Deposition, annealing, and integration of ferroelectric HfxZr1xO2 (HZO) thin films on the high-mobility semiconductor InAs using atomic layer deposition are investigated. Electrical characterization reveals that the HZO films on InAs exhibit an enhanced remanent polarization compared to films formed on a reference TiN substrate, exceeding 20μC/cm2 even down to an annealing temperature of 370°C. For device applications, the thermal processes required to form the ferroelectric HZO phase must not degrade the high-κ/InAs interface. We find by evaluation of the capacitance–voltage characteristics that the electrical properties of the high-κ/InAs are not significantly degraded by the annealing process, and high-resolution transmission electron microscopy verifies a maintained sharp high-κ/InAs interface.

Ferroelectric films are currently attracting consideration for memory, neuromorphic, and steep-slope transistor applications.1–5 Films based on HfO2 deposited by atomic layer deposition (ALD) are found to be ferroelectric after annealing at elevated temperatures, opening a path for integration in Si-based electronics.6–8 While most work so far has been based on deposition on Si or metal electrodes, there are a few examples of ferroelectric HfO2-based films on III–V substrates. The III–V materials provide outstanding electron transport properties that have enabled performance improvements for both high-speed and low-power electronics.9–11 For integration of ferroelectric HfO2 on III–V semiconductors, the limitations on the thermal budget are stricter since the III–V surface can decompose at temperatures exceeding 400°C.12 It is hence vital to investigate strategies to integrate HfO2-based ferroelectric gate stacks also on III-V materials and to evaluate their electrical properties.

In this paper, we study HfxZr1xO2 (HZO) deposited on the III-V material InAs by electrical and structural characterization of metal-insulator-semiconductor (MIS) capacitors and compare the results with conventional TiN metal-insulator-metal (MIM) capacitors as references. HZO was selected due to its property of achieving ferroelectricity at comparably low annealing temperatures,13 whereas InAs was chosen because of its technological relevance as channel material in high-frequency MOSFETs14 and tunnel field effect transistors.10 In addition, InAs is one of the III–V semiconductors that are most sensitive to thermal treatments,12 and thus, it constitutes a good test case to evaluate the feasibility of integrating ferroelectric HZO also on other III–V semiconductors. We demonstrate that HZO films on InAs exhibit ferroelectric behavior even under treatment with a restricted thermal budget below 400°C, i.e., close to typical annealing temperatures beneficial for defect reduction at high-κ/InAs interfaces.15 Our findings demonstrate that there is a viable path for integration of ferroelectric HZO on III–V technology platforms.

Metal-insulator-semiconductor (MIS) capacitors consisting of a TiN–HZO–InAs stack were fabricated by first growing a 100-nm-thick epilayer of unintentionally doped InAs (5×1017cm3) upon an InAs(100) wafer (3×1018cm3) using Metalorganic Vapor Phase Epitaxy (MOVPE). The native oxide of the epitaxial InAs was removed using BOE (1:10), and immediately, thereafter, 12 nm HZO was deposited at 200°C using a thermal ALD reactor with a 1:1 alternation between the precursors Tetrakis(dimethylamido)hafnium (TDMAHf) and Tetrakis(ethylmethylamido)zirconium (TEMAZr), using water as the oxygen source. A 14-nm-thick TiN layer was subsequently sputtered at an argon pressure of 2.7 mTorr, and the samples were annealed in a rapid thermal process that had been calibrated relative to the Au-Si eutectic point (363±3°C) to yield accurate temperature values.16 Top electrodes were defined by optical lithography and thermal evaporation of 5 nm Ti and 200 nm Au, after which lift-off was performed. Finally, TiN between the Ti–Au electrodes was wet etched. Reference MIM capacitors were fabricated by the same process, with the only difference being that Si(100) substrates with a 14 nm sputtered TiN bottom electrode were used instead of InAs as both the substrate and the bottom electrode.

Electrical characterization of the devices was performed in a CRX-4K cryogenic probe station using a Keysight B1500A Parameter Analyzer equipped with a B1530A waveform generator module to enable pulsed measurements. The relation between polarization and the electric field (the PE curve) was measured by a conventional positive-up-negative-down (PUND) voltage sequence.17 A first negative pulse was used as a preset pulse in order to set the polarization before the actual measurement. Before the measurements, all devices were exposed to a wake up sequence consisting of a square wave with an amplitude of 3.5 V and a repetition of 1000 times. An Agilent 4294A Impedance Analyzer was used to measure the capacitance–voltage (CV) characteristics of the structures, which were measured by sweeping the bias voltage from −3.5 V to 3.5 V while keeping the oscillation frequency and amplitude at 10 MHz and 50 mV, respectively. Grazing incidence x-ray diffraction (GIXRD) was applied using a Bruker D8 Diffractometer with a Cu K-α source and an incidence angle of 0.5°. Transmission electron microscopy was performed on a Ga ion-beam milled lamella in a JEOL JEM-3000F TEM.

Typical PE characteristics as obtained from PUND measurements are shown in Fig. 1, and the data display clear hysteresis loops for both MIS and MIM structures, with coercive fields ranging between 1 and 1.5 MV/cm and remanent polarization of about 20μC/cm2 for films on InAs and somewhat lower for MIM devices. The device-to-device spread for a certain annealing temperature is about 1μC/cm2. The hysteresis loops for films on the InAs substrate annealed above 345°C in Fig. 1(a) indicate no major change between different annealing temperatures; however, a slightly shrinking coercive field at negative voltages can be observed at higher annealing temperatures. In Fig. 1(b), the same trend is not visible for the MIM samples and it is suggested that this coercive voltage shift in the MIS samples might be related to a degradation of the HZO/InAs interface at elevated annealing temperatures. At negative fields, an ideal InAs surface will be depleted and this depletion layer will act as a series capacitor that reduces the voltage drop across the oxide. However, a high interface defect density might lead to a less effective depletion of InAs and an increased voltage would then drop across the HZO instead of the InAs depletion layer. Thus, the seemingly lower electric field at higher annealing temperatures might be related to defect density at the HZO/InAs interface rather than the oxide itself. The PE curves for the MIS samples in Fig. 1(a) have steeper polarization switches than the MIM reference samples in Fig. 1(b). This is interpreted as being a result of the MIM structures having more domains with different distinct internal bias fields than the MIS structures. However, the broadening could also be caused by an interfacial non-ferroelectric layer at the electrodes.18 The observed steeper polarization switch on the InAs electrode might therefore be due to a more beneficial crystallization or a better interface quality.

In Fig. 2, remanent polarization is displayed as a function of annealing temperature for both HZO films on InAs and the MIM reference samples. We observe plateaus of about 21μC/cm2 for MIS structures and 14μC/cm2 for the MIM reference structures above a certain temperature threshold, 370°C for MIS and 420°C for MIM. Below these thresholds, the remanent polarization is rapidly reduced to essentially zero, and it is clear that the MIS structures exhibit a ferroelectric behavior at considerably lower annealing temperatures than the reference MIM-structures. The reason for this 50°C difference is not entirely understood; however, similar results of a lower crystallization temperature have been obtained previously when comparing InAs nanowires with Si substrates.19 The lower annealing limit for forming ferroelectric HZO in MIM structures is typically reported to be about 400°C,20,21 similar to the values we find here. However, a recent study indicates that the formation of nanocrystals in an as-deposited plasma enhanced ALD film at 300°C can lower the required annealing temperature to 300°C.13 Much research is needed to determine the underlying mechanism for such an effect; nevertheless, the fact that the remanent polarization values of the MIM structures (which are lower than for the MIS) are in line with other values reported in the literature13 and that these samples were processed in parallel with the presented MIS devices strengthens the hypothesis that the InAs substrate plays an active role in increasing the remanent polarization in the annealed HZO films.

Figure 3 shows the GIXRD data of the as-deposited and annealed samples, clearly revealing the appearance of additional diffraction peaks after annealing the samples above 370°C. At about 37°,43°, and 62°, minor peaks corresponding to cubic phase TiN can be seen in all four spectra except for the annealed samples at 62° where they are obscured by HZO peaks on each side. A major peak at 50° is also observed for the MIS structures due to an InAs(311) reflection. Except for these peaks, the patterns are consistent with the existence of a crystalline tetragonal or orthorhombic HZO phase, with a minor peak at 28° indicating a small portion of monoclinic phase. By comparing the intensity of the 28° peak with that of the 30° peak, the MIS sample annealed to 470°C is estimated to have a monoclinic phase composition of about 3% compared to 9% for the corresponding MIM sample, indicating a somewhat more beneficial crystallization. The similarity of the diffraction patterns for the orthorhombic and tetragonal crystalline phases makes them difficult to distinguish. However, the distinction between these two phases is important since the orthorhombic phase is believed to contribute to ferroelectricity in HZO films, whereas the tetragonal phase is seemingly connected to antiferroelectricity.22 Nevertheless, the measured suppression of the monoclinic HZO phase in favor of a tetragonal/orthorhombic phase is commonly used as an indicator for potential ferroelectricity.23 The inset of Fig. 3 shows the peak about 30.5° for annealing temperatures of 345°C and 370°C, i.e., just below and above the onset of ferroelectricity in Fig. 2. This sharp onset of both electrically measured ferroelectricity and crystallization around the same annealing temperature indicates that the measured characteristics are not a misinterpretation related to defects or leakage currents, i.e., phenomena that easily can be misinterpreted as ferroelectricity.24,25 Furthermore, there is a decrease in the polarization measured at about 470°C for the reference MIM structures in Fig. 2. The cause of this is unknown; yet, it has been observed consistently in repeated experiments for MIM structures annealed at about 470°C. GIXRD measurements have not shown a clear crystalline difference between the samples other than the expected increasing monoclinic phase composition with increasing annealing temperature26 (see the supplementary material for further discussion).

A MIS capacitor annealed at 470°C was investigated by high-resolution TEM, as displayed in Fig. 4. Both HZO and InAs exhibit clear crystallinity in Fig. 4(a) with an about 2–3 atomic layers thick interfacial layer. At lower magnification in Fig. 4(b), the single crystallinity of InAs is more apparent, whereas a grain boundary can be seen in the right of the HZO layer. From a limited diffraction study along HZO, a typical grain size of about 50–60 nm was obtained with a single grain of 100 nm. Finally, the long range sharp HZO/InAs interface in Fig. 4(b) and similar not published micrographs indicate that a comparable quality interfacial layer can be expected across the sample. The EDX measurements indicate a film stoichiometry of about Hf0.4Zr0.6O2.

Finally, the structures were investigated by CV-measurements at 10 MHz. In Fig. 5(a), both the annealed MIM- and MIS-structures do at room temperature show CV-curves with the two peaks that are characteristic of ferroelectricity.7 For the MIS structures, the capacitance is measured across both the oxide and the semiconductor, making an estimation of the film permittivity complex. The relative permittivity of the MIM structures has, however, been estimated to be 20–27 (1.5–2.0 μC/cm2). At room temperature, the as-deposited MIM- and MIS-structures provide a capacitance independent of the applied voltage, which is expected for a MIM-structure. However, the absence of capacitance modulation in the MIS structure can be attributed to the narrow bandgap of InAs, which gives rise to thermally excited minority carriers and the comparably thick equivalent oxide thickness (EOT), which results in the limited movement of the Fermi level at the interface.27 The hypothesis of thermally excited minority carriers is strengthened by low temperature CV-characterization at 13 K in Fig. 5(b), in which InAs can be driven to depletion when the measurement frequency exceeds the minority carrier as well as border trap response times. Ideally, the CV-curve is hysteresis free; however, a clockwise hysteresis for the as-deposited HZO on InAs can be observed, likely caused by the occurrence of border traps close to the HZO/InAs interface.27 However, when annealing the MIS structure at 470°C, the low temperature CV-measurement has a smaller and to some extent counterclockwise hysteresis. This is an indication for the onset of ferroelectric switching, which will give rise to a counterclockwise hysteresis due to the poling of a polarization. The fact that the CV-curve is not entirely counterclockwise hysteric may, as before, be ascribed to the interplay of border traps and ferroelectric poling. Most importantly, there is no major difference in the ability to deplete InAs before and after annealing, which indicates that the HZO/InAs interface quality is not severely degraded by the annealing process at 470°C.

In summary, HZO deposited on the III-V semiconductor InAs has demonstrated ferroelectric behavior after ferroelectric phase formation at annealing temperatures below 400°C with enhanced remanent polarization compared to reference MIM structures. We confirm the appearance of diffraction peaks corresponding to the tetragonal/orthorhombic HZO phases after annealing at 370°C and above. TEM investigations indicate that the formation process does not visibly degrade the high-κ/InAs interface, and it is demonstrated that the MIS structure retains its CV-characteristics even after HZO crystallization. These findings will enable a future integration of ferroelectric HZO on III–V technology platforms.

See the supplementary material for further structural and electrical characterization.

The authors acknowledge Filip Lenrick for the help to FIB for the TEM lamella and Crispin Hetherington for the TEM measurements. This work was supported by the Swedish Research Council and the Swedish Foundation for Strategic Research.

1.
J.
Müller
,
T.
Böscke
,
S.
Müller
,
E.
Yurchuk
,
P.
Polakowski
,
J.
Paul
,
D.
Martin
,
T.
Schenk
,
K.
Khullar
,
A.
Kersch
,
W.
Weinreich
,
S.
Riedel
,
K.
Seidel
,
A.
Kumar
,
T.
Arruda
,
S.
Kalinin
,
T.
Schlösser
,
R.
Boschke
,
R.
van Bentum
,
U.
Schröder
, and
T.
Mikolajick
, in
IEEE International Electron Devices Meeting
(
2013
), p.
10.8.1
.
2.
H.
Mulaosmanovic
,
J.
Ocker
,
S.
Müller
,
M.
Noack
,
J.
Müller
,
P.
Polakowski
,
T.
Mikolajick
, and
S.
Slesazeck
, in Symposium on VLSI Technology (
2017
), p.
T176
.
3.
S.
Salahuddin
and
S.
Datta
,
Nano Lett.
8
,
405
(
2008
).
4.
A.
Chernikova
,
M.
Kozodaev
,
A.
Markeev
,
D.
Negrov
,
M.
Spiridonov
,
S.
Zarubin
,
O.
Bak
,
P.
Buragohain
,
H.
Lu
,
E.
Suvorova
,
A.
Gruverman
, and
A.
Zenkevich
,
ACS Appl. Mater. Interfaces
8
,
7232
(
2016
).
5.
Z.
Krivokapic
,
U.
Rana
,
R.
Galatage
,
A.
Razavieh
,
A.
Aziz
,
J.
Liu
,
J.
Shi
,
H.
Kim
,
R.
Sporer
,
C.
Serrao
,
A.
Busquet
,
P.
Polakowski
,
J.
Müller
,
W.
Kleemeier
,
A.
Jacob
,
D.
Brown
,
A.
Knorr
, and
B. S.
Carter
, in
IEEE International Electron Devices Meeting (IEDM)
(
2017
), p.
15.1.1
.
6.
T.
Böscke
,
J.
Müller
,
D.
Bräuhaus
,
U.
Schröder
, and
U.
Böttger
,
Appl. Phys. Lett.
99
,
102903
(
2011
).
7.
J.
Muller
,
T. S.
Böscke
,
U.
Schröder
,
S.
Mueller
,
D.
Brauhaus
,
U.
Bottger
,
L.
Frey
, and
T.
Mikolajick
,
Nano Lett.
12
,
4318
(
2012
).
8.
M.
Dogan
,
S.
Fernandez-Peña
,
L.
Kornblum
,
Y.
Jia
,
D. P.
Kumah
,
J. W.
Reiner
,
Z.
Krivokapic
,
A. M.
Kolpak
,
S.
Ismail-Beigi
,
C. H.
Ahn
, and
F. J.
Walker
,
Nano Lett.
18
,
241
(
2018
).
10.
E.
Memisevic
,
J.
Svensson
,
E.
Lind
, and
L.-E.
Wernersson
,
IEEE Trans. Electron Devices
64
,
4746
(
2017
).
11.
O.-P.
Kilpi
,
J.
Svensson
,
E.
Lind
, and
L.-E.
Wernersson
,
IEEE J. Electron Devices Soc.
7
,
70
(
2019
).
12.
N.
Kobayashi
and
Y.
Kobayashi
,
Jpn. J. Appl. Phys., Part 2
30
,
L1699
(
1991
).
13.
T.
Onaya
,
T.
Nabatame
,
N.
Sawamoto
,
A.
Ohi
,
N.
Ikeda
,
T.
Nagata
, and
A.
Ogura
,
Microelectron. Eng.
215
,
111013
(
2019
).
14.
J.
Wu
,
Y.
Fang
,
B.
Markman
,
H.-Y.
Tseng
, and
M. J.
Rodwell
,
IEEE Electron Device Lett.
39
,
472
(
2018
).
15.
J.
Wu
,
E.
Lind
,
R.
Timm
,
M.
Hjort
,
A.
Mikkelsen
, and
L.-E.
Wernersson
,
Appl. Phys. Lett.
100
,
132905
(
2012
).
16.
H.
Okamoto
and
T.
Massalski
,
Bull. Alloy Phase Diagrams
4
,
190
(
1983
).
17.
M.
Fukunaga
and
Y.
Noda
,
J. Phys. Soc. Jpn.
77
,
064706
(
2008
).
18.
T.
Schenk
,
E.
Yurchuk
,
S.
Mueller
,
U.
Schroeder
,
S.
Starschich
,
U.
Böttger
, and
T.
Mikolajick
,
Appl. Phys. Rev.
1
,
041103
(
2014
).
19.
T.
Rieger
,
T.
Jörres
,
J.
Vogel
,
A.
Biermanns
,
U.
Pietsch
,
D.
Grützmacher
, and
M.
Lepsa
,
Nanotechnology
25
,
405701
(
2014
).
20.
M.
Hyuk Park
,
H.
Joon Kim
,
Y.
Jin Kim
,
W.
Lee
,
T.
Moon
, and
C.
Seong Hwang
,
Appl. Phys. Lett.
102
,
242905
(
2013
).
21.
S. J.
Kim
,
D.
Narayan
,
J.-G.
Lee
,
J.
Mohan
,
J. S.
Lee
,
J.
Lee
,
H. S.
Kim
,
Y.-C.
Byun
,
A. T.
Lucero
,
C. D.
Young
,
S. R.
Summerfelt
,
S.
Tamer
,
L.
Colombo
, and
J.
Kim
,
Appl. Phys. Lett.
111
,
242901
(
2017
).
22.
M. H.
Park
,
Y. H.
Lee
,
H. J.
Kim
,
Y. J.
Kim
,
T.
Moon
,
K. D.
Kim
,
J.
Mueller
,
A.
Kersch
,
U.
Schroeder
,
T.
Mikolajick
, and
C. S.
Hwang
,
Adv. Mater.
27
,
1811
(
2015
).
23.
S. J.
Kim
,
J.
Mohan
,
S. R.
Summerfelt
, and
J.
Kim
,
J. Miner., Met. Mater. Soc.
71
,
246
(
2019
).
24.
J.
Scott
,
J. Phys.: Condens. Matter
20
,
021001
(
2007
).
25.
K.
Endo
,
K.
Kato
,
M.
Takenaka
, and
S.
Takagi
,
Jpn. J. Appl. Phys., Part 1
58
,
SBBA05
1
(
2019
).
26.
T.
Mittmann
,
F. P.
Fengler
,
C.
Richter
,
M. H.
Park
,
T.
Mikolajick
, and
U.
Schroeder
,
Microelectron. Eng.
178
,
48
(
2017
).
27.
A. S.
Babadi
,
E.
Lind
, and
L.-E.
Wernersson
,
Appl. Phys. Lett.
108
,
132904
(
2016
).

Supplementary Material