J. Appl. Phys., Vol. 96, No. 1, 1 July 2004 Robert et al. 253of the technology, and the results obtained are useful in dem-onstrating that spin coating onto a curved substrate is a fea-sible approach to transducer fabrication.A. Solution preparation P͑VDF-TrFE͒ powder of 3.75 g with a 75/25 molar ratio͑MSI Inc., Valley Forge, PA͒ was disolved in 40 ml MEK ina bottle with a screw-top lid. The lid was securely fastenedand the solution was spun in a mixing machine for 24 h topromote homogeneity. The mixture was then heated in anoven at 70 °C for 30 min to help increase the dissolution ofthe solute. The solution could be kept and used for severalmonths after it had been made. FIG. 1. Drawing of the screw clamp assembly ball bearing used to press focus the aluminum substrate used as a transducer backing.B. Substrate centering technique An aluminum substrate was used to form the curved sur- The curved aluminum substrate was ﬁrst centered on topface in this work. The preparation of the aluminum substrates of the vacuum chuck of the spin coater ͑Chemat Technologywas crucial to the success of the spin coating process. The Inc., Northridge, CA͒. One drop of copolymer was then ap-design goal for the transducer was a 3 mm aperture with an f plied to a curved substrate spinning at 2000 rpm and accel-number of 3, and these design criteria mandated that the erated to a speed of 3500 rpm for 30 s. The ﬁlm was coveredsubstrates had a 3 mm diameter and a centered indentation with the spin coater lid to slow the evaporation process andwith a spherical diameter of 18 mm. There were two major was allowed to dry for at least 30 min before another layerrequirements of the substrate fabrication process. was added in the same manner. A scanning electron micros-͑1͒ The spherical indentation must be centered in order to copy ͑SEM͒ photograph of a cross section of the resulting ensure that there are no ‘‘ﬂat spots’’ around the edge of device is shown in Fig. 2. the transducer surface. Such areas would negatively af- Several factors contribute to obtain a high-quality ﬁlm of fect the quality of the acoustic signal. uniform thickness and acceptable homogeneity. It was cru-͑2͒ The surface of the indentation must not contain any cial to use a substrate with an extremely smooth ﬁnish, ide- scratches with a depth greater than 0.5 m to make it ally with no irregularities greater than 0.5 m in depth. Mi- conducive for the formation of a uniform ﬁlm with ac- croscratches could alter the acoustic properties of the ceptable piezoelectric properties. transducer by allowing variation in ﬁlm thickness across the surface. Another important factor to control was the rate of The centering method that best fulﬁlled the aforemen- copolymer application. The viscosity of the copolymer solu-tioned requirements consisted of several steps. An aluminum tion created drops that were approximately the size of therod of 3.18 mm diameter was machined to have a spherically aperture of the ﬁnished device. When more than one dropshaped concave end using a 6.36 mm ball end mill. A 5 mmlength was parted off the original rod, and the back of thedisk was sanded until it was ﬂat. A stainless steel ball bearingwith a 9 mm radius was used to create a substrate with thedesired f number of 3. The ball bearing was placed on theprepared aluminum piece and was automatically centereddue to the spherically shaped impression in the substrate. Theassembly was pressed using a screw clamp jig in an arborpress, as shown in Fig. 1. After the piece was sphericallymolded, it was lapped using glycerine with 12.5 and 3.0 maluminum oxide powder and 3M Finesse-It II ﬁnishing ma-terial over another 9 mm radius ball bearing. The lappedsurface of the substrate prepared as described was extremelysmooth, and the spherical indentation was centered.C. Spin coating procedure Little detail on spin coating a piezoelectric copolymerﬁlm directly onto a focused substrate for the purpose ofbuilding a transducer can be found in the literature.7 At the FIG. 2. A ﬁlm with two spin coated layers of copolymer shows a thicknesstime of this study no models or concrete experimental data of Ϸ6 m. The silver particles embedded in epoxy shown in the photographexisted to provide a clue as to how the geometric shape of are not part of the ﬁnished device, but were needed for proper photographicthe substrate would alter the spin coating process. contrast.
254 J. Appl. Phys., Vol. 96, No. 1, 1 July 2004 Robert et al.was added to the substrate surface per ﬁlm layer, the subse-quent drops interfered with the natural ﬂuid outﬂow withinthe critical ﬁrst few seconds of the spin coating process. Thisinterference created bubbles in the ﬁlm during drying thathad subsequently popped and created craters within the ﬁlmduring evaporation. After analyzing these results, it became apparent thatboth the substrate surface condition and the method by whichthe copolymer was applied could impact the ﬁnal outcome ofthe thickness and quality of the ﬁlm. Therefore the optimalprocess required dropping a discrete amount of P͑VDF-TrFE͒/MEK solution all at once onto a polished, roundedsurface. The speciﬁc volume applied was not critical to thesuccess of the process because the excess ﬂuid was removedfrom the aluminum substrate due to centripetal force. It wascrucial, however, that the amount of P͑VDF-TrFE͒/MEK so-lution be added without interruption, either as a single dropor as a ﬁnite steady stream. In these tests, each layer wasadded as a single drop that was approximately the diameterof the 3.18 mm spinning substrate. The ﬁnal thickness ofeach layer was Ϸ5– 6 m after evaporation. All transducersbuilt for this study used two layers of copolymer ﬁlm, butadditional layers could be added if a thicker copolymer ﬁlmwas needed for a lower frequency application without a sig-niﬁcant loss of uniformity across the surface of the ﬁlm. Acomputer program was written to evaluate the uniformity ofthe ﬁlm based on SEM images as shown in Fig. 3, whereadditional layers were added to create a thickness of over 80m as can be seen in the graph of thickness versus position.D. Curing and poling FIG. 3. ͑a͒ A program was written to measure uniformity by having a user deﬁne the edges of the copolymer ﬁlm by tracing along the edges of the ﬁlm After the spin-coated ﬁlm had been given at least 30 min as shown on the computer screen. ͑b͒ The program displayed a plot ofto fully evaporate, it was cured in a 120 °C oven for at least thickness vs position to illustrate the uniformity of the ﬁlm.3 h. Curing or annealing the ﬁlm helped to promote uniformchemical and mechanical properties across the surface. Thesubstrate was properly masked by placing the substrate in the E. Transducer characterizationcenter of a brass ring of the same height and ﬁlling the gap Pulse-echo and insertion loss measurements were per-with EPO-TEK 301 epoxy ͑Epoxy Tech., Billerica, MA͒. formed on the fabricated transducers to assess their perfor-The addition of the epoxy ensured that the two sides of the mance. The transducers were excited with a Panametricsﬁlm remained electrically isolated. The device was then sput- ͑Waltham, MA͒ 5900 PR pulser/receiver. The experimentaltered with Ϸ1000 Å of gold and chrome to create a top setting of the Panametrics unit used for the measurements areelectrode across the copolymer ﬁlm. A hole was drilled into listed in Table I. A 50 ⍀ 30 cm cable was used to connect thethe bottom of the aluminum, and a wire was attached using a transducer to the pulser/receiver. The device was placed in aconductive epoxy. The polymer was poled by applying a 20 degassed water bath with a quartz crystal with its ﬂat surfaceV/m voltage across the thickness of the ﬁlm in a 90 °C placed perpendicular to the beam at the focus of the trans-oven for 30 min to align the dipoles. The temperature was ducer that was 9 mm from the center of the transducer. Thedropped from 90 to 25 °C while maintaining a constant elec- received echo was displayed on a LeCroy ͑Chestnut Ridge,tric ﬁeld across the copolymer ﬁlm. Once the device had NY͒ LC 534 Oscilloscope with 50 ⍀ coupling. Insertion losscooled and the electric potential was removed, the two sides was measured using an approach reported by Sherar andof the ﬁlm were shorted for at least 12 h to relax the excess Foster.8 More details of the experimental protocols can becharge. The ﬁnished transducers were housed in a modiﬁed found in Snook et al.9SMA connector ͑Fig. 4͒. Two devices of 3 mm diameterwere designed to have f number of 3 and built following the III. RESULTSfabrication procedure as outlined above. Device 1 had a cen-ter frequency of 43 MHz and Device 2 had a center fre- The pulse-echo response and bandwidth of Device 2 arequency of 41 MHz. shown in Fig. 5. All relevant characterization parameters
J. Appl. Phys., Vol. 96, No. 1, 1 July 2004 Robert et al. 255FIG. 4. Design cross section of a spin-coated P͑VDF-TrFE͒ transducer. Thecenter conductor of the SMA connector was electrically connected to thenegative electrode of the P͑VDF-TrFE͒ ﬁlm through the aluminum backing.A sputtered layer of 1000 Å in thickness of chrome/gold was used to con-nect the positive electrode of the P͑VDF-TrFE͒ to the brass housing and toelectrically shield the entire device.were calculated for the two devices and are summarized inTable II. Results on insertion loss measurements showed aloss of 37 dB at 43 MHz after compensating for the attenu-ation in water. Copolymer devices generally exhibit a higherinsertion loss and a lower sensitivity than lead titanate zir-conate ͑PZT͒ devices due to their lower electromechanicalcoupling coefﬁcent.9 FIG. 5. Time-domain echo response ͑top͒ and normalized frequency spec-IV. CONCLUSION trum for device 2. Spin coating is a simple concept, but an extremely com-plex process. Environmental concerns compound the difﬁ- The uniformity of a copolymer ﬁlm coating is of para-culty in achieving consistent results with spin coating be- mount importance in the success of the ﬁnished device be-cause a diverse set of factors—from table vibrations to cause irregularities will impact negatively on the quality ofhumidity levels—can alter the outcome. Such factors were the acoustic signal produced by the transducer. SEM showednot controlled in this feasibility study, but a methodical that uniform coatings were possible on the curved aluminumanalysis of their impact could help to further understand the substrates. The thicknesses of these ﬁlms were appropriateﬂuid mechanics of the ﬁlm formation. For all these reasons, for high frequency applications.even spin coating on ﬂat substrates can be a challenging task. One of the biggest challenges of the fabrication proce-Spin coating on a curved substrate truly adds a new dimen- dure was to create a centered, spherical surface on the alu-sion to the already complicated spin coating problem. The minum parts. This obstacle was overcome through the use ofﬂuid mechanics becomes much more complex as a result of a combination of techniques, including press focusing andhaving new forces such as gravity that play a role in the ﬁnal lapping over a curved surface.shape and thickness of the ﬁlm. The transducers produced by following the fabrication procedure as described had center frequencies of over 40 MHz and average bandwidths of 75%. The performance ofTABLE I. The operational settings for the Panametrics 5900 PR pulser/ the devices are consistent with copolymer transducers pro-receiver used for pulse-echo testing. duced using other fabrication methods, and it is expected thatParameter Value TABLE II. Measured transducer performance.Pulse repititionFrequency 1 kHz Parameter Device 1 Device 2Input energy 1 JDamping 50 ⍀ Center frequency 43 MHz 41 MHzAttenuation 5 dB Focal distance 8.54 mm 9.22 mmGain 40 dB F-Number 2.6 2.8Low pass ﬁlter 200 MHz Ϫ6 dB bandwidth 67.25% 83%High pass ﬁlter 2 MHz Signal amplitude 1.42 V 2.22 V
256 J. Appl. Phys., Vol. 96, No. 1, 1 July 2004 Robert et al.the devices would have a lower sensitivity due to the nature The materials used in the fabrication of this device wereof the active element material. However, the bandwidth selected for a variety of reasons. It is possible that the use ofachieved ͑75%͒ is at least in par with if not better than those other materials may lead to better performance characteris-obtained with transducers fabricated from other materials. tics or to new applications. For example, the use of a differ-There are many advantages to this approach over ones that ent backing material such as a polymer, silicon, or glass mayhave been used in the past, the greatest of which is that the serve to improve both the sensitivity and the bandwidth ofproblem of ﬁlm handling is eased considerably and the ﬁlm the device.10 Aluminum was chosen for this study due to itsis less likely to be damaged. Another advantage is that the conductive nature so that it could be used as an electrode forprocessing of these devices is less complex in many regards, poling. Silicon backing is especially attractive because it of-which may lead to a greater success rate in transducer pro- fers the possibility of the integration of imaging electronicsduction. with the ultrasonic sensor, which is necessary in the design There are many applications for which P͑VDF-TrFE͒ de- of arrays in the frequency range from a few hundred MHz tovices such as the ones reported in this study would be well GHz.suited, including but not limited to, ultrasound backscattermicroscopy6 with clinical applications in imaging anterior ACKNOWLEDGMENTSsegments of the eye and skin. The authors would like to thank Eugene Gerber and JayV. FUTURE WORKS Williams for their technical advice and assistance. This work The goal of this research was to demonstrate that it was has been supported by NIH Grant No. P41-EB2182.possible to directly spin coat a piezoelectric copolymer layeronto a curved substrate and produce an operational high fre- 1 L. F. Brown, R. L. Carlson, and J. M. Sempsrott, 1997 Proceedings ofquency transducer. Although there are many ways in which IEEE Ultrasonics Symposium ͑IEEE, New York, 1997͒, p. 1725.the proposed process can be improved, it serves as a good 2 H. Kawai, Jpn. J. Appl. Phys. 8, L975 ͑1969͒. 3starting point for future research in the area. L. F. Brown and A. M. Fowler, 1998 Proceedings of IEEE Ultrasonics Symposium ͑IEEE, New York, 1998͒, p. 607. Matching layers are almost always used with ceramic 4 T. Yamada, T. Ueda, and T. Kitayama, J. Appl. Phys. 52, 948 ͑1981͒.active elements because of the large acoustic mismatch be- 5 H. Ohigashi and K. Koga, Jpn. J. Appl. Phys. 21, L455 ͑1982͒.tween the ceramic material ͑around 34 MRayl͒ and the tis- 6 L. S. Foster, C. J. Pavlin, G. R. Lockwood, L. K. Ryan, K. A. Harasiewicz, L. Berube, and A. M. Rauth, IEEE Trans. Ultrason. Ferroelectr. Freq.sues of the human body ͑around 1.5 MRayl͒. They are not Control 40, 608 ͑1993͒.required in the design of a copolymer transducer because the 7 K. Kimura and H. Ohigash, J. Appl. Phys. 61, 4749 ͑1987͒.acoustic impedance ͑Ϸ4.5 MRayl͒ is much closer to that of 8 M. D. Shearer and F. S. Foster, Ultrason. Imaging 11, 75 ͑1989͒. 9the biological medium. Matching layers were not utilized in K. A. Snook, J. Z. Zhao, C. H. F. Alves, J. M. Cannata, W. H. Chen, R. J. Meyer, T. A. Ritter, and K. K. Shung, IEEE Trans. Ultrason. Ferroelectr.this work. However, use of a matching layer could serve to Freq. Control 49, 169 ͑2002͒.further enhance the performance of the devices due to the 10 L. F. Brown, IEEE Trans. Ultrason. Ferroelectr. Freq. Control 47, 1377better coupling between the two media. ͑2000͒.