High-Rate Reactive Sputtering

William D. Sproul, Sputtered Films Inc., Santa Barbara, Calif -- Semiconductor International, 4/1/1999

Today's hard disk drive heads are true thin film devices. The detector in a magnetoresistive (MR) head is a nickel iron film on the order of a few hundred angstroms, and in the newer giant magnetoresistive (GMR) heads, the individual metal layers are on the order on a few tens of angstroms. These very thin films, which are quite sensitive to changes in magnetic field, are the heart of the device, but surrounding either the MR or GMR sensor are other materials, the main one being aluminum oxide. In fact, a typical head is composed of 90% to 95% aluminum oxide, where it is used as an electrically insulating and protective material. The base coat is typically on the order of 5-10 µm, whereas the topcoat is typically on the order of 40-60 µm. Although the trend today is for thinner base and topcoats, aluminum oxide still makes up most of the thin film head.

In the manufacture of the head, the aluminum oxide is usually deposited by rf sputtering, an old technique. It works, but it is very slow, requiring up to 20 hrs to deposit the topcoat. It can also create particle problems; the target material is hot pressed aluminum oxide, which upon sputtering can generate particles that get trapped in the film. And, unless oxygen is added to the sputtering atmosphere, the deposited film will be oxygen deficient with a resulting in property degradation.

Aluminum oxide can also be deposited by reactive sputtering. It seems like a simple enough process. All one has to do is to sputter a pure aluminum target in an argon atmosphere while at the same time oxygen is added to the sputtering. As oxygen is added, an oxide film will form on the substrate, but--and this is where problems arise--aluminum oxide will also form on the target surface. When the oxide covers all of the target surface, the target is said to be 'poisoned,' and the deposition rate from the poisoned state is much slower than the deposition rate from the metallic target state.

Fig. 1. Typical hysteresis loop for the reactive sputtering of a metal in an argon/oxygen atmosphere using flow control of the reactive gas.

When the deposition rate drops, consumption of the reactive gas is reduced, because there is less metal to react with the reactive gas, and the partial pressure of the reactive gas in the chamber increases. The amount of this increase can be quite significant as shown in Figure 1. Initially, when the flow of the reactive gas is low, all reactive gas is consumed by the reaction with the metal or is pumped away. However, when the target becomes poisoned, the sputtering rate drops rapidly. Due to the lower sputtering rate, less reactive gas is consumed, and its partial pressure jumps to a much higher level (from points A to B).

This makes it difficult to control the sputtering rate and/or film composition by simply controlling the flow of the reactive gas. Instead, as you will soon see, it is more desirable to use a control methodology based on partial pressure measurements of the reactive gas.

The advantages of pulsed dc power

With the reactive sputtering of a non-conducting material such as aluminum oxide, one must use either rf or pulsed dc power. Conventional dc power will charge up the poisoned layer on the target surface, which will break down in an arc when the charge is sufficiently high. This arcing leads to droplet ejection, which degrades the quality of the deposited film, and it can also damage the power supply. RF power overcomes the charging and arcing problems, but rf power is not as efficient as dc power. For an equivalent amount of power, the deposition rate from dc power is approximately twice that from rf power.

In the past few years, it has been shown that pulsed dc power can overcome the arcing problem that were associated with the reactive dc sputtering of non-conducting materials. Pulsed dc power supplies a negative pulse to sputter the target surface, but then it switches to a positive pulse for a short time to attract electrons back to the target surface to discharge it. This switching back and forth between the negative and positive pulses allows dc power to be used very successfully for the reactive sputtering of non-conducting materials such as titanium oxide or aluminum oxide.

Partial pressure control

Pulsed dc power will also increase the reactive deposition rate for oxides compared with rf power, but when the target becomes poisoned, the rate falls quickly with either rf or dc power. To overcome the loss of deposition rate during reactive sputtering, it is necessary to control the partial pressure of the reactive gas. Instead of controlling the flow of the reactive gas, a signal that is representative of the partial pressure of the reactive gas is used as the feedback signal to the closed loop gas controller. The opening and closing of the gas inlet valve then responds to changes in partial pressure, not changes in flow. The sensor for such a control scheme can be a device such as a mass spectrometer or an optical emission spectrometer. The only requirement for the sensor is that it produces a signal proportional to the reactive gas partial pressure and that it produces this signal in a timely manner.

Fig. 2. Typical hysteresis loop for the reactive sputtering of a metal in an argon/oxygen atmosphere using partial pressure control of the reactive gas.

When partial pressure of the reactive gas is used, the shape of the flow/partial pressure curve changes as shown in Figure 2 for the reactive sputtering of an oxide such as aluminum oxide or titanium oxide. Initially, with partial pressure control, the response is similar to what it is with flow control. There is a large change in the flow for small changes in the partial pressures of the reactive gas. However, at the nose of the hysteresis loop (Point A in Fig. 2), the partial pressure does not increase suddenly, because the consumption of the reactive gas is balanced by its availability. Once the nose of the hysteresis curve has been passed, the flow of the reactive gas initially decreases, as the partial pressure increases. Physically, the edges of the 'race track' region of the sputtering target, where the sputtering rate is lower (the assumption here is that we are doing magnetron sputtering), are being converted to an oxide layer. The higher erosion areas of the target (toward the center of the racetrack) remain in a metallic state, because there is a limited amount of reactive gas in the chamber. Again, consumption of reactive gas is limited by availability.

Eventually, as the partial pressure of the reactive gas is increased to a sufficiently high value, the whole target surface will become fully poisoned, which is at Point B in Figure 2. Here, the deposition rate is very low compared to the rate at Point A, and any further increases in the partial pressure of the reactive gas just led to a linear increase in the flow.

Using a process control technique based on the partial pressure of the reactive gas, it is possible to operate at any point between points A and B, and there are no forbidden compositions as with flow control-based techniques. In practice, the partial pressure of reactive gas that produces a coating with desirable composition and properties usually falls somewhere between Points A and B. The closer one can operate to Point A, the higher the deposition. However, one should not lower the partial pressure too much, or the film will be substoichiometric.

When partial pressure control of the reactive gas is combined with pulsed dc power, it is possible to reactively sputter deposit oxide films at very high rates. The pulsed dc power prevents the debilitating problems of conventional dc, and the partial pressure control prevents the target surface from being completely poisoned. For example with both pulsed dc power and partial pressure control of the reactive gas, stoichiometric aluminum oxide can be deposited at 75% of the metal deposition rate. Compared to 2% of the metal rate using flow control of the reactive gas and RF power, partial pressure control and pulsed DC power give much higher deposition rates for oxides.

William D. Sproul is the General Manager of Magnetic Products at Sputtered Films. Prior to joining Sputtered Films, he was the manager of the Vapor Deposition Coatings Group at BIRL, Northwestern University's industrial research laboratory, and an Adjunct Professor in the Department of Materials Science and Engineering. He is the inventor of the high-rate reactive sputtering process. He holds eight US patents. Dr. Sproul is a past president of the American Vacuum Society.

Talkback

We would love your feedback!

Post a comment

» VIEW ALL TALKBACK THREADS

Related Content

• Topics
• Author

Related Content

 

By This Author

There are no other articles written by this author.

SPONSORED LINKS