We propose crystalline ZnSnO3 as a new channel material for field-effect transistors. By molecular-beam epitaxy on LiNbO3(0001) substrates, we synthesized films of ZnSnO3, which crystallizes in the LiNbO3-type polar structure. Field-effect transistors on ZnSnO3 exhibit n-type operation with field-effect mobility of as high as 45 cm2V−1s−1 at room temperature. Systematic examination of the transistor operation for channels with different Zn/Sn compositional ratios revealed that the observed high-mobility reflects the nature of stoichiometric ZnSnO3 phase. Moreover, we found an indication of coupling of transistor characteristics with intrinsic spontaneous polarization in ZnSnO3, potentially leading to a distinct type of polarization-induced conduction.

The successful development of In-Ga-Zn-O field-effect transistors1 (FETs) has given a boost to the continuing search for new transparent oxide semiconductors. On the basis of the general concept that heavy typical elements, which produce s-orbital derived conduction bands, are beneficial for attaining high electron mobility, various multi-cation oxides such as In-Zn-O, Cd-Sn-O, and Zn-Sn-O have been proposed.2–5 In those oxides, crystalline phases are known to exist at certain stoichiometric compositions, but attention has been paid to search for optimal compositions in their amorphous alloy phase diagrams towards high-performance FETs, rather than properties of specific phases.

In the field of fundamental solid-state chemistry and geophysics, various crystalline phases in ternary oxides and their structural transitions under applied pressure have been systematically investigated.6–8 Recently, using a high-pressure technique, Inaguma et al. has successfully stabilized LiNbO3-type ZnSnO3 shown in Fig. 1(a) 9,10 and discovered ferroelectricity9 arising from the broken inversion symmetry (space group: R3c).11,12 The uniqueness of this new ferroelectric oxide lies in its d10 electronic configuration, distinct from conventional d0 (e.g., Ti perovskites) and ((n − 1)d)10(ns)2 (e.g., BiFeO3) configurations. As a consequence of d10 closed shell, the conduction band is composed mainly of spatially extended Sn 5s orbitals.13,14 The calculated electron effective mass of me = 0.32m0 (m0 is the free electron mass)13 is comparably small to me = 0.2 – 0.4m0 of BaSnO3, which exhibits high electron mobility15 owing to the Sn 5s derived conduction band.16 In addition, if electrical conduction is induced on the polar crystal surface, this system could potentially offer a new route to understanding of polarization-induced conduction as observed at ferroelectric domain walls.17,18 In this paper, we describe FET operation of crystalline ZnSnO3 films synthesized on LiNbO3 substrates by molecular-beam epitaxy (MBE). We demonstrate high field-effect mobility (μFE) exceeding 40 cm2V−1s−1 at room temperature, and observe peculiar FET characteristics depending on the poled direction of substrates, which imply coupling of interface conduction with spontaneous polarization.

FIG. 1.

(a) Crystal structure of LiNbO3-type ZnSnO3.9,10 (b) Out-of-plane XRD pattern for a ZnSnO3 film grown on a –Z surface of LiNbO3(0001) substrate. The asterisks indicate the forbidden (0003(2n − 1)) (n: natural number) reflections of LiNbO3. (c) Typical AFM topographic image of ZnSnO3 films.

FIG. 1.

(a) Crystal structure of LiNbO3-type ZnSnO3.9,10 (b) Out-of-plane XRD pattern for a ZnSnO3 film grown on a –Z surface of LiNbO3(0001) substrate. The asterisks indicate the forbidden (0003(2n − 1)) (n: natural number) reflections of LiNbO3. (c) Typical AFM topographic image of ZnSnO3 films.

Close modal

ZnSnO3 films were grown on LiNbO3(0001) substrates by oxygen-plasma assisted MBE. We used two-side polished, pre-poled (+Z and −Z), congruent LiNbO3 single crystal substrates. Prior to the film growth, the LiNbO3 substrates were annealed in an oxygen ambient at 900 °C, resulting in atomically smooth surfaces with a step-and-terrace structure both on +Z and –Z surfaces. Elemental Zn and Sn were supplied by effusion cells with high-purity 7N and 5N metal sources, respectively. The optimal growth conditions were the substrate temperature of 500 °C and oxygen pressure of 7.0 × 10−3 Pa. The relationship between beam flux and the actual Zn/Sn composition in the film was determined by X-ray photoelectron emission spectroscopy. We fabricated top-gate FETs on 100-nm-thick ZnSnO3 channels patterned by photolithography and Ar ion milling. The channel width and length were W = 100 μm and L = 250 μm, respectively. A parylene gate dielectric layer (relative dielectric constant of 3.15) was deposited at room temperature. Source, drain, and gate Ti electrodes were prepared by electron-beam evaporation. For the ohmic source and drain contacts, the ZnSnO3 surface was irradiated with Ar+ ions19,20 before the electrode formation. Electrical measurements were carried out at room temperature in air with a semiconductor parameter analyzer (Agilent Technologies 4155C) and a source-measure unit (Keithley 2614B).

X-ray diffraction (XRD) pattern for a Zn-Sn-O film grown at a condition of Zn/Sn ratio close to unity is shown in Fig. 1(b). Intense diffraction peaks are observed near LiNbO3 (0006) and (00012), indicating the formation of highly oriented film. Given that the Zn-Sn-O film crystallizes in a corundum-derivative rhombohedral structure and the above peaks are indexed as (0006) and (00012), the c-axis lattice parameter is estimated to be c = 14.28 Å, which is close to, but slightly larger than c = 14.00 Å of LiNbO3-type ZnSnO3 bulk.9 The absence of (0003) and (0009) reflections, which are allowed for another corundum derivative, non-polar ilmenite-type structure, excludes ilmenite-type ZnSnO3. Therefore, it is most probable that our Zn-Sn-O film is of LiNbO3-type ZnSnO3. Detailed structural characterization, which will be reported elsewhere, shows that the c-axis ZnSnO3 film is composed of two domains of three-fold symmetry, rotated by 60° in the film plane. At the domain boundaries, LiNbO3-type ordering is preserved along the c-axis direction, but atomic rearrangements would form edge-sharing ZnO6 (SnO6) octahedra, which are corner-sharing in regular LiNbO3-type ZnSnO3, making the system ilmenite-like locally. In related compounds, MnSnO3 and FeTiO3, the structural transformation from LiNbO3-type to ilmenite-type is known to be associated with increases of c-axis parameter as large as 2.4–3.4%.7,21,22 The two-domain ZnSnO3 film should contain such an elongated ilmenite-like local structure at the domain boundaries, and the observed c-axis elongation of 2% is less than those lattice variations. Hence we speculate that local modification of octahedral network should play a role in the increased c-axis parameter. The surface of c-axis oriented ZnSnO3 films thus fabricated has a characteristic step-and-terrace structure over a wide area as observed by atomic force microscopy shown in Fig. 1(c).

We here focus on ZnSnO3 FET fabricated on negatively poled LiNbO3 substrate plane, called –Z surface. We confirmed that film crystallinity does not depend on +Z and –Z surfaces by XRD, surface morphology observation, and compositional analysis. Figure. 2(a) presents output characteristics (drain current ID versus drain voltage VD curves), where clear saturation behavior is observed at high VD regions. Note that the large operation voltages are simply due to low capacitance of the thick parylene gate dielectric layer (540 nm for the device shown in Fig. 2).23,24 The sheet carrier density under applied gate voltage VG of 100 V is of the order of 1012 cm−2, which is comparable to values obtained by other gate dielectric materials.19,25 Transfer characteristics shown in Fig. 2(b) demonstrate that channel conduction is enhanced with positively increasing VG (i.e., electron-type carrier) while gate leakage current IG remains as low as 10−11 – 10−12 A. These characteristics are essentially identical to those of conventional n-type oxide semiconductor FETs,19,23,25 indicating that band bending and resulting carrier accumulation occur at the surface of this polar oxide. Although finite conduction is present at VG = 0 V, it can be completely suppressed by applying negative VG. In the depleted state, ID becomes comparably low to IG, yielding ID on/off ratio of as high as 106 – 107 with the present device geometry. According to procedures generally used for analysis of oxide FETs,19,23,25 we calculated μFE from the linear regime characteristics (VD = 1 V), as displayed in Fig. 2(c). The μFE value of 43 cm2V−1s−1 surpasses typical values in amorphous and phase-mixed polycrystalline Zn-Sn-O systems (μFE = 10 – 30 cm2V−1s−1).26–33 Hall effect measurement will be needed in future to understand the relation between μFE, Hall mobility, and the actual carrier density.

FIG. 2.

(a) Typical output characteristics of a ZnSnO3 FET fabricated on a –Z surface of LiNbO3 substrate. The thicknesses of ZnSnO3 channel and parylene gate dielectric are 100 nm and 540 nm, respectively. (b) Transfer characteristics measured at VD = 1 V. (c) μFE calculated from the linear regime transfer characteristics presented in (b).

FIG. 2.

(a) Typical output characteristics of a ZnSnO3 FET fabricated on a –Z surface of LiNbO3 substrate. The thicknesses of ZnSnO3 channel and parylene gate dielectric are 100 nm and 540 nm, respectively. (b) Transfer characteristics measured at VD = 1 V. (c) μFE calculated from the linear regime transfer characteristics presented in (b).

Close modal

To inspect that such high μFE does not come from a mere admixture of Zn and Sn oxides, we prepared films with various Zn/Sn ratios by varying Zn beam flux while fixing Sn beam flux. No traces of segregation and/or formation of secondary phases were discerned in their XRD patterns (see Fig. S1 of the supplementary material), indicating that excess Zn (Sn) atoms are introduced while retaining the corundum derivative framework. In the upper panel of Fig. 3, the threshold gate electric field Eth for inducing channel conduction is plotted as a function of Zn/Sn ratio. Here, gate electric field EG is converted from VG for fair comparison of Eth in devices with different parylene gate dielectric thicknesses (400 – 540 nm), and Eth is defined as EG at which ID exceeds 10−9 A. Eth is relatively small for almost stoichiometric samples (Zn/Sn ratio = 0.96 and 1.04), but negatively increases as nonstoichiometry becomes significant. This is probably caused by the generation of excess electron charge carriers from defects although the detailed defect species are not clear at present. Linked to this, μFE also reaches a peak around the stoichiometric composition as shown in the lower panel, where μFE at the equivalent carrier accumulation condition of EG = Eth + 3MVcm−1 is plotted (hence, the plotted μFE does not mean the maximal μFE at each composition). These observations underpin that stoichiometric LiNbO3-type ZnSnO3 is the origin of the superior FET properties.

FIG. 3.

Eth (top) and μFE (bottom) as a function of Zn/Sn ratio. Stoichiometric ZnSnO3 corresponds to Zn/Sn ratio = 1. Error bars represent standard deviation for measured data. μFE at EG = Eth + 3 MVcm−1 is plotted (black, filled circles) for comparison. For reference, maximal μFE values obtained for Zn/Sn ratio = 1.04 are also included (red, filled triangle).

FIG. 3.

Eth (top) and μFE (bottom) as a function of Zn/Sn ratio. Stoichiometric ZnSnO3 corresponds to Zn/Sn ratio = 1. Error bars represent standard deviation for measured data. μFE at EG = Eth + 3 MVcm−1 is plotted (black, filled circles) for comparison. For reference, maximal μFE values obtained for Zn/Sn ratio = 1.04 are also included (red, filled triangle).

Close modal

By taking advantages of poled LiNbO3 substrate, we compared FETs on +Z and –Z surfaces, as schematically depicted in Fig. 4(a). Figure 4(b) shows transfer characteristics of two devices on almost stoichiometric ZnSnO3 films fabricated with identical growth parameters. Interestingly, substrate polarization is found to induce a large shift in Eth for transistor operation at the ZnSnO3 channel surface, which is separated by 100 nm from the LiNbO3 surface. Since the electric field generated by substrate surface polarization is not likely to penetrate into the whole ZnSnO3 film with the grounded source electrode, it is natural to consider that the observed Eth shift arises from spontaneous polarization of ZnSnO3. Based on an assumption that polarization direction of ZnSnO3 predominantly follows that of substrate via the film crystallization process, the increased (decreased) number of charge carriers should be accumulated on the +Z (–Z) type ZnSnO3 channel, giving rise to Eth shift to the negative (positive) gate electric field. We note that the above consideration can apply for our c-axis oriented ZnSnO3 films with two in-plane rotated domains, because ionic displacement of Zn ions along the c-axis is proposed to be responsible for ferroelectricity in ZnSnO3.34 We therefore attribute the observed Eth shift to the intrinsic polarization of ZnSnO3. The effect of specific polarization geometry is reminiscent of ferroelectric domain wall systems, where asymmetric band bending profiles at head-to-head and tail-to-tail type domain walls result in different transport properties.18 In the present system, similar charge imbalance by surface polarization could give rise to the increased (decreased) channel conduction on the +Z (−Z) surface. Direct observation of ferroelectricity in our films, e.g., by polarization reversal,34 would provide further evidence to the coupling of interface conduction with spontaneous polarization in this polar oxide.

FIG. 4.

(a) Schematics of ZnSnO3 films grown on +Z and –Z surfaces of LiNbO3 substrate. (b) Transfer characteristics of ZnSnO3 FETs on +Z (blue) and –Z (red) surfaces of LiNbO3 substrate. The data presented in Fig. 2(b) is replotted for comparison. The parylene thicknesses are 450 nm and 540 nm for the devices on the +Z and –Z surface, respectively.

FIG. 4.

(a) Schematics of ZnSnO3 films grown on +Z and –Z surfaces of LiNbO3 substrate. (b) Transfer characteristics of ZnSnO3 FETs on +Z (blue) and –Z (red) surfaces of LiNbO3 substrate. The data presented in Fig. 2(b) is replotted for comparison. The parylene thicknesses are 450 nm and 540 nm for the devices on the +Z and –Z surface, respectively.

Close modal

In conclusion, LiNbO3-type ZnSnO3 is a promising material not only for developing high-performance FETs but also for exploring coupling of electrical transport with ferroelectric polarization. These findings demonstrate the potential of LiNbO3-type oxides for research on polarization-induced conduction at polar interfaces35 as well as ferroelectric domain walls.17,18 Expanding the variety of related oxides should be important to design LiNbO3-type heterostructures.

See supplementary material for XRD patterns of films with various Zn/Sn ratios (Fig. S1).

The authors thank N. Shibata for valuable discussions, J. Shiogai and K. Miura for their assistance with sample preparation, and NEOARK Corporation for the use of a maskless lithography system PALET. This work was performed under the Inter-University Cooperative Research Program of the Institute for Materials Research, Tohoku University (Proposal No. 17G0415). This work was partly supported by JSPS KAKENHI (JP15H02022) and Kato Foundation for Promotion of Science (KJ-2607).

1.
K.
Nomura
,
H.
Ohta
,
K.
Ueda
,
T.
Kamiya
,
M.
Hirano
, and
H.
Hosono
,
Science
300
,
1269
(
2003
).
2.
T. J.
Coutts
,
D. L.
Young
,
X.
Li
,
W. P.
Mulligan
, and
X.
Wu
,
J. Vac. Sci. Technol. A
18
,
2646
(
2000
).
4.
A. J.
Freeman
,
K. R.
Poeppelmeier
,
T. O.
Mason
,
R. P. H.
Chang
, and
T. J.
Marks
,
MRS Bull.
25
,
45
(
2000
).
5.
E.
Fortunato
,
P.
Barquinha
, and
R.
Martins
,
Adv. Mater.
24
,
2945
(
2012
).
6.
Y.
Shono
,
S.
Akimoto
, and
K.
Kohn
,
J. Phys. Soc. Jpn.
26
,
993
(
1969
).
7.
K.
Leinenweber
,
W.
Utsumi
,
Y.
Tsuchida
,
T.
Yagi
, and
K.
Kurita
,
Phys. Chem. Minerals
18
,
244
(
1991
).
8.
A.
Navrotsky
,
Chem. Mater.
10
,
2787
(
1998
).
9.
Y.
Inaguma
,
M.
Yoshida
, and
T.
Katsumata
,
J. Am. Chem. Soc.
130
,
6704
(
2008
).
10.
K.
Momma
and
F.
Izumi
,
J. Appl. Crystallogr.
44
,
1272
(
2011
).
11.
N. A.
Benedek
and
C. J.
Fennie
,
J. Phys. Chem. C
117
,
13339
(
2013
).
12.
M.
Ye
and
D.
Vanderbilt
,
Phys. Rev. B
93
,
134303
(
2016
).
13.
H.
Wang
,
H.
Huang
, and
B.
Wang
,
Solid State Commun.
149
,
1849
(
2009
).
14.
K. P.
Ong
,
X.
Fan
,
A.
Subedi
,
M. B.
Sullivan
, and
D. J.
Singh
,
APL Mater.
3
,
062505
(
2015
).
15.
H. J.
Kim
,
U.
Kim
,
H. M.
Kim
,
T. H.
Kim
,
H. S.
Mun
,
B.-G.
Jeon
,
K. T.
Hong
,
W.-J.
Lee
,
C.
Ju
,
K. H.
Kim
, and
K.
Char
,
Appl. Phys. Express
5
,
061102
(
2012
).
16.
W.-J.
Lee
,
H. J.
Kim
,
J.
Kang
,
D. H.
Jang
,
T. H.
Kim
,
J. H.
Lee
, and
K. H.
Kim
,
Annu. Rev. Mater. Res.
47
,
391
(
2017
), and references therein.
17.
J.
Seidel
,
L. W.
Martin
,
Q.
He
,
Q.
Zhan
,
Y.-H.
Chu
,
A.
Rother
,
M. E.
Hawkridge
,
P.
Maksymovych
,
P.
Yu
,
M.
Gejek
,
N.
Balke
,
S. V.
Kalinin
,
S.
Gemming
,
F.
Wang
,
G.
Catalan
,
J. F.
Scott
,
N. A.
Spaldin
,
J.
Orenstein
, and
R.
Ramesh
,
Nat. Mater.
8
,
229
(
2009
).
18.
D.
Meier
,
J.
Seidel
,
A.
Cano
,
K.
Delaney
,
Y.
Kumagai
,
M.
Mostovoy
,
N. A.
Spaldin
,
R.
Ramesh
, and
M.
Fiebig
,
Nat. Mater.
11
,
284
(
2012
).
19.
T.
Sato
,
K.
Shibuya
,
T.
Ohnishi
,
K.
Nishio
, and
M.
Lippmaa
,
Jpn. J. Appl. Phys.
46
,
L515
(
2008
).
20.
S.
Harashima
,
C.
Bell
,
M.
Kim
,
T.
Yajima
,
Y.
Hikita
, and
H. Y.
Hwang
,
Phys. Rev. B
88
,
085102
(
2013
).
21.
22.
T.
Yamanaka
,
Y.
Komatsu
, and
H.
Nomori
,
Phys. Chem. Miner.
34
,
307
(
2007
).
23.
H.
Nakamura
,
H.
Takagi
,
I. H.
Inoue
,
Y.
Takahashi
,
T.
Hasegawa
, and
Y.
Tokura
,
Appl. Phys. Lett.
89
,
133504
(
2006
).
24.
V.
Podzorov
,
M. E.
Gershenson
,
Ch.
Kloc
,
R.
Zeis
, and
E.
Bucher
,
Appl. Phys. Lett.
84
,
3301
(
2004
).
25.
M.
Katayama
,
S.
Ikesaka
,
J.
Kuwano
,
Y.
Yamamoto
,
H.
Koinuma
, and
Y.
Matsumoto
,
Appl. Phys. Lett.
89
,
242103
(
2006
).
26.
R. L.
Hoffman
,
Solid-State Electron.
50
,
784
(
2006
).
27.
P.
Görrn
,
P.
Hölzer
,
T.
Riedl
,
W.
Kowalsky
,
J.
Wang
,
T.
Weimann
,
P.
Hinze
, and
S.
Kipp
,
Appl. Phys. Lett.
90
,
063502
(
2007
).
28.
S.-J.
Seo
,
C. G.
Choi
,
Y. H.
Hwang
, and
B.-S.
Bae
,
J. Phys. D: Appl. Phys.
42
,
035106
(
2009
).
29.
J.
Heo
,
S. B.
Kim
, and
R. G.
Gordon
,
Appl. Phys. Lett.
101
,
113507
(
2012
).
30.
J. S.
Rajachidambaram
,
S.
Sanghavi
,
P.
Nachimuthu
,
V.
Shutthanandan
,
T.
Varga
,
B.
Flynn
,
S.
Thevuthasan
, and
G. S.
Herman
,
J. Mater. Res.
27
,
2309
(
2012
).
31.
M. S.
Rajachidambaram
,
A.
Pandey
,
S.
Vilayurganapathy
,
P.
Nachimuthu
,
S.
Thevuthasan
, and
G. S.
Herman
,
Appl. Phys. Lett.
103
,
171602
(
2013
).
32.
S.
Han
and
S. Y.
Lee
,
Appl. Phys. Lett.
106
,
212104
(
2015
).
33.
K. M.
Niang
,
J.
Cho
,
S.
Heffernan
,
W. I.
Milne
, and
A. J.
Flewitt
,
J. Appl. Phys.
120
,
085312
(
2016
).
34.
J. Y.
Son
,
G.
Lee
,
M.-H.
Jo
,
H.
Kim
,
H. M.
Jang
, and
Y.-H.
Shin
,
J. Am. Chem. Soc.
131
,
8386
(
2009
).
35.
Y.
Kozuka
,
A.
Tsukazaki
, and
M.
Kawasaki
,
Appl. Phys. Rev.
1
,
011303
(
2014
).

Supplementary Material