Enhancing Thermal Efficiency of HAMR NFTs with an AIN Heatsink

Heat-Assisted Magnetic Recording (HAMR) stands at the forefront of emerging technologies for enhancing the capacity of hard disk drives (HDDs). It leverages precise laser heating to locally increase the temperature of a high-anisotropy recording medium, enabling data writing. The near field transducer (NFT) plays a crucial role in HAMR by focusing the laser light to generate a confined, precise hotspot on the recording medium. This project addresses the critical challenge of NFT self-heating. During operation, as the NFT focuses light to create this confined hotspot, it also experiences a significant temperature rise, often reaching hundreds of degrees. This self-heating can impact the efficiency and lifespan of the HDD. To mitigate this, one strategy involves using materials with high thermal conductivity to aid in heat dissipation from the NFT. Aluminum nitride (AlN) is a promising dielectric candidate, offering both low optical loss and high thermal conductivity under appropriate deposition conditions. The aim of this dissertation is to assess whether AlN can serve as the primary heatsink at the air bearing surface (ABS), offering a high thermal efficiency design option for future HAMR head designs. 

We begin our study by proposing and analyzing a feasible ridge waveguide NFT structure heatsinked by AlN with FEM simulation. In this design, AlN was part of the light delivery system and also served as the heatsink. Leveraging AlN’s high thermal conductivity and the ridge waveguide NFT’s substantial interface area, our design can achieve thermal efficiency (TE), defined as the ratio of peak temperature rise in the medium to that in the NFT as high as 2.93. Different NFT excitation methods are examined, with an impinging dielectric waveguide excitation leading to the highest TE. We note that surface plasmon is confined inside the ridge waveguide NFT cavity, while heat escapes through the outer surface of the NFT, separating the dielectric interfaces responsible for optical propagation and heat transfer. This allows for further reduction in thermal interface resistance by adding an interface adhesion layer like Cr at the NFT/AlN interface without sacrificing optical properties. 

After demonstrating a practical use of AlN for reducing temperatures in NFTs, we then proceed with experimental demonstration of low loss AlN waveguides, benchmarked by amorphous SiNx. We fabricated ring resonators on SiNx and AlN films and extracted propagation loss from the quality factor and free spectral range of the resonance pattern. Deposition conditions play a crucial role in attaining low optical loss in AlN. This was evident when we analyzed two AlN films deposited using different tools, which exhibited a significant difference in waveguide propagation loss despite undergoing the same patterning process flow. Through comprehensive characterization and analysis of the two films, we found both films have good crystalline qualities and low oxygen concentrations, but differ greatly in grain size. This suggests that scattering and absorption induced by grain boundaries could be among the primary reasons for the high optical loss in AlN. With the low-loss film, we were able to fabricate waveguides on crystalline AlN with loss down to 1.2 dB/cm. The validation of the AlN film’s good crystalline quality is considered essential for achieving its high thermal conductivity. Therefore, we affirmed that AlN waveguide with both low optical loss and high thermal conductivity is achievable. 

Subsequently, we established an experimental platform to investigate the optical heating of NFTs. This platform enables us to demonstrate the integrated functionality of Au-on-AlN devices, using Au-on-SiNx as a comparative group. We fabricated on-chip waveguides equipped with grating couplers and adiabatic tapers for efficiently coupling external light into these devices. This design allows the optical power to be evanescently transferred to the Au NFT, which is strategically positioned atop and in direct contact with the SiNx/AlN core, facilitating effective heat transfer. We observed the Au NFTs on AlN could tolerate 1.5-2.5× power compared with the ones with SiNx as its underlayer, which justified the remarkable heatsinking capability of AlN. 

Being able to measure the temperature of NFTs can further facilitate the understanding of the precise temperature reduction achieved by AlN. To address this need, we utilized the two-color method for temperature measurement. Calibration was performed using an incandescent light bulb with a tungsten filament, with its ground truth temperature determined from its radiation spectrum in the range of 600-1700 nm for high temperatures (1100 - 2800 K) and extrapolated into the lower temperature range (300-1100 K) using a resistance-based method. The two-color method was then applied using both a Si photodetector (PD) and a single photon detector (SPD), iv with the temperatures measured by both detectors showing reasonable agreement with the ground truth. After benchmarking the experimental setup against a standard body, we calibrated and measured the temperature of a micro-scale on-chip tungsten heater, a much weaker radiator, finding good agreement with simulation values. Regarding the optically heated on-chip NFTs, the small emission areas and intense scattered pump light necessitate a better signal-to-noise ratio (SNR) than what we achieved. Our extrapolation suggests that robust measurements could be feasible if the NFT temperature exceeds 1200 K or if the current signal collection efficiency can be improved, along with an additional 30 dB pump light suppression beyond what a single bandpass filter can achieve. 

In summary, this study presents a feasible solution for NFT temperature reduction and supports the key aspect of this design (AlN for heatsinking) with experimental confirmation. Specifically, we proposed a thermally efficient ridge waveguide NFT design heatsinked by AlN. We then experimentally verified the viability of low-loss crystalline AlN waveguides, ensuring no compromise in optical loss or coupling efficiency for high thermal conductivity. The heatsink capability of AlN was further demonstrated with an optical heating based testing platform, which can achieve non-uniform heating with thermal gradients in a fine time scale. We observed a 1.5-2.5× improvement in power tolerance for Au NFTs on AlN compared with their counterpart on SiNx under optical heating, highlighting the substantial thermal advancement offered by AlN. Lastly, we demonstrated the temperature measurement of incandescent light bulb and micro-scale on-chip tungsten heaters using the two-color method, which showed a good agreement with the simulation results. Regarding nano-scale NFT temperature measurement, although the current SNR is insufficient for two-color measurement, our extrapolation indicates that achieving sufficient SNR is possible if the NFT temperature exceeds 1200 K and there is at least an additional 30 dB attenuation of scattered pump light. Further improvements in the signal collection efficiency could extend the capability to conduct such measurements on NFTs into lower temperature ranges.