Nickel is an excellent ohmic-contact metal on 4H-SiC. This paper discusses the formation mechanism of nickel ohmic contact on 4H-SiC by assessing the electrical properties and microstructural change. Under high-temperature annealing, the phase of nickel-silicon compound can be observed with X-ray diffraction, and the contact resistance also changes. A comparative experiment was designed to use X-ray diffraction and energy-dispersive spectroscopy to clarify the difference of ohmic-contact material composition and elemental analysis between samples prepared using pulsed laser annealing and rapid thermal annealing. It is found that more Ni2Si and carbon vacancies formed at the interface in the sample prepared using pulsed laser annealing, resulting in a better ohmic-contact characteristic.

  • The ohmic contact resistance of 4H-SiC samples is measured using a transmission-line-model structure.

  • The nickel–silicon compound phase in both pulsed laser annealing (PLA) and rapid thermal annealing (RTA) samples is Ni2Si, which is confirmed by X-ray diffraction.

  • The elemental analysis results prove that PLA samples have more Ni2Si and thus lower contact resistance.

Silicon carbide (SiC) is a wide-bandgap semiconductor material with high thermal conductivity, high breakdown field, high-saturation electron drift velocity, high chemical stability, strong mechanical strength, and other excellent properties, all of which allow the development of high-power electronics applications.1–4 Ohmic contact (OC) plays an important role in SiC power devices, with lower OC resistance being effective in reducing device losses and obtaining better performance.5 There are many ways to reduce OC resistance, such as by changing the OC metal, changing the annealing process, and adopting special surface treatment technologies.6–8 Nickel (Ni) is an excellent OC metal on 4H-SiC. Han et al.9 studied the OC formation mechanism of Ni on n-type 4H-SiC, and herein we study the properties of Ni-based OCs under different annealing processes.

The processes of rapid thermal annealing (RTA) and pulsed laser annealing (PLA) have their own advantages in chip manufacturing, with RTA being relatively inexpensive and widely used. Vassilevski et al.10 deposited 50-nm-thick Ni on SiC and calculated a specific contact resistance (SCR) of 9 × 10−5 Ω cm2 after 2 min of annealing at 1000 °C. Perez et al.11 deposited 100-nm-thick Ni and 30-nm-thick Ti layers on SiC and obtained an SCR of 3 × 10−5 Ω cm2 after 3 min of annealing at 900 °C. Liu et al.12 used a 100-nm-thick Ni layer with 1 min of annealing at 975 °C in an N2 atmosphere and obtained an SCR of 6 × 10−5 Ω cm2.

Most previous research involving RTA has been focused on the OC mechanism and has involved using different contact metals and characterizing the surface morphology. However, wafer processing with RTA is somewhat complicated and involves four steps: (i) contact metal processing on the rear side of the wafer, (ii) Schottky structure formation on the front side, (iii) metal deposition on the rear side, and (iv) passivation layer deposition and curing on the front side. To optimize the process sequence, PLA is introduced, which involves only two steps: (i) Schottky structure formation and passivation on the front side of the wafer, and (ii) contact formation and metal deposition on the rear side.13 Under PLA, the rear side of the wafer can be processed after completing the front structure, and the temperature on the rear surface can be greater than 950 °C and less than 600 °C within 10 μm without affecting the metal properties on the front of the wafer.14 de Silva et al.14 used a laser density of 4.7 J/cm2 to obtain an SCR of 4 × 10−4 Ω cm2, Cheng et al.15 used a laser density of 6 J/cm2 to obtained an SCR of 1.97 × 10−3 Ω cm2, and Rascunà et al.16 studied Ni-based OCs with laser annealing (4.7 J/cm2) to deal with 4H-SiC and obtained an SCR of 5 × 10−5 Ω cm2.

Herein, we seek to characterize the material composition and element proportions of nickel silicide (Ni2Si) to evaluate which annealing method gives the best OC properties and why. To assess the similarities and differences of Ni/4H-SiC interfaces after RTA and PLA, we use X-ray diffraction (XRD) and energy-dispersive spectroscopy (EDS).

N-type nitrogen-doped 4H-SiC C-face (0001) bulk substrates with a thickness of 360 μm and a doping concentration of 8.0 × 1018 cm3 were used in the experiments. The samples were cleaned with standard cleaning fluids of RCA, 1#, 2#, and buffered oxide etchant (BOE) cleaning, then rinsed with deionized water and blow-dried with N2. An Ni layer (100 nm) was deposited on the SiC and heat treated by RTA or PLA under different experimental conditions. Each RTA sample was annealed at 950 °C for 2 min in N2 ambient, while each PLA sample was treated under the following parameters: the laser wavelength was 355 nm with a 100-μm beam diameter, the laser power was 2.8 J/cm2, and the scanning speed was 660 mm/s. To measure the SCR, the transmission line model (TLM) structure was employed. The fabrication process of the TLM structure can be divided into three steps: (i) remove the excess Ni layer to leave only Ni2Si; (ii) finish the OC photo-etching and deposit an oxide layer; (iii) deposit a Ti/Ni/Ag metal layer by vacuum evaporation and metal photo-etching. The TLM model consists of five contact pads (200 μm × 60 μm) with spacings of d1 = 30, d2 = 40, d3 = 50, and d4 = 100 μm.

Through data fitting and calculation of the TLM data for the samples, the SCR values for the PLA and RTA samples were 5.1 × 10−5 and 2.1 × 10−4 Ω cm2, respectively. The XRD data show that the Ni/SiC interface in each type of sample was Ni2Si, and the EDS data show that the PLA sample contained more Ni2Si because of the higher proportions of Ni. Based on the analysis, the PLA sample had more free carbon distributed on the SiC surface, therefore there may have been many carbon vacancies inside the SiC. Both of these phenomena could reduce the OC resistance, which is probably why the PLA sample had the better OC characteristic.

A TLM test pattern was prepared to measure the electrical properties of the Ni/SiC contacts on the PLA and RTA samples. As shown in Fig. 1(b), each contact pad of the TLM structure was laterally isolated.17 The IV characteristic curves (d = 30 μm) of the PLA and RTA samples are shown in Fig. 1(a), and each annealing method obtained OC characteristics. The SCR Rc is calculated from13 

Rtotal=2Rc+Rsh×dZ,
(1)

where the contact width is Z = 200 μm, the contact gap d varies from 30 to 100 μm, and Rsh is the sheet resistance. According to the test and data fitting of Rtotal vs contact distance in Fig. 1, the PLA sample had the better OC characteristic (OC resistance: 0.428 Ω) than the RTA sample (OC resistance: 1.745 Ω). The SCR values of the PLA and RTA samples were calculated as 5.1 × 10−5 and 2.1 × 10−4 Ω cm2, respectively. As indicated, for the same wafer processed with different annealing methods, the SCR of the PLA sample was lower than that of the RTA sample. In future work, we will change the PLA parameters (e.g., laser wavelength, beam diameter, laser power, scanning speed) to optimize the interface temperature distribution and obtain better OC quality.

FIG. 1.

Total resistance vs contact distance d (30–100 μm) for rapid thermal annealing (RTA) and pulsed laser annealing (PLA) samples. Inset a: IV curves for d = 30 μm. Inset b: schematic of transmission line model (TLM) structure.

FIG. 1.

Total resistance vs contact distance d (30–100 μm) for rapid thermal annealing (RTA) and pulsed laser annealing (PLA) samples. Inset a: IV curves for d = 30 μm. Inset b: schematic of transmission line model (TLM) structure.

Close modal

We then removed the metal layer to expose the Ni/SiC interface to the surface, and the XRD patterns for the RTA and PLA samples are shown in Figs. 2(a) and 2(b), respectively. The broad peak around 22° in the XRD data for the RTA and PLA samples is because each sample was fixed to the sample table with paraffin. The other diffraction peaks for the RTA and PLA samples belong to SiC (JCPDS#29-1129) and Ni2Si (JCPDS#48-1339). The presence of some low-intensity impurity peaks in the XRD patterns for the RTA and PLA samples might be due to other forms of nickel silicide (e.g., NiSi or NiSi2), but it is clear that Ni2Si is the predominant phase. Note that the diffraction peaks for the PLA sample are much stronger in intensity than those for the RTA sample, which means that the Ni2Si in the PLA sample had better lattice arrangement, and the PLA sample may have had more Ni2Si in its Ni/SiC interface.

FIG. 2.

X-ray diffraction (XRD) patterns for (a) RTA and (b) PLA samples, with standard patterns for (c) SiC JCPDS#29-1129 and (d) Ni2Si JCPDS#48-1339.

FIG. 2.

X-ray diffraction (XRD) patterns for (a) RTA and (b) PLA samples, with standard patterns for (c) SiC JCPDS#29-1129 and (d) Ni2Si JCPDS#48-1339.

Close modal

To determine whether the PLA sample had more Ni2Si in its Ni/SiC interface, we consider the EDS spectra of the PLA and RTA samples as shown in Fig. 3. The integral intensity of the Ni peak for the PLA sample is obviously larger than that for the RTA sample, and the elemental composition ratios (Table I) lead to the same conclusion: the PLA sampling region contained 25.30% Ni, whereas the RTA one contained only 8.33% Ni, which indicates that more Ni2Si was generated in the PLA sample.

FIG. 3.

Energy-dispersive spectroscopy (EDS) results for (a) PLA and (b) RTA samples. For each spectrum, the inset shows the corresponding sampling region.

FIG. 3.

Energy-dispersive spectroscopy (EDS) results for (a) PLA and (b) RTA samples. For each spectrum, the inset shows the corresponding sampling region.

Close modal
TABLE I.

Elemental composition ratios [%] in PLA and RTA sampling regions.

SamplesSiCONi
PLA 27.61 12.73 34.36 25.30 
RTA 38.67 14.78 38.22 8.33 
SamplesSiCONi
PLA 27.61 12.73 34.36 25.30 
RTA 38.67 14.78 38.22 8.33 

According to the literature, the OC characteristic is due to the formation of Ni2Si.18,19 During high-temperature annealing, Ni reacts with SiC to produce Ni2Si, and at the same time free carbon forms and aggregates in the Ni/SiC interface, free carbon being a highly efficient conductor of electricity.20 Therefore, more graphite is distributed in the interface of the PLA sample, so its OC characteristic is better. On the other hand, the free carbon moves to the SiC surface, so there might be many carbon vacancies inside the SiC that can act as donors with positive charges, and the width of the depletion layer and the height of the effective tunneling barrier for electrons transport decrease simultaneously, resulting in the reduction of SCR.21,22 Based on these two phenomena, using PLA can enhance the OC characteristic of 4H-SiC devices by lowering the SCR.

Prepared samples were treated with either PLA (2.8 J/cm2) or RTA (950 °C for 2 min). The PLA sample had a better OC characteristic, with an OC resistance of 5.1 × 10−5 Ω cm2, whereas the SCR of the RTA sample was only 2.1 × 10−4 Ω cm2. Material composition and element analysis using XRD and EDS showed that the Ni/SiC interface of the PLA sample contained more Ni2Si, indicating that more carbon clusters and carbon vacancies were produced. Carbon clusters help to increase the conductivity, and carbon vacancies help to increase the tunneling probability. Therefore, using PLA instead of RTA results in 4H-SiC devices with a better OC characteristic with lower SCR.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

This work was supported by Shenzhen Science and Technology Program (Grant No. KQTD2017033016491218).

1.
Koizumi
A
,
Jun
SD
,
Kimoto
T
.
Temperature and doping dependencies of electrical properties in Al-doped 4H-SiC epitaxial layers
.
J Appl Phys
2009
;
106
(
1
):
013716
.
2.
Liu
G
,
Tuttle
BR
,
Dhar
S
.
Silicon carbide: A unique platform for metal-oxide-semiconductor physics
.
Appl Phys Rev
2015
;
2
(
2
):
021307
.
3.
Mikamura
Y
,
Hiratsuka
K
,
Tsuno
T
, et al 
Novel designed SiC devices for high power and high efficiency systems
.
IEEE Trans Electron Devices
2014
;
62
(
2
):
382
389
.
4.
Palmieri
R
,
Boudinov
H
,
Radtke
C
,
da Silva
 Jr 
EF
.
Effect of the oxidation process on SiO2/4H-SiC interface electrical characteristics
.
Appl Surf Sci.
2008
;
255
(
3
):
706
708
.
5.
Hallin
C
,
Yakimova
R
,
Pecz
B
, et al 
Improved Ni ohmic contact on n-type 4H-SiC
.
J Electron Mater
1997
;
26
(
3
):
119
122
.
6.
Konishi
R
,
Yasukochi
R
,
Nakatsuka
O
, et al 
Development of Ni/Al and Ni/Ti/Al ohmic contact materials for p-type 4H-SiC
.
Mater Sci Eng, B
2003
;
98
(
3
):
286
293
.
7.
Marinova
T
,
Kakanakova-Georgieva
A
,
Krastev
V
, et al 
Nickel based ohmic contacts on SiC
.
Mater Sci Eng, B
1997
;
46
(
1-3
):
223
226
.
8.
Lu
W
,
Mitchel
WC
,
Thornton
CA
, et al 
Ohmic contact behavior of carbon films on SiC
.
J Electrochem Soc
2003
;
150
(
3
):
G177
G182
.
9.
Han
SY
,
Kim
KH
,
Kim
JK
, et al 
Ohmic contact formation mechanism of Ni on n-type 4H–SiC
.
Appl Phys Lett
2001
;
79
(
12
):
1816
1818
.
10.
Vassilevski
K
,
Zekentes
K
,
Tsagaraki
K
, et al 
Phase formation at rapid thermal annealing of Al/Ti/Ni ohmic contacts on 4H-SiC
.
Mater Sci Eng, B
2001
;
80
(
1-3
):
370
373
.
11.
Perez
R
,
Mestres
N
,
Tournier
D
, et al 
Ni/Ti ohmic and Schottky contacts on 4H-SiC formed with a single thermal treatment
.
Diamond Relat Mater
2005
;
14
(
3-7
):
1146
1149
.
12.
Liu
S
,
He
Z
,
Zheng
L
, et al 
The thermal stability study and improvement of 4H-SiC ohmic contact
.
Appl Phys Lett
2014
;
105
(
12
):
122106
.
13.
de Silva
M
,
Ishikawa
S
,
Miyazaki
T
, et al 
Formation of amorphous alloys on 4H-SiC with NbNi film using pulsed-laser annealing
.
Appl Phys Lett
2016
;
109
(
1
):
012101
.
14.
de Silva
M
,
Kawasaki
T
,
Miyazaki
T
, et al 
Formation of epitaxial Ti-Si-C Ohmic contact on 4H-SiC C face using pulsed-laser annealing
.
Appl Phys Lett
2017
;
110
(
25
):
252108
.
15.
Cheng
Y
,
Lu
W
,
Wang
T
,
Chen
Z
.
Fabrication of Ohmic contact on semi-insulating 4H-SiC substrate by laser thermal annealing
.
J Appl Phys
2016
;
119
(
22
):
225705
.
16.
Rascunà
S
,
Badalà
P
,
Tringali
C
, et al 
Morphological and electrical properties of nickel based ohmic contacts formed by laser annealing process on n-type 4H-SiC
.
Mater Sci Semicond Process
2019
;
97
:
62
66
.
17.
Vang
H
,
Lazar
M
,
Brosselard
P
, et al 
Ni–Al ohmic contact to p-type 4H-SiC
.
Superlattices Microstruct
2006
;
40
(
4-6
):
626
631
.
18.
Kuchuk
A
,
Kladko
V
,
Guziewicz
M
, et al 
Fabrication and characterization of nickel silicide ohmic contacts to n-type 4H silicon carbide
.
J Phys: Conf Ser
2008
;
100
(
4
):
042003
.
19.
Vivona
M
,
Greco
G
,
di Franco
S
, et al 
Comparative study of the current transport mechanisms in Ni2Si ohmic contacts on n- and p-type implanted 4H-SiC
.
Mater Sci Forum
2014
;
778-780
:
665
668
.
20.
Siad
M
,
Vargas
CP
,
Nkosi
M
, et al 
Characterisation of Ni and Ni/Ti contact on n-type 4H-SiC
.
Appl Surf Sci
2009
;
256
(
1
):
256
260
.
21.
Cao
Y
,
Nyborg
L
,
Yi
DQ
,
Jelvestam
U
.
Study of reaction process on Ni/4H–SiC contact
.
Mater Sci Technol
2006
;
22
(
10
):
1227
1234
.
22.
Han
SY
,
Shin
JY
,
Lee
BT
,
Lee
J L
.
Microstructural interpretation of Ni ohmic contact on n-type 4H–SiC
.
J Vac Sci Technol, B
2002
;
20
(
4
):
1496
1500
.

Ziwei Zhou, Research and Development Department, Shenzhen BASiC Semiconductor Ltd., China. Her research interests include material characteristics and failure analysis of SiC power devices.

Weiwei He, Research and Development Department, Shenzhen BASiC Semiconductor Ltd., China. His research interests include high-power semiconductor devices such as IGBTs and MOSFETs, and their applications.

Jun Sun, Research and Development Department, Shenzhen BASiC Semiconductor Ltd., China. His research interests include chip design and the fabrication of SiC power devices.