1. Introduction

Thin film resistors are used extensively in the field of microelectronics and MEMS industry because they exhibit a wide range of sheet resistance and a low temperature coefficient of resistance (abbreviated as TCR) compared to polysilicon or diffused resistors. One of the most widely used thin film resistors is nickel-chromium (NiCr) film, which has a low TCR between-50 and +50 ppm/℃^[1-3]. Considering the material properties, an addition of silicon to the NiCr film is particularly suitable for reducing TCR since silicon forms several phases with nickel and chromium^[4]. Gawalek prepared NiCrSi film by a casting alloy target of Cr₅₀Ni₄₅Si₅. The sputtering was performed at an argon working pressure of 1.3 Pa with a magnetron power of 1 kW. The sheet resistance of the as-deposited films was about 25 Ω /□. After annealing at 370 ℃ in air for 40 h, the film showed a near-zero TCR with a standard deviation of ± 5 ppm/℃^[5]. Schippel deposited NiCrSi films in a range of the film composition from 5 to 20 at.% Ni, from 20 to 40 at.% Cr and from 50 to 65 at.% Si. After annealing the films in air for 16 h at 250 ℃, the temperature coefficient of resistance was found to be in the range of ± 50 ppm/℃^[4].

Although DC magnetron is more preferable for the deposition of both binary and ternary compound thin film resistors, RF magnetron sputtering has several advantages over DC sputtering, including higher densification and lower defect and so on. Lee fabricated Ni₅₁Cr₄₁Si₈ thin films by DC/RF magnetron sputtering and studied the effect of the process parameters on the electrical properties. For RF and DC power source, resistivity was 172 and 209 μΩ·cm, TCR were-52 and-25 ppm/℃, respectively^[6].

Zhang prepared CrSiNi resistive films on glass and n-type Si (100) substrates by RF magnetron sputtering from a casting alloy target of Cr₁₇Si₈₀Ni₃. Experimental results indicated that the electrical resistivity of the films on glass substrates were higher than that of on Si substrates at the same annealing temperature^[7]. Chiang sputtered the NiCrSi thin film resistor on Al₂O₃ substrate using DC and RF magnetron sputtering techniques. The uniform large grain size with low sheet resistance was obtained by using a high annealing temperature of 450 ℃. In addition, a stable temperature coefficient of resistance (8 ppm/℃) was attained by using RF sputtering with annealing temperature at 400 ℃^[8].

Cheng used Ni₅₅Cr₄₀Si₅ as the target and sputtered the thin-film resistors on Al₂O₃ substrates. NiCrSi thin-film resistors with different thicknesses of 30.8--334.7 nm were obtained by controlling deposition time. The as-deposited thin film resistors were annealed at 400 ℃ under different durations in N₂ atmosphere using the rapid thermal annealing (RTA) process. The annealed NiCrSi thin-film resistors had a low TCR between 0 and +50 ppm/℃^[9].

The above NiCrSi thin films with low temperature coefficient of resistance can meet most demands of microelectronics and MEMS devices. However, these films cannot meet high demands of some sensors, such as thermal converters^[10-13]. Thermal converters are the most accurate standards for the transfer of alternating voltage and current to the equivalent DC quantities. A well-designed thermal converter needs a heater with a TCR of only several ppm/℃, a long-term stability of better than 0.1% per year and small Thomson and Peltier coefficients.

Thin film resistors are very sensitive to the substrate, such as coefficient of thermal expansion, surface morphology, and film stress. Therefore, interest in the direct deposition of NiCrSi thin films on the silicon dioxide film at the surface of single-crystalline silicon wafers had been spurred by their vital importance in microelectronics and MEMS devices. But most of the above mentioned films were deposited on glass or Al₂O₃ substrates except to the film deposited by Zhang^[7].

In this paper, Ni_24.9Cr_72.5Si_2.6 films are deposited on silicon dioxide film by DC and RF magnetron sputtering. Then films are annealed at 450 ℃ under different durations in N₂ atmosphere. The purpose of the present study is to prepare films with a TCR less than 1 ppm/℃.

2. Experimental procedures

The original substrates are (100) n-type, double-sided polished silicon wafers with thickness of 380 μm and resistivity of 3-6 Ω·cm. The substrates are cleaned by the standard procedure and dried by blowing nitrogen gas over them. A 1.0 μm thick silicon dioxide film is thermally grown on both sides of the wafers at 1100 ℃.

Ni_24.9Cr_72.5Si_2.6 films are prepared by DC and RF magnetron sputtering technique from a 24.9 wt% Ni-72.5 wt% Cr-2.6 wt% Si alloy target. The distance between the centre of target and the substrates is 94 mm. The sputtering chamber is initially evacuated to about 6 ×10^-4 Pa. The sputtering gas Ar with a purity of 99.999% at flow of 50 sccm is introduced into the chamber. The target is pre-sputtered for 5 minutes to remove any surface impurities. The sputtering is performed at an argon working pressure of 0.3 Pa with a magnetron power of 150 W. During deposition the substrates are heated to 70 ℃.

After deposition, the wafers' deposited Ni_24.9Cr_72.5Si_2.6 films are placed on a temperature controllable platform. The sheet resistance of thin films is measured by the four-point probe technique at temperature of 20, 50, 100, 150, and 200 ℃. Then these thin films are pushed into an annealing furnace. The films are annealed at 450 ℃ for 30 minutes in nitrogen atmosphere. The sheet resistance of thin films is measured again. Then the films are annealed again at 450 ℃ for 30 min in nitrogen atmosphere. The sheet resistance of thin films is measured for the third time.

Finally, the as-deposited films are patterned and etched using an etching solution (Nitric acid: ammonium ceric nitrate: DI water = 10 ml : 10 g : 30 ml) at temperature of 20 ℃. The thickness of the Ni_24.9Cr_72.5Si_2.6 thin films is measured by a Dektak 150 surface profiler. The experiment results are summarized in Table 1.

3. Result and discussion

3.1 Deposition rate

Fig. 1 shows the thickness of deposited Ni_24.9Cr_72.5Si_2.6 thin films by DC and RF magnetron sputtering as a function of deposition time. The sputtering method influences the deposition rate strongly, i.e. the deposition rate of RF magnetron sputtering is significantly lower than that of DC magnetron sputtering for the same discharge power of 150 W.

This can be interpreted in terms of the lower discharge voltage on the target for the RF magnetron discharge compared to a DC sputtering process because the deposition rate strongly depends on the energy of the sputter gas atoms arriving at the target. The difference in the discharge voltage has been explained by the different roles of the secondary electrons from the target for DC and RF excitation of magnetron plasma. For DC sputtering, ionization in the plasma is effected by fast electrons from the cathode whereas for RF sputtering it is effected by the RF excitation directly^[14].

3.2 Sheet resistance of Ni_24.9Cr_72.5Si_2.6 films before and after annealing

Fig. 2 shows the sheet resistance as a function of film's thickness before and after annealing, respectively. It can be seen from Fig. 2 that the sheet resistance decreases with increasing the thickness of Ni_24.9Cr_72.5Si_2.6 films. In addition, whether before or after annealing, the sheet resistance of Ni_24.9Cr_72.5Si_2.6 films deposited by DC magnetron sputtering is lower than that by RF magnetron sputtering of identical thickness.

Fig. 3 shows the sheet resistance before and after annealing as a function of film's thickness deposited by DC magnetron sputtering and RF magnetron sputtering, respectively. It can be seen from Fig. 3 that the sheet resistance after annealing is reduced compared with that of before annealing except for thinner films deposited by RF magnetron sputtering.

In general, freshly deposited films contain a high concentration of unstable or metastable structures. Annealing can partially eliminate the lattice defects and residual stress. It also results in grain growth and the grain boundary area reducing. All these factors play a role in reducing the sheet resistance of the Ni_24.9Cr_72.5Si_2.6 films. On the other hand, although the Ni_24.9Cr_72.5Si_2.6 thin films are annealed in N₂ atmosphere, a thin oxide layer is still formed on the surface of the Ni_24.9Cr_72.5Si_2.6 films when the wafers are put into or drawn out from the annealing furnace. The oxidation consumes partial Ni_24.9Cr_72.5Si_2.6 films and plays a role in increasing the sheet resistance. Especially for thinner films, the oxidation may have a greater influence than the change of lattice defects, residual stress, grain growth, etc. As a result, the sheet resistance of thinner films may increase after annealing.

3.3 Sheet resistance of Ni_24.9Cr_72.5Si_2.6 films before and after annealing

Fig. 4 shows the dependence of the sheet resistance of Ni_24.9Cr_72.5Si_2.6 films of 126.2 nm deposited by DC magnetron sputtering and films of 66.9 nm deposited by RF magnetron sputtering on temperature before and after 30 minutes annealing, respectively. It can be seen from Fig. 4 that the plotted curve of sheet resistance against temperature after annealing has a better linear relationship than that of before annealing. The reasons are that annealing partially eliminates the lattice defects and residual stress, as a result, the structure of the film shifts from metastable state to stable state.

3.4 Sheet resistance of Ni_24.9Cr_72.5Si_2.6 films before and after annealing

The TCR of thin films are calculated using the following equation.

$\text{TCR}({{t}_{1}})=\frac{1}{{{R}_{\square }}({{t}_{1}})}\frac{{{R}_{\square }}({{t}_{2}})-{{R}_{\square }}({{t}_{1}})}{{{t}_{2}}-{{t}_{1}}},$

(1)

where R_□ (t₁), R_□(t₂) is the sheet resistance of the films at temperature t₁ and t₂, respectively. The R_□(t_i) is measured at temperatures 20, 50, 100, 150, and 200 ℃, respectively.

Fig. 5 shows the temperature coefficient of resistance as a function of film's thickness before annealing, after 30 minutes annealing and 60 min annealing, respectively. The TCR values of Ni_24.9Cr_72.5Si_2.6 thin-films deposited by DC magnetron sputtering are always negative before and after annealing. The TCR values of Ni_24.9Cr_72.5Si_2.6 thin-films deposited by RF magnetron sputtering are first shifted to small positive value as the thin films' thickness increased and then shifted to large negative value before and after 30 min annealing. But that of after 60 min annealing are always negative.

3.5 Optimized process conditions for Ni_24.9Cr_72.5Si_2.6 films with lowest TCR

The annealing response of the TCR is the result of competition between a negative from lattice defects, residual stress, grain growth, etc., and a positive contribution from oxidation. It can be seen from Fig. 5 that the TCR of Ni_24.9Cr_72.5Si_2.6 film with the thickness of 89.1 nm deposited by RF magnetron sputtering is the lowest. Especially, the TCR of the film after 30 min annealing is only-0.86 ppm/℃. It means the negative contribution from lattice defects, residual stress, grain growth, etc., almost equals to the positive contribution from oxidation. After exposing an unprotected film in air for 12 h at 125 ℃ the TCR is still less than-1 ppm/℃.

4. Conclusion

In this study, NiCrSi films of different thickness are deposited on silicon dioxide film by DC and RF magnetron sputtering from a 24.9 wt% Ni-72.5 wt% Cr-2.6 wt% Si alloy target and annealed at 450 ℃ under different durations in N₂ atmosphere. It is found that deposition rate of RF magnetron sputtering is significantly lower than that of DC magnetron sputtering under the same power. The sheet resistance of Ni_24.9Cr_72.5Si_2.6 films deposited by DC magnetron sputtering is lower than that by RF magnetron sputtering of identical thickness. Most importantly, the NiCrSi film with low TCR of only-0.86 ppm/℃ can be achieved by RF sputtering from a 24.9 wt% Ni-72.5 wt% Cr-2.6 wt% Si alloy target and annealing at 450 ℃ for 30 min in nitrogen atmosphere.