Self-heating temperature measurement in AlInN/GaN HEMTs by using CeO2 and TiO2 micro-Raman thermometers

https://doi.org/10.1016/j.microrel.2022.114693Get rights and content

Abstract

Thermal characterization of GaN-based components is an important and challenging issue that requires the measurement of the self-heating temperature of the channel, the metal contact surface and the substrate with high accuracy. This paper points out that all these thermal parameters can be measured by combining confocal Raman spectroscopy and micro-Raman thermometers. To prove these assertions, thermal resistance and Thermal Boundary Resistance (TBR) were accurately estimated experimentally. In addition, temperature measurements with TiO2 and CeO2 micro-Raman thermometers were carried out to determine which is more accurate and reliable in measuring the surface self-heating temperature of these components.

Introduction

GaN-based high electron mobility transistors (HEMTs) are excellent candidates for high power, high frequency and high temperature applications [1] such as satellite or RF telecommunications devices [2]. However, the increase in power dissipation and the reduction in size of GaN-based devices requires increasingly higher operating temperatures [1]. In these conditions, their electrical properties and their reliability can be degraded by the temperature increase [3]. It is therefore crucial to estimate the self-heating temperature (ΔT) of GaN-based components with a high accuracy in order to optimize their thermal management and consequently avoid their electrical performances and reliability degradation. In addition, a suitable temperature measurement of these devices in real operating condition is required to validate thermal simulations used to predict their thermal behaviors.

Infrared (IR) thermography is the most used technique to measure self-heating temperature of biased components [4]. Spatial and thermal resolutions are 2–3 μm and 0.05 K, respectively. This technique can lead to an underestimation of the device temperature [3].

Literature also reports thermoreflectance [5] and Raman spectroscopy [3] as valuable techniques to achieve this purpose. Both techniques have a sub-micrometer spatial resolution [5], [6].

Thermoreflectance is usually used to measure the surface temperature of a component. It's based on the relative surface reflectivity variation (ΔR) as a function of temperature variation (ΔT) as shown in Eq. (1) [1]:RR=CTH·Twhere CTH is the thermoreflectance coefficient. CTH is material and wavelength dependent and has to be determined experimentally for each point of measurement [7]. This technique is particularly suitable for metallization temperature estimation [5] but also allows semiconductor self-heating measurement. However, subsurface layers may impact the reflectance signal, resulting in low accuracy of the measurement [1]. This technique can also be used to map the transient temperature distribution over the entire device active area [7] with a time resolution of a few hundred nanoseconds [8].

Micro-Raman thermometry is based on the variation study of phonon frequencies (δ) assigned to the GaN semiconductor with temperature T. The relation between δ and T is given by Eq. (2) [1]:δT=δ0AeBħδOkBT1where δ(T) and δ0 are the Raman phonon frequency at the temperature T and T0 = 0 K, respectively. A and B are fitting parameters, ħ is the reduced Planck constant and kB is the Boltzmann constant. Nevertheless, this technique requires calibration curves to estimate the operating temperature [1], [3]. Since the semiconductor materials used for GaN-based components are usually transparent to visible excitation sources used in Raman spectroscopy, the evaluation of ΔT of GaN surface is not possible. Indeed, ΔT is averaged over the probed zone of the component. However, the optical penetration depth of a UV excitation source is low and therefore it is possible to estimate ΔT of a zone close to the device surface [9]. It is interesting to note that there are a few reports concerning the surface temperature mapping of biased components using Raman spectroscopy [10]. Thus, Raman thermometry does not allow to directly measure the operating temperature of biased devices on the semiconductor surface and on the drain, source and gate contact surface. Furthermore, it is not possible to directly measure the self-heating temperature of the GaN semiconductor for a component with field plates or air bridges [1]. The use of micro-Raman thermometers, such as CeO2 [1], TiO2 [11] or diamond [12] microparticles is the only solution to overcome these technological locks. In this way, the self-heating temperature can be estimated on the metallization surface of the device. Moreover, Simon et al., [12] have already measured the transient self-heating temperature on the gate contact using diamond micro-Raman thermometers, thus demonstrating the interest of this technique.

Self-heating measurements allow determination of thermal dissipation bottleneck such as TBR which induces a thermal discontinuity between the GaN layer and the substrate. A high TBR value leads to device performances degradation and also reduces its lifetime [13].

In this study, TiO2 and CeO2 micro-Raman thermometers have been compared to estimate the surface self-heating temperature of biased AlInN/GaN HEMTs. Thus, the ΔT of the volumetric GaN layer, the GaN semiconductor surface and the thermal resistance of the device were evaluated. At the same time, TBR at the GaN/sapphire interface has been studied using confocal Raman microscopy. The thermal contact between the component surface and CeO2 or TiO2 microparticles was also analysed.

Section snippets

Experimental details

The AlInN/GaN layers were grown on a 300 μm sapphire substrate by Metal Organic Chemical Vapor Deposition. They consist of a 3 μm thick GaN buffer layer, a 1 nm AlN spacer layer and an 11 nm thick Al0.82In0.18N barrier layer. Ohmic contacts were deposited by evaporating Ti/Al/Ni/Au then annealed at 900 °C for 30 s under nitrogen atmosphere. Ni/Au metals were used to form a Schottky contact. The gate length was defined by electron beam lithography and is 0.35 μm. The gate width, gate-drain and

Results and discussion

It is imperative that the deposition of microparticles on the device surface doesn't result in any change in their DC electrical characteristics, as reported in [1]. The self-heating temperature of biased AlInN/GaN HEMTs was evaluated at the surface of the AlInN/GaN heterostructure between the drain and gate contacts by using TiO2 and CeO2 microparticles that act as micro-Raman thermometers. At the same time, the volumetric temperature of the GaN layer was also estimated using a GaN Raman mode.

Conclusion

We have shown that the surface self-heating temperature of biased AlInN/GaN HEMTs can be estimated accurately using TiO2 and CeO2 micro-Raman thermometers. These microparticles are deposited on the device surface allowing the self-heating temperature measurement of the semiconductor surface and of the metal contacts independently of their surface roughness. However, CeO2 thermometers allow the evaluation of ΔT from δ(CeO2) or Γ(CeO2) unlike TiO2 thermometers, which is a considerable advantage.

CRediT authorship contribution statement

R. Strenaer: Experimental studies, Data curation, Formal analysis, writing – original draft preparation, Review. Y. Guhel: Methodology, Data curation, Formal analysis, Supervision, Review. G. Brocero: Experimental studies, data curation, formal anlysis, C. Gaquière: Ressources, Formal analysis, Review. B. Boudart: Methodology, Data curation, Formal analysis, Funding acquisition, Supervision, Review.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

The authors would like to thank the European project “ETHNOTEVE” which is funded by the European Union within the framework of the Operational Programme ERDF/ESF 2014-2020.

References (18)

  • A. Fletcher et al.

    Superlattices Microstruct.

    (2017)
  • M. Zhao

    Opt. Commun.

    (2013)
  • O. Lancry

    Solid State Electron

    (2010)
  • G. Brocero

    IEEE Trans. Electron Devices

    (2019)
  • M. Kuball et al.

    IEEE Trans. Device Mater. Reliab.

    (2016)
  • B. Chatterjee

    IEEE Electron Device Lett.

    (2021)
  • M. Kuball

    CSIC Symp.

    (2007)
  • G. Pavlidis

    IEEE Trans. Electron Devices

    (2018)
  • G. Pavlidis
There are more references available in the full text version of this article.

Cited by (0)

View full text