IEEE TRANSACTIONS ON COMPONENTS AND PACKAGING TECHNOLOGIES, VOL. 28, NO. 4, DECEMBER 2005 637

Noncontact Transient Temperature Mappingof Active Electronic Devices Using the

Thermoreflectance MethodMihai G. Burzo, Member, IEEE, Pavel L. Komarov, Member, IEEE, and Peter E. Raad, Member, IEEE

Abstract—This work presents a demonstration of the applica-bility and efficacy of an experimental system capable of nonin-vasively and nondestructively scanning the transient surface tem-perature of pulsed microelectronic devices with submicron spatialand sub-microsecond temporal resolutions. The article describesthe features of the experimental setup, provides details of the cal-ibration process used to map the changes in the measured surfacereflectivity to absolute temperature values, and explains the dataacquisition procedure used to measure the transient temperatureover a given active region. This thermoreflectance thermometrysystem is shown to be particularly suited for directly measuring thesurface temperature field of devices undergoing the fast transientsthat are typical of next generation microelectronic devices. To il-lustrate the experimental approach, both quasisteady and tran-sient temperature measurement results are presented for standardMOSFET devices.

Index Terms—MOSFET devices, thermoreflectance thermom-etry system.

I. INTRODUCTION

KNOWLEDGE of the thermal behavior of a microelec-tronic device is crucial for improving its performance and

reliability. As a result, there is an increased demand for methodsto determine the temperature of submicron level features. Tem-perature can be measured by a diverse array of methods. Allmethods infer temperature by sensing some change in a physicalcharacteristic. However, in order to measure the active junctionsof modern devices, which are frequently powered in a pulsedmode, a method needs to have superior spatial and temporal res-olutions. The use of contact methods presents the added difficul-ties of having to access features of a submicron device with anexternal probe, or in the case of embedded features, fabricate ameasuring probe into the device, and then having to isolate andexclude the influence of the measuring probe itself. Even then,since in the case of submicron devices the thermal capacitanceof the junction is extremely small, contact methods cannot beused if accurate measurements are desired. Consequently, non-contact, optical methods are usually preferred.

Among the various optical methods, the thermoreflectancemethod possesses important advantages and is so far the onlymethod that has been employed to make submicron temperature

Manuscript received November 1, 2004; revised July 4, 2005. This work wasrecommended for publication by Associate Editor C. Lasance upon evaluationof the reviewers’ comments.

The authors are with the Nanoscale Electro-Thermal Sciences Laboratory,Mechanical Engineering Department, Southern Methodist University, Dallas,TX 75275-0337 USA (e-mail: praad@smu.edu).

Digital Object Identifier 10.1109/TCAPT.2005.859738

mappings [1]–[7]. The main advantage of the thermoreflectancemethod is that it is a noncontact and nondestructive optical ap-proach for probing steady-state and transient surface tempera-ture, providing accurate results for submicron features of micro-electronic devices with good temporal resolution.

The most challenging aspect for thermoreflectance measure-ments is the small value of the thermoreflectance coefficient ofthe top layer material, , which defines the rate of changein the surface reflectivity as a function of the change in surfacetemperature, i.e., 1 . The coeffi-cient needs to be sufficiently high in order to obtain an appro-priate signal-to-noise ratio in the measurements. Usually, it mustbe higher than 10 per Kelvin for thermoreflectance temper-ature measurements to be obtainable with good accuracy. Themost important factors that influence are the material undertest, the wavelength of the probing laser [8]–[11], and the com-position of the sample (if multilayered) [8], [12], [13].

The aim of this article is to report on the development andcharacterization of an experimental methodology capable ofnoninvasively scanning the transient surface temperature ofpulsed devices with submicron spatial and sub-microsecondtemporal resolutions. The system’s capabilities are demon-strated by scanning the active area of typical MOSFET devicesof differing gate widths and lengths. Both quasisteady-state andtransient temperature measurement results are presented. Thetemporal and spatial limitations of the methodology are alsodiscussed.

II. EXPERIMENTAL METHODOLOGY

This section describes the features of the experimentalsetup, provides details of the calibration process used to mapthe changes in the measured surface reflectivity to absolutetemperature values, addresses the mitigation of undesirableout-of-focus and interferometric parasitic effects, and explainsthe data acquisition procedure used to measure the transienttemperature over a given active region with minimum randomnoise.

A. Experimental Setup

The newly-built temperature mapping experimental system atthe SMU NETS Laboratory is depicted schematically in Fig. 1.The probing light source is an Ar–Ion CW laser with a lin-early polarized, single-mode irradiation beam at a wavelength of488 nm. The beam is delivered to the microscope assembly viaa polarization preserving, fiber optic cable with mode.

1521-3331/$20.00 © 2005 IEEE

638 IEEE TRANSACTIONS ON COMPONENTS AND PACKAGING TECHNOLOGIES, VOL. 28, NO. 4, DECEMBER 2005

Fig. 1. Schematic of the experimental setup.

Delivering the probing laser energy by a fiber makes it possibleto easily incorporate lasers of different wavelengths to maxi-mize the . This is an important issue since, as shown byTessier et al. [8], the thermoreflectance coefficient of a specificmaterial is extremely sensitive to the wavelength of the laser andvaries strongly between materials. The data presented for Al,Au, Cu, and Ni by Rosei and Lynch [9], Hanus et al. [10], andScouler [11] confirm the fact that the coefficient is wave-length dependent.

A beamsplitter cube and 4 retardation plate are used tominimize losses through the probing optical path. The micro-scope objective lens focuses the laser light perpendicularly tothe device under test (DUT). The probing beam reflects from theheated surface back along the optical path to the sensitive area ofa photodiode (PD). The intensity of the reflected light dependson the reflectivity (temperature) of the sample’s surface. ThePD signal, containing the change in surface reflectivity causedby the temperature variations of the DUT, is acquired with an8-b resolution via a digital oscilloscope at a rate of up to twoGiga-samples per second. This sampling rate allows the mea-surement of a transient temperature field of a DUT activated inthe pulse mode with frequencies of up to 50 MHz, which pro-vides 40 data points to describe a full heating and cooling cycle.

The integrated microscope and CCD camera system ismounted on a Cascade Microtech Alessi precision probingstation, making it possible to view the DUT and to positionthe laser beams on its surface with a resolution of 0.5 m.The continuous zoom microscope provides a maximum mag-nification of 4000 . The smallest probing spot is that can beachieved with the current system and a 100 objective lensis 0.7 m. The device under test is placed on a Temptronic,8-in, liquid-stabilized thermal chuck, capable of maintainingthe bottom of the device substrate at an isothermal condition,in the range of 0–200 C.

B. Calibration and Parasitic Effects

Initially, the above thermal chuck was used to generate a tem-perature-reflectivity calibration curve (from which one can de-termine the thermoreflectance coefficient, ). However, un-controlled movements of the chuck surface resulted in signifi-cant planarity distortion, which in turn made it impossible to dis-tinguish the desired thermal reflectance. To avoid the parasiticeffects associated with the thermal expansion of the relativelymassive thermochuck, a smaller (30 30 5 mm) thermoelec-tric (TE) device was used to calibrate the measurements. Thecalibration approach consists of determining the dependence be-tween the change in the reflectance and the change in the sur-face temperature. The change in reflectance was measured bya differential scheme involving two identical PDs in order tominimize the influence of fluctuations in the energy output ofthe probing laser. In this approach, the laser light is divided intotwo beams, one collected on the reference PD and the other col-lected on the second PD after being reflected from the samplesurface. The sample temperature is measured with a K-type ther-mocouple. The calibration must be performed for each of thematerials on the surface of each device where a mapping of thetemperature is carried out. The is measured at room tem-perature only, since it has been previously shown that theof polysilicon is constant for a wide range of temperatures [14].

The reflectance coefficient is normally very small (on theorder of 10 for gold covered with a silicon oxide passiva-tion layer [8]), and therefore parasitic effects must be minimizedin its measurements. Notably, Dilhaire et al. [15] have pointedout that the thermoreflectance calibration procedure can be hin-dered by the movement of the device under test as a result of thethermal expansion of the heater used to heat and position the de-vice. This movement generates both interferometric (due to theFabry–Perot interferometric effect [16]) and out-of-focus para-sitic effects in the photodetector signal. The out-of-focus effect

BURZO et al.: NONCONTACT TRANSIENT TEMPERATURE MAPPING 639

Fig. 2. Experimental data of actual out-of-focus and interference parasiticeffects for the 10� and 50� objective lenses.

is more significant if a high numerical aperture (NA) objectivelens is used in the measurements. The out-of-focus effects in-duce a change in the reflectivity because when the sample is notin the focal plane of the objective lens, a portion of the light re-flected back from the sample is cut off by various componentsof the optical system (including the aperture of the objectivelens and the active area of the photodetector). The portion of thelight is cut off either because the beam radius increases and itcannot fit through the optical components, or because the beamangle changes due to the movement of the sample (induced bythe thermal expansions of the sample holding system), or both.Given the potential negative implications of the aforementionedparasitic effects, tests were designed and carried out to help as-sess the magnitude of these effects in the present system.

First, a reflective sample was held on a -axis translationstage (i.e., along the optical axis of the objective lens) whoseresolution is 0.1 m. While moving the sample up and downfrom the plane of focus of the objective lens, the resulting signalwas measured by a differential scheme that uses two identicalPDs. The results for two objective lenses (10 and 50 ) withdifferent numerical apertures (NA 0.28 and 0.42, respectively)are presented in Fig. 2.

As expected, the measurements with the high NA objectivelens are more sensitive to the out-of-focus effect than those withthe low NA lens. However, unlike in the case of the system re-ported on by Dilhaire et al. [15], no interferometric effect was de-tected in the PD signal for either objective lens. Specifically, inorder to have an interferometric effect, a regular sinusoidal pat-tern in the PD signal must be present. For a Fabry–Perot interfer-ometer, the expected period of the pattern should equal to 2,which in the SMU system is 244 nm. Even with the coarse reso-lution of the available translation stage (100 nm), which is com-parable to the 2 period, the PD signal should exhibit a telltale“zigzag” pattern when the interferometric effect is present in thesystem. The absence of such a pattern indicates that the interfer-ometric effect is negligible or altogether nonexistent.

To further confirm the absence of the parasitic Fabry–Perotinterferometric effect in the SMU system, a second, simple, butconclusive test was performed. The idea for the test originatesfrom the fact that any interferometer must have at least two arms

whose lights interfere on the sensitive area of a photodetector.As mentioned by Dilhaire et al. [15], for the thermoreflectancesystem one arm includes the sample surface while the other armincludes the input and output planes of the objective lens. Thisimplies that if one arm were to be deactivated, the PD woulddetect the light from the other arm. Therefore, the test consistsof blocking the reflected light from the sample and measuringthe output PD signal that should be produced from the secondarm, i.e., from the two planes of the objective lens. Observingthat the output PD signal was nearly zero for either the 10 orthe 50 objective lenses leads to the conclusion that there isno second arm interference present in the system used in thisinvestigation.

Combining the results of this investigation with that ofDilhaire et al. [15] yields important experimental guidelines;namely, that the presence of parasitic interferometry effectsshould be investigated for each new thermoreflectance system.If such effects are found to be present, they can be mitigatedby minimizing back reflection and increasing the distance fromthe objective lens to the PD.

C. Temperature Data Acquisition

Extracting the temperature value at a given point on thesurface of an active device in the framework of the thermore-flectance method requires the measurement of the surfacereflectivity at that point. This can be achieved by capturing onthe photodetector the level of laser energy reflected back fromthe sample and comparing it to the calibrated data. However,given the small value of the coefficient of reflectivity, suchan approach would have a very weak signal-to-noise ratio. Toovercome this limitation, the activation voltage of the deviceis modulated, resulting in a modulated photodetector signalthat can more easily yield the useful signal from the raw pho-todetector signal. The outcome of each data collection aftera pulsed activation is a transient waveform, an example ofwhich is drawn with open symbols in Fig. 3, superimposedon the modulated activation voltage. After averaging over 256waveforms, each containing 500 data samples, the transientreflectivity signature at a physical location is obtained withgood accuracy (in the range of 1% to 2%). Once the reflectivityis measured, the corresponding absolute temperature value canbe calculated by scaling with the thermoreflectance coefficient.

The temperature field over a region of interest can be mappedby repeating the above procedure at multiple physical locations.The SMU thermoreflectance temperature scanning system is de-signed to acquire the temperature at a point, which can be posi-tioned with a resolution of 0.5 m (in the and directions),and then automatically repeat the process over a grid coveringthe physical region of interest. This scanning process yields atransient temperature field over the desired surface area with asubmicrosecond temporal resolution.

While the extent of the region of interest and the number ofmeasurement points in it can be specified, acquisition time is alimiting factor. Presently, the system can generate the averagedtemperature waveform (shown in Fig. 3) at a physical location in9 s. Faster post-processing and the use of photodetector arrayscan potentially speed up the measurements by at least an orderof magnitude.

640 IEEE TRANSACTIONS ON COMPONENTS AND PACKAGING TECHNOLOGIES, VOL. 28, NO. 4, DECEMBER 2005

Fig. 3. Transient normalized temperature and modulation signal (at hottestpoint on a MOSFET device).

Fig. 4. Geometry of the measured MOSFET Device (field oxide omitted;vertical scale magnified by 20�).

III. RESULTS AND DISCUSSION

To demonstrate the temperature scanning methodology, ex-periments were conducted on typical MOSFET devices. The ge-ometry and structure of one of the devices is shown in Fig. 4. Theoxide passivation layer, omitted in the schematic, is transparentto the probing laser used (488 nm), and hence, does not figurein the optical measurements. In scanning across the channel,for example, from drain to source, the probing laser detects thesignal from several different materials including aluminum, sil-icon, and polysilicon. As previously discussed, the thermore-flectance coefficient differs between these materials, leading todifferent measurement signal levels. With the 8-b oscilloscopeADC, the polysilicon covered region yielded data with the req-uisite level of accuracy. This is consistent with expectationssince polysilicon has a superior value (order of 10 ) toeven that of silicon (order of 10 ) [15] or that of contact metal(order of 10 ) [5], [6].

Fig. 5 shows the channel region of the device with the mea-surement locations depicted along rows crossing from the drainto the source. The scanned area of the poly gate contact cov-ering the channel of the device is also marked. For the casedescribed in conjunction with Fig. 3, the reflectivity reaches a

Fig. 5. Direction and placement of temperature scans over the polysilicon gateregion covering the MOSFET device channel.

Fig. 6. Profiles of normalized temperature maxima along the area of thepolysilicon gate contact area that covers the channel of the device (from Drainto Source) of the activated 18� 15 �m device shown in Fig. 5 at differentlocations over the channel width.

Fig. 7. Scan Area and corresponding temperature contours on an activatedMOSFET device at the time of peak temperature (channel dimensions are15� 5 �m).

maximum every 200 s starting at around 100 s. The ampli-tude of the waveform can be easily extracted and represents therise in the reflectivity of the device at the measurement location.The reflectivity rise profile is recorded along the length of thechannel (from drain to source) for 12 different paths across thechannel width. Then, the profiles are normalized with respect tothe highest reflectivity change recorded in the channel region.

BURZO et al.: NONCONTACT TRANSIENT TEMPERATURE MAPPING 641

Fig. 8. Temperature contours for the activated device shown in Fig. 7 during the heating phase of a pulse cycle.

In addition, the 12 profiles are averaged to yield a single rep-resentative reflectivity profile along the channel. Once normal-ized, the reflectivity rise and temperature rise profiles becomeidentical. The 12 temperature rise profiles and their representa-tive average are plotted in Fig. 6. As can be expected, the tem-perature distribution along the channel exhibits asymmetricalbehavior with the peak temperature shifting toward the drain[17], [18].

Next, in order to demonstrate the method’s capability of cap-turing the transient behavior of the absolute surface tempera-ture field, a MOSFET device with a smaller channel was acti-vated and scanned. As depicted in the bottom left half of Fig. 7,the scanned area of 50 15 m includes the channel as well asthe gate polysilicon region. At each measurement location, 256waveforms of the photodetector signal were averaged as in theprevious case in order to reduce the random noise inherent in thesystem. Since the overall sampling capability of the system ismore than 50 MHz (20 ns) and the timescale of the heat transfer

(conduction) in microelectronic devices is usually in the mi-croseconds (and up to milliseconds), the system has sufficienttime resolution to capture the fastest possible heating/coolingprocesses. This fast temporal capability of the thermoreflectancetemperature scanning system is evident graphically in the tem-perature waveform shown in the center of Fig. 8 (open symbolsfitted by a solid line). The full heating and cooling cycle in thiscase takes around 0.2 ms.

The electric current used to activate this second smaller-channel (5 15 m) device was pulse-modulated at 5 KHzand 50% duty cycle. The two-dimensional (2-D) temperaturecontours are shown on the right hand side of Fig. 7 at the timewhen the device experiences the highest temperature rise. Toproduce the absolute temperature rise values plotted in Fig. 7,the measured reflectivity field was scaled with the thermore-flectance coefficient. The latter was measured as described inSection II-B, but with a 10 objective lens in order to eliminatethe previously discussed out-of-focus parasitic effect.

642 IEEE TRANSACTIONS ON COMPONENTS AND PACKAGING TECHNOLOGIES, VOL. 28, NO. 4, DECEMBER 2005

The temperature rise ranges from zero to about 22 C in thechannel, with the peak occurring closer to the drain of the de-vice. The white gap area near the bottom of the contour plot inFig. 7 corresponds to the back contact metallization strip. Sincethis strip sits on top of the passivation layer, it is well insulatedfrom the heat produced in the active channel and hence experi-ences a negligible rise in temperature.

To further demonstrate the capability of the system to providetransient surface temperature behavior, snapshots of an anima-tion of the temperature contours are shown in Fig. 8. The ninesnapshots represent the surface temperature field at different in-stances in the heating and cooling phases of the full pulsed cycle.The time during the cycle of each of the contour plots is identi-fied on the contour plot and on the waveform graph shown in thecentral part of Fig. 8. The snapshot times were chosen to graph-ically represent the transient thermal behavior of the MOSFETdevice. This approach can provide equally rich understandingof the transient thermal behavior of devices that are operated atmuch higher pulse rates and/or at other duty cycles.

Accounting for the uncertainties associated with the calibra-tion (i.e., obtaining ) and the actual temperature scanning,the overall random uncertainty of the results is estimated to beless than 13%.

The results obtainable through the use of the thermore-flectance temperature scanning system would make it possiblefor analysts and designers of both existing and next generationdevices to make decisions that are based on actual behavior asopposed to approximate simulations or reduced-order models.And, even when the latter simulations and models are availableand their use is desirable, direct measurements can be invalu-able in validating the results of such computational techniquesand substantiating the consistency of such models.

IV. CONCLUSION

This article introduced a new experimental system of ther-moreflectance thermometry that is ideally suited to noninva-sively and nondestructively measure the surface temperaturefield of devices undergoing the ultrafast transients that are typ-ical of next generation microelectronic devices. Specifically, wehave demonstrated for the first time a system that is capable ofcapturing the thermal transients of a device pulsed at speeds ofup to 50 MHz, which far exceed the physical scales of temper-ature response. The system probes with a 488 nm Ar-Ion laser,which provides a submicron, diffraction-limited, spatial reso-lution. The system’s efficacy was demonstrated by presentingthe results of surface temperature scans of pulsed MOSFET de-vices. Accounting for the uncertainties associated with the cal-ibration (i.e., obtaining ) and the actual temperature scan-ning, the overall random uncertainty of the results is estimatedto be less than 13%.

The ultimate power of a direct thermography approach is as apart of an integrated approach whose aim is to characterize thetransient thermal behavior of fully-featured three-dimensional(3-D) structures. The overall approach would use the describedexperimental technique to noninvasively scan the surface tem-perature where possible and then would solve an inverse nu-merical problem with an ultra-fast adaptive technique to deter-

mine the temperature distribution over the entire 3-D device,including important embedded features whose response is oth-erwise impossible to measure directly.

ACKNOWLEDGMENT

The authors wish to thank Dr. S. Banerjee and L. Weltzer,University of Texas at Austin, for providing the MOSFETdevices.

REFERENCES

[1] K. E. Goodson and Y. S. Ju, “Short-time-scale thermal mapping ofmicrodevices using a scanning thermoreflectance technique,” ASME J.Heat Transfer, vol. 120, pp. 306–313, 1998.

[2] S. Grauby, S. Hole, and D. Fournier, “High resolution photothermalimaging of high frequency using visible charge couple device cameraassociated with multichannel lock-in scheme,” Rev. Sci. Instrum., pp.3603–3608, 1999.

[3] V. Quintard, S. Dilhaire, T. Phan, and W. Claeys, “Temperature mea-surement of metal lines under current stress by high resolution laserprobing,” IEEE Trans. Instrum. Meas., vol. 48, no. 1, pp. 69–74, Feb.1999.

[4] S. A. Thorne, S. B. Ippolito, M. S. Ünlü, and B. B. Goldberg, “High-resolution thermoreflectance microscopy,” in Proc. Materials ResearchSoc., vol. 738, Nov. 2002, pp. 12.9.1–12.9.6.

[5] J. Christofferson, D. Vashaee, A. Shakouri, P. Melese, X. Fan, G. Zeng,C. Labounty, J. E. Bowers, and E. T. Croke III, “Thermoreflectanceimaging of superlattice micro refrigerators,” in Proc. 17th Annu. IEEESymp. Semiconductor Thermal Measurement Management, San Jose,CA, Mar. 2001, pp. 58–62.

[6] J. Christofferson, D. Vashaee, A. Shakouri, and P. Melese, “Real timesub-micron thermal imaging using thermoreflectance,” in Proc. Int.Mechanical Engineering Congress Exhibition (IMECE’01), New York,Nov. 2001.

[7] Z. Bian, J. Christofferson, A. Shakouri, and P. Kozodoy, “High-poweroperation of electroabsorption modulators,” Appl. Phys. Lett., vol. 83,no. 17, pp. 3605–3607, 2003.

[8] G. Tessier, S. Hole, and D. Fournier, “Quantitative thermal imagingby synchronous thermoreflectance with optimized illumination wave-lengths,” Appl. Phys. Lett., vol. 78, pp. 2267–2269, 2001.

[9] R. Rosei and D. W. Lynch, “Thermomodulation spectra of Al, Au, andCu,” Phys. Rev. B, vol. 5, pp. 3883–3893, 1972.

[10] J. Hanus, J. Feinleb, and W. J. Scouler, “Low-energy interband transi-tions and band structures in nickel,” Phys. Rev. Lett., vol. 19, pp. 16–20,1967.

[11] W. J. Scouler, “Temperature modulated reflectance of gold from 2 to 10eV,” Phys. Rev. Lett., vol. 18, pp. 445–448, 1967.

[12] J. Heller, J. W. Bartha, C. C. Poon, and A. C. Tam, “Temperature depen-dence of the reflectivity of silicon with surface oxides at wavelengths of633 nm and 1047 nm,” Appl. Phys. Lett., vol. 75, no. 1, pp. 43–45, 1999.

[13] V. Quintard, G. Deboy, S. Dilhaire, D. Lewis, T. Phan, and W. Claeys,“Laser beam thermography of circuits in the particular case of passivatedsemiconductors,” Microelectron. Eng., vol. 31, pp. 291–298, 1996.

[14] X. Xu, C. P. Grigoropoulos, and R. E. Russo, “Transient temperatureduring pulsed excimer laser heating of thin polysilicon films obtainedby optical reflectivity measurement,” Trans. ASME J. Heat Transf., vol.117, pp. 17–24, 1995.

[15] S. Dilhaire, S. Grauby, and W. Claeys, “Calibration procedure for tem-perature measurements by thermoreflectance under high magnificationconditions,” Appl. Phys. Lett., vol. 84, pp. 822–824, 2004.

[16] P. Pirani, G. Appenzeller-Combe, and W. Lukosz, “Thermal-expansion-induced submilliwatt optical bistability in air-spaced Fabry–Perot res-onators with metal mirrors,” Opt. Commun., vol. 75, p. 326, 1990.

[17] G. Kaytaz, P. L. Komarov, and P. E. Raad, “A new simulation model ofelectrothermal degradation for MOSFET devices subjected to hot car-rier injection stress,” in Proc. 9th Int. IEEE Workshop THERMal INves-tigations IC Systems (THERMINIC’03), Aix-en-Provence, France, Sep.24–26, 2003, pp. 251–256.

[18] I. Ahmad, V. Kasisomayajula, M. Holtza, J. M. Berg, S. R. Kurtz, C.P. Tigges, A. A. Allerman, and A. G. Baca, “Self-heating study of anAlGaN/GaN-based heterostructure field-effect transistor using ultravi-olet micro-raman scattering,” Appl. Phys. Lett., vol. 86, no. 17, 2005.

BURZO et al.: NONCONTACT TRANSIENT TEMPERATURE MAPPING 643

Mihai G. Burzo (M’03) received the M.S. degree(with honors) in engineering from the TechnicalUniversity of Cluj-Napoca, Cluj-Napoca, Romania,in 1995 and the Ph.D. degree in mechanical engi-neering from Southern Methodist University (SMU),Dallas, TX, in 2001.

He is currently a Postdoctoral Fellow in the Me-chanical Engineering Department, SMU. He workedon projects involving computational fluid dynamics(CFD) and computational heat transfer, thermal prop-erties measurements of electronics materials and non-

contact, in-situ measurements of the temperature of active devices. His currentresearch, in the Nanoscale Electrothermal Sciences Laboratory, SMU, focuseson the laser-based thermal properties measurements of thin film materials andtemperature measurements of microelectronic devices. He has published morethan 20 articles, conference papers, and reports in the professional literature.His research interests have covered areas of heat transfer as diverse as forcedand natural convection, refrigeration techniques and cooling of electronics.

Dr. Burzo received the Outstanding Graduate Student Award (2001) from theNorth Texas Section of ASME, a Leadership Award from Southern MethodistUniversity (2002), a scholarship award from the ME Department of SMU(1997–1998), a Mercedes Benz scholarship (1995), and Valedictorian Award(1995). He is a member of ASME, APS, and SAE.

Pavel L. Komarov (M’85) received the B.S. andM.S. degrees in mechanical engineering fromMoscow Power Engineering University, Moscow,Russia, and the Ph.D. degree in physics and math-ematics from the Institute for High Temperatures,Russian Academy of Sciences, Moscow.

In 1982, he joined the Institute for High Tem-peratures, where he was a Researcher in the Heatand Mass Transfer Laboratory until 1997. Duringthat time, he extensively used laser doppler (LDA)and hot wire (HWA) anemometry to investigate the

heat transfer of separated turbulent flows in channels. He designed specialsoftware for the LDA and HWA measurements, as part of an effort to fullyautomate fluid dynamics and heat transfer measurement systems. In 1991,he had achieved the grade of System Programmer after passing through thecourses in the Training Center, Soviet–American Joint Venture “Dialogue.”This accomplishment allowed him to work as a Software Designer with theLaser Measurement System Company, Moscow, along with his main positionat the Institute for High Temperatures. In 1997, he joined the NanoscaleElectrothermal Sciences Laboratory, Southern Methodist University (SMU),Dallas, TX, as a Post-Doctoral Fellow. In the next three years, he helped designand oversee the successful development of the experimental systems for boththermal property measurements of thin-films and temperature measurementof active MMICs, in collaboration with experts from the local electronicsand telecom industry. He has been the Lead Designer and Developer of themeasurement automation systems and analytical performance estimators of thetransient thermoreflectance method. Since 1999, he has held the position ofa Research Faculty member in the same Laboratory. He directs the graduateresearch assistants and is responsible for maintaining research activities andfurther development of laser-based measurement techniques. He has published13 journal articles and 17 conference papers.

Peter E. Raad (M’97) received the B.S.M.E. degree(with honors), and the M.S. and Ph.D. degrees inmechanical engineering from the University ofTennessee, Knoxville.

In 1986, he joined the Mechanical EngineeringDepartment, Southern Methodist University (SMU),Dallas, TX. He currently holds the Linda WertheimerHart Professorship and is Director of the Linda andMitch Hart eCenter, SMU and Executive Director ofthe Guildhall, SMU. The eCenter aims to stimulate,facilitate, and support innovative interdisciplinary

activities that enable the creative and responsible development and use of in-teractive network technologies. Prior to becoming the Founding Director of theeCenter, he served as the Associate Dean for the SMU School of Engineering.He has taught courses in the thermal and fluid sciences, and his research in theseareas combines computational and experimental investigations. From 1990 to1993, he held the J. Lindsay Embrey Trustee Professorship in Engineering (anendowed chair for an outstanding junior faculty member). He has publishedover 40 journal articles and over 100 symposium and conference papers.He has received over $2.5M in funding and support from, among others,NSF, TI, Raytheon, TriQuint Semiconductor, Chrysler Technologies, Isonics,and Marlow Industries. In 1995, he founded the Nanoscale ElectrothermalSciences Laboratory dedicated to the adaptive thermal modeling of submicronelectronic devices and laser-based measurements for thin-film materials usedin high-performance ICs. He has served as an Associate Editor for the ASMEJournal of Fluids Engineering as well as a frequent Reviewer for NSF, ASMETRANSACTIONS, the Journal of Computational Physics, and Physics of Fluids.He also does research in tsunami mitigation and fluid wave interactions withsolid structures.

Dr. Raad received the Outstanding Graduate Faculty Award four times, theOutstanding Undergraduate Faculty Award two times, the Sigma Xi Faculty Re-search Award, and was named the ASME North Texas Section Engineer of theYear in 1999 and 2000. He is a member of ASME, APS, Sigma Xi, and TauBeta Pi. He is a Licensed Professional Engineer in the State of Texas.