Thermal Performance of High Power LED on Boron Doped Aluminium Nitride Thin Film Coated Copper Substrates

Aims: The paper is aim to study the thermal performance of Boron doped Aluminium Nitride (B-AlN) thin film coated over Copper (Cu) substrate to improve surface configuration of the interface between two materials with different synthesis parameters. Study Design: Synthesis of Boron doped AlN thin film by sputtering and post processed for various temperatures. The processed samples were characterized to study the behavior of B doping as well as the annealing temperature in changing the properties of B doped AlN thin film. The structural and surface properties were studied and reported. Place and Duration of Study: Nano Optoelectronic Research Laboratory, School of Physics, University Sains Malaysia, Penang 11800, Malaysia, between December 2013 and July 2014. Methodology: B-AlN thin films were prepared with five different gas ratios over Cu substrates by DC-RF coupled sputtering method and suggested for thin film based thermal interface material Original Research Article Ong et al.; JSRR, 5(2): 109-119, 2015; Article no.JSRR.2015.078 110 (TIM). 3W green LED package was tested with B-AlN thin films coated Cu substrates through thermal transient and surface analysis. The results are compared with the performance of bare Cu substrates. Results: The thin film prepared with gas ratio of Ar 7: N2 13 coated at 200°C showed the lowest thermal resistance Rth (53.27 K/W), board to ambient thermal resistance RthB-A (36.03 K/W) and the lowest junction temperature Tj(120.98°C) at higher driving current (700 mA). Surface analysis results show that the thin film mentioned above exhibits low in surface roughness (8 nm) and range of valley depth (30-80 nm), which contribute in thermal performance of LED. Conclusion: Overall, B-AlN films coated with gas ratio of Ar 7: N2 13 are more favorable in reducing both total thermal resistance and junction temperature (Tj) of LED.


INTRODUCTION
In the advancement of the luminescence technology, evolution of Light Emitting Diodes (LEDs) have significant impact on lighting solutions which provides better optical performance than incandescent/halogen lighting technology, with its efficiency of about 10-20%. However, the efficiency of LEDs strongly depends on the junction temperature and improper thermal management would cause excessive rise in junction temperature that eventually compromising the overall performance [1,2]. Junction temperature of the LEDs need to be maintained below 120°C, if it exceeds which may lead towards device failure or damage [3]. Around 80% of the heat generated at the LED junction when forward biased must be removed to the ambient [4]. In order to extend the life of the LED device, a proper heat path from the junction to ambient must be established and addressed.
In recent PCB technology, Fire Retardant 4 (FR4) has been commonly used as electrical circuit board for LEDs. Recently high thermal conductive ceramics such as Alumina (Al 2 O 3 ) and Beryllia (BeO) are being employed as dielectric materials in circuit board. Alumina is popular employed as insulating base, but it has its own drawbacks such as incapability of radiating heat emitted by LED chip. The beryllia type ceramic exhibits an excellent heat-radiating property, butit is classified as toxic and expensive [5]. To meet out the demands in electronic industry, highly conductive materials with reasonable production cost are preferred.
Group III nitrides are good in thermal conductivity, high thermal stability, high mechanical strength, excellent in electrical insulation, and highly recommended for electronic applications. Theoretically, thermal conductivity of AlN is reported to be about 280 Wm -1 K -1 [6], however the typical value of AlN thin film is significantly lower at the range of 0.4 -26 Wm -1 K -1 and vary depending upon the deposition process conditions and their properties such as crystallite structure, film thickness, and impurities [7]. Although bulk BN has relatively high thermal conductivity (1300 Wm -1 K -1 ) but it is commercially difficult to produce [8]. Very few works on combination of group III elements such as boron and aluminum in nitride thin film were reported and found that the boron doping can shrink the lattice of AlN and thus improves the thermal conductivity [9,10]. Several works had been reported on LED performance of AlN and BN coated substrates and noticeable improvement was achieved at high driving currents [11][12]. Though many papers have been reported on thermal properties of AlN and BN separately, B-AlN has not been reported so far.
In this study, B-AlN thin film were deposited for about one hour at different process conditions by DC-RF coupled sputtering method on copper (Cu) substrates. The thermal performance of 3W LED fixed on B-AlN thin film coated Cu substrates were conducted using thermal transient analysis method and the results based on cumulative structural function are presented here.

THEORITICAL BACKGROUND
The device junction temperature in the test condition can be determined by: Where T j0 = initial device junction temperature (ºC), ΔT j = change in junction temperature due to heater power application (ºC). It should be noted that the relationship between ΔT J and power dissipation is usually linear under some specific conditions, and it may vary considerably at the extremes of device operation. The method itself is independent of the environment of the device under test (DUT), thus it is required to pay careful attention on environmental conditions in order to assure the significant test results. Static mode was applied using still air box for all measurement, which applies heating power to the DUT on a continuous basis while the T j was monitored through measurement of temperaturesensitive parameter.
The efficiency of heat dissipation in LEDs is measured in terms of thermal resistance (R th ) which serves to the determination of junction temperature that arises in the LEDs under various operating conditions. It also helps to determine the maximum allowable external reference temperature for a given internal temperature and power dissipation. According to JEDEC standard 51-1 [13], the R th of heat flow from the junction to ambient (total thermal resistance) of the LED is generally defined as in Eq.2: Where R th is the thermal resistance from device junction to the ambient, T J is the device junction temperature under steady state condition, T A is an ambient temperature and P H is the heat power dissipated in the device. By reducing the R th , the device can be protected from overheating and led to operate for a longer lifetime.

B-AlN Thin Film Synthesis
B-AlN thin films were deposited on Cu substrates (23 cm (l) x 25 cm (b) x 1.5 cm (h)) by sputtering method using DC-RF coupled sputtering machine (Edwards make, Model-Auto 500). The sputtering process was conducted using pure Al (99.99% purity) target (3 inch in diameter and 4 mm in thickness) and Boron (B) (99.99% purity) target (3 inch in diameter and 4 mm in thickness). DC power source and RF power source were used to sputter Al and B target respectively. High pure Ar (99.999%) and N 2 (99.999%) were used as sputtering gas and reactive gas (nitrogen source) respectively. The Cu substrates were initially cleaned by rinsing in ultrasonic bath of acetone and isopropyl alcohol for 10 minutes.
The deposition chamber was initially pumped down to high vacuum of 3.56 x 10 -5 mbar by using a turbo molecular pump backed by a rotary pump and fixed as base pressure for coating. In order to clean the surface oxidation of the targets, pre-sputtering process was carried out for about 5-10 minutes prior to film deposition. The discharge power of DC and RF were kept constant at 100W and 50 W respectively under the pressure of 9.54 x 10 -3 mbar. A rotary drive assembly was used (25 RPM) to ensure the uniform film thickness. The B-AlN films were prepared on Cu substrates for about 1 hr at three different Ar and N 2 gas mixture ratios. The selective thin film samples were sputtered at two different temperatures for comparison.
The sputtering parameters used and the thicknesses obtained are summarized in Table 1.
The surface morphology of all thin film samples was also analyzed using Bruker AXS make dimension edge atomic force microscope (AFM) system and the results were analyzed by using Nanoscope analysis software. In order to understand easily, the sample name (Sample 1 to 5) will be used for the discussion of our results in this paper.

Thermal Transient Analysis
In order to test and compare the performance of the thin film as thermal interface material on Cu substrates, 3W green LED package with starboard MCPCB were used and fixed on the B-AlN thin film coated substrates for thermal transient analysis. The measurement was carried out by Thermal Transient Tester (T3Ster) in a still air chamber (300 x 300 x 300 mm) at room temperature of 25±1°C according to JEDEC JESD 51-2A standards on natural air convection measuring method [14]. The schematic diagram of boundary conditions for LED testing is given in Fig. 1.
Before the real measurement, the LED was thermally calibrated using dry thermostat and T3Ster as the power supply. The product of K and the difference in temperature-sensing voltage (referred to as ΔV F ) produces the device junction temperature rise: During the calibration process, the LED was driven with lower operating current at 1mA to prevent self-heating effect at the junction. The ambient temperature of the LED was fixed at 25 ˚C and the voltage drop across the junction was recorded once the LED reaches thermal equilibrium with the temperature of the thermostat. Later, the ambient temperature of the LED was varied from 25°C to 85°C at 10°C step (35°C, 45°C, 55°C, 65°C, 75°C and 85°C) and the voltage drop across the junction was noted at each ambient temperature. From the calibration process, the K-factor of the LED was determined (2.289×10 -3 V/°C) from the graph of junction voltage (voltage drop) against ambient temperature as shown in Fig. 2.
During the thermal measurement, the LED was forward biased at three different driving currents (100 mA, 350 mA and 700 mA) for 900 s until it reached the steady state. Once the steady state was achieved, the transient cooling curve of heat flow was captured for additional 900 s. The thermal transient curve under different measurement conditions is captured based on the electrical test method JEDEC JESD51-51 [15]. The recorded cooling profiles of the measurement were processed for structure functions using T3Ster Master Software. Fig. 3(a-c) represents the complete heat transfer profile in the form of cumulative structural function from junction to ambient which is captured for the given LED fixed on B-AlN thin film coated Cu substrates at (a) 100 mA, (b) 350 mA, (c) 700 mA. From the Fig. 3(c), the curve region between point A and B is the thermal resistance of junction to board which is the substrate, (R thJ-B ) of LED package at 700mA. The R thJ-B value was recorded as 17.88 K/W based on JEDEC JESD51-14 standard [16]. According to the product datasheet, the R thJ-B of the given LED at 700 mA is 17 K/W [17], this may due to different measurement method and conditions practiced by the industry. The region between the point B and C widely covered the profile and it explains the heat path from substrate to ambient.

Thermal Transient Measurement
From the analysis of cumulative structural function, the thermal resistance (R th ) is recorded and presented in Table 2. From the measurement of R th shown in Fig. 3, it is found that Sample 1 shows the highest R th value in all driving currents (100 mA, 350 mA, 700 mA). This may due to defects in crystal structure grows with gas ratio of Ar 6: N 2 14 and hence influences the heat conductance. At 100 mA, Sample 2 shows the lowest R th value at 53.87 K/W. While Sample 5 achieved the lowest thermal resistance at high driving currents (350 mA and 700 mA).
Thermal resistances of board to ambient (R thB-A ) from the cumulative structural function are also analysed and the values are summarized in Table 2. From the measured results, it is observed that the R thB-A value decreases as the driving current increases. It obeys the approach of Debye-Model, where the rise in driving current increases the thermal conductivity as the inverse proportionality between thermal resistance and conductivity shifts the transient towards lower resistance [18]. From Table 2, the lowest thermal resistance of R thB-A (36.03 K/W) is also observed with Sample 5.    Table 3 shows the rise in junction temperature of LED on thin film coated and bare substrates. Overall, Sample 5 performs well on reducing Tj of LED at higher driving currents. As observed from the measured results, Sample 1 exhibits highest Tj for all three driving currents, which agrees well with the result of thermal resistance. There is a difference in junction temperature (ΔTj) of 5.77°C between Sample 5 and Sample 1 when LED is driven at 700 mA. As the current increase from 350 mA to 700 mA, there is an increment on measured Δ T j between Sample 1 and Sample 5. This may attribute to the effect of temperature on changing the thermal conductivity of thin films [22]. In addition, there is a noticeable difference in T j variation pattern between100mA and higher driving currents (≥350 mA). This may due to the greater effect of molecular agitation of free electrons within the material at higher T j when LED operates at higher driving current [23].

Surface Analysis
In order to analyze the contact resistance of B-AlN/Cu with MCPCB board (check which LED used), the surface morphology was recorded by using AFM with tapping mode at scan range of 10μm and analyzed the surface parameters. The captured surface morphology of B-AlN/Cu is presented in 3D images as shown in Fig. 5. According to the surface roughness (R a ) of B-AlN thin films plotted in Fig. 6, the surface roughness of bare Cu substrate is 11nm. Sample 1 exhibits the highest value in surface roughness (14 nm) and this shows that the B-AlN coating with gas ratio of Ar 6: N 2 14 increase the surface roughness of Cu substrate, which leads to increase of the R th and T j of the given LED. This may due to disordered crystal structure at interface region contributed by stress and strain developed during synthesis, hence formed rougher surface and resulted in highest R th at all current. According to Yovanovich et al. [24], contact conductance is high for the surface with low surface roughness. Even though the lowest surface roughness is achieved by Sample 2 (7 nm), it does not support in reducing both R th and T j of given LED at higher input current (350 mA and 700 mA), though its R th is the lowest at 100 mA. It may attribute to the defects formed in thin film structure and hence reduced the thermal conductance for coated B-AlN on Cu. A relatively low value in roughness (8 nm) is also achieved with Sample 5. It shows that the coating prepared with gas ratio of Ar 7: N 2 13 at 200 °C reduces surface roughness of metal substrate and conducts more heat than the coatings prepared at other parameters. The AFM images of all samples were processed for depth-valley analysis using AFM software and the observed results are shown in Fig. 7. The thermal conductance is related to the no. of points made contact with the surface of the material, if the no. of contact points is higher, these will result in increased contact resistance. From Fig. 7

CONCLUSION
B-AlN thin film was prepared on Cu substrate by DC and RF coupled sputtering at different gas ratios and processed at various temperatures. B-AlN thin films were used as thermal interface material for high power LED. The R th derived from the thermal transient curve was lower for B-AlN coated Cu substrates synthesized with gas ratio Ar 7: N 2 13 at 200°C than for bare Cu substrate measured at 700 mA. Consequently, noticeable reduction in T j of the given LED was also achieved by the same boundary condition. Surface analysis showed the influence of synthesis and processing parameters on the surface of B-AlN thin film which was affected the thermal performance of high power LED. Overall, it is suggested that the introduction of temperature (200°C) on preparation of B-AlN coating using gas ratio of Ar 7: N 2 13 over Cu substrate as an effective thermal interface material for enhanced solid state lighting performance.