Sergey Mishin

Method of controlling coupling coefficient of Aluminum Scandium Nitride deposition in high volume production

In this paper, we present our studies of the influence of the stress on Aluminum Nitride containing various concentrations Scandium (Sc). Coupling coefficient (kt2) was measured across the wafer and wafer to wafer as a function of stress and Sc content of the film. Previous studies demonstrate a considerable increase in kt2 as a function of Sc content of the film [1], [2], [4], [5]. Unfortunately, when deposited on 200 mm wafers we observed that coupling coefficient varies significantly more than that of a standard Aluminum Nitride (AlN). Both stress and concentration of Sc must be controlled across the wafer to achieve uniform coupling coefficient acceptable for production of Bulk Acoustic Resonator (BAW) devices [3], [6], [7]. We were able to control coupling coefficient across the wafer and wafer-to-wafer by adjusting magnetic fields in dual magnetron configuration as well as adjusting concentration of Sc in our two sputtering targets.

Thickness control by ion beam milling in acoustic resonator devices

In this paper, practical aspects of production worthy methods for film uniformity adjustment (trimming) used in manufacturing of Film Bulk Acoustic Resonator (FBAR) filters [1], [2] have been presented. Two-step trimming in conjunction with thickness “smoothing” technique control total thickness range to within less than 8A on product wafers with variable surface film etch rates even with difficult to measure film thickness. Trimming processes were used to allow using one wafer from a batch to provide compensation feedback in the FBAR devices. Combining ion mill with deposition in the same tool produces <0.1% uniformity in the deposited films.

Characterization of reversed c-axis AlN thin films

Background: It is desired to grow AlN in a reversed c-axis configuration to fabricate R-FBARS (Reversed c-axis Film Bulk Acoustic Resonator) with reactively sputtered, thin film Aluminum Nitride (AlN). Previous methods of growing reversed c-axis AlN, result in films with low electro-acoustic coupling constant, low Q, or inability to withstand the Avago Technologies FBAR release process. Conribution/Methods: An AMS Inc. deposition tool, modified to allow independent control and unique processes, was used to deposit AlN on Al, Mo or W bottom electrodes used in this study. AlN films of thickness 1.2 microns were deposited over patterned bottom electrodes, with additional processing used to reverse the c-axis of the AlN. A top electrode was deposited, patterned, and either processing stopped at the transducer point, or the R-FBAR was released from the silicon wafer with HF acid. Results: The Avago Technologies Acoustic Imaging Microscope interferometer (AIM) is used in point mode to determine the AlN c-axis orientation. The transducer or R-FBAR is driven with a 40 kHz sine wave, and the phase of the top surface motion is observed. Both the orientation and the piezoelectric coefficient were determined. To evaluate the material constants of the AlN for the R-FBAR structure, the input RF reflection coefficient vs frequency is measured. From a one-dimensional Mason model for the R-FBAR stack, the AlN material parameters — coupling constant kt2, resonant frequency, velocity, and attenuation, were determined by varying them in the model, to backfit the measured data. R-FBAR resonators with reversed c-axis orientation, termed type CN (“Compression Negative”), as well as FBARS with normal c-axis orientation termed Type CP, (“Compression Positive”), were fabricated. A strong piezocoupling constant was observed, depending on the deposition parameters used. The voltage shift (“Voltco”) coefficient of the resonator resonant frequencies, fS or fP, was observed, compared to the interoferometric observations, and found to be a reliable indicator of c-axis polarity. For the FBARS and R-FBARS reported here, for the Type CP films, a Voltco of +40 kHz/Volt was observed, and for Type CN films, −30 kHz/Volt was observed. Low frequency RBARS and high frequency RSBARS stacked resonator structures were fabricated, and preliminary results are given.

Manufacturability of highly doped aluminum nitride films

There have been several investigations [1], [2], [3], that demonstrated benefits of adding dopants such as (Sc) or combination of other materials, like Zr/Mg for example, to the aluminum nitride (AlN) films in order to increase coupling coefficient (kt^2) of the Bulk Acoustic Wave (BAW) devices. For concentrations below 10% atomic Sc, it is possible to use a single composite target with a standard magnetron design [4]. Most R&D systems that performed initial investigations on AlScN films with high concentration of Sc dopant, used two separate targets with two separate magnetrons: one with pure Al and one with pure Sc with different applying power to compensate for the large difference in sputtering rates of the two materials and get stoichiometric composition. Unfortunately, depositing from two different targets is only viable for low volume R&D experiments. The system described in this article uses standard dual conical magnetron with AC deposition source. Targets are cut into multiple segments as shown in Figure 1 [5]. Based on simple geometry of targets surface, deposited film composition is proportional to the surface of specific pieces of target material. Unfortunately, Al is eroded at much higher rate than Sc at the same potential and same magnetic field. Over the target life, concentration of Sc increases in the deposited films. In order to maintain same Sc composition over the entire target life, it is necessary to vary magnetic field locally over the surface of the Al and Sc pieces to provide same erosion rate of Al vs. Sc at the same target potential. Adjusting magnetic field for each segment of both Al and Sc allows for constant deposited film composition over the entire target life solves this problem.

Improving frequency control of temperature compensated surface acoustic wave devices

In this paper, we demonstrated improvement of frequency uniformity and stability for Temperature Compensated Surface Acoustic Wave (SAW) devices. SiO2 has been used to obtain low Temperature Coefficient (TempCo) in SAW devices for more than three decades [1]. One of the big issues is that SAW devices have to be processed at temperatures below 300C. When low temperature SiO2 is exposed to the ambient environment, it interacts with ambient humidity [2], [3], [5], [6]. Such interactions can change frequency of the SAW devices and can make frequency trimming with a focused Ion Beam [7], [8] extremely challenging. UV and steam treatment of SiO2 improved trimming rate stability on the first trimming [4] and [9], but was not sufficient to provide tight frequency control required for SAW devices after the second trimming. Using silicon nitride (Si3N4) capping layer on top of SiO2 showed some improvement in frequency control after trimming process. Most improvement was obtained using aluminum nitride (AlN) capping layer on top of SiO2 followed by two trimming steps.

Production issues in using Silicon Dioxide films for temperature compensated Bulk and Surface Acoustic Wave devices

In this paper, production oriented aspects of using Silicon Dioxide (SiO2) films in manufacturing of Temperature Compensated Bulk Acoustic Wave (BAW)/ Film Bulk Acoustic Resonator (FBAR) [1], [2] and Surface Acoustic Wave (SAW) devices have been presented. SiO2 has been used to obtain low Temperature Coefficient (TempCo) in acoustic wave devices for more than three decades [3]. One of the big issues is that depending on the method of deposition and the amount of times SiO2 is exposed to the ambient environment [4], it can significantly alter temperature compensating properties of the film as well as etch rate in a thickness trimming process with focused Ion Beam. Plasma Enhanced Chemical Vapor Deposition (PECVD), RF diode and RF magnetron depositions with in-situ thickness trimming and capping layers were tested on the temperature compensated FBAR and SAW structures. Repeatability of the results was tested with different amount of time before processing steps. Best results were obtained using RF diode sputtered SiO2 with in-situ trimming process [5] and in-situ aluminum nitride (AlN) sputtered capping layer.

Improving coupling coefficient distribution on BAW filters manufactured on 200mm wafers

In the past, most BAW and FBAR filters were produced on 150mm wafers. Typical yield on 150mm wafers was > 90% [1], [2]. In the last few years, most filter manufacturers started migrating to 200mm wafers in order to reduce the cost of manufacturing. Unfortunately, due to variation of coupling coefficient (kt^2) [3], [4], [5], yields on 200mm wafers were significantly lower than on 150mm wafers. Stress in the piezoelectric material is the largest variable that changes coupling coefficient [6]. Figure 1. Shows how coupling coefficient varies with stress for devices that use piezoelectric aluminum nitride (AlN) and molybdenum (Mo) electrodes. Depending on the exact processing involved in making a filter, the same AlN deposition process will produce a completely different stress/kt^2 variation across the wafer. Figure 2 illustrates a difference coupling coefficient variation across wafer, using tungsten (W) vs. Mo electrode material while leaving everything else the same. In order to be able to obtain the best coupling coefficient with different processing on 200mm wafer, it is critical to have independent control of film stress across wafer. In this paper, we will demonstrate a method of controlling coupling coefficient across wafer within less than +/-1% (total range) of the target.

Investigation of 20% scandium-doped aluminum nitride films for MEMS laterally vibrating resonators

This paper reports on the investigation of 1 μm thick films of 20% Scandium-doped Aluminum Nitride (ScAlN) for the making of piezoelectric MEMS laterally vibrating resonators (LVRs). The ScAlN films, which can be sputter-deposited such as undoped Aluminum Nitride (AlN) films, were used to demonstrate high performance resonators. These devices showed quality factor (Qs) in excess of 1000 in air centered around 250 and 500 MHz and enhanced electromechanical coupling (kt2) in the range of 3.2–4.5%. This kt2 is double the value of what has been achieved on similar resonators made out of AlN films. A 3-dB Qs of 1300 has been recorded both for 1-port and 2-port resonators at 250 and 500 MHz, while a maximum Qs of 1500 has been recorded for a 1-port resonator at 500 MHz. Along with experimental results from actual devices, this work also reports the etching characteristics of the piezoelectric material under Cl2/BCl3 chemistry to attain high selectivity and straight sidewall with a SiO2 hard mask. More broadly, enhancement of resonators design and fabrication process, suppression of spurious modes and increase in the concentration of Sc (theoretically up to 40%) will lead to significant performance improvements for many classes of piezoelectric MEMS, especially tunable filters.

AlN film stress and uniformity for BAW filters on 200mm wafers

Most standard processes only require controlling average stress on the sputter deposited films.Recently, MEMS devices have moved into high volume applications. Devices such as FBAR filters, MEMS microphones, and cantilever structures have much more sensitivity to stress variation across wafer and across the thickness of the deposited film. Sputter deposited aluminum nitride (AlN) films have polycrystalline and strongly oriented grain structure, starting with extremely compressive film at the initial few thousand angstroms thickness and becoming more tensile as grains get larger. Also most deposition systems have high variation of film stress across wafer due the magnetic field strength and electrical discharge at each racetrack not being selected for the best stress control across wafer. Because most systems have been designed with film thickness uniformity and average stress on a wafer as primary goals, stress across wafer and across film thickness is sometimes highly nonuniform. In this paper we will show how it is possible to control stress for all desired characteristics: average stress, stress uniformity across wafer and across entire film by using appropriate magnetic field design and in-situ ion beam trimming as part of the deposition process. Average stress control cross wafer is achieved by precise adjustment of magnetic field of inner and outer target ring magnetrons. Stress variation across wafer is achieved by varying the ratio of magnetic field strength between two magnetrons as shown in the paper. Stress variation across deposited films is achieved by using multistep deposition/ion mill trimming processes. As result, film stress variation across wafer and across film is maintained at less than +/-100MPa range with thickness uniformity on 200mm wafers maintained at below 0.2% one standard deviation.