† Corresponding author. E-mail:
Project supported by the Collaborative Research in Engineering, Science & Technology (Grant No. P28C2-13).
Miniaturization of electronic package leads to high heat density and heat accumulation in electronics device, resulting in short life time and premature failure of the device. Junction temperature and thermal resistance are the critical parameters that determine the thermal management and reliability in electronics cooling. Metal oxide field effect transistor (MOSFET) is an important semiconductor device for light emitting diode-integrated circuit (LED IC) driver application, and thermal management in MOSFET is a major challenge. In this study, investigations on thermal performance of MOSFET are performed for evaluating the junction temperature and thermal resistance. Suitable modifications in FR4 substrates are proposed by introducing thermal vias and copper layer coating to improve the thermal performance of MOSFET. Experiments are conducted using thermal transient tester (T3ster) at 2.0 A input current and ambient temperature varying from 25°C to 75°C. The thermal parameters are measured for three proposed designs: FR4 with circular thermal vias, FR4 with single strip of copper layer and embedded vias, and FR4 with I-shaped copper layer, and compared with that of plain FR4 substrate. From the experimental results, FR4I–shaped shows promising results by 33.71% reduction in junction temperature and 54.19% reduction in thermal resistance. For elevated temperature, the relative increases in junction temperature and thermal resistance are lower for FR4I–shaped than those for other substrates considered. The introduction of thermal vias and copper layer plays a significant role in thermal performance.
Thermal management in electronic systems has become a major challenge since its inception in the 20th century. According to the Moore’s law in electronics, the number of transistors in an electronic circuit would be doubled every two years and adversely leading to miniaturization of the electronic components. Thermal management in high power density electronic boards is absolutely necessary for efficient, reliable, durable and perpetual operation. Metal oxide field effect transistor (MOSFET) is a semiconductor device which is widely used in driver circuits and predominantly in light emitting diode-integrated circuit (LED IC) driver circuits due to certain advantages.[1] There are various types of MOSFET packages such as surface mount, through hole and PQFN packages which are classified according to its applications.[2,3] Reduction in size and increase in power density of the MOSFET are analogous resulting in high demand in terms of thermal management.
Good thermal management techniques are required to achieve durable device without exceeding the recommended junction temperature.[4,5] The miniaturization of the device causes the difficulty of thermal management which emerges as a major problem to sustain the device operation in different environments and in rugged system.[6–8] Hence, special attention is paid by researchers around the world, to studying and managing thermal problems in electronic systems as the devices become smaller and smaller over time.
Junction temperature and thermal resistance of semiconductor device are considered as the critical parameters for reliability, device performance and lifetime. In order to measure the junction temperature and thermal resistance of MOSFET, electrical test method is the common and standard measurement method. By using the electrical test method, it is possible to evaluate internal heat flow path of any semiconductor device. Therefore, the heat flow path enables us to detect any internal defect in the device such as void and degradation of thermal interface material.
There are a few methods to evaluate thermal transient such as thermocouple technique, infrared camera and thermal transient measurement.[9–11] Among these three methods, thermal transient measurement is a common tool in thermal characterization of power semiconductor and internal structure analysis.[10] Many researchers and industries focus on thermal transient method in order to characterize thermal performance of any semiconductor device because it is a reliable and non-destructive measurement. Thermal resistance is the quantity measured to evaluate the heat dissipation capacity. Farkas and Simon[12] carried out the studies on the thermal properties of the channel resistance and parasitic elements of insulated gate devices by using the thermal transient measurement. The change in thermal resistance due to internal serial resistance or in resistance due to individual material was reported. Variation in thermal resistance coefficients due to changes in power dissipation and input current was reported by Lalith et al.[13] Andras et al.[14] conducted the investigations on the simulation based method to eliminate the effect of electrical transients from thermal transient measurements. They suggested that using thermal simulations, the missing initial part of measured thermal transient can be accurately retrieved. Zoltan et al. presented the temperature change induced degradation of SiC MOSFET device.[15]
Printed circuit boards (PCBs) are integral part of any electronic system where the components are mounted on specific substrate in order to obtain the target functions.[16] The choices of MCPCBs and insulated substrates are preferred by PCB manufacturers.[17,18] MCPCBs offer good thermal performances at reasonable cost. In the case of insulated substrates, FR4 and ceramics are widely used substrates as its edges are higher than MCPCBs in terms of weight and cost of the system without compromising the thermal performance.[19] In recent years, many researchers and electronic manufacturers have evolved the thermal vias and copper coated thermal vias for better cooling electronics for small chip packaging. Cho et al. carried out the experiment and simulation on thermal characteristics of glass interposed copper layer through-package vias.[20] From their study the comparison has been made between silicon and glass with vias drilled on both. They found that by the fabrication of thermal vias the thermal performance of interposers has improved. Copper-core MCPCB with thermal vias was investigated by Eveliina at al on high-power COB LED modules.[21] From their investigation, Cu MCPCB showed the outstanding performance on thermal behavior of multichip module by introducing copper-filled microvias. Though many studies were reported in terms of increasing the thermal performance of substrate by suitable coating, confined usage of coating material was not focused due to manufacturing constraints.
In this study, the thermal performances of MOSFETs mounted on FR4 substrate and modified FR4 substrate are discussed, respectively. Three design modifications in FR4 substrates are proposed: FR4 with circular thermal vias, FR4 with single strip of copper layer and embedded vias and FR4 with I-shaped copper layer. The proposed substrates are manufactured by screen-printing process. The thermal performances of MOSFETs mounted on the proposed substrates are determined by thermal transient method by utilizing thermal transient tester (T3ster) system. The Junction temperature and thermal resistance obtained from the experimental measurement are discussed in detail. Heat dissipation capacity and thermal performance for the modified FR4 substrates are compared with those for plain FR4 substrates in order to draw conclusions.
The thermal transient measurement is an electrical test method widely used to characterize the thermal performances of electronic components. The T3ster (Mentor Graphics Corporation) is an instrument to measure the thermal transient which conforms to the JEDEC standards.[22] According to this method, it is not required to remove encapsulation to evaluate the thermal transient. Thermal transient measurement utilizes step function evaluation to gain heating or cooling curve. The heating or cooling curve is defined by temperature measured against response time.
There are four steps in order to obtain thermal transient curve for an electronic component. In the first step, the device under test (DUT) must be calibrated in order to increase the accuracy of the thermal measurement. In the calibration process, a small constant sensor current (1 mA) is needed to calibrate the DUT. Every semiconductor device such as MOSFET, diode, etc, has a temperature-sensitive parameter (TSP). Usually this TSP is responsible for sensing the temperature during the thermal transient testing. For all the device categories, thermal measurement is based on temperature dependence of voltage.[23] The common temperature range for the calibration is from 25 °C to 75 °C and this graph slope is known as K-factor. The voltage over the DUT is recorded in steps of 10 °C until it reaches a steady state with the ambient temperature. Theoretically, K-factor is the correlation of forward voltage (Vf) at constant forward current with junction temperature (TJ), which is given by
Prior to thermal characterization of any device understanding the operation of thermal transient measurement is mandatory. This equipment is compliant to JEDEC thermal testing standard for electrical test method. According to JEDEC 51 the current jump measurement is suitable for the diode-based device. Current jump can be understood by changing power level to fast switching ON to switching OFF. Figure
From the thermal junction measurement response, the cumulative structure function describing the thermal resistance is extracted.[24] The cumulative structure function is a one-dimensional description of the thermal path from the heat source to the ambience. Characteristic parts of the thermal resistance curve can be used to identify thermal domains of the measured device.[25] In many cases the thermal domains are difficult to distinguish because the heat flow path is comprised of materials with similar thermal conductivity and good interface quality between the domains. The differential structure function is the derivative of the cumulative structure function which shows even the small changes in thermal resistance as peaks and valleys.
By definition, the thermal resistance from chip junction to the specific environment, Rth, is the temperature difference between the junction and the reference ambient, ΔT, divided by the heating power, Pheat.[26] For integrated circuit (IC) driver components, thermal resistance (Rth) from the junction to solder point is typically given by[27,28]
Prior to thermal measurement testing, the description of each substrate is explained in detail. MOSFET of 135 W power is used in this study. The detailed specification of MOSFET is given in Table
In this study, the MOSFETs are mounted respectively on four substrates and their thermal performances are experimentally evaluated. The specifications and geometric dimensions of different FR4 substrate are given in Table
The prototypes of the proposed FR4 substrates mounted with MOSFET are attached to the thermostat of T3ster system by using thermal pad. The thickness and thermal conductivity of thermal pad are 0.35 mm and 0.67 W/mK, respectively. In this study, MOSFET is driven at constant current of 2.0 A and voltage of 4.5 V. The MOSFET is subjected to thermal test with above-mentioned operating parameters for duration of 30 s heating time and 120 s cooling time to attain thermal equilibrium for the measurement. During the test, the ambient temperature of the DUT varies from 25 °C to 75 °C by suitable thermostat regulation. It is important to note that after 30 s of heating time, the MOSFET is switched to sensing current of 1 mA during the cooling time. In this study, the superiority in thermal performance of modified substrates and the influence of ambient temperature on the substrates are reported.
As explained earlier, the MOSFET is calibrated to find the K-factor of DUT. The sensitivity coefficient for the DUT heat source is determined in a calibration measurement by using a temperature controlled thermostat cold plate. The DUT is driven with a small sensor current (1 mA) in order to generate only a negligible amount of heat. The voltages over the DUT are recorded at different temperatures, for example, from 25 °C to 75 °C in steps of 10 °C until it reaches a steady state with the ambient temperature. All calibration curves obtained for investigation on different designs of FR4 substrates are illustrated in Fig.
The values of K-factor for MOSFET mounted on FR4 with thermal vias, stacked substrate design and heat spreader design are almost the same each with a minimum deviation of 0.33%. This is expected because the MOSFET packages used are of the same type, and should exhibit similar characteristics.
Thermal performances of MOSFETs mounted on FR4plain, FR4vias, FR4ss_vias, and FR4I–shaped design for various ambient temperatures are measured using T3Ster. The measured data are processed using T3ster Master Software. Smoothed response curves, differential structure functions and cumulative structure functions obtained for the DUTs are given in Figs.
Figures
The value of temperature rise (ΔTJ) at the junction of the MOSFET die can be obtained from smoothed response. By using Eq. (
As illustrated in Fig.
Lastly, the extension of Cu layer on FR4 is designed in order to analyze the effect of higher in-plane conduction without the thermal vias. Increased heat spreader surface area of Cu layer is expected to reduce the TJ value drastically. From Fig.
To gain an insight into the thermal behaviors of the modified FR4 substrates, evaluations of differential and cumulative structure functions are absolutely necessary to identify the different regions in thermal flow path of DUT. A peak-to-peak region in a differential structure function refers to difference in layer material within the DUT. The internal structure having been analyzed as shown in Fig.
Additionally, the partial thermal resistance of MOSFET is found to express a similar reduction in accordance with the performance of the respective DUT. The input current applied to the DUT is not completely utilized by the component where a small portion of input power is used for functional specification while most of the remaining power is eliminated as heat transfers through the junction. Increase in Rthjs refers to less heat dissipation to ambience.
From the results, it is evident that at a constant input current of 2.0 A, Rthja–FR4plain is larger than the Rthja–FR4vias, Rthja–FR4SSvias, and Rthja–FR4I–shaped. Furthermore, the derived structure functions for the proposed substrates shift from left to right in the order of Rthja–FR4plain, Rthja–FR4vias, Rthja–FR4SSvias, and Rthja–FR4I–shaped. The observation affirms that the thermal conductivity and heat dissipation capacity of the respective substrates contribute to reduction in Rthja. The introduction of thermal vias drastically reduces the Rthja, and significant reduction is observed between the peak 3 and peak 5 which was contributed by the presence of thermal vias in FR4 substrates. In the case of differential structure functions of FR4ss_vias and FR4I–shaped, partial thermal resistance for substrates decreases due to higher thermal conductivity of Cu. Further observation shows that the FR4ss_vias possesses higher partial thermal resistance due to the presence of thermal vias whereas lower thermal resistanceis reported for FR4I–shaped. Although the surface area of Cu layer in FR4I–shaped is higher than in FR4ss_vias, the presence of thermal vias in FR4ss_vias contributes to higher partial thermal resistance which is evident due to the presence of mild shift in peak and plateau towards right in differential and cumulative structure function as shown in Figs.
Figure
In this study, the thermal performances of MOSFETs mounted on different FR4 substrates designed are reported. Thermal behaviour of MOSFET mounted on plain FR4 (FR4plain) substrate is compared with those on modified FR4 substrates such as FR4 with circular thermal vias (FR4vias), FR4 with single strip Copper layer and thermal vias (FR4ss_vias) and FR4 with I–shaped copper heat spreader layer (FR4I–shaped). Thermal performances of different substrates are measured by The T3ster. The junction temperature (TJ) and total thermal resistance (Rthja) are obtained by smoothed response curve and by differential structure functions and cumulative structure functions respectively. For constant input current (2.0 A) and ambient temperature (25 °C), FR4I–shaped shows lower TJ and Rthja of 62.44 °C and 21.52 K/W respectively. The presence of thermal vias in FR4ss_vias limits its superiority in thermal performance whereas the utilization of high thermal conductivity material on the substrate proves to reduce the partial thermal resistance within the DUT. Additionally, the influence of ambient temperature on thermal performance of proposed substrate is reported. The Rthja relatively decreases with the increase of ambient temperature due to predominant heat conduction in the substrate, thereby increasing TJ. This study provides an insight into the enhancement in thermal performance of electronic component by introducing thermal vias and high conductive copper layer in substrates.
[1] | |
[2] | |
[3] | |
[4] | |
[5] | |
[6] | |
[7] | |
[8] | |
[9] | |
[10] | |
[11] | |
[12] | |
[13] | |
[14] | |
[15] | |
[16] | |
[17] | |
[18] | |
[19] | |
[20] | |
[21] | |
[22] | |
[23] | |
[24] | |
[25] | |
[26] | |
[27] | |
[28] | |
[29] | |
[30] | |
[31] | |
[32] | |
[33] | |
[34] | |
[35] | |
[36] | |
[37] | |
[38] | |
[39] |