Characterization of the surface acoustic wave devices based on ZnO/nanocrystalline diamond structures

reprint

Phys. Status Solidi A 210, No. 8, 1575–1583 (2013) / DOI 10.1002/pssa.201228631 p s sa

statu

s

soli

di

www.pss-a.comph

ysi

ca

applications and materials science

Characterization of the surfaceacoustic wave devices based on ZnO/nanocrystalline diamond structures

Hua-Feng Pang1,2, Luis Garcia-Gancedo3, Yong Qing Fu*,2, Samuele Porro4, Yan-Wei Gu2,J. K. Luo5, Xiao-Tao Zu**,1, Frank Placido2, John I. B. Wilson4, Andrew J. Flewitt3, and W. I. Milne3

1School of Physical Electronics, University of Electronic Science and Technology of China, Chengdu 610054, P. R. China2Thin Film Centre, Scottish Universities Physics Alliance (SUPA), University of the West of Scotland, Paisley PA1 2BE, UK3Electrical Engineering Division, Department of Engineering, University of Cambridge, 9 JJ Thomson Avenue,

Cambridge CB3 0FA, UK4School of Engineering and Physical Sciences, Scottish Universities Physics Alliance (SUPA), Heriot-Watt University,

Edinburgh EH14 4AS, UK5Centre for Material Research and Innovation, University of Bolton, Deane Road, Bolton BL3 5AB, UK

Received 16 September 2012, revised 6 March 2013, accepted 8 March 2013

Published online 5 April 2013

Keywords diamond, nanocrystalline materials, sputtering, surface acoustic waves, thin films, ZnO

*Corresponding author: e-mail [email protected], Phone: þ141 8483563, Fax: þ44-141-8483627** e-mail [email protected], Phone: þ86 028 8320 2130, Fax: þ86 028 8320 2130

Nanocrystalline ZnO films with strong (0002) texture and fine

grains were deposited onto ultra-nanocrystalline diamond

(UNCD) layers on silicon using high target utilization

sputtering technology. The unique characteristic of this

sputtering technique allows room temperature growth of

smooth ZnO films with a low roughness and low stress at high

growth rates. Surface acoustic wave (SAW) devices were

fabricated on ZnO/UNCD structure and exhibited good trans-

mission signals with a low insertion loss and a strong side-lobe

suppression for the Rayleigh mode SAW. Based on the

optimization of the layered structure of the SAW device, a

good performance with a coupling coefficient of 5.2% has been

realized, promising for improving themicrofluidic efficiency in

droplet transportation comparing with that of the ZnO/Si SAW

device. An optimized temperature coefficient of frequency of

�23.4 ppm 8C�1 was obtained for the SAW devices with the

2.72mm-thick ZnO and 1.1mm-thick UNCD film. Significant

thermal effect due to the acoustic heating has been redcued

which is related to the temperature stability of the ZnO/UNCD

SAW device.

� 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1 Introduction Surface acoustic wave (SAW) deviceshave found broad applications in communications, chemicalandbiochemical sensors aswell asmicrofluidics [1–5], and theconventional SAW devices are generally made of either bulkmaterials such as LiNbO3, LiTaO3, and quartz, or thin filmssuch as ZnO and AlN [6–8]. Recently, there have been greatinterests in depositing ZnO thin films on various substratesor interlayers with intrinsic high acoustic velocities such assapphire, diamond, nanocrystalline diamond (NCD) ordiamond-like carbon (DLC) in order to increase the phasevelocity of the acoustic devices [9–13]. These interlayersbeneath the piezoelectric film (i.e. ZnO, AlN) not only narrowthe bandwidth of SAW filters at high frequencies of hundredsof MHz or GHz, but also improve the mass sensitivity of the

SAW sensors [14, 15]. Diamondmaterials includingNCD andultra-nanocrystalline diamond (UNCD) have the highestSAWpropagation velocity, largest elastic modulus and lowestthermal expansion coefficients despite their lack of piezo-electricity [16–19]. Theoretical simulation showed thatacoustic energy can be confined within the ZnO and diamondlayers, and experimental results confirmed that the propa-gation loss of the SAW can be significantly reduced onpolycrystalline diamond films with a small grain size andsmooth surface [7, 10, 15]. Currently, improvement inquality of the ZnO thin films and prevention of significantdissipation of acoustic energy into the substrate are the twomain challenges for the successful fabrication of ZnO-baseddevice for microfluidics and lab-on-chip applications.

� 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1576 H.-F. Pang et al.: Characterization of the surface acoustic wave devicesp

hys

ica ssp st

atu

s

solid

i a

Theoretical modelling based on Campbell and Jonesmethod indicated that ZnO/diamond SAW device couldapproach a phase velocity in the range of 7180–10,568m s�1, a coupling coefficient (k2) of 1.56–7.01%and a temperature coefficient of frequency (TCF) of 22–30 ppm 8C�1 [20]. Based on the finite element analysisresults, Fujii reported a high phase velocity of 11,600m s�1

and a TCF of 22 ppm 8C�1 for ZnO/diamond SAW devices[16]. Wu et al. [21] showed that the ZnO/diamond SAWdevices have phase velocities of �4600–11,400m s�1 and ak2 value of 1.1–2.7%. However, much experimental work isstill needed to optimize ZnO film properties for the bestperformance of the SAW devices for a specific application[7, 20–22]. In microfluidic applications, such as mixing,pumping, and nebulization, the main challenges are toachieve an optimal coating design for a maximum couplingcoefficient and a lowest TCF of the SAWdevices, in order toimprove the efficiency of liquid mixing and ejection atrelatively low frequencies of tens or hundreds of MHz [14,23]. Large coupling coefficient and good thermal stability ofthe ZnO/UNCD SAW devices could be realized through thedeposition of high-quality ZnO and UNCD films as well asoptimization of ZnO/UNCD layered structures.

In previous work, ZnO/UNCD devices with �6mm-thick ZnO film and �1mm UNCD layer were used formicrofluidic applications such as streaming, pumping, andjetting of microdroplets [24]. Further comparison of themicrofluidic efficiency for the ZnO/UNCD SAW deviceswith different thicknesses depends on the performance ofthose devices, including the acoustic properties and thermaleffect of the SAWs. Improvement of the performance ofthose devices operated at a relatively low frequency is asignificant and effective route to enhance the microfluidicefficiency for manipulating the liquid in the SAW-basedmicrofluidics. In this paper, effective optimization of thelayered structure is reported to improve the performance of theZnO/UNCD device using an advanced sputtering techniqueknown as high target utilization sputtering (HiTUS) and hot-filament chemical vapor deposition (HFCVD). Phase velocity,TCF value and coupling coefficient of the ZnO/UNCD SAWdevices were characterized as a function of the thickness of theZnO film and UNCD layer. The thermal effect of the ZnO/UNCD SAW device was also evaluated in consideration ofliquid heating in the droplet manipulation.

2 Experimental UNCD films with thicknesses in therange of 0.5–4mm were grown on four-inch (100) orientedsilicon substrates using a commercial sp3 Diamond Tech-nology system based on an HFCVD technique. Parallelarrayed tungsten filaments with a diameter of 0.12mm werefixed at a distance of 20mm from the substrate surface andheated up to�2000 8C by Joule effect in a vacuum chamber.A gaseous mixture of methane (CH4) and hydrogen (H2) wasintroduced as theworking gas, with 5%concentration ofCH4

in H2 and a chamber pressure of 8 Torr. The substratetemperature was�600 8C. Further details of the UNCD filmdeposition process were reported elsewhere [24, 25].

� 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

ZnO films with various thicknesses ranging from 2.7 to7.7mm were deposited on the UNCD layers using a HiTUStechnique [24, 26–29]. The sputter chamber was pumpeddown to a pressure of 1.0� 10�6mbar and argon gas was thenfed into the chamber. An RF power of 1000W and a bias DCpower of 800W were applied to an antenna coil and target,respectively. A remarkable characteristic of theHiTUS systemis that a remote and high-intensity plasma with large iondensity (typically around 1013 cm�3) and low ion energy(typically lower than 50 eV) is generated in a side arm adjacentto the deposition chamber. The plasma is then launched intothe chamber and steered onto the target using electromagnets.In this way, the plasma is never in direct contact with thesubstrate, avoiding undesired Arþ ion bombardment duringdeposition [26]. Therefore, the resulting films have low stress,low surface roughness and good stoichiometry. ZnO filmswere sputtered from a metallic Zn target (with purity of99.999%) in an Ar:O2 gas flow ratio of 10:7. This conditionprovided optimum film properties at a deposition rate of�50 nmmin�1. To make a comparison with the �6mm ZnOfilm on �1mm UNCD layer, the �6mm ZnO film was alsodirectly deposited on a silicon substrate.

Root mean square (RMS) roughness of the surface of theUNCD film was measured using atomic force microscopy(AFM,Agilent 5500SPM). Cross sectional microstructure ofthe deposited ZnO/diamond films was characterized using ascanning electron microscopy (SEM, Hitachi S-4100).Crystalline structures of the films were determinedusing X-ray diffraction (XRD; Siemens D5000), and theaverage grain sizes were calculated using the Debye–Scherrer formula with the removal of the equipment inducedbeam broadening [30, 31]. The c-axis stress values inthe ZnO films were estimated using the following equation[32]:

sfilm ¼ 2c213 � c33ðc11 þ c12Þ2c13

� cfilm � cbulkcbulk

; (1)

where the elastic stiffness constants c11, c12, c13 and c33 ofthe ZnO are 208.8, 119.7, 104.2, and 213.8 GPa, respect-ively [33]. The last term (cfilm–cbulk)/cbulk is the strain in theZnO film along c-axis, and cfilm is the lattice constant of thefilm obtained from the position of the (0002) diffractionpeak, cbulk is the unstrained lattice parameter of the bulkZnO taken as 0.52069 nm.

SAW delay lines, in the form of split-finger interdigitaltransducers (IDTs), were patterned by a standard photo-lithography process. The IDTs were formed by Cr/Aucontacts with a thickness of �7/50 nm. The two ports of theIDTs are composed of 60 pairs of fingers, an aperture of4900mm and a spatial periodicity of 64mm. The separationbetween the IDTs is 10mm. The transmission and reflectioncharacteristics of the delay lines were characterized using anHP8752A RF network analyzer. The phase velocity wasdetermined from the equation: V¼ f0� l, where f0 and l arethe central frequency and wavelength, respectively.

www.pss-a.com

Phys. Status Solidi A 210, No. 8 (2013) 1577

Original

Paper

The SAW devices were placed inside an environmen-tally-controlled chamber with a temperature regulator andheated from 20 to 160 8C. Reflection signals (S11) of theRayleigh peaks were recorded from the network analyzerusing a data acquisition system when the temperature wasstable for at least 1min. The TCF values were calculatedusing the following equation [31, 34]:

www

TCF ¼ � Df

DT � f0; (2)

Figure 1 AFM images of (a) 6.34mm thick ZnO film on Sisubstrate, (b) 6.06mm thick ZnO film on the 1.14mm thick UNCDlayer, (c) 5.95mm thick ZnO film on the 0.57mm thick UNCDlayer, and (d) 0.57mm thick UNCD film; cross-section SEMimages of ZnO/diamond films on Si wafer with differentthicknesses (hUNCD and hZnO) of the UNCD layer and ZnO film:(e) hUNCD¼ 1.07mm, hZnO¼2.72mm; (f) hUNCD¼ 1.11mm,hZnO¼3.83mm; (g) hUNCD¼ 1.14mm, hZnO¼5.96mm; and(h) hUNCD¼1.06mm, hZnO¼7.68mm.

Figure 2 XRD patterns of ZnO/UNCD films with different thick-nesses of (a) ZnO film and (b) UNCD layer.

where Df and DT are frequency and temperature variations,respectively.

An RF signal generated from a Marconi 2024 signalgenerator was amplified using a power amplifier (Amplifierresearch, 75A250) and then applied onto the IDTs of theSAW device put on a large aluminium holder. The inputpower on the IDTwas controlled bymodifying the amplitudeof the RF signal. The surface temperature of the SAWdeviceunder different powers was detected and recorded using aFlir T400 thermal imaging infrared camera.

3 Results and discussion3.1 Morphology and microstructure Figure 1a–c

show the surface morphologies of three ZnO films measuredusing AFM, one with a thickness of 6.34mm on the Sisubstrate, second onewith a thickness of 6.06mmZnOon the1.14mm thick UNCD and the last one with a 5.95mm thickZnO on the 0.57mm thick UNCD film, respectively. Theircorreponding surface roughnesses are 18.1, 19.2, and17.2 nm. The roughness and lateral feature size of the0.57mm-thick UNCD film (Fig. 1d) are 6.8 nm and 100–300 nm, respectively. The UNCD film has a low roughnesswithout need for surface polishing, and UNCD layers withthicknesses of 0.57–3.32mm will not cause much variationof the surface roughness of the ZnO films. Furthermore, therelatively smooth surface of the ZnO films is beneficial toreduce the propagational loss of the SAW. Figure 1e–hpresent cross-sectional SEM images of the ZnO filmsdeposited on a �1mm-thick UNCD film. The compactcolumnar structure can be observed for all the ZnO films onthe diamond layers. The ZnO films typically grow ascolumns along the c-axis, which results in apparent columnargrain structures. The thicknesses of ZnO films are 2.72, 3.83,5.96, and 7.68mm, and the corresponding UNCD layers,which were designed to be one micron thick, were measuredas 1.07, 1.11, 1.14, and 1.06mm, respectively. The averagethickness and standard error of the UNCD layers are 1.1mmand 18.5 nm, respectively.

XRD diffractograms of the ZnO films of differentthicknesses deposited on the UNCD layers are shown inFig. 2. Only the peaks corresponding to the (0002) ZnOplaneare observed, confirming the preferential c-axis growth of theZnO films. The accurate positions and values of full-width athalf maximum (FWHM) of the (0002) peak, as well as theestimated stress values of all the ZnO films are listed inTable 1. Results reveal that the ZnO films possess relatively

.pss-a.com

lower stresses on the UNCD layer, compared with a largerstress of the ZnO film directly deposited on the Si substrate.This may be attributed to the higher quality of crystallinity oftheZnOfilmon the diamond layer compared to those directlydeposited on the Si. With the increase in film thickness, theZnO film stress decreases on the �1mm thick UNCD layer,whereas the stress of the ZnO filmwith a thickness of�6mmincreases slightly with the increased thickness of the UNCDlayer. The low stress in the ZnOfilm is beneficial not only forthe good adhesion of ZnOfilm on the diamond layer, but alsofor good piezoelectricity of the ZnO films due to the goodcrystallinity.

� 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1578 H.-F. Pang et al.: Characterization of the surface acoustic wave devicesp

hys

ica ssp st

atu

s

solid

i a

Table 1 Characterization results of film thickness hUNCD and hZnO of nanocrystal diamond and ZnO films from XRD analysis; 2u andFWHM value of the ZnO (0002) peak, calculated lattice constant c, stress, and grain size of the ZnO films.

hUNCD (mm) hZnO (mm) 2u (8) FWHM (8) c (nm) sfilm (GPa) grain size (nm)

0 6.34 34.26 0.208 0.52304 �1.051 40.130.57 6.06 34.32 0.195 0.52216 �0.657 42.161.14 5.95 34.30 0.198 0.52242 �0.774 41.521.79 6.17 34.32 0.194 0.52216 �0.657 42.383.32 6.08 34.28 0.201 0.52272 �0.908 40.901.07 2.72 34.26 0.226 0.52300 �1.033 36.371.11 3.83 34.30 0.198 0.52242 �0.774 41.521.06 7.68 34.32 0.181 0.52216 �0.657 45.42

Figure 3 Frequency responses of (a) reflection coefficient S11 and(b) transmission coefficient S21 for the ZnO/UNCD/Si deviceDU0.5;phase velocity of Rayleigh mode and Sezawa mode as functions ofthicknesses of (c) UNCD layer and (d) ZnO film; the data (�) of thephase velocity versus hk are from Ref. [6].

Table 1 also shows that the average grain size of the ZnOfilms (obtained from the Debye–Scherrer formula) increasesslightly from 36.37 to 45.42 nm with the increased thicknessof the ZnO film on the �1mm-thick UNCD layer, but it didnot show much change for the ZnO film of �6mm with theincreased thickness of the UNCD layer. With the increase inthe ZnO film thickness, the corresponding increase in grainsize and decrease in the stress of the ZnO film on the�1mm-thick UNCD layer are attributed to the thermodynamicallydriven coalescence of the crystallites during the recrystalli-zation [35, 36]. Based on the unique configuration of theHiTUS, the grain boundaries of small grains could bemovedinduced by the energies of the continuously sputtering atomsor ions at a high growth rate of 50 nmmin�1, and small grainsmerge to minimize the surface energy, which help eliminat-ing the defects and improving the quality of the ZnO film.

3.2 Surface acoustic wave properties The trans-mission characteristics, phase velocities and TCFs of theZnO/UNCD SAW devices have been evaluated. In order todescribe conveniently those device, the devices withdifferent UNCD thicknesses beneath the �6mm ZnO filmwere labelled as DU0, DU0.5, DU1.1Z5.9, DU1.7, DU3.3,respectively; DZ2.7, DZ3.8, DU1.1Z5.9, and DZ7.6 denote thedevice with different ZnO thicknesses on �1mm UNCDfilms. A typical frequency response of the ZnO/UNCDSAWdevice DU0.5 is shown in Fig. 3a. The resonant peak of thereflection spectrum (S11) at 65.5MHz corresponds to thefundamental Rayleigh mode wave with a phase velocity of4200m s�1. A second peak at 110.7MHz corresponds to aphase velocity of 7100m s�1. This is a guided Sezawa wavepropagating in the interlayer between ZnO and diamond,which occurs when the bulk transverse velocity in thesubstrate (diamond and silicon) is larger than that in the toplayer (ZnO). For the Sezawa mode SAW, a large level ofspurious ripples was observed from the Sezawa modesignal, which is mainly due to the reported triple transiteffect [37–39]. The reflection effects under the metalizedsurface of the ZnO film cause the slight shift of the centerfrequency for the Sezawa mode SAW [37].

Figure 3b shows the corresponding transmission signalS21 of the Rayleigh mode (0th mode) for the same deviceDU0.5. The symmetrical curve of the signal suggests that thefinger-reflection effect can be neglected, which agrees well

� 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

with the predicted results of the coupling-of-modes (COM)theory without consideration of the finger-reflection effect[37]. For the Rayleigh mode wave, the center peak locatesat �65.5MHz with a low insertion loss of �19.8 dB and aside-lobe suppression of 20.4 dB. The insertion lossof ��28 dB and the side-lobe suppression of 8–10 dB forthe transmission signal were reported for the ZnO/NCDSAW device with similar film thickness parameters [18]. LeBrizoual et al. [40] have fabricated the ZnO/free-standingdiamond SAW device with a transmission signal consistingof an insertion loss of��37 dB and a side-lobe suppressionof 10–12 dB for the Rayleigh mode. Results show thatimprovement of the acoustic performance of the ZnO/diamond SAW device depends not only on the UNCD filmbut also on the ZnO film quality and layered structure.Therefore, it is expected that the characteristics of thetransmission signal was improved based on the high qualityZnO film from the HiTUS technique compared with thoseusing conventional sputtering technique.

www.pss-a.com

Phys. Status Solidi A 210, No. 8 (2013) 1579

Original

Paper

The phase velocities of Rayleigh and Sezawa modeSAWs for the device DU0 are 4100 and 6600m s�1,respectively. Adding the UNCD layer beneath the ZnO filmcould increase the velocities of both the waves as expecteddue to the large elastic modulus of diamond material. Withthe increase of the UNCD thickness from 0.57 to 3.32mm,the phase velocities of the Rayleigh mode SAWs wereincreased from 4186 to 4227m s�1, and those of the Sezawamode SAWs were increased from 7000 to 8000m s�1. Thethickness effect of the UNCD layer on the phase velocity ofthe Sezawa mode SAW is more significant than that on theRayleigh mode SAW as shown in Fig. 3c, which is mainlybecause the Sezawa mode is a wave propagating along theinterface layers among the ZnO, diamond layers, and siliconsubstrates. A thicker UNCD layer can effectively reduce theacoustic energy loss during the propagation of the Rayleighmode and Sezawamode SAWs in the UNCD layer, and keepmuch acoustic energy within the ZnO and UNCD layers [10,24]. The results correlate well with the theoretical calcu-lations from finite element analysis, which predict that thepenetration depth of the fundamental mode decreases withincreasing the thickness of the diamond layer in the ZnO/UNCD/Si SAW device [40].

The phase velocities of the Rayleigh and Sezawa modeSAWs decrease with the normalized thickness hk of ZnOfilm as shown in Fig. 3d, where h is the thickness of ZnO filmand k¼ 2p/l, in which l is the wavelength of the IDTs. Itindicates that the thinner is the ZnO film, the higher is thephase velocity of the SAWdevice, especially for those of theSezawa modes. The maximum phase velocity of 8610m s�1

was obtained from the Sezawa mode SAW of the device,DZ2.7. Whereas the maximum value was 4672m s�1 for theRayleigh mode SAW for the device DU0.5. Previouslypublished velocity results of Rayleigh and Sezawa modewaves for the ZnO film SAW on silicon substrates (ZnO filmwas deposited using a standard magnetron sputter) have alsobeen present for comparison [5, 6]. The velocities of theZnO/UNCD/Si SAW devices have higher frequency com-pared to those of ZnO/Si SAW devices with the same ZnOfilm thickness because of the large acoustic velocity indiamond materials. Clearly for both the ZnO/Si SAWdevices and ZnO/diamond/Si devices, there is a thicknesseffect, i.e. the phase velocities of the SAWdevices decreaseswith increase of hk. However, the phase velocities of theZnO/UNCD/Si SAW devices did not significantly decreasewhen the hk value is below 0.8. These can be attributed to theexistence of theUNCD layer, which could help to restrain theSAW from dissipating into the silicon substrate, thusreducing the substrate effect [10, 15, 20].

To estimate the efficiency of the energy conversion fromthe electric signal to the SAW, the electromechanicalcoupling coefficient (k2) was calculated using the followingequation [44–46]:

Figure 4 Electromechanical coupling coefficients of the Rayleighmode SAW and Sezawa mode SAWs as functions of thickness for(a) ZnO film and (b) UNCD layer.

www

k2 ¼ p

4N

G

B

� �f¼f0

; (3)

.pss-a.com

where N is the finger pairs, and G and B are the radiationconductance and susceptance at the central frequency,respectively. G and B were measured from the Smith chartsof the reflection coefficients at the central frequency of theSAW signals. For the ZnO/UNCD SAW device DU0.5, theimpedance at the center frequency of 65.5MHz (theRayleigh mode SAW) was obtained as 10.22þ 3.43jmS(in the form of Gþ jB). The value of k2 can be calculatedbased on Eq. (3), which is 3.90%. The SAW device for theSezawa mode SAW at a frequency of 110.7MHz has animpedance of 20.88þ 13.74jmS, corresponding to thecalculated k2 value of 1.99%. However, it should be notedthat the calculation of k2 for the Sezawa mode SAW basedon Eq. (3) may not be very accurate due to the triple transiteffect, but the data provides useful information in theconsideration of designing the SAW devices for micro-fluidics and biosensing [6, 14, 24].

For the SAW device DU0, the calculated couplingcoefficients of the Rayleigh and Sezawa mode SAWs are1.08% and 0.98%, respectively. As shown in Fig. 4a, afteradding a diamond layer beneath the ZnO film, the couplingcoefficents have been increased. For the Rayleigh modeSAWs, there is a linear trend of increased couplingcoefficientswith the increasedZnOfilm thickness. However,

� 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1580 H.-F. Pang et al.: Characterization of the surface acoustic wave devicesp

hys

ica ssp st

atu

s

solid

i a

Figure 5 (a) Fequency shift with temperature for ZnO/UNCDSAW device DU1.7; TCF as functions of (b) UNCD thickness ofZnO/diamond devices with ZnO thickness of �6mm, (c) ZnOthickness of ZnO/diamond devices with UNCD thickness of�1mm.

there is no clear trend for the data from the Sezawa modeSAWs. The coupling coefficients of the Sezawamode SAWsbecome larger than those of the Rayleigh mode SAWs whenthe hk values are 0.27 and 0.38, which agrees with thetheotical simulation results [20, 21]. However, the couplingcoefficients of the Sezawa mode SAWs are smaller thanthose of Rayleigh mode SAWs when the hk values are 0.58and 0.75. Themaximum value of the Rayleighmode SAW is5.2% at the hk value of 0.75.

Figure 4b shows the coupling coefficients of theRayleigh mode and Sezawa mode SAWs as a function ofUNCD thickness. It indicates that after adding aUNCD layerbeneath the ZnO film, the coupling efficiency has beenimproved signficantly. When the UNCD thickness isincreased from 0.57 to 3.32mm, the coupling coefficientsof the Rayleighmode and Sezawamode SAWsvaried but arestill close to each other. Results show that an UNCD layerwith a sufficiently large thickness could reduce the acousticwave penetrating into the silicon, which decreases the loss ofthe acoustic energy and enhances the conversion of theelectric energy into acoustic energy. However, there is anoptimum thickness combination between the UNCD layerand ZnO film for the performance, and the maximum valueof the coupling coefficient for the Rayleigh mode is�5.04%for the SAW device DU1.7. This improvement is importantfor enhancing the acoustic-liquid interaction in the micro-fludics and biosensing applications [24].

3.3 Thermal stability and acoustic heatingeffect The temperature stability of the ZnO/UNCD/SiSAW devices was evaluated based on the measured TCFvalues. The typical frequency shift and difference ofinsertion loss with the increased temperature are shown inFig. 5a for the ZnO/UNCD SAW device DU1.7. Within thetemperature range studied, a linear relationship between thefrequency shift versus temperature can be observed, whichhas been used to calculate the TCF values. For the ZnO(�6mm) SAW devices with UNCD thicknesses of 1.14 and1.79mm, the TCF values are quite lowwith average values of�28.6 and�26.9 ppm 8C�1 as shown in Fig. 5b. With eitherdecrease of the UNCD layer to 0.57mm or increase of theUNCD layer up to 3.32mm, the values of the TCFs decreaseto �55.9 and �63.8 ppm 8C�1, respectively. Results fromthis study showed that in order to obtain the smallest value ofTCF, the thickness of theUNCD should be�1–1.8mmwhenthe ZnO film is about 6mm. Generally, the TCF isdetermined by the thermal expansion and the changes ofthe stiffness coefficient of the film and substrate [41]. For theZnO/Si SAW devices, the reported values of the TCF are�45 ppm 8C�1 [42]. The introduction of the UNCD inter-layer could effectively increase the temperature stability dueto its lowest thermal expansion coefficient of0.8� 10�6 K�1, compared with 3.2� 10�6 K�1 for the Sisubstrate [5, 43].

Figure 5c shows the changes of the TCF values as afunction of the ZnO film thickness on a �1.0mm-thickUNCD layer for the ZnO/diamond devices. The lowest TCF

� 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

is �23.4 ppm 8C�1, which was obtained from the ZnO/UNCD/Si SAWdeviceDZ2.7. This TCF value is very close tothe theoretical value of�22 ppm 8C�1 in the ZnO/diamond/Si structure [15, 16]. From Fig. 5, it is observed that thetemperature stability has been improved significantly whenthe ZnO thickness is decreased from 7.68 to 2.72mm in thestudied ZnO film thickness range. As there is a largedifference in the thermal expansion coefficients of the ZnOfilm and diamond layers, with the increase of thickness of theZnO layer, the thermal effects of ZnO film on the frequencyshifts gradually become apparent. Considering the micro-fluidic applications, a relatively low TCF is sufficient whenthe SAW device has the largest coupling coefficient.

Significant acoustic heating effect could occur on thesurface of the SAW device, and it is quite critical for SAWapplications in sensing and microfluidics [6, 47, 48]. Atypical thermal image of the ZnO/UNCD SAW device,DU3.3, is shown in Fig. 6a when the IDT was applied with anRF power of 3.6W. The thermal image clearly shows that thetemperature near the IDT is much higher than those in theother area of the SAW device due to the acoustic heatingeffect from SAW [5]. Themeasured temperature versus timeat different RF powers for this device is shown in Fig. 6b,

www.pss-a.com

Phys. Status Solidi A 210, No. 8 (2013) 1581

Original

Paper

Figure 6 (a) The infrared thermal image of the ZnO/UNCD SAWdevice DU3.3 driven by a power of 3.6W, the spot in the centerof the cross shows the temperature of the area in front of the IDT;(b) the temperature variation with time for the ZnO/UNCD SAWdevice DU3.3 by applying different powers from low value to highone.

Figure 7 The temperature variations of the ZnO/UNCD SAWdevices with different thicknesses of the ZnO films and UNCDlayers excited under different powers, the solid lines represent thebest linear fit.

which shows that there is a linear relationship between thetemperature increase and input RF power. The residual heatexisted in the SAWdevice after a series of themeasurementsfrom the lower power to the higher ones makes the baselinetemperature increase continuously as shown in Fig. 6b. Onthe aluminium holder, the temperature became stable whenthe power on the SAW device was applied after �1 s. Themaximum value of the surface temperature of the SAWdevice is 58.2 8C excited by the input power of 23W.

To estimate the heating rate of the thermal effect in theSAW device, the temperature variations as a function of theapplied power has been linearly fitted as shown in Fig. 7. Arelatively low heating rate of 0.67 8C/Wwas obtained for theZnO/UNCD SAW device DZ2.7, which has a goodtemperature stability with a TCF value of �23.4 ppm 8C�1.A relatively large heating rate of 1.26 8C/W is observed forthe ZnO/UNCD SAW device DU3.3, which possesses lowtemperature stability with a TCF value of �63.8 ppm 8C�1.This indicates that significant thermal effect due to theacoustic heating are closely related to the temperaturestability of the ZnO/UNCD SAW device, which is possiblyoriginated from the large energy dissipation of the

www.pss-a.com

interactions with the thermal lattice vibrations during theSAW propagation [49, 50]. In addition, the defects in theZnO films can also cause heat generation with the highfrequency vibration, which is related to the crystallinestructure, grain boundaries, and large surface roughness forthe deposited ZnO films [6]. Therefore, the improvement ofthe temperature stability could effectively reduce the thermaleffect of the ZnO/UNCD SAW device.

4 Conclusions The layered structure of ZnO/UNCDfilm on the silicon wafer is optimized using a novel coatingprocess combining the newHiTUS technologywithHFCVDtechnique. The unique characteristics of the sputteringtechnique allows growth of a smooth ZnO film with a lowroughness of �17–19 nm and low stress under the highgrowth rate. The ZnO/UNCD films with strong (0002)texture and fine grains were successfully used to fabricateSAW devices. The transmission signal obtained for theRayleigh mode SAW is better than those using theconventional magnetron sputtering technique. The phasevelocities of the Rayleigh mode and Sezawa mode SAWsincrease with the decrease in the ZnO films thickness and theSezawa mode SAW showed a relatively more significantthickness effect. The TCFs of the SAW devices have beenmeasured and a value of less than�30 ppm 8C�1 for the TCFcan be obtained when the thickness of the UNCD layerranges from 1.14 to 1.79mm. For the SAW device with ZnOthickness of 7.68mm and UNCD thickness of 1.06mm, theRayleigh mode SAW achieves a coupling coefficient of5.2%. The large coupling coefficient and strong signals arevery important on improving the microfluidic efficiency inthe droplet manipulation comparing with that of the ZnO/SiSAWdevice. The investigation of the surface temperature ofthe ZnO/UNCD SAW device suggests that the improvementof the temperature stability could effectively reduce thethermal effect. The low TCF of �23.4 ppm 8C�1 are

� 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1582 H.-F. Pang et al.: Characterization of the surface acoustic wave devicesp

hys

ica ssp st

atu

s

solid

i a

obtained for the SAW devices with the 2.72mm-thick ZnOand 1.1mm-thick UNCD film. Significant thermal effect dueto the acoustic heating has been redcued which is related tothe temperature stability of the ZnO/UNCD SAW device.

Acknowledgements The authors acknowledge supportfrom Scottish Sensing Systems Centre (S3C), Royal Society-Research Grant (RG090609), Carnegie Trust Funding, RoyalSociety of Edinburgh, Royal Academy of Engineering-ResearchExchanges with China and India Awards, the FundamentalResearch Funds for the Central Universities (ZYGX2009J046and ZYGX2009x007), and the Sichuan Young ScientistsFoundation (2010JQ0006). L. Garcia-Gancedo, J.K. Luo, A.J.Flewitt, and W.I. Milne acknowledge the financial support of theEPSRC, through grants number EP/F063865/1 and EP/F06294X/1.

References

[1] C. C. W. Ruppel, R. Dill, A. Fischerauer, G. Fischerauer, W.Gawlik, J. Machui, F. Mueller, L. Reindl, W. Ruile, G.Scholl, I. Schropp, and K. C. Wagner, IEEE Trans. Ulrason.Ferroelelect. Freq. Contr. 40, 438 (1993).

[2] S. Bender, F. L. Dickert, W. Mokwa, and P. Pachatz, Sens.Actuators B 93, 164 (2003).

[3] K. Lange, B. E. Rapp, and M. Rapp, Anal. Bioanal. Chem.391, 1509 (2008).

[4] T. Franke, A. R. Abate, D. A. Weitz, and A. Wixforth, LabChip 9, 2625 (2009).

[5] X. Y. Du, Y. Q. Fu, S. C. Tan, J. K. Luo, A. J. Flewitt, W. I.Milne, D. S. Lee, N. M. Park, J. Park, Y. J. Choi, S. H. Kim,and S. Maeng, Appl. Phys. Lett. 93, 094105 (2008).

[6] Y. Q. Fu, J. K. Luo, X. Y. Du, A. J. Flewitt, Y. Li, G. H.Markx, A. J. Walton, and W. I. Milne, Sens. Actuators B 143,606 (2010).

[7] H. Nakahata, S. Fujii, K. Higaki, A. Hachigo, H. Kitabayashi,S. Shikata, and N. Fujimori, Semicond. Sci. Technol. 18, S96(2003).

[8] W. Hinsberg, F. A. Houle, J. Hoffnagle, M. Sanchez, G.Wallraff, M. Morrison, and S. Frank, J. Vac. Sci. Technol. B16, 3689 (1998).

[9] T. Mitsuyu, S. Ono, and K. Wasa, J. Appl. Phys. 51, 2464(1980).

[10] V. Mortet, O. A. Williams, and K. Haenen, Phys. StatusSolidi A 205, 1009 (2008).

[11] I.-T. Tang, Y. C. Wang, W. C. Hwang, C. C. Hwang, N. C.Wu, M.-P. Houng, and Y.-H. Wang, J. Cryst. Growth 252,190 (2003).

[12] J. K. Luo, Y. Q. Fu, S. B. Milne, H. R. Le, J. A. Williams, S.M. Spearing, A. J. Flewitt, and W. I. Milne, J. Micromech.Microeng. 17, S147 (2007).

[13] Y.-F. Chiang, C.-C. Sung, and R. Ro, Appl. Phys. Lett. 96,154104-1-3 (2010).

[14] M. F. Hribsek, S. S. Ristic, and B. M. Radojkovic, Acta Phys.Pol. A 117, 794 (2010).

[15] S. Fujii, Phys. Status Solidi A 208, 1072 (2011).[16] S. Fujii, S. Shikata, T. Uemura, H. Nakahata, and H. Harima,

IEEE Trans. Ultrason. Ferroelectr. Freq. Contr. 52, 1817(2005).

[17] S.-H. Seo, W.-C. Shin, and J.-S. Park, Thin Solid Films 416,190 (2002).

[18] W.-C. Shih, M.-J. Wang, and I. N. Lin, Diam. Relat. Mater.17, 390 (2008).

� 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

[19] L. Le Brizoual, T. Lamara, F. Sarry, M. Belmahi, O. Elmaz-ria, J. Bougdira, M. Remy, and P. Alnot, Phys. Status Solidi A202, 2217 (2005).

[20] H. Nakahata, A. Hachigo, K. Higaki, and S. Fujii,IEEE Trans. Ultrason. Ferroelectr. Freq. Contr. 42, 362(1995).

[21] S. Wu, R. Ro, and Z. X. Lin, Appl. Phys. Lett. 94, 032908(2009).

[22] B. Bi, W.-S. Huang, J. Asmussen, and B. Golding, Diam.Relat. Mater. 11, 677 (2002).

[23] M. Maezawa, H. Nomura, and T. Kamakura, Proc. 20th Int.Congr. on Acoustics, ICA 2010 (23–27 August 2010, Sydney,Australia).

[24] Y. Q. Fu, L. Garcia-Gancedo, H. F. Pang, S. Porro, Y. W. Gu,J. K. Luo, X. T. Zu, F. Placido, J. I. B. Wilson, A. J. Flewitt,and W. I. Milne, Biomicrofluidics 6, 024105 (2012).

[25] S. Porro, G. De Temmerman, S. Lisgo, P. John, I. Villal-pando, J. W. Zimmer, B. Johnson, and J. I. B. Wilson, Diam.Relat. Mater. 18, 740 (2009).

[26] L. Garcıa-Gancedo, J. Pedros, Z. Zhu, A. J. Flewitt, W. I.Milne, J. K. Luo, and C. J. B. Ford, J. Appl. Phys. 112, 014907(2012).

[27] J. Pedros, L. Garcıa-Gancedo, C. J. B. Ford, C. H. W. Barnes,J. P. Griffiths, G. A. C. Jones, and A. J. Flewitt, J. Appl. Phys.110, 103501 (2011).

[28] A. J. Flewitt, J. D. Dutson, P. Beecher, D. Paul, S. J.Wakeham, M. E. Vickers, C. Ducati, S. P. Speakman, W.I. Milne, and M. J. Thwaites, Semicond. Sci. Technol. 24,085002 (2009).

[29] L. Garcia-Gancedo, J. Pedros, G. M. Ashley, X. B. Zhao,A. J. Flewitt, W. I. Milne, J. K. Luo, J. R. Lu, andC. J. B. Ford, in: IEEE Ultrasonics Symposium (2010).pp. 1064–1067.

[30] A. L. Patterson, Phys. Rev. 56, 978 (1939).[31] F. Benedic, M. B. Assouar, P. Kirsch, D. Moneger, O. Brinza,

O. Elmazria, P. Alnot, and A. Gicquel, Diam. Relat. Mater.17, 804 (2008).

[32] S. Maniv, W. D. Westwood, and E. Colombini, J. Vac. Sci.Technol. 20, 162 (1982).

[33] Y. Yoshino, M. Takeuchi, K. Inoue, T. Makino, S. Arai, andT. Hata, Vacuum 66, 467 (2002).

[34] J.-J. Chen, F. Zeng, D. M. Li, and F. Pan, Thin Solid Films485, 257 (2005).

[35] J. C. Lodder, T.Wielinga, and J.Worst, Thin Solid Films 101,61 (1983).

[36] F. C. M. van de Pol, F. R. Blom, and Th. J. A. Popma, ThinSolid Films 204, 349 (1991).

[37] C. K. Campbell, Surface Acoustic Wave Devices for Mobileand Wireless Communications (Academic Press, New York,USA 1998).

[38] R. Marcelli, M. Rossi, and P. De Gasperis, J. Appl. Phys. 73,3082 (1993).

[39] B. Lewis and R. G. Arnold, IEEE Trans. Ultrason. Ferroe-lectr. Freq. Contr. SU-32, 409, (1985).

[40] L. Le Brizoual, F. Sarry, F. Moreira, and O. Elmazria, Phys.Status Solidi A 203, 3179 (2006).

[41] N. Dewan, K. Sreenivas, and V. Gupta, Semicond. Sci.Technol. 23, 085002 (2008).

[42] D.-T. Phan, H.-C. Suh, and G.-S. Chung, Microelectron. Eng.88, 105 (2011).

[43] M. N. R. Ashfold, P. W. May, C. A. Rego, and N. M. Everitt,Chem. Soc. Rev. 23, 21 (1994).

www.pss-a.com

Phys. Status Solidi A 210, No. 8 (2013) 1583

Original

Paper

[44] W. R. Smith, H. M. Gerard, J. H. Collins, T. M. Reeder, andH. J. Shaw, IEEE Trans. Microw. Theory Tech. MTT-17,856, (1969).

[45] Y.-J. Hsiao, T.-H. Fang, Y.-H. Chang, Y.-S. Chang, andS. Wu, Mater. Lett. 60, 1140 (2006).

[46] S.-Y. Chu, T.-Y. Chen, and W. Water, IEEE Trans. Ultrason.Ferroelectr. Freq. Contr. 5, 2308 (2005).

[47] J. Kondoh, N. Shimizu, Y. Matsui, M. Sugimoto, andS. Shiokawa, Sens. Actuators A 149, 292 (2009).

www.pss-a.com

[48] J. Kondoh, N. Shimizu, Y. Matsui, M. Sugimoto, andS. Shiokawa, Proc. IEEE Ultrasonic Symp. (2005),pp. 1023–1027.

[49] D. Morgan, Surface Acoustic Wave Filters, 2nd ed. (Elsevier,New York, 2007), pp. 100–102.

[50] C. C. W. Ruppel and T. A. Fjeldly, Advances inSurface Acoustic Wave Technology, Systems and Appli-cations, Vol. 2 (World Scientific Inc., Singapore, 2001),pp. 890–891.

� 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim