Piezoelectric Alloy Films

Abstract

A thin film X.sub.yAl.sub.(1-y)N alloy preferably deposited with an intrinsic tensile stress significantly enhances the piezoelectric properties of AlN. The alloy contains y percent of the compound XN, where X is selected from the group consisting of Yb, Ho, Dy, Lu, Tm, Tb, and Gd. The percentage of XN preferably lies in the range 10-60%, and the stress is preferably in the range 200 MPa-1.5 GPa. The film is useful in MEMS devices.

Claims

1. A piezoelectric thin film made of an alloy Al.sub.(1-y)X.sub.yN, where X is selected from the group consisting of: Yb, Ho, Dy, Lu, Tm, Tb and Gd; where y is the fraction of XN.

2. The piezoelectric thin of claim 1, wherein the fraction y lies in the range 5 to 60%.

3. The piezoelectric thin film of claim 1, wherein the thickness of the film is at least 50 nm.

4. The piezoelectric thin film of claim 3, wherein the film is subjected to an intrinsic tensile stress of at least 200 MPa to increase the piezoelectric coefficient.

5. The piezoelectric thin film of claim 4, wherein the film is subjected to an intrinsic tensile stress of about 1 GPa.

6. The piezoelectric thin film of claim 5, wherein X is selected from the group consisting of: Yb and Gd.

7. A method of making a piezoelectric thin device comprising depositing on a substrate a thin film of Al.sub.(1-y)X.sub.yN, where XN is selected from the group consisting of: Yb, Ho, Dy, Lu, Tm, Tb, and Gd; and y is the fraction of XN by co-reactive sputtering using targets of Al and X.

8. The method of claim 7, wherein the substrate is silicon.

9. The method of claim 8, wherein the fraction y lies in the range 10 to 60%.

10. The method of claim 9, wherein the thickness of the film is at least 50 nm.

11. The method of claim 11, comprising subjecting the thin film to an intrinsic tensile stress to increase the piezoelectric coefficient.

12. The method of claim 11, wherein the thin film is subjected to an intrinsic tensile stress of about 1 GPa.

13. The method of claim 12, wherein X is Dy.

14. A piezoelectric device, comprising the thin film of claim 1 deposited on a substrate.

15. The piezoelectric device of claim 14, wherein the atomic percentage y lies in the range 10 to 60%.

16. The piezoelectric device of claim 14, wherein the thickness of the film is at least 50 nm.

17. The piezoelectric device claim 16, wherein the film is subjected to a tensile stress of up to 1 GPa.

18. The piezoelectric device of claim 17, which is a MEMS device.

19. A piezoelectric thin film made of an alloy Al.sub.(1-y)Sc.sub.yN; where y is the fraction of ScN, and which is subject to an intrinsic tensile stress of at least 200 MPa.

20. A piezoelectric thin film as claimed in claim 19, which is Al.sub.0.50Sc.sub.0.50N.

21. A piezoelectric thin film made of an alloy Al.sub.(1-y)X.sub.yN, where X is selected from the group consisting of: Yb, Ho, Dy, Lu, Tm, Tb, Sc, and Gd; where y is the fraction of XN, wherein the film is subjected to an intrinsic strain of about of 0.2 to 1.5%.

22. A MEMS device comprising the piezoelectric thin film of claim 1.

23. A MEMS device comprising the piezoelectric thin film of claim 21.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The invention will now be described in more detail, by way of example only, with reference to the accompanying drawings, in which:—

[0021] FIG. 1 is a plot of the d.sub.33 coefficient vs. tensile stress for pure AlN;

[0022] FIG. 2 is a plot of the d.sub.33 coefficient vs. tensile stress for Al.sub.0.5Sc.sub.0.5N;

[0023] FIG. 3 shows a related 3×3×3 supercell, and the positioning of the corresponding atoms in a quasi random structure;

[0024] FIG. 4 is a cross sectional view schematically illustrating a piezoelectric device;

[0025] FIG. 5 is a schematic view of a co-reactive sputtering chamber; and

[0026] FIG. 6 is a schematic cross section of a MEMS device containing a piezoelectric film in accordance with embodiments of the invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0027] Embodiments of the invention provide new aluminum nitride (AlN) based thin film alloys incorporating the selected heavy rare earth elements (SHREEs), namely ytterbium gadolinium, dysprosium, holmium, thulium, terbium, and lutetium, that have higher piezoelectric do coefficients than pure aluminum nitride. This improvement can be enhanced by applying a tensile stress of preferably at least 200 MPa and more preferably in the order of 1.0 GPa on the thin film. The results of DFT simulations show that these alloys present piezoelectric d.sub.33 coefficients of up to 18 pm/V more than three times the piezoelectric coefficient of aluminum nitride with no tensile stress applied. In addition, DFT simulations show that the addition of tensile stress on the thin film can potentially increase by 70% the piezoelectric coefficients of the alloys.

[0028] The DFT simulation is started by finding a proper special quasi random structure to represent an alloy and generating 4f electron in core pseudopotentials for the lanthanides with commonly available software. This structure is then stretched and relaxed with DFT simulation software for +−0, 1% of the c-axis lattice parameter for the fully relaxed system. The electric polarization and stress inside the structure is then calculated with Berry phase calculation for these relaxed structures and the d.sub.33 coefficient is calculated from the equation:

d.sub.33≈e.sub.33/c.sub.33=(difference in polarization)/(difference in stress).

[0029] This is a good approximation for high values of d.sub.33. See Tasnadi, F et al. supra, the contents of which are herein incorporated by reference. As a verification of the model, the results for AlN and Al.sub.0.5Sc.sub.0.5N have been reproduced.

[0030] The following table shows the results of DFT simulations for selected alloys.

TABLE-US-00001 TABLE 1 Simulation results for selected SHREE-nitride alloyed with AlN Calculated Calculated DFT effective Alloy d.sub.33 Calculated c.sub.33 e.sub.33 band gap investigated (pm/V) (GPa) (pC/m.sup.2) (eV) Al.sub.0.5Gd.sub.0.5N 12.6 158.4 2.0 1.3 Al.sub.0.5Dy0.sub..5N 18.3 134.9 2.5 2 Al.sub.0.5Ho.sub.0.5N 15.6 152.6 2.4 2.1 Al.sub.0.5Tm.sub.0.5N 16.2 150.7 2.4 2.2 Al.sub.0.5Yb.sub.0.5N 16.8 148.7 2.5 2.2 Al.sub.0.5Lu.sub.0.5N 17.7 145.6 2.6 2

[0031] Simulations for terbium have not yet been completed, but similar results are expected based on its electronic configuration.

[0032] The bandgaps obtained in the DFT calculations are known to be underestimated. In practice, they are expected to be in the vicinity of 4 eV. See Dixit H. et al. Electronic structure of transparent oxides with the Tran-Blaha modified Becke-Johnson potential. J. Phys.: Condens. Matter. 24 (2012) 205503 (9 pp), the contents of which are herein incorporated by reference. The bandgap is significantly above zero, which is a necessary condition for high resistivity of the alloys.

[0033] The output structure of the simulation has been examined to confirm a wurtzite structure. For example, FIG. 3 shows a relaxed 3×3×3 supercell, and the positioning of the corresponding atoms in the special quasi random structure. In FIG. 3, the larger gray spheres represent the SHREE atoms, the darker medium-sized spheres represent Al atoms, and the smaller gray spheres represent nitrogen atoms, as indicated in the figure.

[0034] FIG. 1 shows the effect of applying tensile stress to pure AlN. There is a gradual improvement in the d.sub.33 coefficient as apparent from the following table.

TABLE-US-00002 TABLE 2 Stress (GPa) d.sub.33 (pm/V) 9.06 7.99 3.16 5.99 0 5.24 −1.22 4.99 −8.51 3.8

[0035] FIG. 2 shows the effect of applying tensile stress to Al.sub.0.5Sc.sub.0.5N.

TABLE-US-00003 TABLE 3 Stress (MPa) d.sub.33 (pm/V) 8 27.1 600 33.9 1200 42.5

[0036] The effect of tensile stress is significantly greater than for pure AlN, but as previously noted scandium is an expensive material to work with. Nevertheless, these results show that existing scandium-based alloys can be improved by providing an intrinsic stress. The effect occurs almost immediately with increasing stress, but a practical lower limit is 200 MPa. The applicants have demonstrated that certain elements in the lanthanide series exhibit a similar phenomenon. The following table shows the results for Al.sub.0.5Yb.sub.0.5N.

TABLE-US-00004 TABLE 4 Stress (MPa) d.sub.33 (pm/V) 0 16.7 1000 19.5

[0037] This shows that there is an improvement in the d.sub.33 coefficient with tensile stress, although not as great as with scandium.

[0038] The following table shows the simulation results for lanthanum (La), which is a not member of the selected SHREE elements and lutetium (Lu). They were obtained for a smaller 2×2×2 supercell simulation domain, which tends to overestimate results by 28%.

TABLE-US-00005 TABLE 5 d.sub.33 (pm/V) (2 × 2 × 2) d.sub.33 (pm/V)(2 × 2 × 2) Al.sub.0.5La.sub.0.5N 9.9 Al.sub.0.5Lu.sub.0.5N 22.8

[0039] Lu has a full 4f shell electronic configuration that is much easier to handle in simulations. Simulations using 4f electrons in core pseudopotentials and the usual 4f electrons as valence pseudopotentials give piezoelectric coefficients of 23.2 pm/V and 22.8 pm/V respectively for a given 2×2×2 supercell simulation domain. The similarity of the results support the validity of our simulation method. The validity of the results is further supported by the strong correlation between the calculated SHREE-nitride lattice parameters and the experimental values.

[0040] The simulations for La show that not all lanthanides can be alloyed heavily and give significant increase in d.sub.33. The results for Lanthanum are only marginally better than for pure AlN, and the configuration would not be stable if the alloy were fabricated. It would probably separate in cubic non-piezoelectric material phases.

[0041] A comparison of the results for Al.sub.0.5La.sub.0.5N and Al.sub.0.5Lu.sub.0.5N show the effect of change in structure. Al.sub.0.5Lu.sub.0.5N remains almost wurtzite like (same structure as base AlN), whereas Al.sub.0.5La.sub.0.5N does not.

[0042] The alloys in accordance with embodiments of the invention are manufactured generally in accordance with the techniques disclosed in U.S. Pat. No. 7,758,979, the contents of which are herein incorporated by reference. However, the SHREE-based alloys should be less expensive to manufacture than Scandium-based materials. Scandium is extremely scarce and hard to refine.

[0043] Piezoelectric materials with larger d.sub.33 piezoelectric coefficients are essential for advanced piezoelectric MEMS devices, such as sensors, resonators, piezoelectric accelerometers, and gyroscopes.

[0044] The piezoelectric device 1 shown in FIG. 4 comprises a thin film 2 of Al.sub.(1-y)X.sub.yN, where X is selected from the group consisting of: Yb, Ho, Dy, Lu, Tm, Tb, and Gd; and y is the atomic fraction of XN deposited on a substrate 2. The material of the substrate 2 could, for example, be single-crystal silicon or the like, but other suitable materials may be employed such as sapphire, molybdenum or platinum. Buffer layers (not shown) could also be incorporated between the substrate 3 and the film 2, to better control the properties of the film 2.

[0045] In order to manufacture the device in accordance with the invention, as shown in FIG. 5, the temperature controlled substrate holder and substrate 3 is placed in a sputtering chamber 6 with two targets 7, 8, an inlet port 9 for the sputtering gas, and an outlet port 10. In this example, the sputtering gas is a mixture of nitrogen, which is the reactive gas, and argon, which helps the sputtering, although it will be appreciated that other mixtures could be employed. The first target 7 is aluminum, and the second target 8 is the SHREE element X, for example, ytterbium.

[0046] The Al atoms from the target 7 and the X atoms from the target 8 are deposited on the silicon substrate and react with the reactive gas, in this case nitrogen, to form the Al.sub.(1-y)X.sub.yN film 2 on the substrate 1.

[0047] The tensile stress can be controlled by changing the deposition parameters during sputtering. For example, lower adatom mobility (lower substrate temperature) to control crystallite island growth can lead to induce tensile stress in polycrystalline films.

[0048] Stress can also be generated by using a templating substrate with a higher lattice parameter instead of a bulk substrate, for example, a silicon bulk substrate with an epitaxially grown buffer layer.

[0049] The piezoelectric alloys according to embodiments of the invention show an increase of up to 300% of the piezoelectric coefficient d.sub.33 of the aluminum nitride following incorporation SHREE elements, such as ytterbium or gadolinium. This improvement can reach up to about 500% upon the application of a tensile stress of 1 GPa on the thin layer in the planar direction.

[0050] The alloys can also be made at reduced cost of manufacture compared to scandium-based alloys, which is a consequence of the lower price of the SHREE (except for Lu) targets when compared to a scandium target. For example, in the case of 4″ diameter sputtering targets, the price of ytterbium and gadolinium targets is only of the order $1000 compared to $14,000 for a scandium target.

[0051] The current results are based on simulations. Although the methodology allows accuracy within a few % of the experimental values, the microstructure can be optimized during the deposition steps.

[0052] By way of example, a MEMS device in the form of an inertial sensor (accelerometer) is shown in FIG. 6. This comprises three stacked silicon wafers, namely MEMS substrate wafer 21, membrane wafer 22, and TSV (Though Silicon Via) wafer 23.

[0053] Cavities 26 to contain inertial masses 27 are etched in the Si substrate wafer 21. The second Si wafer 22 is bonded to the first wafer 21, then ground and polished to form a thin silicon layer. The piezoelectric film 28 in accordance with embodiments of the invention and top electrode 29 are deposited and patterned, followed by anisotropic etching of vias 30 and silicon springs (not shown) to form the inertial mass 27.

[0054] Cavities 31 to contain the inertial masses 27 are prepared on the third Si wafer 23, which is bonded to the membrane wafer 22. The wafer 23 is ground and polished then AlCu contacts 32 are deposited on the third wafer 23.

[0055] In the presence of an acceleration the mechanical deformation of the piezoelectric film 28 produces a electrical signal. It will be appreciated that the actual layout (as seen from the top) of the device varies according to the intended application.

[0056] All references introduced above are herein incorporated by reference.