Hydrogen-assisted nitrogen incorporation in zinc oxynitride ﬁlms grown by rf sputtering

Zinc oxynitride has good electrical characteristics applicable for thin- ﬁlm transistors in display devices. Since nitrogen has lower reactivity compared to oxygen, tiny amount of background oxygen can prohibit nitrogen incorporation into the ﬁlms. In order to minimise the effects of oxygen, hydrogen gas was introduced during sputtering growth. The main hypothesis was that the hydrogen gas would react with oxygen to lower the oxygen density in the vacuum chamber. The ﬁlms were grown by rf sputtering using zinc metal target and nitrogen gas. Energy dispersive spectroscopy results showed that nitrogen incorporation in the grown ﬁlms was increased by the additional hydrogen. It was also shown that nitrogen pulsing was an effective method to maintain a reli- able target condition.


Introduction:
Recently, zinc oxynitride (ZnON) has attracted the attention for the development of display devices where high mobility thinfilm transistors are required for high-speed data manipulation [1][2][3]. Nitrogen incorporation into ZnON film is affected by many factors, such as background oxygen contamination. Sputtering chambers may have oxygen contamination originated from previous oxide growths or exposure to air. Introducing hydrogen gas during sputtering growth could potentially address this problem, as a large portion of the ionised oxygen would react with the hydrogen, thereby decreasing oxygen density in the vacuum. The function of hydrogen during sputtering has been investigated before [4][5][6]. In [4], ArH + energy distributions showed a large high-energy tail, which increased the sputtering rate. In [5], it was explained that hydrogen converted the insulating surface of the metal target into a conductive surface, which could increase metal emission from the target.
Nitrogen pulsing: One important complication in reactive sputtering is change in the target surface condition or target contamination [7,8]. When gases react with the metallic target, the target surface can become covered with oxide or nitride. This results in a decreased emission of metal from the target. The formation of the cover on the target surface can be prevented by using low flows of reacting gases. Another method to prevent the target contamination is gas pulsing [9]. In case of sputtering using nitrogen gas and Zn metal target, Argon (Ar) gas supply is always on, and N 2 supply is switched between on and off. When N 2 gas is off, the nitrogen-Zn compound on the target surface would be etched away by the Ar ions. This process can also function as a source for N ions since it is difficult to break the triple bond of N 2 molecules. When N 2 supply is on, the nitrogen-Zn compound can be formed again on the target surface.
While N 2 supply is switched between on and off, the condition of the target surface may change significantly, and this change will modify rf energy coupling to the plasma. In order to regulate rf energy transmission, sputtering systems have an impedance matching circuit.

Fig. 2 C-load measured during sputtering growth of samples A (circles and solid line) and C (triangles and dashed line). N 2 flow is shown at the bottom
shows a schematic diagram of an rf sputtering system. During sputtering growth, the matching circuit automatically adjusts series-(C-tune) and parallel-(C-load) capacitances to match the shifting rf condition. By monitoring these capacitances, it would be possible to estimate the condition of the target surface.
Details of experiments: Our sputtering system (SCIEN Tech, Korea) has a 2-inch diameter Zn metal target with a 13.56 MHz rf power source. The target-substrate distance was 9 cm, and rf power during growth was set at 60 W. High vacuum was achieved by a diffusion pump. The pressure during growth was 3 mTorr at 5-sccm Ar, 1.5-sccm N 2 , and 2-sccm H 2 flows. The films were grown on p+ Si substrates with a thermal oxide of 150 nm thickness. Growth parameters for the three samples are listed in Table 1. Ar (5 sccm) and H 2 flows were kept constant. N 2 flow (1.5 sccm) was turned on for 30 s, and remained off for 90 s. Eight periods were grown, and total growth time was 16 min. Film thickness' were measured by a Tencor profiler. X-ray diffractions (XRD) were measured by a diffractometer (Rigaku Ultima III), using Cu Kα line as a source. Energy dispersive spectroscopy (EDS) was measured by a scanning electron microscope (JEOL JSM-6380).
Changes in matching capacitances: Figure 2 shows the matching capacitances measured during the growth of samples A and C. Sample A was grown with N 2 and H 2 , and sample C was grown with N 2 only. The two capacitors in our matching circuit can vary in the range of 0 and 1000 pF. When N 2 was turned on and off, C-load showed stepwise changes. C-tune remained almost constant at 340 pF. Figure 2 shows a large capacitance jump when N 2 is turned on, and we think that this represents the formation of the nitride cover on the target surface. When N 2 is turned off, the capacitance keeps decreasing, which is the result of nitride etch from the target surface. The stepwise capacitance changes were observed in both samples regardless of H 2 flow. Average of C-load for sample A is smaller than that of sample C, and we explain that the additional H 2 removed part of contaminations on the target surface. Our sputtering system does not have a load-lock chamber, and at every opening, the wall is exposed to air. It is possible that the oxygen can also contribute to the formation of the cover on the target. Figure 3 shows XRD measured in samples A, B, and C. Samples A and B were grown with additional H 2 . Data of sample C was magnified by a factor of 5. Although oxygen was not supplied during the growth, all samples show strong ZnO peaks (a: 31.8 o (100),   (101)), which is the result of reduction by the additional H 2 . ZnN peaks were not observed, and this could be due to low N concentration as shown below. The peaks of samples A and B are stronger than those of sample C, and this indicates that the additional H 2 is also helpful for the growth rate.  Table 1, the sam-ple grown at 200°C (sample B) shows larger film thickness. This can be explained as higher reaction rate at the higher temperature.

Conclusion:
We showed that adding hydrogen gas during sputtering growth increased nitrogen concentrations in ZnON films. EDS results showed increased N concentrations and decreased O concentrations by the hydrogen. During sputtering growth, nitrogen pulsing was used to prevent target contamination. Capacitors in the rf matching circuit showed stepwise changes by the nitrogen pulsing, and we explained it by using a model made of target and target cover. XRD data showed strong Zn and ZnO peaks. It appears that there is a significant amount of oxygen contamination in our growth chamber, which could be the result of O 2 adsorption on the chamber wall. It is not clear at this point how the additional hydrogen contributed during sputtering growth. Decrease of oxygen density is one factor, but there are other possibilities. Increased energy profile of Ar ions, target surface modification by the hydrogen, or formation of nitrogen-hydrogen bonding can also occur.