 (2008) 1755–1764

positive dipole case of Fig. 7, the negative dipole case shown in


Fig. 8 exhibits a polarity inversion of VBI and inversion of the
curvature of the interfacial semi-conductor band bending. This
VBI polarity and semi-conductor curvature inversion is a
consequence of the fact that the magnitude of the negative
dipole is greater than that of VBI for the ideal charge transfer
case shown in Fig. 6.
Fig. 9 is the heterojunction analog to the Schottky barrier
case shown in Fig. 4, illustrating energy band diagrams for
heterojunction formation according to ideal theory when
Φ2 N Φ1. Non-ideal heterojunctions corresponding to Fig. 9 are
also shown in Figs. 10 and 11 [18]. Fig. 10 features a large
positive microscopic dipole which inverts the polarity of ΔEC
and also inverts the sense of the band bending in the semi-
conductor interfacial regions, whereas Fig. 11 is characterized
by a negative microscopic dipole which increases the magnitude
of ΔEC and the extent of the semi-conductor interfacial band
bending. The conduction band discontinuity trends presented in
Figs. 10 and 11 are most readily understood with the aid of the
quantitative formulation discussed below. Qualitatively, the
ideal macroscopic dipole contribution to ΔEC is equal to (χ1 Fig. 9. Energy band diagrams for (a) two isolated n-type semi-conductors,
− χ2) which is negative; a positive microscopic dipole reduces (valence bands not shown) with Φ2 N Φ1 and χ2 b χ1, and (b) the corresponding
heterojunction band structure for an ideal interface (i.e., charge exchange is
and then inverts the polarity of ΔEC if the dipole magnitude is
Fermi level mediated, as indicated by the arrow shown in (a)). Ideal charge
sufficiently large (Fig. 10), whereas a negative microscopic exchange results in the formation of a macroscopic positive dipole, as indicated
dipole correction increases the magnitude of ΔEC (Fig. 11). As by the arrow labeled ξ in (a).
observed multiple times in the previous Schottky barrier and
heterojunction formation discussion, the presence of the
microscopic dipole in Figs. 10 and 11 renders the local vacuum A quantitative formulation for the conduction band discon-
level discontinuous at the interface. tinuity according to the non-ideal heterojunction model is given
by
DEC
¼ v1  v2 þ DHJ ; ð7Þ
q
where ΔHJ is the heterojunction microscopic dipole correction
due to non-ideal charge transfer mediated by misalignment of
semi-conductor charge neutrality levels [15], and which is given
by
DHJ ¼ ð1  S12 ÞðUCNL2  UCNL1 Þ; ð8Þ
where S12 is the overall interface parameter, which is assumed
to be equal to
1
S12 ¼ h i; ð9Þ
ðϵl1 1Þ2 ðϵl2 1Þ2
1 þ 0:1 ðϵl1 1Þ2 þðϵl2 1Þ2

where ϵ∞1 and ϵ∞2 refer to the high frequency dielectric


constants of individual semi-conductors. A corresponding
quantitative formulation for the built-in potential relevant to
conduction band electrons for the non-ideal heterojunction
model is given by
Fig. 8. Energy band diagrams for (a) two isolated n-type semi-conductors, nHJ
VBI ¼ U1  U2 þ DHJ : ð10Þ
(valence bands not shown) with Φ2 b Φ1, χ2 b χ1, and ΦCNL2 N ΦCNL1, and (b) the
corresponding heterojunction band structure for a non-ideal interface (i.e.,
Several comments are warranted with regard to the non-ideal
charge exchange is Fermi level mediated (upper arrow in (a)) and also occurs
due to misalignment of the semi-conductor charge neutrality levels, as indicated heterojunction model quantitative formulation given by Eqs.
by the bottom arrow shown in (a)). Non-ideal charge exchange results in the (7)–(10). First, the ideal heterojunction model obtains if the
formation of a microscopic negative dipole, as indicated in (b). microscopic dipole correction term, ΔHJ, is equal to zero.
J.F. Wager / Thin Solid Films 516 (2008) 1755–1764 1761

where IP1 and IP2 are semi-conductor ionization potentials


(equal to a sum of the semi-conductor electron affinity and band
gap). A corresponding quantitative formulation for the built-in
potential relevant to valence band holes for the non-ideal
heterojunction model is given by
pHJ
VBI ¼ U2  U1  DHJ : ð12Þ

Table 1 presents a summary of non-ideal Schottky barrier


and heterojunction interface formation equations. Note that
these equations are explicitly partitioned into their macroscopic
and microscopic dipole contributions, involving, respectively,
Fermi level mediated ideal charge transfer and charge neutrality
level mediated non-ideal charge transfer.

4. Transparent electronics interfaces

To date, the primary Schottky barrier/heterojunction inter-


face that has been employed in the context of transparent
electronics is the source-drain injecting contact of a transparent
Fig. 10. Energy band diagrams for (a) two isolated n-type semi-conductors, thin-film transistor (TTFT). Note that a TTFT source-drain
(valence bands not shown) with Φ2 N Φ1, χ2 b χ1, and ΦCNL2 NΦCNL1, and (b) the contact is referred to as an ‘injecting contact’ [23], rather than an
corresponding heterojunction band structure for a non-ideal interface (i.e., ‘Ohmic contact’ in order to underscore the fact that the channel
charge exchange is Fermi level mediated (upper arrow in (a)) and also occurs
layer of a well-functioning TTFT has a very low carrier
due to misalignment of the semi-conductor charge neutrality levels, as indicated
by the bottom arrow shown in (a)). Non-ideal charge exchange results in the concentration.
formation of a microscopic positive dipole, as indicated in (b). In almost all of the TTFTs reported to date, indium tin oxide
(ITO) has served as the ‘metal’ or heavily doped n-type semi-
Second, assessment of Eq. (8) reveals that non-ideal charge conductor. Specifically, ITO is employed as a TTFT source-
transfer depends upon misalignment of the semi-conductor drain contact in conjunction with the following channel layers:
charge neutrality levels, i.e., (ΦCNL2 − ΦCNL1), and also upon (1
− S12), which physically is related to the electronic screening
properties of both semi-conductors, as reflected by the
magnitude of their high frequency relative dielectric constants.
Third, the microscopic dipole contribution to the built-in
potential due to non-ideal interfacial charge transfer is, in
general, accommodated in both semi-conductors as a change
in the interfacial band bending, as evident from inspection of
(Figs. 7, 8, 10, and 11).
The formulation of Eq. (9) is motivated [19,20], in part, by
recognizing that a heterojunction becomes a Schottky barrier in
the limit of high carrier concentration of semi-conductor 1, such
that this degenerate semi-conductor can be considered to be a
metal; i.e., in the strongly degenerate semi-conductor doping
limit, ϵ∞1 → ∞, so that S12 → S2; additionally, ΦCNL1 → ΦM, and
ΔEC → ϕBn; making notational changes in order to be con-
sistent with Schottky barrier nomenclature: S2 → S, ΦCNL2 →
ΦCNL, and χ2 → χS; this results in Eq. (7) → Eq. (1) and Eq.
(8) → Eq. (2). Thus, Schottky barrier and heterojunction
formation are very similar phenomena since a Schottky barrier
is a limiting case of a heterojunction.
A corresponding quantitative formulation of the valence
band discontinuity for the non-ideal heterojunction model may Fig. 11. Energy band diagrams for (a) two isolated n-type semi-conductors,
be obtained by recognizing from energy band diagram analysis (valence bands not shown) with Φ2 N Φ1, χ2 b χ1, and ΦCNL2 b ΦCNL1, and (b) the
corresponding heterojunction band structure for a non-ideal interface (i.e.,
that ΔEG = ΔEV + ΔEC [21,22], thereby resulting in
charge exchange is Fermi level mediated (upper arrow in (a)) and also occurs
due to misalignment of the semi-conductor charge neutrality levels, as indicated
DEV
¼ IP2  IP1  DHJ ; ð11Þ by the bottom arrow shown in (a)). Non-ideal charge exchange results in the
q formation of a microscopic negative dipole, as indicated in (b).
1762 J.F. Wager / Thin Solid Films 516 (2008) 1755–1764

Table 1 function is reported in Ref. [25] to vary between 4.1 V for an ‘as
A summary of non-ideal Schottky barrier and heterojunction interface formation received’ film and 4.7 V for an oxygen plasma-treated film.
model equations
Oxygen plasma treatment of an ITO surface is often employed to
Equation Macroscopic dipole Microscopic dipole increase its work function for organic light-emitting device hole
Schottky barrier injection applications [25]. Therefore, we choose a value of
ϕBn = ΦM − χS + ΔSB ΦM − χS ΔSB = (1 − S)(ΦCNL − ΦM) 4.1 V for our calculations since this most likely corresponds to
BI = ΦM − ΦS + ΔSB
VnSB ΦM−ΦS ”
the TTFT situation. This leads to an interfacial energy band
ϕBp = IPS − ΦM − ΔSB IPS − ΦM ”
BI = ΦS − ΦM − ΔSB
VpSB ΦS−ΦM ” picture similar to that shown in Fig. 5, with an estimated negative
Schottky barrier height of − 0.11 V (note that the Schottky barrier
Heterojunction height illustrated in Fig. 5 is slightly positive), and a positive
DEC χ1 − χ2 ΔHJ = (1 − S12)(ΦCNL2 − ΦCNL1) dipole correction ΔSBn = + 0.36 V. Thus, this calculation
¼ v1  v2 þ DHJ
q confirms that the ITO–ZnO interface should behave as an
BI = Φ1 − Φ2 + ΔHJ
VnHJ Φ1−Φ2 ” injecting contact, as found experimentally [2,4,8].
DEV IP2 − IP1 ” Next, consider the ITO–SnO2 interface of a TTFT source-
¼ IP2  IP1  DHJ
q drain from a Schottky barrier theory perspective. The first
BI = Φ2 − Φ1 − ΔHJ
VpHJ Φ2 − Φ1 ” obstacle precluding assessment of this interfacial barrier is that
the charge neutrality level of SnO2 has not been reported.
Charge neutrality level assessment is most reliably accom-
plished either theoretically by calculating the zero of the
ZnO [2,4,8], InGaO3(ZnO)5 [3], SnO2 [5], indium gallium zinc Green's function of the bulk band structure over the Brillouin
oxide (IGZO) [7], zinc tin oxide (ZTO) [9], zinc indium oxide zone [13,14,37,38] or from the complex band structure [13,39],
(ZIO) [10], In2O3 [11], and indium gallium oxide (IGO) [11,12]. or experimentally from valence band offset data [13,39]. A
Two homojunctions source-drain contact TTFT examples are crude estimate of this parameter may be obtained by
known, involving a gallium-doped ZnO layer as the contact to a recognizing that the charge neutrality level may be calculated
ZnO channel [6] and ITO with a low carrier concentration In2O3 as the Green's function of the band structure taken over the
channel [11]. Brillouin zone, and that the conduction band minimum and
Table 2 summarizes Schottky barrier/heterojunction para- valence band maximum provide the largest contributions to the
meters required for interfacial assessment of ITO source-drain Green's function so that [14]
contacts for ZnO or SnO2 channel layer TTFTs. There appears Z
N ðEÞdE NC NV
to be some uncertainty regarding some of these parameters. GðECNL Þ ¼ g þ ¼ 0;
Specifically, the electron affinities of SnO2 and In2O3 do not ECNL  E ECNL  EG ECNL  0
BZ
appear to be well established [26,31–33], nor does the high
ð14Þ
frequency dielectric constant of In2O3 [24]. For the interface
assessment undertaken herein, the high frequency dielectric which leads to
constant of In2O3 and ITO are assumed to be equal. The
parameter values included in Table 2 are believed to be the most NV EG
ECNL g ; ð15Þ
reliable of those reported in the literature. NV þ NC
In Table 2, note that the TTFT channel layer work function, where NV and NC are the valence and conduction band effective
ΦS, is approximated as ≳(χS + 0.15). This estimate is established density of states, respectively, EG is the band gap, and ECNL is
by recognizing that the channel layer Fermi level, EF, with the charge neutrality level position with respect to the top of the
respect to the conduction band minimum, EC, is equal to valence band, EV, which is assumed to be a zero reference
  energy. As a check to the viability of Eq. (15), if the electron and
kB T nchannel
EF  EC ¼ ln ; ð13Þ
q NC

where the channel layer electron density, nchannel, is assumed to Table 2


be less than ∼ 1016 cm− 3 in order to insure that the channel is Schottky barrier and heterojunction assessment parameters for ITO–ZnO and
ITO–SnO2 TTFT source-drain contacts
highly insulating, and the effective conduction band density of
states is approximately equal to 3 × 1018 cm− 3 since the electron Material ϵ∞ S ΦS (V) χS (V) ΦCNL (V)
density of states relative effective mass for ZnO, SnO2, and Contact
In2O3 is ∼ 0.25 [34–36]. ITO 3.7 ± 0.3 [24] 0.58 ± 0.05 4.1–4.7 [25] 4.45 [26] ?
With regard to interfacial barrier assessment, first consider
Channel
the ITO–ZnO interface of a TTFT source-drain. Since the ITO ZnO 3.85 [27] 0.55 ≳(χS + 0.15) 4.57 [28] 4.9 [29]
contact is degenerately doped, it makes sense to model ITO as a SnO2 3.9 [30] 0.54 ≳(χS + 0.15) 4.5 [31] ?
metal and then use Schottky barrier theory, as prescribed by Eqs. Parameters: ϵ∞ = high frequency relative dielectric constant, S = interface
(1)–(4), to assess the ITO–ZnO interface. In order to accomplish parameter, ΦS = work function, χS = electron affinity, ΦCNL = charge neutrality
such a calculation, it is necessary to recognize that the ITO work level.
J.F. Wager / Thin Solid Films 516 (2008) 1755–1764 1763

hole density of states relative effective masses of ZnO are taken level electrons in the ITO see a small positive barrier ∼ +0.15 V,
to be 0.24 [34] resulting in NC = 3.0 × 1018 cm− 3 and 0.8 ± 0.5 so that this interface would function as an injecting contact.
[40] yielding NV = 1.8 × 1019 cm− 3, respectively, then Eq. (15) Next, consider the heterojunction assessment portion of
yields ECNL ≃ 2.9 eV, which is consistent with the experimen- Table 3. Heterojunction assessment of the ITO–ZnO interface
tally deduced value of ECNL = 3.04 ± 0.21 eV, as reported by indicates that the conduction band discontinuity is negative, i.e.,
Mönch [29]. − 0.25 V, characteristic of an injecting contact. Fig. 11 is the
Table 3 presents a Schottky barrier and heterojunction energy band diagram corresponding to the ITO–ZnO interface,
assessment summary of ITO–ZnO and ITO–SnO2 TTFT except that virtually all of the band bending occurs in the ZnO
contacts. For the ZnO channel, Mönch's experimentally since this layer has a much lower carrier concentration
determined value of the charge neutrality level is employed compared to ITO. This negative barrier is consistent with the
[29]. The In2O3 charge neutrality level is estimated in order to experimental result that the ITO–ZnO interface is an injecting
facilitate heterojunction assessment of ITO–ZnO and ITO– contact for a TTFT with a ZnO channel layer [2,4,8]. A
SnO2 interfaces, where the ITO charge neutrality level is comparison of the calculated values of ϕBn = − 0.11 V and ΔEC /
assumed to be equal to that of In2O3. Effective density of states q = − 0.25 V for the ITO–ZnO interface shows that these values
estimates included in Table 3 are obtained from corresponding are somewhat similar, and both are indicative of an injecting
relative density of states effective masses, which are taken to be contact. However, microscopic dipole corrections are distinctly
0.275 [35] and 0.78 [41], and 0.3 [42] and 0.6 [36,43] for different, i.e., ΔSB = + 0.36 V and ΔHJ = − 0.13 V. The hetero-
electrons and holes of SnO2, and In2O3, respectively. junction dipole correction is quite small. This is a consequence
First, consider the Schottky barrier assessment portion of of Eq. (9), in which the high frequency dielectric constants are
Table 3. As discussed previously for the ITO–ZnO interface, the combined in parallel, in analogy with capacitors in series, and is
Schottky barrier height is calculated to be negative, character- also due to the fact that the high frequency dielectric constants
istic of an injecting contact. In contrast, Schottky barrier height for these oxides are rather small, and that the charge neutrality
assessment of the ITO–SnO2 interface leads to the prediction of level energies are predicted to be very similar for ITO and ZnO.
a slightly positive barrier, i.e., +0.08 V. Even though the Heterojunction assessment of the ITO–SnO2 interface results
Schottky barrier height is calculated to be positive, characteristic in a predicted conduction band discontinuity of − 0.11 V,
of a rectifying contact, energy band diagram analysis demon- consistent with an injecting contact, as found experimentally for
strates that a + 0.08 V barrier would function as a satisfactory the SnO2 TTFT [5], and is characterized by an energy band
source-drain injecting contact in a TTFT application, as diagram similar to the one shown in Fig. 11. The heterojunction
observed experimentally [5]. To see this, note that a very small dipole correction term is predicted to be small for the ITO–
positive barrier is indicated in Fig. 5, similar to that expected for SnO2 interface, ΔHJ = − 0.06 V.
the + 0.08 V barrier case. Since (EC − EF) / q ≈ 0.15 V, as
discussed previously, conduction band electrons in the SnO2 5. Conclusions
actually see a negative barrier equal to − 0.07 V, while Fermi
Two main topics are addressed.
First, an explication of modern Schottky barrier and
Table 3
heterojunction theory is presented from a charge transfer,
Schottky barrier and heterojunction assessment of ITO–ZnO and ITO–SnO2 energy band diagram perspective. For both situations, interface
TTFT source-drain contacts behavior is classified and is quantitatively formulated in terms
Schottky barrier assessment of ideal (Fermi level mediated) and non-ideal (charge neutrality
level mediated) charge transfer, giving rise to a macroscopic
Channel NC NV ECNL χS ΦCNL ΔSB ϕBn
(cm− 3) (cm− 3) (eV) (V) (V) and a microscopic dipole for the ideal and non-ideal case,
ZnO 3.0 × 1018 1.8 × 1019 3.04 4.57 4.9 +0.36 − 0.11 respectively. This approach facilitates recognition of the close
SnO2 3.6 × 1018 1.7 × 1019 3.05 4.5 5.15 +0.48 +0.08 correspondence between Schottky barriers and heterojunctions,
and provides a simple perspective from which to investigate the
Heterojunction assessment
device physics of interface formation.
Layer NC NV ECNL χS ΦCNL (V) ΔHJ DEC Second, Schottky barrier and heterojunction theory are
(cm− 3) (cm− 3) (eV) ðV Þ
q applied to the problem of ITO–ZnO and ITO–SnO2 interfaces
Contact layer
used for source-drain contacts in transparent thin-film transis-
tors. Both interfaces are expected to function as injecting
In2O3 4.1 × 1018 1.2 × 1019 2.7 4.45 5.35 NA NA
contacts according to Schottky barrier and heterojunction theory.
Channel layer
Acknowledgements
ZnO 3.0 × 1018 1.8 × 1019 3.04 4.57 4.9 − 0.13 − 0.25
SnO2 3.6 × 1018 1.7 × 1019 3.05 4.5 5.15 − 0.06 − 0.11
This work was funded by the U.S. National Science
For the ZnO channel, Mönch's [29] experimentally determined value of the
charge neutrality level is employed. Charge neutrality levels for SnO2 and In2O3 Foundation under Grant No. DMR-0245386, the Army Research
are estimated using Eq. (15); the ITO charge neutrality level is assumed to be Office under Contract No. MURI E-18-667-G3, and the Hewlett-
equal to that of In2O3. Packard Company.
1764 J.F. Wager / Thin Solid Films 516 (2008) 1755–1764

References conduction band notch results if either Φ2 N Φ1 and χ2 N χ1, or if Φ2 b Φ1


and χ2 b χ1. Similarly, a valence band step occurs if either Φ2 N Φ1 and
[1] J.F. Wager, Science 300 (2003) 1245. IP2 b IP1, or if Φ2 b Φ1 and IP2 N IP1; in contrast, a valence band notch
[2] R.L. Hoffman, B.J. Norris, J.F. Wager, Appl. Phys. Lett. 82 (2003) 733. results Φ2 N Φ1 and IP2 N IP1, or if Φ2 b Φ1 and IP2 b IP1.
[19] A.W. Cowley, S.M. Sze, J. Appl. Phys. 36 (1965) 3212.
[3] K. Nomura, H. Ohta, K. Ueda, T. Kamiya, M. Hirano, H. Hosono, Science
300 (2003) 1269. [20] Eq. (9) may be derived along the lines of derivations presented in references
[4] B.J. Norris, J. Anderson, J.F. Wager, D.A. Keszler, J. Phys., D, Appl. Phys. [13,19] by first assuming capacitive coupling of interface states across a
thin interfacial layer of atomic dimensions, yielding S12 = 1/(1 + Css‖/Ci),
36 (2003) L105.
[5] R.E. Presley, C.L. Munsee, C.-H. Park, D. Hong, J.F. Wager, D.A. Keszler, where Css‖ denotes a parallel combination of the interface state capacitance
J. Phys., D, Appl. Phys. 37 (2004) 2810. for each semiconductor and Ci is the capacitance of the thin interfacial layer.
[6] E. Fortunato, A. Pimentel, L. Pereira, A. Goncalves, G. Lavareda, H. Aguas, I. Next, employ Mönch's empirical relation from reference [13], Css/Ci = 0.1
(ϵ∞ − 1)2 for each semiconductor interface capacitance to obtain Eq. (9).
Ferreira, C.N. Carvalho, R. Martins, J. Non-Cryst. Solids 338–340 (2004) 806.
[7] K. Nomura, H. Ohta, A. Takagi, T. Kamiya, M. Hirano, H. Hosono, Nature [21] C.M. Wolfe, N. Holonyak, G.E. Stillman, Physical Properties of
432 (2003) 488. Semiconductors, Prentice Hall, Englewood Cliffs, 1989, p. 277.
[8] Y. Kwon, Y. Li, Y.W. Heo, M. Jones, P.H. Holloway, D.P. Norton, Z.V. [22] The expression ΔEG = ΔEV + ΔEC is explicitly equal to EG2 − EG1 = (EV1 −
EV2) + (EC2 − EC1). The subscript reversal of ΔEV compared to ΔEG and
Park, S. Li, Appl. Phys. Lett. 84 (2004) 2685.
[9] H.Q. Chiang, J.F. Wager, R.L. Hoffman, J. Jeong, D.A. Keszler, Appl. ΔEC is confusing, and is a consequence of the convergence of several
Phys. Lett. 86 (2005) 013503. energy band diagram sign conventions which are implicitly employed in
Eqs. (7) and (11); electron energy is positive going upward, hole energy is
[10] N.L. Dehuff, E.S. Kettenring, D. Hong, H.Q. Chiang, J.F. Wager, R.L.
Hoffman, C.-H. Park, D.A. Keszler, J. Appl. Phys. 97 (2005) 064505. positive going downward, electrostatic potential is positive going
[11] H.Q. Chiang, D. Hong, C.M. Hung, C.-H. Park, R.E. Presley, G.S. downward, and band gap is a positive quantity.
Herman, J.F. Wager, D.A. Keszler, J. Vac. Sci. Technol., B (2006) 2702. [23] R.L. Hoffman, J.F. Wager, Thin Solid Films 436 (2003) 286.
[24] T. Gerfin, M. Grätzel, J. Appl. Phys. 79 (1996) 1722.
[12] R.E. Presley, D. Hong, H.Q. Chiang, C.M. Hung, R.L. Hoffman, J.F.
Wager, Solid-State Electron. 50 (2006) 500. [25] D.J. Milliron, I.G. Hill, C. Shen, A. Kahn, J. Schwartz, Appl. Phys. Lett. 87
[13] W. Mönch, Semiconductor Surfaces and Interfaces, 3rd ed. Springer, (2000) 572.
Berlin, 2001. [26] E.Y. Wang, L. Hsu, J. Electrochem. Soc. 125 (1978) 21328.
[27] Z.-C. Jin, I. Hamberg, C.G. Grandqvist, J. Appl. Phys. 64 (1988) 5117.
[14] J.R. Robertson, J. Vac. Sci. Technol., B 18 (2000) 1785.
[15] H. Vázquez, W. Gao, F. Flores, A. Kahn, Phys. Rev., B 71 (2005) 041306. [28] R.K. Swank, Phys. Rev. 153 (1967) 844.
[16] It might seem puzzling that the positive microscopic dipole of Fig. 2 and [29] W. Mönch, Appl. Phys. Lett. 86 (2005) 162101.
the negative macroscopic dipole of Figs. 1 and 2) yield identical, rather [30] A.E. Taverner, C. Rayden, S. Warren, A. Gulino, P.A. Cox, R.G. Egdell,
Phys. Rev., B 51 (1995) 6833.
than opposing barrier height trends. However, it is found for both Schottky
barriers and heterojunctions that identical (opposing) macroscopic– [31] F. Möllers, R. Memming, Ber. Bunsen-Ges. 76 (1972) 469.
microscopic dipole polarities produce opposing (identical) trends in [32] J. Fritsche, D. Kraft, A. Thiben, T. Mayer, A. Klein, W. Jaegermann, Thin
Solid Films 403–404 (2002) 252.
terms of barrier height. Perhaps the easiest way to understand this subtle,
counterintuitive trend is to employ an interfacial layer of atomic [33] A. Klein, Appl. Phys. Lett. 77 (2000) 2009.
dimensions to account for the finite separation distance of the microscopic [34] W.S. Baer, Phys. Rev. 154 (1967) 785.
dipole (e.g., see Ref. [17]). Next, recognize that the separation of charge [35] K.J. Button, C.G. Fonstad, W. Dreybrodt, Phys. Rev., B 4 (1971) 4539.
[36] I. Hamberg, C.G. Granqvist, K.-F. Berggren, B.E. Sernelius, L. Engström,
across this interfacial layer due to the presence of a microscopic dipole
essentially takes charge away from the metal side and places it onto the Phys. Rev., B 30 (1984) 3240.
semiconductor side of the interfacial layer. Since this charge is now in [37] J. Tersoff, Phys. Rev. Lett. 52 (1984) 465.
closer physical proximity to the semiconductor, its effect is enhanced, in [38] M. Cardona, N.E. Christensen, Phys. Rev., B 35 (1987) 6182.
[39] A.A. Demkov, L.R.C. Fonseca, E. Verret, J. Tomfohr, O.F. Sankey, Phys.
terms of barrier height modification, in the same sense as original the
macroscopic dipole. Rev., B 71 (2005) 195306.
[17] S.M. Sze, K.K. Ng, Physics of Semiconductor Devices, 3rd ed. Wiley, [40] B. Enright, D. Fitzmaurice, J. Phys. Rev. B 100 (1996) 1027.
[41] A.E. Rakhshani, Y. Makdisi, H.A. Ramazaniyan, J. Appl. Phys. 83 (1998)
New York, 2007.
[18] Note that a notch conduction band discontinuity is displayed in all of the 1049.
heterojunction energy band diagrams presented in Figs. 6–11. A step [42] Z.M. Jarzebski, Phys. Status Solidi, A Appl. Res. 71 (1982) 13.
conduction band discontinuity is also possible. A conduction band step [43] L. Gupta, A. Mansingh, P.K. Srivastava, Thin Solid Films 176 (1989) 33.
occurs if either Φ2 N Φ1 and χ2 b χ1, or if Φ2 b Φ1 and χ2 N χ1; in contrast, a

S-ar putea să vă placă și