**2. AlN-based SAW devices**

Bulk piezoelectric single crystals, such as LGS and GaPO4, can be used for the implementation of non dispersive SAW devices, such as delay lines, filters and resonators, and the SAW propagation characteristics, such as phase velocity, electroacoustic coupling efficiency K2 and TCD, depend on the crystal cut and SAW propagation direction, as well as on the geometry of the IDTs . The SAW propagation is excited by IDTs located at the free surface of the piezoelectric substrate and directly exposed to the surrounding environment, as shown in figure 1.

Fig. 1. SAW delay line on a piezoelectric substrate

AlN can be grown in thin film form onto non piezoelectric substrates, such as silicon or sapphire, thus allowing the realization of dispersive electroacoustic devices. Moreover, if the AlN film is sandwiched between the IDTs and the ground electrode, four piezoelectric coupling configurations can be obtained by placing the IDTs at the substrate/film interface or at the film surface, with and without the floating electrode opposite the IDTs. These four structures will be mentioned hereafter as substrate/film/IDT (SFT), substrate/IDT/film/metal (STFM)*,* substrate/IDT/film (STF) andsubstrate/metal/film/IDT (SMFT), respectively. Figure 2a shows the top view of a dispersive SAW delay line, and figure 2b shows the cross sections of the four coupling configurations.

When the IDTs are located at the substrate/film interface, the piezoelectric film plays the role of both the acoustic wave transductor and protective layer of the underlying IDTs.

Surface Acoustic Wave Devices for Harsh Environment 583

Fig. 3a. The K2 vs h/λ for SAW propagating along zx-Al2O3/AlN for the four coupling

Fig. 3b. K2 vs h/λ for SAW propagating along zy-Al2O3/AlN for the four coupling

Si/AlN/Pt, being the Pt thickness the running parameter.

Figures 4a and 4b show the K2 vs h/λ for SAW propagation along Si/Pt/AlN and

configurations

structures

These four configurations show frequency dispersive SAW propagation characteristics, that are no longer solely determined by the geometry of the IDTs, the crystals cut, and the SAW propagation direction, but also by the film thickness and the electrical boundary conditions. For SAW propagating along layered structures, the achievable K2 value is sometimes larger than that of the individual piezoelectric materials; it is frequency dispersive and depends on the type and orientation of the piezoelectric material, and it is drastically affected by the location of the IDTs and counter electrode with respect to the piezolectric layer. As an example, figures 3a and 3b show the K2 vs film thickness to wavelength ratio, h/λ, for SAW propagation along *zx*- and *zy*-Al2O3/AlN for the four coupling configurations: the highest K2 values obtainable are about 0.50 and 0.67% for STF x and y propagation (at h/λ ∼ 0.65 and 0.60), 0.49 and 0.64 % for STFM x and y propagation (at h/λ ∼ 0.67 and 0.62), being 0.3% the AlN K2 value.

Fig. 2a. Dispersive SAW delay line

Fig. 2b. The four electroacoustic coupling configurations

These four configurations show frequency dispersive SAW propagation characteristics, that are no longer solely determined by the geometry of the IDTs, the crystals cut, and the SAW propagation direction, but also by the film thickness and the electrical boundary conditions. For SAW propagating along layered structures, the achievable K2 value is sometimes larger than that of the individual piezoelectric materials; it is frequency dispersive and depends on the type and orientation of the piezoelectric material, and it is drastically affected by the location of the IDTs and counter electrode with respect to the piezolectric layer. As an example, figures 3a and 3b show the K2 vs film thickness to wavelength ratio, h/λ, for SAW propagation along *zx*- and *zy*-Al2O3/AlN for the four coupling configurations: the highest K2 values obtainable are about 0.50 and 0.67% for STF x and y propagation (at h/λ ∼ 0.65 and 0.60), 0.49 and 0.64 % for STFM x and y propagation (at h/λ ∼ 0.67 and 0.62), being 0.3%

**SMFT** 

**STFM** 

the AlN K2 value.

Fig. 2a. Dispersive SAW delay line

**SFT**

**STF**

Fig. 2b. The four electroacoustic coupling configurations

Fig. 3a. The K2 vs h/λ for SAW propagating along zx-Al2O3/AlN for the four coupling configurations

Fig. 3b. K2 vs h/λ for SAW propagating along zy-Al2O3/AlN for the four coupling structures

Figures 4a and 4b show the K2 vs h/λ for SAW propagation along Si/Pt/AlN and Si/AlN/Pt, being the Pt thickness the running parameter.

Surface Acoustic Wave Devices for Harsh Environment 585

0.52% at ∼h/λ = 0.555 for Pt h/λ = 10-8 to 10-3. The theoretical data shown in figures 3 and 4 have been evaluated using the PC SAW software developed by Mc Gill University [9]. The

phase velocities along the free and electrically short-circuited surfaces of the AlN film. The phase velocity mvph is obtained by the insertion of a perfectly conductive and infinitesimally thin film at the interfaces where the IDTs and the ground plane are located in each of the four coupling structure. The physical data relative to the elastic, piezoelectric and dielectric constants of AlN film are extracted from [10] and [11] and refer to single crystal AlN thin films grown on the basal plane of Al2O3(0001) by metalorganic vapor deposition. The *TCD* of the bulk piezoelectric crystal depends only on the crystal cut and the SAW propagation direction, while that of a layered structures is frequency dispersive. Because the SAW penetration depth inside the propagating medium is about one wavelength, in a layered

increases with respect to *λ*, more and more of the SAW energy is confined to the film. For small AlN film thickness (*h/λ* < 1) the *TCD* value of the multilayer corresponds approximately to that of the substrate. With increasing the AlN film thickness respect to the acoustic wavelength (*h/λ* ≥ 1) the *TCD* reaches the AlN *TCD* value. If the film and the substrate show opposite sign TCD values, there will be a *h/λ* value at which *TCD* = 0 ppm/°C (the temperature compensated point, TCP). This *h/λ* value represents the film thickness for which the two opposite sign *TCD*s of the film and of the substrate equilibrate to form a thermally compensated structure. Thus high-frequency, enhanced coupling, and thermally compensated elctroacoustic devices can be designed at the proper AlN films

Highly c-axis oriented AlN films were grown at 180°C and at 200°C by rf reactive sputtering technique on the polished surface of (0001) oriented single crystal Al2O3 substrate, on SiO2/Si(100) and Pt/SiO2/Si(100) substrates. The AlN deposition process parameters were the following: gas atmosphere of 100% of N2, high purity (99.999%) 4" diameter Al target disc, RF power 200 watt, background vacuum 5⋅10-8 Torr and pressure during the deposition process 3×10-3 Torr. Before starting the sputtering process, a 30 minute pre-sputtering was performed. The substrate temperature was held at 180 °C during the deposition process. The optimized sputtering parameters ensure AlN films showing a high adhesion to the substrate, c-axis orientation, a columnar growth, smooth surface, and high piezoelectricity; the films are also uniform, stress-free and extremely adhesive to the substrates. The Pt sputtering process parameters were the following: substrate temperature 200°C, gas atmosphere of 100% of Ar, high purity (99.99%) 4" diameter Pt target disc, RF power 150 watt, background vacuum 10-7 Torr and pressure during the deposition process 5×10-3 Torr. The deposition process of both Pt and AlN films is performed subsequently without breaking the vacuum in order to avoid any oxidation effects of the layers. Then the obtained samples were heated at 900°C in air at ambient pressure by a quartz tube furnace, for different lengths of time. The *cold* (20°C) sample was abruptly put inside the furnace preheated at 900°C and the annealing time was measured from the set temperature was reached; then the sample was removed from the furnace and brought abruptly to room

2

f m ph ph f ph v v

<< 1, the most of the SAW energy is confined to the substrate, while, as *h*

<sup>−</sup> <sup>⋅</sup> where <sup>f</sup> vph and mvph are the SAW

v

theoretical K2 has been approximated as

medium, for *h/*

thickness values [12].

**3. Materials and methods** 

λ

Fig. 4a. The K2 vs AlN h/λ for SAW propagation along Si/Pt/AlN, for different Pt thickness values normalized to the acoustic wavelength

Fig. 4b. The K2 vs AlN h/λ for SAW propagation along Si/AlN/Pt, for different Pt thickness values normalized to the acoustic wavelength

In figure 4a the highest K2 values obtainable are in the range 0.5 to 0.63% at ∼h/λ = 0.55 for Pt h/λ = 10-8 to 10-1. In figure 4b the highest K2 values obtainable are in the range 0.5 to

Fig. 4a. The K2 vs AlN h/λ for SAW propagation along Si/Pt/AlN, for different Pt thickness

Fig. 4b. The K2 vs AlN h/λ for SAW propagation along Si/AlN/Pt, for different Pt thickness

In figure 4a the highest K2 values obtainable are in the range 0.5 to 0.63% at ∼h/λ = 0.55 for Pt h/λ = 10-8 to 10-1. In figure 4b the highest K2 values obtainable are in the range 0.5 to

values normalized to the acoustic wavelength

values normalized to the acoustic wavelength

0.52% at ∼h/λ = 0.555 for Pt h/λ = 10-8 to 10-3. The theoretical data shown in figures 3 and 4 have been evaluated using the PC SAW software developed by Mc Gill University [9]. The theoretical K2 has been approximated as f m ph ph f v v 2 v <sup>−</sup> <sup>⋅</sup> where <sup>f</sup> vph and mvph are the SAW

ph phase velocities along the free and electrically short-circuited surfaces of the AlN film. The phase velocity mvph is obtained by the insertion of a perfectly conductive and infinitesimally thin film at the interfaces where the IDTs and the ground plane are located in each of the four coupling structure. The physical data relative to the elastic, piezoelectric and dielectric constants of AlN film are extracted from [10] and [11] and refer to single crystal AlN thin films grown on the basal plane of Al2O3(0001) by metalorganic vapor deposition. The *TCD* of the bulk piezoelectric crystal depends only on the crystal cut and the SAW propagation direction, while that of a layered structures is frequency dispersive. Because the SAW penetration depth inside the propagating medium is about one wavelength, in a layered medium, for *h/*λ << 1, the most of the SAW energy is confined to the substrate, while, as *h* increases with respect to *λ*, more and more of the SAW energy is confined to the film. For small AlN film thickness (*h/λ* < 1) the *TCD* value of the multilayer corresponds approximately to that of the substrate. With increasing the AlN film thickness respect to the acoustic wavelength (*h/λ* ≥ 1) the *TCD* reaches the AlN *TCD* value. If the film and the substrate show opposite sign TCD values, there will be a *h/λ* value at which *TCD* = 0 ppm/°C (the temperature compensated point, TCP). This *h/λ* value represents the film thickness for which the two opposite sign *TCD*s of the film and of the substrate equilibrate to form a thermally compensated structure. Thus high-frequency, enhanced coupling, and thermally compensated elctroacoustic devices can be designed at the proper AlN films thickness values [12].

### **3. Materials and methods**

Highly c-axis oriented AlN films were grown at 180°C and at 200°C by rf reactive sputtering technique on the polished surface of (0001) oriented single crystal Al2O3 substrate, on SiO2/Si(100) and Pt/SiO2/Si(100) substrates. The AlN deposition process parameters were the following: gas atmosphere of 100% of N2, high purity (99.999%) 4" diameter Al target disc, RF power 200 watt, background vacuum 5⋅10-8 Torr and pressure during the deposition process 3×10-3 Torr. Before starting the sputtering process, a 30 minute pre-sputtering was performed. The substrate temperature was held at 180 °C during the deposition process. The optimized sputtering parameters ensure AlN films showing a high adhesion to the substrate, c-axis orientation, a columnar growth, smooth surface, and high piezoelectricity; the films are also uniform, stress-free and extremely adhesive to the substrates. The Pt sputtering process parameters were the following: substrate temperature 200°C, gas atmosphere of 100% of Ar, high purity (99.99%) 4" diameter Pt target disc, RF power 150 watt, background vacuum 10-7 Torr and pressure during the deposition process 5×10-3 Torr. The deposition process of both Pt and AlN films is performed subsequently without breaking the vacuum in order to avoid any oxidation effects of the layers. Then the obtained samples were heated at 900°C in air at ambient pressure by a quartz tube furnace, for different lengths of time. The *cold* (20°C) sample was abruptly put inside the furnace preheated at 900°C and the annealing time was measured from the set temperature was reached; then the sample was removed from the furnace and brought abruptly to room

Surface Acoustic Wave Devices for Harsh Environment 587

The piezoelectric AlN film is *c* axis oriented perpendicularly to the growth plane: in all the samples the AlN (002) and (004) peaks, at ∼ 36° and at ∼ 76°, are visible even after 32 hours

plane; with increasing the annealing time, a very small stress-induced shift in the Pt (111) peak position can be observed, while the same peak becomes narrower indicating a growing (111) fiber texture. As a consequence of the temperature-induced improvement of the Pt local epitaxy, the alignment precision of the AlN film crystalline planes also improves. The XRD data of the outer Pt film showed two peaks at 2θ∼ 40° and 46.3° corresponding to the (111) and (200) orientations, and also a small platinum oxide (220) peak, at ∼66°, clearly visible after the 1st annealing, that does not increase in percentage after the successive

annealing [14] resulted improved, starting from 0.391 and 0.415° of the as grown AlN and Pt films, up to 0.24 and 0.28° after 32 hours annealing. The *c* and *a* lattice parameter of AlN and Pt films of the as-grown samples, calculated from the angular position of the (002) and (111) diffraction peaks, are respectively larger (4.997 Å) and smaller (3.914Å) than the values reported for bulk single crystal AlN and Pt [13], as a consequence of the lattice mismatch between the two materials. Figures 6 and 7 show the FWHM of the AlN(002) and Pt(111) peaks, and the *c* and *a* lattice parameters of AlN and Pt films vs the

∼40° corresponds to the Pt film strongly oriented along the (111)

ϑ ϑ

− scan before and after the

annealing. The peak at 2

annealing time.

ϑ

thermal cycles [14]. The AlN and Pt FWHM of the 2

Fig. 6. FWHM of the AlN(002) and Pt(111) peaks vs annealing time

After the first annealings the AlN *c* parameter relaxes to the bulk while the Pt *a* parameter slightly decreases. Further annealings result in a permanent in-plane compressive stress for both the AlN and Pt films. The AlN(002) and Pt(111) peaks' FWHM decreases with increasing the annealing time indicating a decrease of inhomogeneous strain distribution.

temperature. A temperature ramp of 1°C/s was measured by a thermocouple after the insertion of the *cold* sample inside the furnace. The furnace tube was not hermetically sealed, so ambient air was present during the loading and unloading of samples [14]. The structural properties of annealed Pt and AlN films were investigated by by X-ray diffraction measurements (XRD) a Seifert XRD 3003P performed on a a Seifert XRD 3003P diffractometer operating in the Bragg-Brentano geometry using Cu-Kα radiation (λ = 1.5418 Å) and the diffracted intensities were collected in θ-2θ scan mode in the range 20° <2θ < 80° with step size 0.04°. The reflection peaks of the diffractograms were compared with the standards of the JPCDS database. The crystallite size D was calculated from the Scherrer formula 0.9 <sup>D</sup> B cos λ ϑ <sup>⋅</sup> <sup>=</sup> <sup>⋅</sup> where λ is the wavelength, B the θ-2θ full width at half maximum (FWHM) of the (0002) peak in rad and θ the Bragg angle. The *c* and *a* lattice parameters of the AlN and Pt films were calculated from the angular position of the AlN (002) and Pt (111) diffraction peaks of the θ-2θ scan and compared to the value from ref. 13 for Pt and powder AlN (*a* = 3.9231Å, *c* = 4.979Å). Since the electrical resistivity of the thin conducting films influences the device characteristics (such as insertion loss and Joule heating), the electrical resistance and the surface morphology of the outer Pt electrode were investigated at room temperature after each thermal annealing. The annealing effects on the piezoelectric constant d33 of the AlN films were also estimated [14].

### **3.1 Pt/AlN/Pt/SiO2/Si**

AlN films, 3.15 µm thick, were sputtered on bare and Pt (2200 Å thick) -covered SiO2/Si(100) substrates, being ∼2 µm the silicon oxide thickness; a Pt film (2200 Å thick) was sputtered on the AlN free surface and then the Pt/AlN/Pt/SiO2/Si multilayers were heated at 900°C in air for lengths of time ranging from 1 to 32 hours. Figure 5 shows the XRD patterns of the as grown and annealed multilayers [14].

Fig. 5. XRD patterns of Pt/AlN/Pt/SiO2/Si structures: the running parameter represents the annealing time

temperature. A temperature ramp of 1°C/s was measured by a thermocouple after the insertion of the *cold* sample inside the furnace. The furnace tube was not hermetically sealed, so ambient air was present during the loading and unloading of samples [14]. The structural properties of annealed Pt and AlN films were investigated by by X-ray diffraction measurements (XRD) a Seifert XRD 3003P performed on a a Seifert XRD 3003P diffractometer operating in the Bragg-Brentano geometry using Cu-Kα radiation (λ = 1.5418 Å) and the diffracted intensities were collected in θ-2θ scan mode in the range 20° <2θ < 80° with step size 0.04°. The reflection peaks of the diffractograms were compared with the standards of the JPCDS database. The crystallite size D was calculated from the Scherrer

(FWHM) of the (0002) peak in rad and θ the Bragg angle. The *c* and *a* lattice parameters of the AlN and Pt films were calculated from the angular position of the AlN (002) and Pt (111) diffraction peaks of the θ-2θ scan and compared to the value from ref. 13 for Pt and powder AlN (*a* = 3.9231Å, *c* = 4.979Å). Since the electrical resistivity of the thin conducting films influences the device characteristics (such as insertion loss and Joule heating), the electrical resistance and the surface morphology of the outer Pt electrode were investigated at room temperature after each thermal annealing. The annealing effects on the piezoelectric

AlN films, 3.15 µm thick, were sputtered on bare and Pt (2200 Å thick) -covered SiO2/Si(100) substrates, being ∼2 µm the silicon oxide thickness; a Pt film (2200 Å thick) was sputtered on the AlN free surface and then the Pt/AlN/Pt/SiO2/Si multilayers were heated at 900°C in air for lengths of time ranging from 1 to 32 hours. Figure 5 shows the XRD patterns of the

Fig. 5. XRD patterns of Pt/AlN/Pt/SiO2/Si structures: the running parameter represents the

<sup>⋅</sup> <sup>=</sup> <sup>⋅</sup> where λ is the wavelength, B the θ-2θ full width at half maximum

formula 0.9 <sup>D</sup>

**3.1 Pt/AlN/Pt/SiO2/Si** 

annealing time

B cos λ

ϑ

as grown and annealed multilayers [14].

constant d33 of the AlN films were also estimated [14].

The piezoelectric AlN film is *c* axis oriented perpendicularly to the growth plane: in all the samples the AlN (002) and (004) peaks, at ∼ 36° and at ∼ 76°, are visible even after 32 hours annealing. The peak at 2ϑ ∼40° corresponds to the Pt film strongly oriented along the (111) plane; with increasing the annealing time, a very small stress-induced shift in the Pt (111) peak position can be observed, while the same peak becomes narrower indicating a growing (111) fiber texture. As a consequence of the temperature-induced improvement of the Pt local epitaxy, the alignment precision of the AlN film crystalline planes also improves. The XRD data of the outer Pt film showed two peaks at 2θ∼ 40° and 46.3° corresponding to the (111) and (200) orientations, and also a small platinum oxide (220) peak, at ∼66°, clearly visible after the 1st annealing, that does not increase in percentage after the successive thermal cycles [14]. The AlN and Pt FWHM of the 2 ϑ ϑ − scan before and after the annealing [14] resulted improved, starting from 0.391 and 0.415° of the as grown AlN and Pt films, up to 0.24 and 0.28° after 32 hours annealing. The *c* and *a* lattice parameter of AlN and Pt films of the as-grown samples, calculated from the angular position of the (002) and (111) diffraction peaks, are respectively larger (4.997 Å) and smaller (3.914Å) than the values reported for bulk single crystal AlN and Pt [13], as a consequence of the lattice mismatch between the two materials. Figures 6 and 7 show the FWHM of the AlN(002) and Pt(111) peaks, and the *c* and *a* lattice parameters of AlN and Pt films vs the annealing time.

Fig. 6. FWHM of the AlN(002) and Pt(111) peaks vs annealing time

After the first annealings the AlN *c* parameter relaxes to the bulk while the Pt *a* parameter slightly decreases. Further annealings result in a permanent in-plane compressive stress for both the AlN and Pt films. The AlN(002) and Pt(111) peaks' FWHM decreases with increasing the annealing time indicating a decrease of inhomogeneous strain distribution.

Surface Acoustic Wave Devices for Harsh Environment 589

Fig. 8. SEM photo of the as deposited Pt film

Fig. 9. SEM photo of the 32 hours annealed Pt film

Moreover, the as-grown films contain a number of structural defects that anneal out when heat treatment is carried out and, as a result, the film's resistivity decreases. The Pt annealing results in the relaxation of intrinsic stresses, as well as in the redistribution of structural imperfections: grains coalescence take place and the film sheet resistivity

Fig. 7. *c* and *a* lattice parameters of AlN and Pt films vs the annealing time

The measurement of the longitudinal piezoelectric coefficient d33f of the AlN film was done at ambient temperature on the same type multilayer without the outer Pt films, before and after the thermal annealing, with a method described in ref. 15 and based on the direct piezoelectric effect: a longitudinal acoustic wave perturbs the sample via a special probe and the electrical voltage induced in the piezoelectric film is measured. The probe consisted of a metal rod in contact with a Pb(Zr,Ti)O3 (PZT)-based low frequency transducer that was connected to a pulse generator (pulse width 0.1-1.0 ns) to produce longitudinal bursts propagating along the metal rod. The contact between the rod and the piezoelectric film surface resulted in the application of a stress on the surface of AlN films. Stress-induced electrical charges were collected at the piezoelectric film surfaces by electrodes (the metal rod and the conducting substrate) and observed on a scope. The piezoelectric strain constant, d33f, of the tested films was evaluated comparing the film response with the response of a thin single crystal reference sample, whose d33 was known. All the tested films showed to be piezoelectric with a difference in the d33f obtained values not appreciable with this measurement technique because of an error of about 15-20 %. The estimated mean value is in the range from 6.2 and 7.4 pC/N, for both the as grown and all the annealed samples: these values well agree with the corresponding value reported in the available literature [16] that is about 6 pC/N.

The sheet resistivity of the Pt layer deposited on the AlN/Pt/SiO2/Si multilayer free surface was measured at ambient temperature by the four point method, before and after the thermal annealing. It was observed that the sheet resistance values decreases with increasing the annealing time, starting from ∼0.6 Ω/sq, refereed to the unannealed samples, to ∼0.5 Ω/sq, referred to the 32 hours annealed samples [14]. Scanning electron microscopy (SEM) investigations revealed an average grain size increased with increasing the annealing lasting. The as deposited Pt films have small grain size (about 400 nm), as shown in figure 8, and this high density of grain-boundary affects the high resistivity of the films.

Fig. 7. *c* and *a* lattice parameters of AlN and Pt films vs the annealing time

that is about 6 pC/N.

The measurement of the longitudinal piezoelectric coefficient d33f of the AlN film was done at ambient temperature on the same type multilayer without the outer Pt films, before and after the thermal annealing, with a method described in ref. 15 and based on the direct piezoelectric effect: a longitudinal acoustic wave perturbs the sample via a special probe and the electrical voltage induced in the piezoelectric film is measured. The probe consisted of a metal rod in contact with a Pb(Zr,Ti)O3 (PZT)-based low frequency transducer that was connected to a pulse generator (pulse width 0.1-1.0 ns) to produce longitudinal bursts propagating along the metal rod. The contact between the rod and the piezoelectric film surface resulted in the application of a stress on the surface of AlN films. Stress-induced electrical charges were collected at the piezoelectric film surfaces by electrodes (the metal rod and the conducting substrate) and observed on a scope. The piezoelectric strain constant, d33f, of the tested films was evaluated comparing the film response with the response of a thin single crystal reference sample, whose d33 was known. All the tested films showed to be piezoelectric with a difference in the d33f obtained values not appreciable with this measurement technique because of an error of about 15-20 %. The estimated mean value is in the range from 6.2 and 7.4 pC/N, for both the as grown and all the annealed samples: these values well agree with the corresponding value reported in the available literature [16]

The sheet resistivity of the Pt layer deposited on the AlN/Pt/SiO2/Si multilayer free surface was measured at ambient temperature by the four point method, before and after the thermal annealing. It was observed that the sheet resistance values decreases with increasing the annealing time, starting from ∼0.6 Ω/sq, refereed to the unannealed samples, to ∼0.5 Ω/sq, referred to the 32 hours annealed samples [14]. Scanning electron microscopy (SEM) investigations revealed an average grain size increased with increasing the annealing lasting. The as deposited Pt films have small grain size (about 400 nm), as shown in figure 8,

and this high density of grain-boundary affects the high resistivity of the films.

Fig. 8. SEM photo of the as deposited Pt film

Fig. 9. SEM photo of the 32 hours annealed Pt film

Moreover, the as-grown films contain a number of structural defects that anneal out when heat treatment is carried out and, as a result, the film's resistivity decreases. The Pt annealing results in the relaxation of intrinsic stresses, as well as in the redistribution of structural imperfections: grains coalescence take place and the film sheet resistivity

Surface Acoustic Wave Devices for Harsh Environment 591

Figures 10, 11 and 12 show the diffraction patterns of the AlN films, 4.0, 1.5 and 4.7 µm

Fig. 10. XRD pattern of the AlN, 4 µm thick, annealed for 1 (a), 3 (b), 5 (c), and 9 (d) hours

Fig. 11. XRD pattern of the AlN, 1.5 µm thick, annealed for 4, 8 and 10 hours

thick, annealed for different time.

decreases. SEM photo of the 32 hours annealed sample, shown in figure 9, demonstrates that recrystallization of Pt surface occurs [14].

The unannealed outer Pt film was silvery while the annealed films were matt: this color change can be explained with the increase of the Pt grain size. The adhesion strength of the Pt films was good enough to pass a rudimentary "tape test" with a transparent tape after each anneal as well.

### **3.2 c-AlN/(0001)-Al2O3**

Highly c-axis oriented AlN films were grown by rf reactive sputtering technique on the polished surface of (0001) oriented single crystal Al2O3 substrate. Then the as grown films, 0.26 to 4.7 µm thick, were thermally annealed at 900°C in air for 1 to 18 hours and the AlN structural characteristics were evaluated after each thermal cycle. References are available in the literature concerning the high thermal annealing (HTA) of AlN on Si performed at 700 – 1200 °C for 2 to 12 hours in controlled atmosphere (in oxygen or nitrogen flux) or in high vacuum; AlN films on Al2O3 were heated at 950 °C in air for 30 minutes [20] to 1 hour [21], at 900 to 1200 °C for 10 s in flowing N2[22], and at 800 °C for 90 minutes in air [23]; bulk AlN was annealed in oxygen at 900 – 1150 °C for 6 hours [24]. In the present work, the HTA of the AlN films on Si and on Al2O3 substrates have been performed up to 32 and 18 hours, respectively. No damage was observed on the surface of the AlN/Al2O3 film even after 18 hours annealing: the AlN films were still clear, uniform, and extremely adhesive to the substrate. The impact of the annealing on the films structural properties was investigated by XRD before and after undergoing the thermal annealing. The D and *c* parameter of the AlN films thermally annealed for different time periods, are listed in tab. 1.


Table 1. The AlN (002) FWHM, the D and *c* parameter of the AlN films thermally annealed for different time periods

decreases. SEM photo of the 32 hours annealed sample, shown in figure 9, demonstrates that

The unannealed outer Pt film was silvery while the annealed films were matt: this color change can be explained with the increase of the Pt grain size. The adhesion strength of the Pt films was good enough to pass a rudimentary "tape test" with a transparent tape after

Highly c-axis oriented AlN films were grown by rf reactive sputtering technique on the polished surface of (0001) oriented single crystal Al2O3 substrate. Then the as grown films, 0.26 to 4.7 µm thick, were thermally annealed at 900°C in air for 1 to 18 hours and the AlN structural characteristics were evaluated after each thermal cycle. References are available in the literature concerning the high thermal annealing (HTA) of AlN on Si performed at 700 – 1200 °C for 2 to 12 hours in controlled atmosphere (in oxygen or nitrogen flux) or in high vacuum; AlN films on Al2O3 were heated at 950 °C in air for 30 minutes [20] to 1 hour [21], at 900 to 1200 °C for 10 s in flowing N2[22], and at 800 °C for 90 minutes in air [23]; bulk AlN was annealed in oxygen at 900 – 1150 °C for 6 hours [24]. In the present work, the HTA of the AlN films on Si and on Al2O3 substrates have been performed up to 32 and 18 hours, respectively. No damage was observed on the surface of the AlN/Al2O3 film even after 18 hours annealing: the AlN films were still clear, uniform, and extremely adhesive to the substrate. The impact of the annealing on the films structural properties was investigated by XRD before and after undergoing the thermal annealing. The D and *c* parameter of the AlN

films thermally annealed for different time periods, are listed in tab. 1.

Time (hours)

4.7 0 0.384 218 4.995

4 0 0.270 540 4.976

1.5 0 0.289 252 4.978

0.26 0 0.500 220 4.986

Table 1. The AlN (002) FWHM, the D and *c* parameter of the AlN films thermally annealed

FWHMθ-2<sup>θ</sup> (deg)

4 0.281 297 4.977 8 0.289 288 4.978 12 0.290 290 4.978

1 0.269 548 5.024 3 0.264 555 5.026 5 0.262 593 5.027 9 0.261 611 5.037 18 - - -

4 0.332 288 4.984 8 0.350 281 4.989 10 0.410 303 5.001

3 0.422 198 5.048 4 - - -

D (Å) c (Å)

AlN thickness

(µm)

for different time periods

recrystallization of Pt surface occurs [14].

each anneal as well.

**3.2 c-AlN/(0001)-Al2O3** 

Figures 10, 11 and 12 show the diffraction patterns of the AlN films, 4.0, 1.5 and 4.7 µm thick, annealed for different time.

Fig. 10. XRD pattern of the AlN, 4 µm thick, annealed for 1 (a), 3 (b), 5 (c), and 9 (d) hours

Fig. 11. XRD pattern of the AlN, 1.5 µm thick, annealed for 4, 8 and 10 hours

Surface Acoustic Wave Devices for Harsh Environment 593

structural investigation of the annealed films showed that the thermal annealing improved the crystal quality of the AlN films sandwiched between the Pt films, as confirmed by the decreased FWHM of the rocking curves, and the film piezoelectric d33f coefficient resulted unaffected by the temperature. The study of the electrical, morphological and structural characteristics of the Pt electrode revealed a dense surface with a hillock-free morphology, confirming that Pt is the material of choice when a high oxidation resistance is required for metallic components within devices operating at elevated temperatures. The AlN films on sapphire show a lattice elongation, perpendicular to the growth plane, that is usually associated with the compressive stress caused in the growth plane by the lattice mismatch between the film and the substrate. After the first few hours annealing, the FWHM of the (002) AlN peak decreases showing an improvement in the film texture; further annealing results in the FWHM broadening, whose magnitude depends on the film thickness. The structural investigations of thin (0.26 µm) and thick (from 2 to 4.7 µm) annealed AlN films revealed that the film behaviour in harsh environment is strongly affected by the film properties. Thick films, whose structure is more relaxed than the thin one, is able to survive to high temperature without suffering significant deterioration for longer annealing times than the thin one. The obtained results confirm that, during the HTA, the Pt film on the substrate surface is protected by the AlN film, while the Pt film directly exposed to the ambient conditions acts as a protective layer with respect to the AlN film; thus the AlNbased STFM coupling configuration to be implemented on sapphire or silicon substrates is an attractive alternative to langasite and GaPO4 for the development of microwave

The author wishes to thank Mr. P.M. Latino for his technical support in the development of

[2] M. Pereira da Cunha, M.P. Saulo de A. Fagundes, ú *Ultrasonics Symposium*, Vol. 1, 283

[3] Maurício Pereira da Cunha, Eric L. Adler and Donald C. Malocha, ú *Ultrasonics* 

[4] Kar, J.P., Mukherjee, S., Bose, G., Tuli, S., Myoung, J.M., *Materials Science and Technology*,

[5] Zhigang Lin and Yoon, R.J., International Symposium on *Advanced Packaging Materials:* 

[6] C. Zolper, D. J. Rieger, and A. G. Baca S. J. Pearton and J. W. Lee R. A. Stall, *Appl. Phys.* 

[8] M.J.Vellekoop, E.Nieuwkoop, J.C.Hsaartsen, A.Venema, ú *Ultrasonics Symposium*,Vol. 1,

[9] E.L. Adler, G.W. Farnell, J. Slaboszewicz, C.K. Jen,ú *Ultrasonics Symposium,* Vol. 1, 103

[10] K. Tsubouchi, K. Sugai, N. Mikoshiba, ú *Ultrasonics Symposium,* Vol. 1, 375 (1981).

[1] Hauser, R. Reindl, L. Biniasch, J., ú *Ultrasonics Symposium*, Vol. 1, 192 (2003).

[7] S.M.Middelhoek, S.A.Audet, *Silicon sensors*, Academic Press, London 1989.

electroacoustic devices able to work at high temperatures.

**5. Acknowledgements** 

**6. References** 

(1998).

25 1023 (2009).

*Lett.* 69, 538 (1996).

375 (1981).

(1982).

the technological processes and the HTA.

*Symposium*, Vol.1 169 (1999).

*Processes, Properties and Interfaces*, 156 (2005).

Fig. 12. XRD pattern of the AlN, 4.7 µm thick, annealed for 4, 8 and 12 hours

The AlN (0002) peak at approximately 2θ = 36° of the as deposited films, 0.26 to 4.7 µm thick, showed a FWHM (from θ-2θ scan) in the range 0.5° to 0.27°. Both the as grown and the annealed samples exhibit a strong peak at 2θ ∼ 42° due to the (0006) reflection of the Al2O3 substrate, and one small peak at 76°, corresponding to the (0004) reflections of the wurtzite AlN structure, not shown in figures 10 to 12. The 2nd order peaks of AlN(0002) and Al2O3 (00001) are clearly visible at ∼32.4° and ∼37.5°. After 4 hours at 900°C in air, the (0002) peak of the thin AlN film (0.26 µm thick) broadens completely as well as that of the 4 µm thick AlN film after 18 hours annealing. No other AlN phases are present, nor AlN-oxide traces are evident in the spectra after the annealing; only a decrease in the 2θ value of the (0002) peak is observed, resulting in an increase in the c lattice parameter whose values are larger than the bulk lattice parameter [13], indicating the presence of compressive stress in the surface plane. This lattice elongation, perpendicular to the growth plane, is usually associated with the intrinsic compressive stress caused by the lattice mismatch between the AlN and Al2O3, (asapphire-aAlN)/asapphire , that is approximately equal to 30%. Our previous results [25] have shown that the *c*-axis of the as grown AlN films on sapphire relaxes to the bulk value with increasing the film thickness: an interface layer, strained because of the large lattice mismatch, is formed on the sapphire surface and is followed by a columnar AlN layer which runs through the entire thickness of the film. The thinner the AlN film and more is strained and unable to survive to the HTA.

### **4. Conclusions**

Highly c-axis oriented AlN films and thin film stacks of Pt/AlN/Pt were sputtered at 180°C on (0001)Al2O3 substrates and at 200°C on oxidized Si substrates. The multilayers were heated at 900 °C in air up to 32 hours to test their resistance to high temperature. The

Fig. 12. XRD pattern of the AlN, 4.7 µm thick, annealed for 4, 8 and 12 hours

is strained and unable to survive to the HTA.

**4. Conclusions** 

The AlN (0002) peak at approximately 2θ = 36° of the as deposited films, 0.26 to 4.7 µm thick, showed a FWHM (from θ-2θ scan) in the range 0.5° to 0.27°. Both the as grown and the annealed samples exhibit a strong peak at 2θ ∼ 42° due to the (0006) reflection of the Al2O3 substrate, and one small peak at 76°, corresponding to the (0004) reflections of the wurtzite AlN structure, not shown in figures 10 to 12. The 2nd order peaks of AlN(0002) and Al2O3 (00001) are clearly visible at ∼32.4° and ∼37.5°. After 4 hours at 900°C in air, the (0002) peak of the thin AlN film (0.26 µm thick) broadens completely as well as that of the 4 µm thick AlN film after 18 hours annealing. No other AlN phases are present, nor AlN-oxide traces are evident in the spectra after the annealing; only a decrease in the 2θ value of the (0002) peak is observed, resulting in an increase in the c lattice parameter whose values are larger than the bulk lattice parameter [13], indicating the presence of compressive stress in the surface plane. This lattice elongation, perpendicular to the growth plane, is usually associated with the intrinsic compressive stress caused by the lattice mismatch between the AlN and Al2O3, (asapphire-aAlN)/asapphire , that is approximately equal to 30%. Our previous results [25] have shown that the *c*-axis of the as grown AlN films on sapphire relaxes to the bulk value with increasing the film thickness: an interface layer, strained because of the large lattice mismatch, is formed on the sapphire surface and is followed by a columnar AlN layer which runs through the entire thickness of the film. The thinner the AlN film and more

Highly c-axis oriented AlN films and thin film stacks of Pt/AlN/Pt were sputtered at 180°C on (0001)Al2O3 substrates and at 200°C on oxidized Si substrates. The multilayers were heated at 900 °C in air up to 32 hours to test their resistance to high temperature. The

structural investigation of the annealed films showed that the thermal annealing improved the crystal quality of the AlN films sandwiched between the Pt films, as confirmed by the decreased FWHM of the rocking curves, and the film piezoelectric d33f coefficient resulted unaffected by the temperature. The study of the electrical, morphological and structural characteristics of the Pt electrode revealed a dense surface with a hillock-free morphology, confirming that Pt is the material of choice when a high oxidation resistance is required for metallic components within devices operating at elevated temperatures. The AlN films on sapphire show a lattice elongation, perpendicular to the growth plane, that is usually associated with the compressive stress caused in the growth plane by the lattice mismatch between the film and the substrate. After the first few hours annealing, the FWHM of the (002) AlN peak decreases showing an improvement in the film texture; further annealing results in the FWHM broadening, whose magnitude depends on the film thickness. The structural investigations of thin (0.26 µm) and thick (from 2 to 4.7 µm) annealed AlN films revealed that the film behaviour in harsh environment is strongly affected by the film properties. Thick films, whose structure is more relaxed than the thin one, is able to survive to high temperature without suffering significant deterioration for longer annealing times than the thin one. The obtained results confirm that, during the HTA, the Pt film on the substrate surface is protected by the AlN film, while the Pt film directly exposed to the ambient conditions acts as a protective layer with respect to the AlN film; thus the AlNbased STFM coupling configuration to be implemented on sapphire or silicon substrates is an attractive alternative to langasite and GaPO4 for the development of microwave electroacoustic devices able to work at high temperatures.
