High mobility amorphous zinc oxynitride

semiconductor material for thin film

transistors
Cite as: J. Appl. Phys. 106, 074512 (2009); https://doi.org/10.1063/1.3236663
Submitted: 21 July 2009 • Accepted: 30 August 2009 • Published Online: 13 October 2009

Yan Ye, Rodney Lim and John M. White

ARTICLES YOU MAY BE INTERESTED IN

Electronic structure and transport in amorphous metal oxide and amorphous metal oxynitride
semiconductors
Journal of Applied Physics 126, 125702 (2019); https://doi.org/10.1063/1.5096042

2 −1 −1
Amorphous ZnOxNy thin films with high electron Hall mobility exceeding 200 cm V s
Applied Physics Letters 109, 262101 (2016); https://doi.org/10.1063/1.4973203

Ar plasma treated ZnON transistor for future thin film electronics

Applied Physics Letters 107, 122105 (2015); https://doi.org/10.1063/1.4930827

J. Appl. Phys. 106, 074512 (2009); https://doi.org/10.1063/1.3236663 106, 074512

© 2009 American Institute of Physics.

 JOURNAL OF APPLIED PHYSICS 106, 074512 共2009兲

High mobility amorphous zinc oxynitride semiconductor material for thin

film transistors
Yan Ye,a兲 Rodney Lim, and John M. White
Display and SunFab Solar Business Group, Applied Materials, Santa Clara, California 95054, USA
共Received 21 July 2009; accepted 30 August 2009; published online 13 October 2009兲
Zinc oxynitride semiconductor material is produced through a reactive sputtering process in which
competition between reactions responsible for the growth of hexagonal zinc oxide 共ZnO兲 and for the
growth of cubic zinc nitride 共Zn3N2兲 is promoted. In contrast to processes in which the reaction for
either the oxide or the nitride is dominant, the multireaction process yields a substantially
amorphous or a highly disordered nanocrystalline film with higher Hall mobility, 47 cm2 V−1 s−1
for the as-deposited film produced at 50 ° C and 110 cm2 V−1 s−1 after annealing at 400 ° C. In
addition, it has been observed that the Hall mobility of the material increases as the carrier
concentration decreases in a carrier concentration range where a multicomponent metal oxide
semiconductor, indium–gallium–zinc oxide, follows the opposite trend. This indicates that the
carrier transports in the single-metal compound and the multimetal compound are probably
dominated by different mechanisms. Film stability and thin film transistor performance of the
material have also been tested, and results are presented herein. © 2009 American Institute of
Physics. 关doi:10.1063/1.3236663兴

I. INTRODUCTION insulator-semiconductor interface make crystalline ZnO a

challenging and unfavorable choice for use in the economi-
Amorphous silicon 共a-Si兲 thin film deposited through a cal bottom gate TFT structure used in current large-area dis-
plasma enhanced chemical vapor deposition 共PECVD兲 pro- play. More recently, amorphous indium–gallium–zinc oxide
cess is currently the semiconductor material pervasively used 共IGZO兲 has been studied and developed, revealing a new
for the thin film transistors 共TFTs兲 in large-area active-matrix path for producing high mobility amorphous semiconductor
liquid crystal displays 共AMLCDs兲. The technology has material through multicomponent post-transition-metal
proven to be scalable to substrates greater than 2 ⫻ 2 m2, oxides.9,10 Since then, TFT arrays for both AMLCD and
with low fabrication costs and high yields. However, its mo- AMOLED displays were made using IGZO.11,12 Several
bility is limited to less than 1 cm2 V−1 s−1, and stability can other types of multicomponent metal oxide semiconductors
be a problem in some applications. As AMLCD TV advances have also been developed.13,14
to higher definition, larger displays, and higher refresh rates, A zinc oxynitride 共ZnON兲 thin film was studied for its
and as new products such as active-matrix organic light- optical properties in the late 1990s.15 The zinc oxynitride
emitting-diode 共AMOLED兲 displays emerge, a-Si TFT tech- film reported had a distinct hexagonal zinc oxide crystalline
nology appears inadequate to meet the greater mobility and structure, as shown by x-ray diffraction 共XRD兲 spectra. The
threshold voltage stability requirements for these applica- film is similar in structure to the nitrogen-doped zinc oxide
tions. While efforts to improve Si thin film semiconductor 共ZnO:N兲 or nitrogen-metal codoped zinc oxides, which have
performance continue, development of alternative thin film been intensely investigated in an effort to produce p-type
semiconductor materials for the next generation display ZnO films for solar cell and solid state lighting
products has drawn more and more attention. applications.16–21 Since all these films have clear hexagonal
ZnO is widely recognized as a n-type transparent con- ZnO crystalline structures, it is evident that the reaction of
ducting oxide 共TCO兲 although it is also used for other zinc with oxygen is the dominant reaction involved in the
applications.1 In the first several years of this millennium, growth of the films even though nitrogen is present.
studies of ZnO as a TFT channel material have been per- In this study, it has been observed that unique film prop-
formed by several groups.2–5 TFT arrays for AMLCD were erties can be achieved by appropriately controlling a reactive
made using ZnO as a channel material for both top gate and sputtering process to allow both oxygen and nitrogen as de-
bottom gate TFT structures.6,7 It has been found that good termining factors for growth of the film. Since competition
quality ZnO film itself strongly tends to be polycrystalline, between reactions responsible for growth of different crystal-
usually with preferential orientation, under nearly all deposi- line structures, instead of dominance of a single reaction, is
tion conditions. It was reported that traps in the dense grain promoted in the process, a substantially amorphous phase,
boundaries of crystalline ZnO can severely affect the overall rather than a well-structured crystalline phase, is attained.
performance of the TFTs.8 In addition, the control of grain Even though the film from the multireaction process does not
nucleation and uniformity of grain size at the critical gate have a crystalline structure as crystalline Zn3N2, or crystal-
line ZnON or ZnO:N, or crystalline ZnO films, it exhibits a
a兲
Electronic mail: [email protected]. higher Hall mobility than any of these other films, and the

0021-8979/2009/106共7兲/074512/8/$25.00 106, 074512-1 © 2009 American Institute of Physics

074512-2 Ye, Lim, and White J. Appl. Phys. 106, 074512 共2009兲

mobility actually increases as the carrier concentration de- 50

a)
creases over a wide range of the carrier concentration. This

-1
40

Mobility, cm V s
trend contrasts greatly with that observed for a multicompo-

-1
nent oxide semiconductor, IGZO, where the Hall mobility 30

2
decreases as the carrier concentration decreases.10 TFTs us-
20
ing the high mobility zinc oxynitride as the channel material
have been made to demonstrate the film performance. The 10
stability of the film has also been tested and presented herein.
0
II. EXPERIMENTAL 10
22

-3
b)

Carrier concentration, cm
A reactive sputtering process was used to deposit thin
20
film semiconductor materials on a glass substrate in a modi- 10
fied AKT 4300 dc-magnetron physical vapor deposition
18
chamber. In the chamber, a metallic sputtering target of pure 10
zinc, zinc doped with Al, zinc doped with Sn, and several 500 sccm N2, Zn
others was alternatively used for this study. The purity of the 10
16
300 sccm N2, Zn
sputtering targets was 99.99 or better. The chamber was 500 sccm N2, Zn:Al
equipped with a turbomolecular pump and a gas delivering 14
300 sccm N2, Zn:Al
10
system capable of introducing and accurately metering a va- 0 10 20 30 40 50
riety of reactive gases into the chamber. Software and elec- O2 flow rate, sccm
trical control of the chamber were also modified to allow a
recipe with multiple gases and multiple steps to be run, as is FIG. 1. 共a兲 Hall mobility and 共b兲 carrier concentration of films produced
commonly done in a PECVD process chamber. using different O2 flow rates under a high N2 flow rate at 50 ° C with a zinc
target and a zinc target doped with 1 at. % aluminum. Hall measurement
A heated pedestal for holding a 700⫻ 520⫻ 0.7 mm3 was carried out at room temperature.
Corning 1737 glass substrate was used. The space between
the pedestal and the sputtering target can be settable by
recipe control to be between 60 and 110 mm. The tempera- minimize the involvement of residual oxygen into the pro-
ture of the pedestal can be controlled from 30 to 400 ° C. The cess. Figure 1 shows the electrical properties of as-deposited
pedestal temperature was set at 50 ° C in the tests reported films produced at 50 ° C pedestal temperature. As described
here. The actual substrate temperature could have been in the legend, two different N2 flow rates of 300 and 500
slightly lower or higher than the temperature. SCCM 共SCCM denotes cubic centimeter per minute at STP兲,
The electrical properties of the films such as carrier con- respectively, and two types of sputtering targets, pure zinc
centration and Hall mobility were measured using the van and zinc doped with 1 at. % at aluminum, respectively, were
der Pauw technique with Accent HL 5500PC. This electrical tested for comparison. Other process conditions not men-
measurement was carried out in air at room temperature. tioned, such as target-substrate spacing, dc power, and Ar
Crystallographic structure and possible chemical composi- flow rate, were the same in all processes.
tion of the film were analyzed using an XRD system with All films produced here are n-type semiconductors.
Cu K␣ radiation 共X’pert Pro, PANalytical兲. Both ␪-2␪ con- Therefore, the Hall mobility and carrier presented here are
figuration and grazing angle configuration were used, espe- the electron mobility and electron carrier, respectively. Fig-
cially for the films having poor crystalline structure. The ure 1共a兲 shows that for all cases, the mobility increases as the
performance of TFTs using the semiconductor film as chan- O2 flow increases up to a point and then decreases as the O2
nel materials was measured at room temperature using an HP flow increases further. The mobility peaks between 10 and 30
semiconductor parameter analyzer 4140B. SCCM of O2 gas depending on whether the target was doped
or undoped, and the highest mobility from the 50 ° C pro-
III. RESULTS AND DISCUSSION cess, 47 cm2 V−1 s−1, are achieved using a pure zinc target.
Figure 1共b兲 shows that the carrier concentration decreases
A. Film produced from multiple reactions
monotonically as the O2 flow rate increases. The carrier con-
In order to create multiple reactions in a reactive sput- centration in the film deposited from the zinc target doped
tering process of a metallic zinc target, at least two reactive with aluminum is lower than that from the pure zinc target
gases, which are responsible for the growth of hexagonal under the same process conditions. This indicates that the
ZnO and cubic Zn3N2, respectively, were introduced in vari- carrier concentration or conductivity of the film can be ad-
ous ratios into the reactive sputtering chamber. In the tests justed by incorporating a small amount of doping in the zinc
reported in this paper, O2 and N2 were selected as the reac- target.
tive gases. In order to produce films from reactions with The crystalline structures of the films produced from the
oxygen only, or nitrogen only, or both gases, the flow rate of test described in Fig. 1 were analyzed using XRD with the
O2 was adjusted under the same flow rate of N2. Before ␪-2␪ configuration. The sputtering target used was a pure
running a process that was not to include oxygen, certain zinc target. All films analyzed were at least 2800 Å or
measures, such as attaining a low base pressure and remov- thicker, as noted in the figure caption, to ensure sufficient
ing the oxidation layer from the metallic target, were taken to growth of any crystalline structures and sufficient detection
074512-3 Ye, Lim, and White J. Appl. Phys. 106, 074512 共2009兲

1000 1600
a) 0 sccm O2, 500 sccm N2 a) 0 sccm O2, 500 sccm N2

321
400
800
1200
Zn3N2(cubic) Zn3N2(cubic)
Intensity

Intensity

440
332
600

321
400
800

440
332
400

721
400

721

222
200

444
521

631
222

611
444
521

631
611
0 0
400 400
b) 20 sccm O2, 500 sccm N2 b) 20 sccm O2, 500 sccm N2
350
300
Intensity

Intensity
300
200
250

200 100

150 0
1600 1600
c) 50 sccm O2, 500 sccm N2 c) 50 sccm O2, 500 sccm N2
1200 1200
101

101
Intensity

Intensity
ZnO (hex) ZnO (hex)
800 800
100

100
002

002
110

110
103

103
102

112

102

112
400 400

0 0
30 40 50 60 70 30 40 50 60 70
Diffraction angle 2 Diffraction angle 2
FIG. 2. Diffraction spectra from XRD measurement with a ␪-2␪ configura- FIG. 3. Diffraction spectra from XRD measurement with a 0.5° glazing
tion for as-deposited films deposited at 50 ° C using an undoped zinc target angle configuration for as-deposited films using an undoped zinc target de-
deposited 共a兲 at 0 SCCM O2 and 500 SCCM N2 共280 nm thickness兲, 共b兲 at posited at 50 ° C 共a兲 at 0 SCCM O2 and 500 SCCM N2 共280 nm thickness兲,
20 SCCM O2 and 500 SCCM N2 共280 nm thickness兲, and 共c兲 at 50 SCCM 共b兲 at 20 SCCM O2 and 500 SCCM N2 共280 nm thickness兲, and 共c兲 at 50
O2 and 500 SCCM N2 共330 nm thickness兲. Stick patterns of a cubic Zn3N2 SCCM O2 and 500 SCCM N2 共330 nm thickness兲. Stick patterns of a cubic
共Ref. 22兲 and a hexagonal ZnO 共Ref. 23兲 are plotted in 共a兲 and 共c兲 for Zn3N2 共Ref. 22兲 and a hexagonal ZnO 共Ref. 23兲 are plotted in 共a兲 and 共c兲 for
comparison. comparison.

signals. Figure 2共a兲 shows the spectrum of the film from a identification of the films with better detecting sensitivity.
process using 0 SCCM O2 and 500 SCCM N2. A stick pat- The results of the measurement, summarized in Fig. 3共a兲,
tern of a cubic Zn3N2 powder22 is plotted in the figure for confirm that the diffraction spectrum of the film from the 0
comparison. All peaks of the diffraction spectrum of the film SCCM O2 process fits reasonably well with the Zn3N2 pat-
correspond well to the Zn3N2 pattern, which indicates that tern at both small and large 2␪ angles, again supporting the
the film produced has the same cubic crystalline structure as observation that the film produced has the same crystalline
the reference Zn3N2. Figure 2共c兲 presents the spectrum of the structure and the same stoichiometric composition as the cu-
film from the process using 50 SCCM O2 and 500 SCCM bic Zn3N2 reference. Figure 3共c兲 reveals that the diffraction
N2. A stick pattern of a hexagonal ZnO 共Ref. 23兲 is also spectrum of the film from the 50 SCCM O2 process corre-
plotted on the same chart for comparison. Here, a strong sponds to the ZnO stick pattern, although the peaks shift
sharp peak around a diffraction angle 2␪ of 35.4° is observed slightly to a smaller diffraction angle at large 2␪ angles. This
over other small peaks. This suggests that the film has the indicates that the interplanar separation of the diffraction
hexagonal crystalline structure as the crystalline ZnO with a planes, therefore the composition of the film, differs slightly
preferred 共002兲 orientation. Figure 2共b兲 shows the diffraction from that of the ZnO reference. Such a departure from the
spectrum of the film produced using 20 SCCM O2 and 500 stoichiometric composition indicates the presence of nitro-
SCCM N2 which gives a Hall mobility of above gen in the lattice of ZnO. Figure 3共b兲 exhibits more clearly
45 cm2 V−1 s−1. All distinct peaks presented in Figs. 2共a兲 the amorphous scattering and confirms the amorphous phase
and 2共c兲 disappear, and the amorphous scattering appears of the film produced from the process with the 20 SCCM O2.
although the detecting signal is weak. A tiny peak around 2␪ The actual composition of the film is not identified here al-
of 31°, not observed from the other two films, suggests that a though it is certainly different from the other two crystalline
small fraction of the crystalline phase may also exist in the films presented.
amorphous film. From these test results, it is evident that very different
Additional XRD measurement with the 0.5° glancing types of thin films can be produced by adjusting the oxygen
angle configuration 共0.5° ⍀-offset兲 was performed for further availability in a reactive environment that has a much higher
074512-4 Ye, Lim, and White J. Appl. Phys. 106, 074512 共2009兲

availability of nitrogen during sputtering of the metallic zinc 50

target. By setting the oxygen flow rate to zero and removing a)

-1
40

Mobility, cm V s
the residual oxygen in the chamber, the growth of the film is

-1
only a result of a reaction between zinc and nitrogen, pro- 30

2
ducing a Zn3N2 film that has the stoichiometric composition
as the reference. When the oxygen flow is set higher than a 20

certain level, the reaction between zinc and oxygen controls 10

the growth of the film even though nitrogen is present during 20 sccm O2, Zn
the process. The film produced has much the same crystalline 0 20 sccm O2, Zn:Al
22
structure as ZnO with nitrogen incorporated in the crystal 10

-3
lattices, which is similar to crystalline ZnO:N 共Refs. 16–21兲 b)

Carrier concentration, cm
21
10
or crystalline ZnON.15 When the oxygen flow rate is adjusted
to an intermediate level, both nitrogen and oxygen strongly 20
10
influence the growth of the film, resulting in complicated
changes in phase, structure, and chemical composition. The 10
19

film produced has no prominent signs of either cubic Zn3N2

18
or hexagonal ZnO structure and is smooth with a typical 10
root-mean-square roughness of less than 10 Å as measured
17
using an atomic force microscope. It can be in an amorphous 10
0 100 200 300 400 500
phase as presented above, or in a highly disordered nanocrys-
N2 flow rate, sccm
talline phase admixing different crystalline structures, de-
pending on process conditions. FIG. 4. Effect of N2 flow rate on 共a兲 Hall mobility and 共b兲 carrier concen-
It should be noted that by adjusting process parameters tration for the as-deposited film deposited at 50 ° C with a zinc target and a
of a reactive sputtering process, there is certain flexibility to zinc target doped with 1 at. % aluminum. Hall measurement was carried out
tune a film to be either more crystalline or more amorphous at room temperature.
in terms of volume fraction of each phase. However, it is
very difficult to make the film have a highly ordered crystal- the test were n-type semiconductors. As shown in Figs. 4共a兲
line structure as long as reactions for different crystalline and 4共b兲, for the film produced with pure zinc target, the Hall
structures compete with each other during the growth of the mobility started to increase significantly as the flow rate of
film. Although the highly ordered crystalline ZnO and Zn3N2 N2 increased above 200 SCCM, while the carrier concentra-
structures disappear in the film due to the multiple reactions, tion decreased. Apparently, N2 is much less reactive than O2
the Hall mobility of the film from the multireaction process since the significant N2 effect is observed above the 10:1
is higher than those dominated by a single reactant, as shown N2 : O2 ratio, and the peak of the mobility requires a 20:1
in Fig. 1. This phenomenon indicates that the highly ordered ratio or higher for the pure Zn sputtering target. Because of
crystalline structure is not required to achieve the high mo- such a large difference in oxygen and nitrogen reactivity, the
bility in the ZnON film. In fact, nitrogen in the compound oxidation reaction involving the oxygen containing reactant
may affect the constitution of the top part of the valence easily became the dominant one to determine the crystalline
band through N 2p orbitals. The bottom part of the conduc- structure as well as electrical and other properties. Therefore,
tion band may still rely on the direct overlap of spatially it is tricky to promote a reaction from the nitrogen reactant to
spread Zn 4s orbitals between neighboring Zn metals. The balance the reaction from the oxygen reactant. Doping with
high mobility should be present as long as the direct overlap other species such as aluminum in the target helps the N2
of Zn 4s orbitals still exists and various scattering mecha- effect to take place at a lower N2 : O2 ratio as the comparison
nisms limiting electron transport are minimized. Therefore, between undoped zinc and zinc doped with 1% aluminum is
the model proposed by Hosono et al.9 and Nomura et al.,10 presented in Fig. 4. Certainly, the ratio required for revealing
for high mobility amorphous metal oxides containing post- the nitrogen effect can also be affected by other conditions.
transition-metal cations, still applies to a compound incorpo- For example, it was observed that the higher N2 : O2 ratio was
rated with nitrogen. Certainly, other effects resulting from required to attain a similar increase in mobility when a lower
the involvement of both nitrogen and oxygen in the com- flow rate of O2 was used, indicating that it might not be an
pound may additionally contribute to the high mobility ob- effective way to achieve the required N2 : O2 ratio by just
served. reducing the O2 flow rate. Without taking a special care, it is
more likely to obtain a crystalline ZnO:N or crystalline
ZnON film with a hexagonal ZnO crystalline structure, in-
B. Effect of nitrogen reactant
stead of the high mobility highly disordered ZnON film pro-
Figure 4 presents the effect of N2 gas on the electrical duced here.
properties of the film produced in a reactive sputtering pro- Figure 5 shows the electrical properties of the films after
cess with the pedestal controlled at 50 ° C. In the test, the nitrogen annealing in a nitrogen environment at 400 ° C for
flow rate of O2 was as set as is shown in the legend and was 30 min. The mobility and carrier concentration of the corre-
unchanged through each data set. The flow rate of N2 gas sponding as-deposited film are also plotted for comparison.
was increased from 80 to 500 SCCM. All films produced in The annealing significantly increases the mobility to about
074512-5 Ye, Lim, and White J. Appl. Phys. 106, 074512 共2009兲

120 100 20 sccm O2, Zn

20 sccm O2, Zn, annealed
20 sccm O2, Zn:Al
-1
100 20 sccm O2, Zn, as dep
Mobility, cm V s 10 sccm O2, Zn:Al, annealed 10 sccm O2, Zn:Al
-1

80 10 sccm O2, Zn:Al, as dep

2

60
a)

-1
Hall mobility, cm V s
40

-1
Amorphous

2
20
IGZO films
10
0
-3
Carrier concentration, cm

21
10

20
10
b)
19
10 Single-crystalline
IGZO films
18
10 1
16 17 18 19 20 21
10 10 10 10 10 10
17 -3
10 Carrier concentration, cm
0 100 200 300 400 500
N2 flow rate, sccm FIG. 6. Relationship between Hall mobility and carrier concentration for
both the ZnON as deposited at 50 ° C and a multicomponent metal oxide,
FIG. 5. 共a兲 Hall mobility and 共b兲 carrier concentration before and after IGZO. ZnON data are converted from Figs. 4 and 5. The solid line is the
annealing at 400 ° C for 30 min. Films were deposited at 50 ° C using a zinc fitting curve for amorphous IGZO films and the dashed line is for single
target and a zinc target doped with 1 at. % aluminum. Hall measurement crystalline IGZO films 共Ref. 10兲. Hall measurement was carried out at room
was carried out at room temperature. temperature.

110 cm2 / V s for the ZnON film and to about 90 cm2 / V s IGZO films from Nomura et al.10 have also been plotted for
for the aluminum-doped zinc oxynitride 共ZnON:Al兲 film comparison. The mobility of both single crystalline and
without significant change in the carrier concentration in amorphous IGZO films increases as the carrier concentration
general. decreases. It is most likely that the carrier transports in the
It should also be noted that under the conditions de- ZnON and the IGZO films are dominated by different
scribed, all films produced, whether annealed or not, are mechanisms.
n-type semiconductors. This result may seem contrary to the The relationship of mobility and carrier concentration in
popular practice of making a p-type ZnO TCO through ni- the ZnON films can be characterized by the dominance of an
trogen doping. Actually, nitrogen can be incorporated with impurity scattering in carrier transport. The main difference
zinc and oxygen in different states, creating either electron is that the same carrier transport behavior takes place in a
donors or acceptors.24 Although intense efforts have been different range of carrier concentration. For degenerate con-
made to produce ZnO:N for p-type TCO, only limited suc- ducting ZnO films produced under many different condi-
cess has been achieved so far. It appears that for ZnO:N tions, a peak mobility is attained at a carrier concentration of
films, donors are more easily produced than acceptors, and around 1 ⫻ 1020 cm−3, and the carrier concentration that the
the n-type conduction is more stable than the p-type conduc- transport of carriers is dominated by impurity scattering
tion. This is consistent with what was observed here. ranged above that.1 For the ZnON films as deposited at
50 ° C, the carrier concentration for the peak mobility and
C. Carrier transport
the regime dominated by the impurity scattering is shifted
The high mobility amorphous or highly disordered nano- down at least two orders of magnitude. Such a shift suggests
crystalline ZnON films produced here share some similarity that the deviation of the films from the degenerate conduct-
in carrier transport mechanism with multicomponent oxide ing ZnO films may have resulted from lowering the Fermi
semiconductors such as IGZO. Both are composed of post- level farther away from the conduction band bottom associ-
transition metals and have possibly a direct overlap of spa- ated with the decrease in electron carrier concentration.
tially spread ns orbitals of neighboring metal cations. As a The contrary relationship between mobility and carrier
result, a highly ordered crystalline structure is not required to concentration from IGZO films has the characteristics that a
achieve high mobility, as was discussed previously. lattice or a grain boundary scattering dominates the carrier
Figure 6 represents the data of as-deposited films from transport. However, it is unusual that this type of behavior is
the 50 ° C process, shown in Figs. 4 and 5, as the change in extended to such high carrier concentration even for the
mobility with respect to the carrier concentration as the flow single crystalline IGZO. To explain this behavior, a model
rate of N2 changes. The mobility of both doped and undoped describing an insulator-metal transition due to the distribu-
ZnON films increases as the carrier concentration decreases tion of potential barriers formed around the conduction band
until it reaches a peak or saturation point. Fitting curves of bottom has been proposed for the single crystalline IGZO.25
the change in Hall mobility with carrier concentration for It was reported that the Hall mobility of the single crystalline
074512-6 Ye, Lim, and White J. Appl. Phys. 106, 074512 共2009兲

IGZO was independent of temperature in the range of carrier 50

a)
concentrations, as shown by the fitting curves in Fig. 6, in-

-1
40

Mobility, cm V s
dicating that the degenerate conduction was achieved. Since

-1
it has also behaviors different from a degenerate semicon- 30

2
ductor, it was proposed that the potential barriers originated
20
from the random distribution of Ga3+ and Zn2+ ions in the
crystalline structure govern the transition to the degenerate 10
conduction.25 The potential barriers were overcome as the
carrier concentration increased along with the increase in the 0
Hall mobility. It is believed that the similar mechanism is 10
22

-3
b)

Carrier concentration, cm
applicable to the amorphous IGZO since a similar behavior
has been observed.10,25 10
20

It is clear, however, that the model of overcoming poten-

tial barriers in a degenerate semiconductor with increased 10
18

carrier concentration as used for the multicomponent metal

oxide cannot explain the behavior observed in the ZnON 16
10
films. Since the film is composed of a single metal, it has no Film A, as-deposited
random insertion of chemically different cations into the nor- 14
Film B, as-deposited
10
mal structure and consequently no additional modulation to 0 5 10 15 20 25 30
the conduction band bottom which creates additional poten- Days in room air without passivation
tial barriers as the multicomponent metal oxide model pro-
poses. Therefore, the high carrier concentration is not re- FIG. 7. Change in 共a兲 Hall mobility and 共b兲 carrier concentration of an
quired for the achievement of the high mobility in the ZnON as-deposited film deposited at 50 ° C after exposure to room air without any
passivation or encapsulation.
films. Instead, the increase in the mobility with the decrease
in carrier concentration of the films suggests that the higher
mobility achieved may be attributed to the defect reduction Another possible explanation for the degradation of the
in the films. For example, the vacancy created in a single ZnON films in the shelf life test is hydrolyzation. It has been
reaction process may have been taken in the multireaction suggested that Zn3N2 is easily hydrolyzed by air moisture.27
process, reducing both the carrier concentration and the im- Small amounts of Zn共OH兲2 and ZnO were observed by Toy-
purity scattering to improve the carrier transport in the films. oura et al.28 on the outer surface of the zinc nitride formed
by the molten salt electrochemical process in their work.
Supposedly, the water came from the moisture adsorbed on
D. Film stability and shelf life the surface. It seems that the degradation of the ZnON film
observed here follows a similar mechanism since through the
It was observed that an as-deposited film without any x-ray photoelectron spectroscopy 共XPS兲 analysis, a thin zinc
encapsulation could deteriorate after exposing it to room air oxide layer was found on the top of the surface of an as-
for several weeks. Figure 7 shows the results of a monitoring deposited film once the deterioration is detected.29 However,
test of as-deposited films deposited at 50 ° C. The sudden the reaction observed here usually takes place after a pro-
change in mobility or conductivity was observed from one of longed exposure to air, after which the film quickly deterio-
the films after 20 days. More testing results showed that the rates. It suggests that in addition to water, the adsorption of
shelf life of an as-deposited film from a 50 ° C process with- other pollutants in the air could further contribute to the de-
out any passivation ranges from several weeks to several terioration. After a certain time, the accumulation water and
months, depending on process and room air conductions. pollutants may provide enough reactants and catalysis to re-
Kaminska et al.18 annealed Zn3N2 in oxygen at 600 ° C sult in an accelerated deterioration of the film. A simple sili-
to form a p-type ZnO semiconductor. In addition, differential con nitride passivation layer, commonly used for a-Si TFTs,
thermal analysis and thermal gravimetric analysis of ther- should prevent the degradation of the as-deposited film.
mally synthesized Zn3N2 powder in the environment of open It was observed that annealing after the film deposition
air were conducted, as reported by Zong et al.26 The tests can be used to increase not only the mobility but also the
showed that the initial thermal oxidization started at about shelf life of the film. For the film deposited at 50 ° C, its
500 ° C and reached the maximum oxidization rate at about shelf life is significantly improved after annealing. Figure 8
681 ° C. The condition for the thermal conversion of Zn3N2 shows the results of a monitoring test for a film which was
under a rich oxygen environment to ZnO is much more se- deposited at 50 ° C and then annealed in N2 at 400 ° C for 30
vere than the condition experienced in the shelf life test to min. No significant changes in mobility and carrier concen-
explain the observation here. Furthermore, a-Si film proper- tration had been observed after exposing the film in room air
ties start to change significantly due to the dehydrogenation for 660 days without any encapsulation or passivation.
in the temperature range. The amorphous phase of a metal
oxide, such as IGZO, is reported to be thermally stable up to
E. TFT performance
⬃500 ° C in air.10 Such high temperatures as are required for
the thermal transformation will not be experienced in com- Bottom gate TFTs have been made by replacing a-Si
mon TFT applications. layers and n + a-Si layers with the high mobility amorphous
074512-7 Ye, Lim, and White J. Appl. Phys. 106, 074512 共2009兲

120 since the measuring instrument was close to its detection

a)
limit. It should be noted that the field effect mobility of this
-1 100
Mobility, cm V s TFT was lowered here in order to achieve a certain desired
-1

80 threshold voltage, subthreshold swing, and so on. It should

2

60
be possible, through further optimization of all process steps
as-deposited making the TFT, to achieve a better overall TFT performance
40 in the future.
20

10
22 IV. SUMMARY
-3

b)
Carrier concentration, cm

as-deposited
ZnO:N or ZnON films with distinct hexagonal ZnO crys-
20
10 talline structure are often achieved even though nitrogen is
present in the film, as shown in previous literatures. It is
18
10 observed that the reaction of zinc with nitrogen is much
weaker than that with oxygen, and therefore the growth of
16
10 the film is easily dominated by the reaction with oxygen. By
Annealed at 400 C for 30min
as-deposited
carefully adjusting the ratio of oxygen to nitrogen containing
10
14
reactants, the competition between reactions for growth of
0 100 200 300 400 500 600 700 cubic Zn3N2 and for growth of hexagonal ZnO is promoted
Days in room air without passivation in a reactive sputtering process. Since multiple reactions re-
sponsible for different crystalline structures influence the
FIG. 8. 共a兲 Hall mobility and 共b兲 carrier concentration for a film produced
using a zinc target doped with 1 at. % aluminum deposited at 50 ° C and growth of the film, an amorphous or a highly disordered
annealed at 400 ° C after exposure to room air without passivation or nanocrystalline ZnON or ZnO:N film can be obtained. The
encapsulation. film produced through the multireaction process at 50 ° C
exhibited a higher Hall mobility than the films with predomi-
zinc oxynitride film as a channel layer. Figure 9 shows trans- nantly either cubic Zn3N2 or hexagonal ZnO crystalline
fer characteristics of a TFT with the channel width and structure. This indicates that a highly ordered crystalline
length of 40 and 10 µm, respectively. The channel layer was structure is not required to achieve a high mobility of the
deposited through a reactive sputtering process as previously film, and that the model proposed by Hosono et al. for high
described, using a zinc target doped with 2.8 at. % Al at mobility amorphous metal oxides still applies to the com-
50 ° C and annealed at 350 ° C. Cr was used for both gate pounds incorporated with nitrogen, such as metal oxynitrides
metal and source-drain contact metal. A 3300 Å thick Si3N4 and possibly even metal nitrides. In addition, the higher mo-
gate dielectric layer and a 2000 Å SiO2 / Si3N4 passivation bility achieved from the multireaction process suggests that
layer were used in the TFT and were deposited by PECVD at the carrier transport in the film may be improved additionally
a pedestal temperature of 200 ° C. On current, off current, through the involvement of both oxygen and nitrogen.
threshold voltage, and subthreshold swing achieved were Furthermore, it was observed that the relationship be-
about 2 ⫻ 10−4 A, 5 ⫻ 10−13 A, 0.5 V, and 0.8 V/decade, re- tween Hall mobility and carrier concentration in the high
spectively. The saturation mobility was about mobility ZnON is different from that of a multicomponent
10 cm2 V−1 s−1. It is suspected that the small peaks in the metal oxide such as IGZO. The Hall mobility of the ZnON
gate voltage measured between ⫺15 and ⫺5 V are artifact film has been shown to increase as the carrier concentration
decreases in a relatively low carrier concentration range. The
-3
Hall mobility of IGZO follows the opposite trend. The dif-
10
ference is likely due to differences in how conduction band
potentials are modulated in the single and the multimetal
10
-5
compound. For the high mobility ZnON film, no random
insertion of chemically distinct cations occurs. Therefore, no
Drain current, A

-7
additional disturbance or modulation upon the electronic
10
structure around the conduction band bottom could take
place, and thus, no additional potential barriers should be
10
-9
created. As a result, higher carrier concentration is not re-
quired to overcome the extra potential barriers for higher
-11
mobility as is the case for a multicomponent metal oxide.
10 Vds=0.1V
Vds=1V
Instead, the improvement of carrier transport in the ZnON
Vds=10V film most likely relies on the reduction in defects in the film,
10
-13
as indicated by increased mobility coincident with lower car-
-20 -10 0 10 20 30
rier concentration.
Gate voltage, V
The shelf life of the ZnON film has been investigated.
FIG. 9. Transfer characteristics of a TFT using the ZnON doped with 2.8% Over time, the adsorption of moisture and pollutants on an
Al as the channel material. as-deposited film deposited at low temperature may result in
074512-8 Ye, Lim, and White J. Appl. Phys. 106, 074512 共2009兲

9
the deterioration of the film. Annealing of the film signifi- H. Hosono, N. Kikuchi, N. Ueda, and H. Kawazoe, J. Non-Cryst. Solids
198–200, 165 共1996兲.
cantly increased the resistance of the film to the deterioration 10
K. Nomura, H. Ohta, A. Takagl, T. Kamiya, M. Hirani, and H. Hosono,
from several weeks to more than 660 days even without any Nature 共London兲 432, 488 共2004兲.
encapsulation or passivation. A TFT has been made using the 11
J. Lee, D. Kim, D. Yang, S. Hong, K. Yoon, P. Hong, C. Jeong, H. Park,
ZnON as the channel layer. The field effect mobility of the S. Kim, S. Y. Kim, S. K. Lim, and S. S. Kim, SID 08 Digest, 2008, pp.
fabricated TFT was lowered in order to achieve the desired 626–628.
12
K. Son, T. Kim, J. Jung, M. Ryu, K. Park, B. Yoo, J. Kim, Y. Lee, J.
threshold voltage, off current, subthreshold swing, and many Kwon, S. Lee, and J. Kim, SID 08 Digest, 2008, pp. 433–636.
other performance specifications. Optimization of all inte- 13
H. Q. Chiang and J. F. Wager, Appl. Phys. Lett. 86, 013503 共2005兲.
14
grated process steps is required to achieve higher field effect W. B. Jackson, R. L. Hoffman, and G. S. Herman, Appl. Phys. Lett. 87,
mobility, as indicated by the Hall mobility as well as better 193503 共2005兲.
15
M. Futsuhara, K. Yoshioka, and O. Takai, Thin Solid Films 317, 322
overall TFT performance in the future. 共1998兲.
16
J. M. Bian, X. M. Li, X. D. Gao, W. D. Yu, and L. D. Chen, Appl. Phys.
ACKNOWLEDGMENTS Lett. 84, 541 共2004兲.
17
J. G. Ma, Y. C. Lui, R. Mu, J. Y. Zhang, Y. M. Lu, D. Z. Shen, and X. W.
This work was performed with additional funding from Fan, J. Vac. Sci. Technol. B 22, 94 共2004兲.
18
the Applied Materials CTO office. The authors would like to E. Kaminska, A. Poitrowska, J. Kossut, A. Barcz, R. Butkute, W. Dobro-
thank Dr. David Eaglesham, Dr. Omkaram Nalamasu, and wolski, E. Dynowska, R. Jakiela, E. Przezdziecka, R. Lukasiewicz, M.
Aleszkiewicz, P. Wojnar, and E. Kowalczyk, Solid State Commun. 135, 11
Dr. Mark Pinto for providing encouragement and support for 共2005兲.
the initiation of the project as well as helpful discussions. 19
M. L. Tu, Y. K. Su, and C. Y. Ma, J. Appl. Phys. 100, 053705 共2006兲.
20
B. Yao, D. Z. Shen, Z. Z. Zhang, X. H. Wang, Z. P. Wei, B. H. Li, Y. M.
1
T. Minami, MRS Bull. 25, 33 共2000兲. Lv, X. W. Fan, L. X. Guan, G. Z. Xing, C. X. Cong, and Y. P. Xie, J. Appl.
2
Y. Ohya, T. Niwa, T. Ban, and Y. Takahashi, Jpn. J. Appl. Phys., Part 1 40, Phys. 99, 123510 共2006兲.
21
297 共2001兲. D. Klaitabtim, S. Pratontep, and J. Nukeaw, Jpn. J. Appl. Phys., Part 1 47,
3
R. L. Hoffman, B. J. Norris, and J. F. Wager, Appl. Phys. Lett. 82, 733 653 共2008兲.
22
共2003兲. JCPDS Card No. 35-0762.
4 23
P. F. Carcia, R. S. Mclean, M. H. Reilly, and G. Nunes, Jr., Appl. Phys. JCPDS Card No. 36-1451.
24
Lett. 82, 1117 共2003兲. Y. F. Yan, S. B. Zhang, and S. T. Pantelides, Phys. Rev. Lett. 86, 5723
5
E. Fortunato, A. Pimentel, L. Pereira, A. Goncalves, G. Lavareda, H. 共2001兲.
25
Aguas, I. Ferreira, C. N. Caralho, and R. Martins, J. Non-Cryst. Solids K. Nomura, T. Kamiya, H. Ohta, K. Ueda, M. Hirano, and H. Hosono,
338–340, 806 共2004兲. Appl. Phys. Lett. 85, 1993 共2004兲.
6 26
T. Hirao, M. Furuta, H. Furuta, T. Matsuda, T. Hiramatsu, H. Hokari, and F. Zong, H. Ma, C. Xue, H. Zhuang, X. Zhang, H. Xiao, J. Ma, and F. Ji,
M. Yoshida, SID 06 Digest, 2006, pp. 18–20. Solid State Commun. 132, 521 共2004兲.
7 27
T. Hirao, M. Furuta, T. Hiramatsu, T. Matsuda, C. Li, H. Furuta, H. M. Futsuhara, K. Yoshioka, and O. Takai, Thin Solid Films 322, 274
Hokari, M. Yoshida, H. Ishii, and M. Kakegawa, IEEE Trans. Electron 共1998兲.
28
Devices 55, 3136 共2008兲. K. Toyoura, H. Tsujimura, T. Goto, K. Hachiya, R. Hagiwara, and Y. Ito,
8
F. Hossain, J. Nishii, S. Takagi, A. Ohtomo, T. Fukumura, H. Fujioka, H. Thin Solid Films 492, 88 共2005兲.
29
Koinuma, and M. Kawasaki, J. Appl. Phys. 94, 7768 共2003兲. EAG’s report of XPS/ESCA analysis to AKT 共2007兲.