low-temperature growth of gallium oxide thin films by

1

Low-temperature growth of gallium oxide thin films

by plasma-enhanced atomic layer deposition

Running title: Ga2O3 layers grown by PEALD

Running Authors: Mahmoodinezhad et al.

Ali Mahmoodinezhad,1 Christoph Janowitz,1 Franziska Naumann,2 Paul Plate,2 Hassan Gargouri,2 Karsten Henkel,1,3,a) Dieter Schmeißer,3 and Jan Ingo Flege1

1Applied Physics and Semiconductor Spectroscopy, Brandenburg University of Technology Cottbus–Senftenberg, K.-Zuse-Str. 1, 03046 Cottbus, Germany 2SENTECH Instruments GmbH, Schwarzschildstraße 2, 12489 Berlin, Germany 3Applied Physics and Sensor Technology, Brandenburg University of Technology Cottbus–Senftenberg, K.-Wachsmann-Allee 17, 03046 Cottbus, Germany

a) Electronic mail: [email protected]

Gallium oxide (Ga2O3) thin films were deposited by plasma-enhanced atomic layer deposition

applying a capacitively coupled plasma source where trimethylgallium (TMGa) as gallium

precursor and oxygen plasma (O2) were used in a substrate temperature (Ts) range of 80 to 200

°C. TMGa exhibits high vapor pressure and therefore facilitates the deposition at lower substrate

temperatures. The Ga2O3 films were characterized by spectroscopic ellipsometry (SE), X-ray

photoelectron spectroscopy (XPS), and capacitance–voltage (CV) measurements. The SE data

show linear thickness evolution with a growth rate of ~0.66 Å per cycle and inhomogeneity of ≤

2% for all samples. The refractive index of the Ga2O3 thin films is 1.86±0.01 (at 632.8 nm) and

independent of temperature whereas the bandgap slightly decreases from 4.68 eV at Ts of 80 °C

to 4.57 eV at 200 °C. XPS analysis revealed ideal stoichiometric gallium to oxygen ratios of 2:3

for the Ga2O3 layers with the lowest carbon contribution of ~10% for the sample prepared at 150

°C. The permittivity of the layers is 9.7±0.2 (at 10 kHz). In addition, fixed and mobile oxide

charge densities of 2 to 4 x 1012 and 1 to 2 x 1012 cm-2, respectively, were observed in the C-V

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

mailto:[email protected]

2

characteristics. Moreover, the Ga2O3 films show breakdown fields in the range of 2.2 to 2.7

MV/cm. The excellent optical and electrical materials properties are maintained even at low

substrate temperatures as low as 80 °C. Hence, the TMGa/O2 PEALD process is suitable for

electronic and optoelectronic applications where low-temperature growth is required.

I. INTRODUCTION

Among the oxide dielectrics, Ga2O3 has gained tremendous interest in recent years due to

its outstanding materials properties, including, e.g., a wide bandgap (~5 eV)1, a high dielectric

constant2, a high break-down electric field (EBD)3, and a high thermodynamic stability4. Thin

films of Ga2O3 are nowadays widely used in numerous applications including high power and

high voltage field-effect transistors5, photovoltaics6, spintronics7, gate dielectric, and passivation

layers in III–V semiconductor-based devices to reduce the leakage current8-10. Owing to the

possibility to control the electrical conductivity between insulating and semiconducting

(typically n-type) behavior and because of its high optical transparency (~80%), Ga2O3 is a

promising candidate for transparent conducting oxides and thin film transistors11. In addition,

Ga2O3 exhibits gas dependent conducting behavior at elevated temperatures, making it one of the

most applicable materials for gas sensing12. To date, various approaches have been employed for

the deposition of Ga2O3 thin films such as chemical vapor deposition (CVD)7, molecular beam

epitaxy13, magnetron sputtering14, pulsed laser deposition4, and atomic layer deposition

(ALD)6,9,10,15-18.

Within the aforementioned deposition techniques, ALD with its inherent self-saturation

scenario is one of the most promising methods to get precise thickness control, excellent step

coverage, high conformality, and uniformity for the preparation of ultra-thin layers. The use of

plasma-enhanced ALD (PEALD) facilitates low-temperature deposition of high-quality films by

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

3

increasing the rate of the (desired) chemical reaction while decreasing the interlayer diffusion

rate, thereby effectively broadening the ALD temperature window19. To date, different Ga

precursors were reported for the deposition of Ga2O3 in ALD and CVD processes, including

trimethylgallium (TMGa, [Ga(CH3)3])17, dimethylgallium isopropoxide9, and gallium(III)

isopropoxide (GTIP)10 in thermal ALD, TMGa20, [(CH3)2GaNH2]36, triethylgallium21, and

tris(2,2,6,6-tetramethyl-3,5-heptanedionato)gallium(III)22 in PEALD, and gallium

trishexafluoroacetylacetonate23, gallium(III) chloride24 in CVD. It should be noted that most of

these precursors could only be used at high temperature (> 300 °C) typically resulting in

undesired impurities within the films such as carbon and hydrogen, whose concentrations

strongly depend on the growth temperature. However, TMGa could be employed as a gallium

source at lower temperatures because it has a high vapor pressure and is highly volatile25, as

compared to other Ga precursors that require higher growth temperatures to obtain an

appropriate vapor pressure. Hence, a short pulse of TMGa could be adequate to quickly saturate

the surface reaction. Despite this obvious advantage, there are only a few studies about the

PEALD growth of Ga2O3 films using TMGa and O2 plasma20,26,27. Establishing this procedure

may thus pave the way for applications where low process temperatures are required. Also, from

a chemistry point of view, the similarity of TMGa to trimethylaluminum, which is commonly

used as Al source for the ALD of Al2O3 films28, may render this system a reference model for the

ALD of Ga2O3 films.

In this work, high-quality 10 and 30 nanometer thick gallium oxide layers were deposited

in the temperature window of 80 to 200 °C by PEALD where a capacitively coupled plasma

(CCP) source was applied. TMGa and oxygen plasma (O2) were used as gallium precursor and

oxidant co-reactant, respectively. The optical and electrical properties as well as the chemical

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

4

composition of the Ga2O3 thin films were determined. The PEALD process delivered a constant

growth rate in the investigated substrate temperature range; the grown thin Ga2O3 films exhibited

an ideal gallium to oxygen (Ga:O) ratio with lowest carbon contamination for films prepared at

150 °C. Furthermore, the films show excellent optical and electrical properties exhibiting values

close to the ones of bulk crystalline material.

II. EXPERIMENTAL

A. Fabrication

The Ga2O3 thin films were deposited on 4″ n-type Si (100)/1.5 nm SiO2 substrates by

PEALD with alternating pulses of TMGa as Ga precursor and O2 plasma as co-reacting oxidant.

The deposition process was carried out at four different substrate temperatures (80, 100, 150 and

200 °C) in the SI ALD reactor (SENTECH Instruments GmbH)28 equipped with a laser

ellipsometer for in-situ ALD real-time monitoring (SENTECH, ALD Real Time Monitor). The

layers are expected to be amorphous in the investigated substrate temperature range.

High-purity N2 (99.999 %) was used as carrier gas of TMGa with a flow rate of 90 sccm.

During the plasma step the precursor line was closed and the nitrogen flow stopped. In this study,

the plasma was generated by a true remote CCP source in which the substrate is located apart

from the plasma generation area and caused neither high energy ions nor UV-light, which might

result in substrate surface damage. The employed CCP source, driven by a single radio frequency

of 13.56 MHz, is mounted to the upper flange of the reactor28,29; the plasma power and O2 (high

purity of 99.998 %) flow rate were set to 200 W and 150 sccm. One ALD cycle includes a 10 ms

pulse of TMGa followed by a purge pulse of 2 s, an oxygen plasma pulse of 5 s, and a 2 s purge

step. Figure 1 schematically shows the used PEALD pulse sequence.

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

5

The samples are labeled as GaOz°C, where ‘z’ correspond to the substrate temperature

applied and GaO represents ‘gallium oxide’ without taking into account the elemental

composition.

FIG. 1. (Color online) Cycle sequence for the PEALD of Ga2O3 thin films. Each PEALD cycle

consisted of four sub-pulses with 10 ms TMGa vapor pulse (90 sccm N2 carrier gas), 2 s N2

purge pulse, 5 s (150 sccm, 200 W) O2 plasma and 2 s N2 purge pulse. The periodic deposition

cycle was repeated until the desired layer thickness was realized.

B. Characterization

The film thicknesses of the PEALD-Ga2O3 thin films were controlled using in-situ ALD-

RTM. This system allowed monitoring the short ALD cycles in the order of a few seconds with

high temporal resolution (20 ms) and a very high signal-to-noise ratio (δΔ = 0.01°,

Δ ellipsometric angle) at a single wavelength of 632.8 nm with incidence angle of 70°. Here, the

sample architecture was represented by a single, homogeneous film within a one-layer model.

The optical properties of the Ga2O3 thin films were investigated by spectroscopic ellipsometry

(SE) in different spectral ranges. UV/VIS/NIR measurements were performed with a SENTECH

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

6

SER 850 in a wavelength range from 200 to 1000 nm at two incidence angles of 65° and 70°,

respectively. A Tauc-Lorentz oscillator was used for modeling and fitting the SE data to obtain

the film thickness, growth rate per cycle (GPC), refractive index, and optical bandgap of the

Ga2O3 layers. Furthermore, a 1.5 nm thick native oxide layer (SiO2) at the interface to the Si

substrate was considered in the analysis.

The chemical compositions and bonding states of the Ga2O3 films were determined by X-

ray photoelectron spectroscopy (XPS) using a hemispherical energy analyzer (Omicron EA 125)

with a dual anode X-ray source delivering Al Kα (1486.6 eV) as well as Mg Kα (1253.6 eV)

excitation. Prior to chemical analysis of the grown layers, the spectrometer transmission factor of

the analyzer was determined by investigating a Ga2O3 single crystal, cleaved under ultra-high

vacuum (UHV), in two different operation modes of the analyzer (constant analyzer energy

(CAE) and constant retarding ratio (CRR)) in accordance with Ref. 30. The photoelectrons were

collected at a take-off angle of 90° with regard to the sample surface plane to maximize the

sensitivity to the deeper layers while minimizing the contribution of surface contamination that is

inevitably present for ex-situ prepared samples. Nevertheless, to remove ambient adsorbates the

sample surface was sputtered by inert gas sputtering (Ar+, 2 keV, 4×10-6 mbar) applying a cold

cathode ion beam source (ISE 5, Scienta Omicron). The survey scan and high-resolution spectra

were collected with 50 and 20 eV pass energies with energy steps of 0.5 and 0.25 eV,

respectively. The obtained spectra were analyzed using CasaXPS analysis software to calculate

the atomic concentrations and gallium to oxygen (Ga:O) ratio by weighting the fitted peak areas

taking into account the respective atomic photoionization cross sections31 and the transmission

function of the analyzer30. In the peak fitting procedure a Gaussian/Lorentzian line shape, a

Shirley background subtraction, and a full width at half maximum (FWHM) of up to 1.7 eV were

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

7

considered. It should be noted that in the high-resolution scans binding energy differences (ΔE ~

0.3±0.2 eV) for the core level signals (Ga3d, O1s and C1s) were observed pointing to a charging

effect that was reconsidered in the peak decomposition procedure. In order to calculate the

elemental compositions and Ga:O ratio before and after sputtering, the high-resolution core level

spectra (Ga3d, O1s, C1s) were collected using both Al Kα and Mg Kα excitation. To determine

the carbon contribution, Al Kα was used due to the fact that for this excitation neither Auger nor

satellite lines are perturbing the C1s main signal. On the contrary, Mg Kα was employed to

ascertain the Ga:O ratio because at this excitation the O1s core level region exhibits a single peak

without contributions of Auger or satellite lines. For the quantitative analysis, the Ga3d region at

lower binding energy was selected in order to collect electrons of higher kinetic energy

compared to the Ga2p region at higher binding energy, as the lower kinetic electrons of the Ga2p

levels deliver more surface sensitive information, which will still be more influenced by the

presence of surface contamination.

For the electrical measurements, metal–insulator–semiconductor (MIS) stacks were

prepared where the circular aluminum contacts (variable diameter between 300 and 800 µm)

were thermally evaporated through shadow masks on top of the Ga2O3/1.5nmSiO2/Si stack. The

stack sequence is illustrated in the lower inset of Fig. 10. For these measurements, thicker Ga2O3

layers of ~30 nm were employed to minimize the influence of probable leakage currents.

Capacitance-Voltage (C-V) measurements on these MIS structures were carried out with a LCR

meter Agilent 4284A.32 During one C-V loop the stack was biased from the inversion regime of

the semiconductor towards the accumulation counterpart and subsequently vice versa applying a

slow DC voltage ramp of 25 mV/s. The differential capacitance was recorded with a superposed

AC signal of 25 mV at frequencies of 1 MHz, 100 kHz, and 10 kHz. The C-V measurements

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

8

were performed at different positions of the samples with varying contact pad areas. The

breakdown electrical field was determined from current-voltage measurements applying a

combination of Agilent E3649A power supply, HP34401 and PREMA4001 volt- and ammeters,

respectively.32

III. RESULTS AND DISCUSSION

A. Optical and spectroscopic characterization

Figure 2 shows the in-situ measurement of film thickness as a function of elapsed process

time for different substrate temperatures. The film thickness was obtained by modeling the

measured ellipsometric data with a one-layer model with a fixed refractive index of 1.86 (at

632.8 nm). A plot of the ellipsometric parameter delta (°) can be found in the supporting

information (Fig. S1). As typical for ALD, an increment and a reduction trend were observed in

the film thickness due to precursor adsorption and ligand removal. The measurement shows film

growth from the first cycle onwards where the nucleation behavior for all four samples is very

similar and linear film growth was achieved for all applied temperatures.

During opening the ALD line prior to the actual precursor pulse, the film thickness

increased. As soon as the precursor was dosed into the reactor, a second trend of increasing

thickness is witnessed. Remarkably, this double adsorption process was not observed during the

first cycle of the deposition of the films and occurs independent of substrate temperature. We

assume that due to its high volatility a certain amount of the precursor remained in the ALD line

after closing. When the ALD line was reopened and the flow of N2 carrier gas was restarted,

excess material was accidentally introduced into the reactor. Since the ALD lines were evacuated

before initiating the deposition process, this double adsorption does not take place during the

first cycle.

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

9

FIG. 2. (Color online) Initial Ga2O3 film thickness evolution as a function of the process time for

different substrate temperatures within the PEALD process. The thickness was obtained by

modeling the in-situ ellipsometer data with a one-layer model with a constant refractive index of

1.86 (at 632.8 nm).

When investigating the plasma exposure step, upon plasma pulsing we observed an

abrupt reduction in film thickness; afterwards, saturation was attained within a few seconds.

During continuous plasma exposure the layer thickness remained constant, confirming that

neither undesirable film growth nor etching happened during this step.

Figure 3 illustrates the linear increase of the film thickness during PEALD process time

for all applied substrate temperatures. While almost no difference in the growth rate is visible

within the first 2000 s (film thickness of 10 nm), the slopes of the curves progressively deviate

with time up to a final thickness of 30 nm. This deviation results in a decrease of the GPC from

0.00

0.25

0.50

0.75

•

••

•

••

•

•

••

••

••

••

•

•

openning/closing of ALD line

0.00

0.25

0.50

0.75

•

••

•

••

••

•

•

•

•

•

•

•

•

•

•

TMGa (•)

Purge

Plasma

Purge

0.00

0.25

0.50

0.75

Th

ickn

ess [

nm

]

0 20 40 60 80 100

0.00

0.25

0.50

0.75

Time [s]

GaO200°C

GaO100°C

GaO150°C

GaO80°C

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

10

0.70 Å/cycle (80 °C) to 0.63 Å/cycle (200 °C) as depicted in Fig. 4a. This phenomenon can be

explained by the reaction of the terminating hydroxyl groups forming a bridging oxygen

species33 as well as causing a decrease in film thickness and changes in nucleation behavior. This

well-known effect has been already reported for other materials systems like TiO2 (Ref. 34,35).

The achieved GPC (~0.66 Å/Cycle) for our Ga2O3 thin films is higher than that obtained for

other films deposited by the same PEALD process20.

FIG. 3. (Color online) Measured in-situ Ga2O3 film thickness versus prolonged process time up to

a total thickness of about 30 nm.

The optical properties of the as-deposited 30 nm Ga2O3 films were investigated by

considering the influence of various substrate temperatures. The GPC, refractive index,

thickness, inhomogeneity across the 4″ Si wafer (mapped at 69 positions) and the bandgap of the

layers were determined using the Tauc-Lorentz model to fit the SE data.

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

11

The used process parameters lead to similar and very low inhomogeneity values in the

range of 1.5 to 2 % for all substrate temperatures as shown in Fig 4b. These low values indicate a

very good process control over large substrates (4″).

FIG. 4. (Color online) Growth rate (a) and inhomogeneity across a 4″ Si wafer (b) of the Ga2O3

layers depending on substrate temperature.

Fig. 5 shows the refractive index and the bandgap as a function of substrate temperature.

The refractive index and extinction coefficient in the full wavelength region are shown in the

supporting information (Fig. S2). The refractive index of the grown amorphous Ga2O3 films is

relatively constant (1.86±0.01 at 632.8 nm) in the investigated substrate temperature range. The

obtained refractive index value is in good agreement with other reported values using the same

PEALD process20,27 and is higher than values that have been achieved for thin films prepared by

other deposition methods, where values between ~1.6 and 1.8 (at a comparable wavelength) have

been reported36-38. The amorphous nature of as-deposited Ga2O3 layers might be one of the main

reasons why the refractive index remains nearly constant with increasing growth temperature.

80 100 120 140 160 180 2000

2

4

Inhom

og. [%

]

Substrate Temperature [°C]

0.60

0.65

0.70

GP

C [Å

/cycl.]

10 nm

30 nm

(b)

(a)

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

12

The refractive index is dependent on the densification of material39,40; therefore, amorphous

films with low-density might be prone to air trapping in the layer’s pores leading to an effective

reduction of the refractive index (when compared to post-annealed samples reported in

literature6). However, in the investigated substrate temperature range we do not expect any

significant change in the amorphous nature of our samples, consequently resulting in an almost

constant refractive index.

The bandgap slightly decreases with increasing substrate temperature. Taking into

account the error bars of the measurement, the deviation is, however, in the range of ~100 meV.

In this respect, our finding is in agreement with data of Choi et al.10, where no significant

changes of the bandgap in the temperature range of 150 to 250 °C were observed for layers

prepared by thermal ALD using GTIP and water. The obtained optical bandgap of 4.63±0.05 eV

is slightly lower than other values (4.8eV at Ts =200°C) reported in literature for the same

PEALD process26. This fact could be attributed to the amorphous phase of the as-deposited

Ga2O3 films.

A summary of the obtained optical properties can be found in Table I.

In order to avoid charging effects during the XPS measurements, Ga2O3 film of 10 nm

film thickness were prepared using the same process parameters. The achieved growth rates and

inhomogeneities for these samples can be found in the supporting information (Table S1). We

assume that the optical properties obtained from the 30 nm thick films translate to their thinner

counterparts.

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

13

FIG. 5. (Color online) SE model results of the optical band gap and refractive index as a function

of deposition temperature for gallium oxide samples. The measurements were carried out ex-situ

on 30 nm thick samples.

TABLE I. Summary of main determined properties of the Ga2O3 layers characterized in this study.

The tabulated parameters are substrate temperature (Ts), GPC, inhomogeneity (inhomog.) across

4″ Si wafers, refractive index (n), optical energy gap (Eg), Ga:O ratio, C1s concentration,

permittivity (k) at 10 kHz, negative fixed charges (Nfix), and electrical breakdown field (EBD).

The Ga:O ratio and C1s contribution were determined from 10 nm thick films whereas all other

properties were obtained from 30 nm thick layers.

Samples Ts

[°C]

GPC

[Å/cycle]

inhomog.

[%]

n

[@632.8nm]

Eg

[eV]

Ga:O C1s

[%]

k

[@10kHz]

Nfix

[1012cm-2]

EBD

[MV/cm]

GaO80°C

GaO100°C

GaO150°C

GaO200°C

80

100

150

200

0.70

0.67

0.66

0.63

1.6

1.8

1.6

2.0

1.85±0.005

1.85±0.003

1.86±0.004

1.87±0.006

4.68±0.02

4.66±0.03

4.61±0.02

4.57±0.02

0.67

0.65

0.66

0.67

12.2

12.6

10.6

15.9

9.8±0.1

9.4±0.2

9.6±0.1

9.9±0.2

-3.8±0.1

-3.1±0.2

-2.1±0.1

-2.2±0.1

2.2±0.3

2.4±0.2

2.7±0.3

2.4±0.3

80 100 120 140 160 180 2004.0

4.2

4.4

4.6

4.8

Ba

nd

ga

p [

eV

]

Substrate temperature [°C]

t = 31.5 0.4 nm

1.82

1.83

1.84

1.85

1.86

1.87

1.88

1.89

1.90

1.91

1.92

n @

63

2.8

nm

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

14

The atomic composition of the Ga2O3 thin films was determined using XPS. Figure 6

shows the survey spectra of the as-deposited samples. Obviously, the spectra contain the

expected photoemission lines of Ga (Ga2p, Ga3s, Ga3p, Ga3d, GaLMM Auger) and oxygen

(O1s, O2s, OKLL Auger), but also of carbon (C1s, CKVV Auger)9,23,38,41-43.

FIG. 6. (Color online) XPS survey spectra of the as-deposited Ga2O3-PEALD thin films prepared

at different substrate temperatures as indicated. The spectra were recorded with Mg Kα

excitation.

The C1s signals imply that carbon was adsorbed on the sample surface during their

transfer for ex-situ XPS analysis. Therefore, all as-deposited samples were exposed to Ar+ ion

beam sputtering for one minute to remove the surface contamination. The Ga:O ratio deduced

from Ga3d and O 1s core level data as well as the carbon concentrations (C1s data) before and

after sputtering are depicted versus the PEALD substrate temperature in Fig. 7; the values of the

sputtered samples are additionally given in Table I. Remarkably, the Ga:O ratio is very close to

the ideal value of 2:3 for all investigated samples. In other works using the same PEALD process

higher deviations (0.70 and 0.77 for as-deposited and annealed (900°C) films) from that ideal

1000 800 600 400 200 0

GaO200°C

Mg K

GaO150°C

C K

VVIn

ten

sity [

arb

. u

nits]

Binding energy [eV]G

a 2

p1/2

Ga 2

p3/2

O K

LL

O 1

s

Ga L

MM

C 1

s

Ga

3s

Ga

3p

Ga 3

d,

O 2

s

GaO80°C

GaO100°C

Survey Spectra Ga2O

3

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

15

value were observed.27 Obviously, the carbon concentration is significantly decreased after

sputtering, where it is nearly constant (C1s ~ 12.4±0.2%) within the Ts range of 80 to 100 °C.

The lowest and highest values are found at Ts = 150 °C (C1s ~ 10.6±0.1%) and Ts = 200 °C (C1s

= ~ 15.9±0.1%), respectively.

FIG. 7. (Color online) Gallium to oxygen ratio (a) and carbon contamination (b) of the Ga2O3

films vs. substrate temperature.

Figure 8 shows the decomposition of the Ga3d spectra into four sub-peaks after

calibrating the binding energy (BE) of the main (Ga-O) component to 20.5 eV. The appearance

of the highly intense peak confirms the presence of Ga-O bonds, i.e., the formation of a Ga2O3

thin film20,21. A weak peak at higher BE (24.2±0.1 eV) is identified as the O2s core level42,44.

Therefore, this O2s contribution might slightly impact the Ga3d raw area leading to an

overestimation of the Ga atomic concentration, however we expect that this error is small (≤2%).

The two further sub-peaks on the higher and lower BE shoulders of the Ga-O peak (22.5±0.1 eV

and 19±0.1 eV) are attributed to carbonates and Ga-OH species44. The relative intensities of the

0.60

0.65

0.70

0.75

(b)

Al K

Ga3d

/O1s r

atio

ideal

1 min sputtering

as-deposited

Mg K

(a)

80 100 120 140 160 180 2009

12

15

18

21

C1s [%

]


This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

16

different Ga3d components are listed in Table II and remain essentially constant with substrate

temperature; just the hydroxide contribution is slightly higher at Ts = 80 °C and is consistently

getting smaller with increasing substrate temperature, possibly owing to the lower thermal

stability of the hydroxide species.

FIG. 8. (Color online) Ga 3d core level XPS data (Mg Kα) of Ga2O3 layers grown at substrate

temperatures of (a) 80 °C, (b) 100 °C, (c) 150 °C, and (d) 200 °C and their decomposition.

Analogously, the O1s core level XPS spectra were fitted into four discrete peaks (Fig. 9)

upon calibrating the BE to 531 eV. This sharp peak corresponds to Ga-O bonding similar to the

Ga3d spectra20,21,38,45,46. Two weak peaks at 532.8±0.1 eV and 529.5±0.2 eV) are attributed to O-

H bonds and O1s satellite peak, respectively. Furthermore, the peak located at 531.6±0.1 eV

points to oxygen bonded with carbon (O-C)43,47,48. The contributions of these sub-peaks in the

0

1

1min sputtering

0

1

0

1

No

rma

lize

d in

ten

sity

26 24 22 20 18

0

1GaO200°C

GaO150°C

GaO100°C

Mg K

Binding energy [eV]

Ga-OHO 2s

d)

c)

b)

a)

data

⎯⎯ fit

carbonates

Ga 3d

GaO80°CGa−O

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

17

O1s core level are also given in Table II. Similarly to the Ga3d decomposition data, with

increasing substrate temperature the contributions of both carbonyl (oxygen bonded to carbon)

and hydroxyl (oxygen bonded to hydrogen) groups tend to decrease except for Ts = 100 °C where

particularly a strong increase of the carbonyl contribution is observed. It was found that the film

grown at 150 °C exhibits the lowest total carbon (Table I) and carbonyl concentrations in the

O1s region (Table II) as well as it is in the lower carbonate concentration range as derived from

the Ga3d data (Table II) among all our samples.

FIG. 9. (Color online) O1s core level XPS data of Ga2O3 layers deposited at substrate


0

1

0

1

0

1

O (satellite)

No

rma

lize

d in

ten

sity

Mg K

Ga−O

536 534 532 530 528

0

1 GaO200°C

GaO150°C

GaO100°C

Binding energy [eV]

O−CO−H

1min sputtering

d)

c)

b)

a)

data

⎯⎯ fit

O 1s

GaO80°C

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

18

TABLE II. Relative sub-peak intensities in the Ga3d and O1s core level XPS spectra of Ga2O3

PEALD films.

Ga3d O1s

Samples carbonates% Ga-O% Ga-OH% O-H% O-C% Ga-O%

GaO80°C

GaO100°C

GaO150°C

GaO200°C

6.3

5.7

5.8

5.7

76.9

79

78.8

79.7

12.6

11.4

11.3

10.9

5.7

7.8

7.2

6.1

21.8

32.2

16.5

19.1

68.1

55

72.6

70.7

B. C-V characterization

In this section the C-V data of the Ga2O3 PEALD samples are compared. Figure 10

exemplarily depicts the C-V measurements at 10 kHz for the GaO80°C sample for different

contact areas.

FIG. 10. (Color online) C-V measurements of the sample GaO80°C at different contact pad

diameters () as listed beside each curve. The measurement direction is indicated by the arrows.

The lower inset illustrates the MIS stack sequence; the upper inset shows one C-V loop at 1 MHz

( = 500 µm).

0 1 2 3 4

0.0

0.2

0.4

0.6

0.8

1.0

n-Si

1.5 nm SiO2

30 nm Ga2O3

Al2

300

400

500

600

GaO80°C

10 kHz

Ca

pa

cita

nce

[n

F]

Voltage [V]

[µm]

700

1

0 2 4

0

11 MHz

1

2

No

rm.

ca

p.

Voltage [V]

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

19

The C-V data follow the typical behavior for an n-type semiconductor. Also, the expected

scaling of the accumulation capacitance Cacc with the device dimension (contact area) is visible.

However, we also find positive flat-band voltage (VFB) shifts as well as hysteresis effects within

the measurements. The latter is significantly present in the first-time measurement on each

contact, which was carried out at 1 MHz; an example is given in the upper inset of Fig. 10. This

phenomenon will be discussed below.

To determine the permittivity of the Ga2O3 layers, the accumulation capacitances of the

C-V loops determined for different contact areas (A) were normalized and plotted against A (Fig.

11). For this procedure, we chose the C-V data recorded at 10 to 100 kHz to avoid influences of

series resistances on Cacc at higher frequencies. Assuming a combination of two capacitances

(representing the Ga2O3 film and the interfacial SiO2) connected in series, the following equation

can be developed:

A

kd

d

k

dC

SiOOGa

SiO

OGa

OGaacc*

11

1*

232

2

32

32

0 +

=

(1)

where dGa2O3, kGa2O3, dSiO2 (1.5 nm, as modeled in the ellipsometric data – see above), and kSiO2

(3.9) denote the thickness and relative permittivity of the Ga2O3 and the interfacial SiO2 layers,

respectively, and ε0 the vacuum permittivity (cf. Refs. 28, 49 for further details).

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

20

FIG. 11. (Color online) Normalized Cacc against contact area according to Eq. (1) for the samples

GaO80°C, GaO100°C, GaO150°C, and GaO200°C as indicated in the legend. The linear fits used for the

determination of the permittivity are drawn with the corresponding colors. The data are shown

for C-V measurements recorded at 10 kHz.

In Fig. 11, the 10 kHz data are plotted following Eq. (1). Subsequently, these data were

linearly fitted in accordance with Eq. (1); the permittivity kGa2O3 was then deduced from the

slope of the linear fit.

In the investigated substrate temperature range we found just a slight variation of the

permittivity with values of 9.6±0.3 in the frequency range of 10 to 100 kHz (Fig. 12a). Only for

the GaO100°C sample the permittivity is noticeably lower than for the other three samples.

Between 100 °C and 200 °C a slight increase in permittivity with temperature is identified,

consistent with the optical data, where a slight increase of the refractive index with temperature

was observed (see above). Our Ga2O3 layers show similar permittivity values as observed for

thermal ALD and PEALD layers prepared with GTIP and water10 or [(CH3)2GaNH2]3 and

oxygen plasma6. For annealed (700-900°C), crystalline samples fabricated also in the TMGa/O2

plasma process much lower values were reported so far (5 to 5.4 at 1 MHz)27.

0.0 0.1 0.2 0.3 0.40.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

GaO80°C

GaO100°C

GaO150°C

GaO200°C

Contact area [mm2]

Cacc*d

Ga

2O

3

/0 [

mm

2]

10kHzTh

is is

the au

thor’s

peer

revie

wed,

acce

pted m

anus

cript.

How

ever

, the o

nline

versi

on of

reco

rd w

ill be

diffe

rent

from

this v

ersio

n onc

e it h

as be

en co

pyed

ited a

nd ty

pese

t.PL

EASE

CIT

E TH

IS A

RTIC

LE A

S DO

I: 10

.1116

/1.51

3480

0

21

FIG. 12. (Color online) Permittivity (a) at 10 (red open circles) and 100 kHz (blue filled squares)

as well as (b) negative fixed (green open squares) and mobile (black filled triangles) charges

(deduced from C-V measurements at 1 MHz) against substrate temperature of the PEALD

processed Ga2O3 samples.

Recently, Fiedler et al. investigated the static dielectric constant of β-Ga2O3 in

dependence of the principal planes and found an average value over all orientations of

11.2±0.2.50 The authors argued that lower values are observed for amorphous or polycrystalline

80 100 120 140 160 180 2000

1

2

3

4

b)

neg. fixed

mobile

Ch

arg

es [

10

12 c

m-2]

8

10

12

100kHz

10kHz


Pe

rmittivity

a)

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

22

layers since the crystalline network is more densely packed than the amorphous phase.50

According to the Clausius-Mossotti equation51, this densification is accompanied by a higher

relative dielectric constant for the crystalline phase. However, we expect our layers, which were

prepared at temperatures of 200 °C and below, to be amorphous as also confirmed by Ref. 20.

The abovementioned flat-band voltage shifts and hysteresis effects within the C-V loops

are observed for all investigated samples. Thereby, the first-time measurement on every contact

pad performed at 1 MHz showed the highest hysteresis, afterwards it was reduced as can be seen

when comparing the data in the main panel (mostly third measurement - at 10 kHz) and upper

inset (first measurement) in Fig. 10. This reduction of the hysteresis is not due to the change of

the measurement frequency as it was proven by repeated measurements at 1 MHz on selected

contacts (supporting information Fig. S3), where a reduction of the hysteresis was observed as

well, when the first-time and subsequent C-V loops were compared at this frequency. However,

this hysteresis reduction is accompanied by a further positive flat-band voltage shift in such a

way that the subsequent C-V loop is located close to the reverse branch of the first-time C-V

loop (labeled “2” in the upper inset of Fig. 10). Afterwards, i.e., in further subsequent C-V loops,

the C-V characteristics remained constant.

Typically, shifts of positive flat-band voltage point to negative fixed charges inside the

insulator.52 Similar behavior was observed for Ga2O3 layers prepared in PEALD processes using

TMGa27 or [(CH3)2GaNH2]3 (Ref. 53) and oxygen plasma. Liu et al. speculated about oxygen

vacancy formation due to a SiO2 interfacial layer between Si and Ga2O3, finally causing extra

negative charges.53 We have recently discussed the electronic structure of Ga2O3 containing

intrinsic defect states, namely polaronic, excitonic and charge transfer states, formed by mixed

atomic valence states.54 In this picture we developed a model where the excitonic states are states

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

23

that are “easy to be filled/refilled” as they are located close to the Fermi level; conversely, charge

transfer states are highly localized and related to fixed charges.55 Initially, the charge transfer

states and hence the fixed negative charges are created by the PEALD process (it should be noted

that these intrinsic electronic defect states are also existent in single crystals; however, their

abundance depends on the preparation conditions54,55). During the first C-V loop, applying a

positive voltage drives the semiconductor into accumulation. Hence, a large reservoir of

electrons is available at the Ga2O3/Si interface that will be injected into the insulator and trapped

in the charge transfer states leading to an increased positive flat-band voltage position in the C-V

reversed scan (labeled by “2” in Fig. 10). Subsequently applying a negative voltage inside the C-

V loop is not sufficient to depopulate these states again. Therefore, in the next C-V run these

states cannot be filled again, resulting in the location of the C-V loop close to the branch labeled

with “2” in Fig. 10. Ideally, after completion of the charge transfer process the C-V loop should

exhibit no hysteresis, but still, electrons can easily be trapped in and released from the excitonic

defect states since they are located close to the Fermi level, causing the remaining C-V hysteresis

as displayed in the main panel of Fig. 10 as well as in the supporting information (Fig. S3).

To determine the concentration of initially fixed (Nfix) and ‘hysteretic’ (Nhyst) charges, the

first-time C-V measurements at 1 MHz were analyzed. The ideal work function difference of the

metal electrode and the semiconductor substrate ΦMS (-0.1 eV) was calculated assuming the

aluminum work function and silicon electron affinity of 4.2 eV56 and 4.05 eV57, and determining

the semiconductor Fermi level position (0.245 eV below the conduction band minimum) from

the substrate doping (2.8±1.5 × 1015 cm-3, i.e., 1 to 3.5 Ωcm) calculated via the depletion

approximation57 for the 1 MHz C-V data. The MIS flat-band capacitance was calculated using This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

24

the semiconductor substrate doping level57 and Cacc; this value was then used to determine the

flat-band voltage in the 1 MHz C-V data. Finally, Nfix and Nhyst were deduced via

Nfix = (ΦMS / e-VFB) * Cacc / A (2a)

Nhyst = ΔVFB * Cacc / A, (2b)

respectively, where e is the elementary charge.

Fig. 12b shows the resulting Nfix and Nhyst values depending on PEALD substrate

temperature. Nhyst is relatively constant (~ 1×1012 cm-2) with substrate temperature, just for the

GaO100°C sample it is higher (1.9 × 1012 cm-2). Generally, Nfix shows a trend to diminish with

increasing substrate temperature. At 80 °C, it amounts to about 4 × 1012 cm-2; at 150 °C, it

exhibits a minimum of about 2 × 1012 cm-2. The latter value is similar to what has been observed

in other works using the same PEALD process, but following post deposition annealing.27

Combining both XPS and electrical data, we may draw the following two conclusions on

how the specific choice of PEALD process parameters influences the resulting materials

properties. Firstly, at lower Ts (80 to 100 °C) the layers have a higher number of fixed and/or

hysteretic charges than at higher Ts (150 to 200 °C). Comparing the carbonate and carbonyl

contributions in the Ga3d and O1s XPS data (Table II), respectively, these amounts are also

higher in sum at lower Ts than at higher Ts. Hence, we speculate that there exists a correlation

between carbon within the layers and the fixed and mobile charges, suggesting that carbon might

act as a trap center. However, further investigations such as temperature-dependent current-

voltage and deep-level transient spectroscopy measurements are needed to clarify this issue,

which are beyond the scope of the current work. Particularly, the mentioned correlation holds for

the GaO150°C film, which exhibits the lowest total carbon amount (Table I) and the lowest

carbonyl contribution in the O1s data (Table II) as well as the lowest Nfix (Table I). In case Nfix

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

25

and Nhyst should be minimized for the desired application of the Ga2O3 layer, the higher Ts

temperature range (150 to 200 °C) should be chosen. Conversely, a post deposition annealing

step might be reasonable for the reduction of Nfix and Nhyst. This, however, is not applicable when

low process temperatures are required. Recently, we have demonstrated for aluminum nitride

PEALD layers that the specific choice of plasma source might also be an option to influence Nhyst

and Nfix, where we obtained reduced values when using an inductively coupled plasma source

instead of a CCP source (which was also applied in the current work).58 This finding implies that

further tuning of the electrical materials properties of the Ga2O3 films should readily be feasible.

Secondly, the GaO100°C sample with the highest Nhyst exhibits the lowest permittivity

value. Furthermore, this sample shows the highest (albeit small) deviation of the Ga:O ratio from

the ideal value (oxygen-rich, see Fig. 7), which might be caused by the higher concentration of

O–C and O–H bonds in the O1s data of this sample (Fig. 9b and Table II). Therefore, the

polarizability and hence the permittivity (according to the Clausius-Mossotti equation50) of the

material seem to be affected by the concentration of injected charges as well as by changes in the

concentration of local dipoles and their polarizabilty.

Furthermore, the electrical breakdown field EBD was determined; the achieved values are

also given in Table I. Here, values between 2.2 and 2.7 MV/cm were identified. The theoretical

breakdown field for β–Ga2O3 is expected to be 8 MV/cm4,5; however, the typically achieved

experimental values are far below this value11. The breakdown electric field values of our

amorphous layers are similar to those reported for electron beam evaporated59 and

photoelectrochemically deposited60 thin films, where values of 2.1 and 2.8 MV/cm were

obtained, respectively. As-grown PEALD layers prepared with [(CH3)2GaNH2]3 and O2 plasma

have shown very leaky behavior; high-temperature (Ts ≥ 700 °C) rapid thermal post-deposition

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

26

annealing was necessary to decrease the leakage current significantly, but the breakdown field

was only about 1 MV/cm.6 Therefore, a further improvement in the insulating behavior of our

layers may be expected when a PDA step is compatible with the desired process conditions. It

should be noted that higher breakdown fields have recently been observed for amorphous layers

prepared by thermal ALD (6.5 to 7.6 MV/cm at Ts between 150 and 250 °C using GTIP and

water)10 as well as by PEALD (6.8 MV/cm at Ts = 200°C and plasma power of 200 W) applying

TMGa and an O2 plasma.26

IV. SUMMARY AND CONCLUSIONS

This paper documents the optical and electrical properties as well as the chemical

composition of amorphous Ga2O3 thin films deposited by PEALD on Si(100) substrates with

TMGa and O2 plasma in the Ts range of 80 to 200 °C. From ellipsometry a constant growth rate

of ~0.66 Å/cycle is inferred; the refractive index and optical bandgap amount to 1.86±0.01 and

4.63±0.05 eV, respectively. Further, the SE data confirm that the chosen substrate temperature

range lies within the ALD window. The XPS results proved that all samples have an almost ideal

gallium to oxygen ratio of 2:3. The lowest carbon contamination (~10%) is observed at Ts = 150

°C. The high-resolution Ga3d and O1s core level spectra corroborated the overwhelming

presence of Ga-O bonds in the films, whose fraction increases with increasing temperature while

the concentration of carbonates, carbonyl, and hydroxide species decreases. The C-V data

revealed a dielectric constant of 9.7±0.2 at frequency of 10 kHz for the Ga2O3 films. Moreover,

fixed and mobile oxide charge densities of 2 to 4 x 1012 and 1 to 2 x 1012 cm-2 were found. In

addition, the breakdown electric field lies within the range of 2.2 to 2.7 MV/cm, with the highest

value obtained at 150 °C. Summarizing all data, the sample prepared at Ts = 150 °C carries the

best performance. Remarkably, their excellent optical and electrical materials properties do not

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

27

significantly change even at substrate temperatures as low as 80 °C, rendering our Ga2O3 thin

films very well suited for electronic and optoelectronic applications where low-temperature

growth is required.

ACKNOWLEDGMENTS

This work is financially supported by the Federal Ministry for Economic Affairs and Energy

(BMWi) of Germany within the ZIM program (ZF4510602AG7, ZF4245503AG7). A.M.

acknowledges the funding by the Graduate Research School (GRS) of the Brandenburg

University of Technology Cottbus–Senftenberg within the Cluster »Functional Materials and

Film Systems for Efficient Energy Conversion (FuSion)«. We acknowledge Guido Beuckert for

technical support.

1H. Peelaers and C. G. Van de Walle, Phys. Status Solidi B 252, 828 (2015). 2M. Orita, H. Ohta, M. Hirano and H. Hosono, Appl. Phys. Lett. 77 , 4166 (2000). 3Y. Yao, R. F. Davis and L. M. Porter, J. Electron. Mater. 46, 2053 (2017). 4S. J. Pearton, J. Yang, P. H. Cary, F. Ren, J. Kim, M. J. Tadjer and M. A. Mastro, Appl. Phys. Rev. 5,

011301 (2018). 5M. Higashiwaki, K. Sasaki, A. Kuramata, T. Masui and S. Yamakoshi, Appl. Phys. Lett. 100, 013504

(2012). 6F. K. Shan, G. X. Liu, W. J. Lee, G. H. Lee, I. S. Kim and B. C. Shin, J. Appl. Phys. 98, 023504 (2005). 7H. Hayashi, R. Huang, H. Ikeno, F. Oba, S. Yoshioka, I. Tanaka and S. Sonoda, Appl. Phys. Lett. 89,

181903 (2006). 8G. W. Paterson, P. Longo, J. A. Wilson, A. J. Craven, A. R. Long, I. G. Thayne, M. Passlack and R.

Droopad, J. Appl. Phys. 104, 103719 (2008). 9H. Lee, K. Kim, J.-J. Woo, D.-J. Jun, Y. Park, Y. Kim, H. W. Lee, Y. J. Cho and H. M. Cho, Chem.

Vapor. Depos. 17, 191 (2011). 10D.-W. Choi, K.-B. Chung and J.-S. Park, Thin Solid Films 546, 31 (2013). 11S. Stepanov, V. Nikolaev, V. Bougrov and A. Romanov, Rev. Adv. Mater. Sci 44, 63 (2016). 12M. Fleischer and H. Meixner, J. Appl. Phys. 74, 300 (1993). 13M. Holland, C. R. Stanley, W. Reid, R. J. W. Hill, D. A. J. Moran, I. Thayne, G. W. Paterson and A. R.

Long, J. Vac. Sci. Technol. B 25, 1706 (2007). 14M. Fleischer, J. Giber and H. Meixner, Appl. Phys. A 54, 560 (1992). 15M. Nieminen, L. Niinistö and E. Rauhala, J. Mater. Chem. 6, 27 (1996). 16Dezelah, J. Niinistö, K. Arstila, L. Niinistö and C. H. Winter, Chem. Mater. 18, 471 (2006). 17D. J. Comstock and J. W. Elam, Chem. Mater. 24, 4011 (2012). 18Y. S. Lee, D. Chua, R. E. Brandt, S. C. Siah, J. V. Li, J. P. Mailoa, S. W. Lee, R. G. Gordon and T.

Buonassisi, J. Adv. Mater. 26, 4704 (2014).

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

28

19T. Faraz, K. Arts, S. Karwal, H. C. M. Knoops and W. M. M. Kessels, Plasma Sources Sci. T. 28,

024002 (2019). 20I. Donmez, C. Ozgit-Akgun and N. Biyikli, J. Vac. Sci. Technol. A 31, 01A110 (2012). 21H.-Y. Shih, F.-C. Chu, A. Das, C.-Y. Lee, M.-J. Chen and R.-M. Lin, Nanoscale Res. Lett. 11, 235

(2016). 22R. K. Ramachandran, J. Dendooven, J. Botterman, S. Pulinthanathu Sree, D. Poelman, J. A. Martens, H.

Poelman and C. Detavernier, J. Mater. Chem. A 2, 19232 (2014). 23G. A. Battiston, R. Gerbasi, M. Porchia, R. Bertoncello and F. Caccavale, Thin Solid Films 279, 115

(1996). 24R. Binions, C. J. Carmalt, I. P. Parkin, K. F. E. Pratt and G. A. Shaw, Chem. Mater. 16, 2489 (2004). 25M. J. Ludowise, J. Appl. Phys. 58, R31 (1985). 26H.-P. Ma, H.-L. Lu, T. Wang, J.-G. Yang, X. Li, J.-X. Chen, J.-J. Tao, J.-T. Zhu, Q. Guo and D. W.

Zhang, J. Mater. Chem. C 6, 12518 (2018). 27H. Altuntas, I. Donmez, C. Ozgit-Akgun and N. Biyikli, J. Vac. Sci. Technol. A 32, 041504 (2014). 28K. Henkel, H. Gargouri, B. Gruska, M. Arens, M. Tallarida and D. Schmeißer, J. Vac. Sci. Technol. A

32, 01A107 (2014). 29J. Haeberle, K. Henkel, H. Gargouri, F. Naumann, B. Gruska, M. Arens, M. Tallarida and D. Schmeißer,

Beilstein J. Nanotech. 4, 732 (2013). 30P. Ruffieux, P. Schwaller, O. Gröning, L. Schlapbach, P. Gröning, Q. C. Herd, D. Funnemann and J.

Westermann, Rev. Sci. Instrum. 71, 3634 (2000). 31J. J. Yeh and I. Lindau, Atom. Data Nucl. Data 32, 1 (1985). 32K. Henkel, “Electrical Investigations on Praseodymium Oxide/Aluminum Oxynitride Containing Metal-

Insulator-Semiconductor Stacks and on Metal-Ferroelectric-Insulator-Semiconductor Structures

Consisting of Poly[vinylidene fluoride trifluoroethylene],” Ph.D. thesis (Brandenburg University of

Technology, Germany, 2009). 33R. L. Puurunen, Chem. Vapor. Depos. 9, 249 (2003). 34B. Abendroth, T. Moebus, S. Rentrop, R. Strohmeyer, M. Vinnichenko, T. Weling, H. Stöcker and D. C.

Meyer, Thin Solid Films 545, 176 (2013). 35P. Malet and G. Munuera, J. Chem. Soc. Farad. T. 1 85, 4157 (1989). 36E. Kobayashi et al., J. Vac. Sci. Technol. A 36, 021518 (2018). 37O. M. Bordun, I. Yo. Kukharskyy, B. O. Bordun, and V. B. Lushchanets, J. Appl. Spectrosc. 81, 771

(2014)

38C. V. Ramana, E. J. Rubio, C. D. Barraza, A. Miranda Gallardo, S. McPeak, S. Kotru and J. T. Grant, J.

Appl. Phys. 115, 043508 (2014). 39K. Taniguchi, M. Tanaka, C. Hamaguchi and K. Imai, J. Appl. Phys. 67, 2195 (1990). 40W. Rzodkiewicz and A. Panas, Acta Phys. Pol. A 116, S92 (2009). 41M. Valet and D. M. Hoffman, Chem. Mater. 13, 2135 (2001). 42D.-H. Kim, S.-H. Yoo, K.-S. An, H. Yoo and Y. Kim, Bull. Korean Chem. Soc. 23, 225 (2002). 43N. Makeswaran, A. K. Battu, R. Swadipta, F. S. Manciu and C. V. Ramana, ECS J. Solid State Sc. 8,

Q3249 (2019). 44J. F. Moulder, W. F. Stickle, P. E. Sobol and K. D. Bomben, Handbook of X-ray photoelectron

spectroscopy: a reference book of standard spectra for identification and interpretation of XPS data.

(Physical Electronics Division: Perkin-Elmer Corporation, Waltham, Massachusetts 02451, USA, 1992). 45E. J. Rubio, T. E. Mates, S. Manandhar, M. Nandasiri, V. Shutthanandan and C. V. Ramana, J. Phys.

Chem. C 120, 26720 (2016). 46S. Ghose, S. Rahman, L. Hong, J. S. Rojas-Ramirez, H. Jin, K. Park, R. Klie and R. Droopad, J. Appl.

Phys. 122, 095302 (2017). 47J.-C. Dupin, D. Gonbeau, P. Vinatier and A. Levasseur, Phys. Chem. Chem. Phys. 2, 1319 (2000). 48H. Perron, J. Vandenborre, C. Domain, R. Drot, J. Roques, E. Simoni, J. J. Ehrhardt and H. Catalette,

Surf. Sci. 601, 518 (2007).

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

29

49C. Das, K. Henkel, M. Tallarida, D. Schmeißer, H. Gargouri, I. Kärkkänen, J. Schneidewind, B. Gruska

and M. Arens, J. Vac. Sci. Technol. A 33, 01A144 (2015). 50A. Fiedler, R. Schewski, Z. Galazka and K. Irmscher, ECS J. Solid State Sc. 8, Q3083 (2019). 51C. Kittel, Introduction to solid state physics (John Wiley & Sons, New York, USA, 2005). 52E. H. Nicollian and J. R. Brews, MOS physics and technology (John Wiley and Sons, New York, USA,

1982). 53G. X. Liu, F. K. Shan, J. J. Park, W. J. Lee, G. H. Lee, I. S. Kim, B. C. Shin and S. G. Yoon, J.

Electroceram. 17, 145 (2006). 54D. Schmeißer and K. Henkel, J. Appl. Phys. 123, 161596 (2018). 55K. Henkel, M. Kot and D. Schmeißer, J. Vac. Sci. Technol. A 35, 01B125 (2017). 56D. R. Lide, CRC handbook of chemistry and physics, 79th ed (CRC Press, Boca Raton, 1998). 57S. Sze, Physics of Semiconductor Devices, 2nd ed (Jhon Wiley, New York, USA, 1981). 58M. Kot, K. Henkel, F. Naumann, H. Gargouri, L. Lupina, V. Wilker, P. Kus, E. Pożarowska, S. Garain,

Z. Rouissi and D. Schmeißer, J. Vac. Sci. Technol. A 37, 020913 (2019). 59M. Passlack, N. E. J. Hunt, E. F. Schubert, G. J. Zydzik, M. Hong, J. P. Mannaerts, R. L. Opila and R. J.

Fischer, Appl. Phys. Lett. 64, 2715 (1994). 60C.-T. Lee, H.-W. Chen and H.-Y. Lee, Appl. Phys. Lett. 82, 4304 (2003).

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

30

TABLES

TABLE I. Summary of main determined properties of the Ga2O3 layers characterized in this study.

The tabulated parameters are substrate temperature (Ts), GPC, inhomogeneity (inhomog.) across

4″ Si wafers, refractive index (n), optical energy gap (Eg), Ga:O ratio, C1s concentration,

permittivity (k) at 10 kHz, negative fixed charges (Nfix), and electrical breakdown field (EBD).

The Ga:O ratio and C1s contribution were determined from 10 nm thick films whereas all other

properties were obtained from 30 nm thick layers.

Samples Ts

[°C]

GPC

[Å/cycle]

inhomog.

[%]

n

[@632.8nm]

Eg

[eV]

Ga:O C1s

[%]

k

[@10kHz]

Nfix

[1012cm-2]

EBD

[MV/cm]

GaO80°C

GaO100°C

GaO150°C

GaO200°C

80

100

150

200

0.70

0.67

0.66

0.63

1.6

1.8

1.6

2.0

1.85±0.005

1.85±0.003

1.86±0.004

1.87±0.006

4.68±0.02

4.66±0.03

4.61±0.02

4.57±0.02

0.67

0.65

0.66

0.67

12.2

12.6

10.6

15.9

9.8±0.1

9.4±0.2

9.6±0.1

9.9±0.2

-3.8±0.1

-3.1±0.2

-2.1±0.1

-2.2±0.1

2.2±0.3

2.4±0.2

2.7±0.3

2.4±0.3

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

31

TABLE II. Relative sub-peak intensities in the Ga3d and O1s core level XPS spectra of Ga2O3

PEALD films.

Ga3d O1s

Samples carbonates% Ga-O% Ga-OH% O-H% O-C% Ga-O%

GaO80°C

GaO100°C

GaO150°C

GaO200°C

6.3

5.7

5.8

5.7

76.9

79

78.8

79.7

12.6

11.4

11.3

10.9

5.7

7.8

7.2

6.1

21.8

32.2

16.5

19.1

68.1

55

72.6

70.7

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

32

FIGURE CAPTIONS

FIG. 1. (Color online) Cycle sequence for the PEALD of Ga2O3 thin films. Each PEALD cycle

consisted of four sub-pulses with 10 ms TMGa vapor pulse (90 sccm N2 carrier gas), 2 s N2

purge pulse, 5 s (150 sccm, 200 W) O2 plasma and 2 s N2 purge pulse. The periodic deposition

cycle was repeated until the desired layer thickness was realized.

FIG. 2. (Color online) Initial Ga2O3 film thickness evolution as a function of the process time for

different substrate temperatures within the PEALD process. The thickness was obtained by

modeling the in-situ ellipsometer data with a one-layer model with a constant refractive index of

1.86 (at 632.8 nm).

FIG. 3. (Color online) Measured in-situ Ga2O3 film thickness versus prolonged process time up to

a total thickness of about 30 nm.

FIG. 4. (Color online) Growth rate (a) and inhomogeneity across a 4″ Si wafer (b) of the Ga2O3

layers depending on substrate temperature.

FIG. 5. (Color online) SE model results of the optical band gap and refractive index as a function

of deposition temperature for gallium oxide samples. The measurements were carried out ex-situ

on 30 nm thick samples.

FIG. 6. (Color online) XPS survey spectra of the as-deposited Ga2O3-PEALD thin films prepared

at different substrate temperatures as indicated. The spectra were recorded with Mg Kα

excitation.

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

33

FIG. 7. (Color online) Gallium to oxygen ratio (a) and carbon contamination (b) of the Ga2O3

films vs. substrate temperature.

FIG. 8. (Color online) Ga 3d core level XPS data (Mg Kα) of Ga2O3 layers grown at substrate


FIG. 9. (Color online) O1s core level XPS data of Ga2O3 layers deposited at substrate


FIG. 10. (Color online) C-V measurements of the sample GaO80°C at different contact pad

diameters () as listed beside each curve. The measurement direction is indicated by the arrows.

The lower inset illustrates the MIS stack sequence; the upper inset shows one C-V loop at 1 MHz

( = 500 µm).

FIG. 11. (Color online) Normalized Cacc against contact area according to Eq. (1) for the samples

GaO80°C, GaO100°C, GaO150°C, and GaO200°C as indicated in the legend. The linear fits used for the

determination of the permittivity are drawn with the corresponding colors. The data are shown

for C-V measurements recorded at 10 kHz.

FIG. 12. (Color online) Permittivity (a) at 10 (red open circles) and 100 kHz (blue filled squares)

as well as (b) negative fixed (green open squares) and mobile (black filled triangles) charges

(deduced from C-V measurements at 1 MHz) against substrate temperature of the PEALD

processed Ga2O3 samples.

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

This

is the

autho

r’s pe

er re

viewe

d, ac

cepte

d man

uscri

pt. H

owev

er, th

e onli

ne ve

rsion

of re

cord

will

be di

ffere

nt fro

m thi

s ver

sion o

nce i

t has

been

copy

edite

d and

type

set.

PLEA

SE C

ITE

THIS

ART

ICLE

AS

DOI:

10.11

16/1.

5134

800

low-temperature growth of gallium oxide thin films by

Documents