Using Graphene-powder-based Thermal Interface Material for High Lumen LED Array Chip: An Experimental Study of Heat Dissipation

An LED is solid-state lighting with low environmental impact and green technology. It has the advantages of high luminance, long life, structural strength, and short reaction time. With the continuous improvement in luminous efficiency, the heat energy accumulates rapidly, resulting in a chip-on-board (COB) array LED contact surface that easily exceeds the specification temperature Tj and exhibits a sharp drop in light attenuation. Therefore, we blended graphene powder materials and silicone oil to develop excellent graphene grease (10, 15, and 20 wt%) as a contact surface interface medium material with high thermal conductivity, enabling the high-brightness COB array LED to maintain a stable and excellent heat transfer performance. The mixing concentration percentage of the thermally conductive grease is an important influential parameter. Forced convection airflow speeds of 0.5, 1.0, 1.5, 2.0, 2.5, and 3.0 m/s are used to simulate the coolant flow rate. The qualitative and quantitative experimental verifications, discussion, the analysis of temperature changes, and flow visualization using a thermal imager flow field are carried out. The contact surface interface medium material with 15 wt% graphene grease has a relatively good thermal conductivity, but increasing or decreasing the graphene grease concentration causes no clear improvement in thermal conductivity. From the results of forced convection flow visualization, we observed that the airflow speed increases, the thermal boundary layer gradually becomes thinner, and the heat recirculation zone gradually dissipates, which improves the heat transfer effect. In the process of forced convection experiment, we have used the thermocouple temperature (Tcase, Tf, and Ta) sensor and the rotating frequency sensor of axial fan control airflow speed, and a Pitot tube for differential pressure sensing is used to measure and transmit data to a recorder.


Introduction
The lamp is a lighting device and has been the symbol of man-made light since ancient times. With the evolution of technology from the 18th century to the 21st century, inventors have been working hard to change the brightness of man-made lighting devices. (1) The lamp

Principle of High-luminance Chip on Board Array LED
The chip-on-board (COB) array LED is a multichip InGaN/GaN element package, as shown in Fig. 1. The principle of this lighting package is the same as that of the die silicon chip package, because electric energy is directly converted to light and heat energy when a forward bias voltage is applied to the positive and negative ends of the semiconductor. When current flows, electrons are combined with the electromotive force, and the excess energy is released in the form of photons. Depending on the material used, the energy gap of photons can generate light of blue wavelengths, and the light particles passing through the phosphor can produce white light illumination, as shown in Fig. 2. (5)

Graphene Powder and Graphene Thermal Grease
To achieve a better heat dissipation mode of the multichip COB LED, graphene powder (sheet material) is prepared using a newly formulated graphene material with high thermal conductivity. It is combined with methyl phenyl silicone oil as the high-temperature oil [molecular formula: (C 7 H 8 OSi)n, density: 1.102 g/mL at 25 °C (lit.), vapor density: >1 (vs air), and vapor pressure: <5 mm Hg (25 °C)] to prepare the graphene thermal grease used as the interface conductive material in the experiment, as shown in Fig. 3. Graphene thermal grease with different concentrations of graphene powder (10,15, and 20 wt%) is prepared and evenly applied to the interface between the LED substrate and the heat dissipation fin. Then, a type T thermocouple wire is pressed onto the interface, and the LED is locked with the heat sink

Flowchart of experimental research
In this work, the heat conduction from a COB LED to a fin heat sink and the convection of air are studied. The flowchart of COB LED experimental measurement under natural and forced convection study conditions is shown in Fig. 6.

Experimental procedure
1. Use an anemometer and a Pitot tube in the test duct to measure the control values of the frequency and wind speed of the aspirating subsonic wind tunnel, and compare them with each other to ensure the correct wind speed. 2. Apply the heat-conduction graphene grease with the best heat dissipation effect to the interface between the COB LED substrate and the heat sink module, connect the type T thermocouple, and attach the torque plate. 3. Place the COB LED module in the test flow channel, connect two type T thermocouple lines to the COB LED for contact surface temperature and cooling fin surface temperature measurement, and connect one end of two type T thermocouple lines to the GL840 data recorder for real-time temperature monitoring. The recording frequency is once every 2 s. 4. Connect the positive and negative ends of the COB LED to the power supply, adjust the test wattage to 100 W, and monitor the voltage and current constantly to maintain a fixed input power, always paying attention to T case to avoid burn out. 5. At the beginning of the experiment, turn on the COB LED and adjust the wattage to 100 W.
Measure the changes in T case , T f , and T a under different airflow speeds (0.5, 1.0, 1.5, 2.0, 2.5, and 3.0 m/s). Calculate T j and draw the performance curve. The experimental recording time and temperature reach the steady state. 6. After the temperature reaches the steady state, use the infrared thermal imager to capture the temperature distributions of the COB LED and cooling heat sink module under different airflow speeds. 7. Install the digital camera and laser light generator equipment in their correct positions, turn on the laser light generator after operating the infrared thermal imager, and introduce smoke into the airflow channel to observe the change in heat flow visualized near the COB LED heat sink module and the effect at the thermal boundary with the heat flow field state.

Experimental equipment
The thermal experiment and flow visualization of forced convection for the test flow channel of 800 mm L × 200 mm W × 200 mm H is shown in Fig. 7. The equipment used in the experimental measurement is the laser (output power: 300 mW, laser beam: ψ2.5 mm, wavelength: 532 nm for green light, and output stability: ≤ ±5%), the camera recorder with flow visualization (full HD 1080P, 24.72 × 10 6 pixels, and 1/8000 to 30 s speed), the multirange DC power supply (voltage: 0 to 80 V, current: 0 to 40.5 A, and power output: 1080 W), the data acquisition system (GL-840M, 20 channels, and 20 mV to 100 V), the thermocouple (Type T, temperature range: −270 to 370 °C, and accuracy standard: ±1.0 °C or ±0.75%), the infrared thermal imager (measuring range: −20 to 250 °C, thermal image pixels: 80 × 60 pixels, and accuracy: ±2% or ±2 °C), and the digital light meter spectrometer (spectral wavelength range: 360 to 750 nm, resolution: 12 nm, measurement time: 8 to 1000 ms, illumination: 70 to 70000 lux, CIE 1931 and CIE 1976, relative color temperature accuracy: ±2%, and illumination accuracy: ±5%). The convection environmental control room was used to test the LED temperature, as shown in Fig. 8.

Forced convection performance evaluation in environmental control room
Forced convection air cooling is used to test the change in T case under different airflow speeds. A standard wind tunnel is used to control the forced air convection speed to determine the optimal air cooling speed for dealing with the hot spots of heat accumulation on the multichip COB LED. The forced convection is applied in the wind tunnel channel test and device and frame experiments. Graphene grease has good thermal conductivity and reliability. The COB LED is tested in a forced convection wind tunnel with an ambient temperature of 25 °C for a test time of 3000-3600 s. T case values are 57, 56, and 55 °C at airflow velocities  of 1.5-1.7, 2.0, and 3.0 m/s, respectively. When the airflow speed is close to 1.7 m/s, the heat dissipation of the COB LED tends to be stable, and increasing the airflow speed has little effect on heat dissipation. Therefore, the forced convection wind speed of 1.7 m/s is selected as the benchmark of forced convection airflow speed. Because natural convection cooling does not solve the hot spot problem of the 100 W COB LED, forced convection cooling is promising as the heat transfer method. The concentration of graphene powder in the graphene grease used as the interface thermal conduction medium material for forced convection is 15%. The aluminum extrusion and cold forging heat sinks are configured on the basis of heat capacity shown in Fig.  9(a).
In the experiment, after T case becomes stable, the temperature distribution of the heat sink module under different airflow speeds, the observation of the heat flow field, and the effect of (1)

Aluminum extrusion heat sink module
The interface between the 100 W COB LED alumina substrate and the aluminum extrusion heat sink is made conductive by applying self-developed graphene thermal grease in forced convection (1.5-1.7 m/s). COB LED T case is 57-59 °C, which is lower than the specification temperature of 75 °C for the COB LED, and COB LED T j is 90-92 °C, which is lower than the specification temperature of 125 °C for the COB LED, as shown in Fig. 9(b). The graphene thermal grease is a suitable thermally conductive (grease) interface material that will not be oxidized or affected by the environment. The self-developed graphene grease is used at the interface between the 100 W COB LED alumina substrate and the aluminum extrusion heat sink, and liquid graphene is sprayed on the surface of the aluminum extrusion heat sink module to induce the forced convection (1.5-1.7 m/s) mode. After 1,200 s of forced convection, COB LED T case reaches 64 °C, and the temperature continues to rise below the specification temperature of 75 °C for the COB LED. COB LED T j reaches 92 °C, which is lower than the specification temperature of 125 °C for the COB LED. Overall, it is 2 °C higher than the temperature in the case without the spraying of liquid graphene on the surface of the aluminum extruded heat sink. It shows that graphene is a poor material for the surface coating of the heat sink module surface. It causes an undesirable heat transfer effect of black-body thermal-radiation-absorbing heat energy, as shown in Fig. 9(c).

Cold forging aluminum heat sink module
When the 100 W COB LED of the forging heat sink module is activated (the interface has self-developed graphene grease), the COB LED temperature rises faster and becomes more stable than that of the aluminum extrusion heat sink module. The voltage and current of the COB LED can be accurately controlled within the set range so that T case of the COB LED can be controlled very stably between 59 and 61 °C, which is lower than the specification temperature of 75 °C for the COB LED. T j is between 90 and 91 °C, which is lower than the specification temperature of 125 °C for the COB LED. However, compared with the cooling temperature of the aluminum extrusion heat sink module, the stable temperatures of the two materials are close to each other, and there is no significant difference between them. The self-developed graphene grease can help the heat transfer at the interface and can cool the convection airflow (1.7 m/s), as shown in Fig. 9(d).

Change in COB LED interface temperature
The change in COB LED interface temperature is examined by forced convection testing under the following conditions: 15% graphene grease at the interface of the 100 W COB LED, ambient temperature T a of 25 °C, and airflow speeds of 0.5, 1.0, 1.5, 2.0, 2.5, and 3.0 m/s.

Airflow speed of 0.5 m/s
The COB LED interface temperature T case is stable from 1600 to 3600 s, and time is measured between 0 and 3600 s when the convection airflow speed is 0.5 m/s from the start of forced convection, as shown in Fig. 10(a). It is found that the interface temperature T case is 73.38 °C, which is lower than the specification temperature of 75 °C for the COB LED. The calculated chip junction temperature T j is 94.67 °C, which is lower than the specification temperature of 125 °C for the COB LED. The heat sink fin module temperature T f is 45.78 °C.

Airflow speed of 1.0 m/s
The COB LED interface temperature T case is stable from 1200 to 2600 s, and time is measured between 0 and 2600 s when the convection airflow speed is 1.0 m/s from the start of forced convection, as shown in Fig. 10(b). It is found that T case is 69.14 °C, which is lower than the specification temperature of 75 °C for the COB LED. The calculated T j is 93.82 °C, which is lower than the specification temperature of 125 °C for the COB LED. T f is 42.16 °C.

Airflow speed of 1.5 m/s
The COB LED interface temperature T case is stable from 1400 to 2400 s, and time is measured between 0 and 2400 s when the convection airflow speed is 1.5 m/s from the start of forced convection, as shown in Fig. 10(c). It is found that T case is 66.89 °C, which is lower than the specification temperature of 75 °C for the COB LED. The calculated T j is 93.37 °C, which is lower than the specification temperature of 125 °C for the COB LED. T f is 40.23 °C.

Airflow speed of 2.0 m/s
The COB LED interface temperature T case is stable from 1200 to 2200 s, and time is measured between 0 and 2200 s when the convection airflow speed is 2.0 m/s from the start of forced convection, as shown in Fig. 10(d). It is found that T case is 63.85 °C, which is lower than the specification temperature of 75 °C for the COB LED. The calculated T j is 92.77 °C, which is lower than the specification temperature of 125 °C for the COB LED. T f is 37.80 °C.

Airflow speed of 2.5 m/s
The COB LED interface temperature T case is stable from 800 to 2200 s, and time is measured between 0 and 2200 s when the convection airflow speed is 2.5 m/s from the start of  forced convection, as shown in Fig. 10(e). It is found that T case is 61.13 °C, which is lower than the specification temperature of 75 °C for the COB LED. The calculated T j is 92.22 °C, which is lower than the specification temperature of 125 °C for the COB LED. T f is 35.61 °C.

Airflow speed of 3.0 m/s
The COB LED interface temperature T case is stable from 600 to 1800 s, and time is measured between 0 and 1800 s when the convection airflow speed is 3.0 m/s from the start of forced convection, as shown in Fig. 10(f). It is found that T case is 58.54 °C, which is lower than the specification temperature of 75 °C for the COB LED. The calculated T j is 91.70 °C, which is lower than the specification temperature of 125 °C for the COB LED. T f is 33.34 °C.
It can be seen from Fig. 11 that with the gradual increase in airflow speed, T case and the calculated T j show a gradually decreasing trend, and when the airflow speed increases, the time required for temperature to reach stability gradually decreases: 3600 s at 0.5 m/s and 1800 s at 3.0 m/s. The difference in T case between the cases of 0.5 and 1.0 m/s is 4.24 °C and that between the cases of 2.5 and 3.0 m/s is 2.59 °C. After the airflow speed reaches a certain value, the cold air molecules cannot completely remove the heat of the hot air molecules on the material surface, whereby the thermal boundary layer continues to exist, but the energy consumption increases, with the consequent result that heat cannot be removed continuously. Therefore, the optimization of the airflow rate and the combination of the heat sink fin area and rough surface will lead to improved heat transfer. The COB LED device is better than a fluorescent lamp in controlling the power wattage, color temperature, lighting, and color rendering index with energy-saving.

Airflow speed of 0.5 m/s
The results obtained under air convection with the airflow speed of 0.5 m/s are shown in Fig. 12(a). When the smoke flows through the cooling module, it can be seen that the flow is disturbed by temperature and thermal buoyancy. The smooth flow field becomes chaotic after the smoke passes through the heat dissipation module, and the thermal buoyancy effect also causes the line of heat flow to drift upward. As shown in Fig. 12(b), it is clear that the heat dissipation module is affected by high temperature, and the surface of the heat sink fin forms a thermal boundary. With higher temperature, the thermal boundary layer is thicker. When the smoke passes between two cooling fins, it is squeezed by the fin spacing and thermal boundary layer, and the closer the smoke is to the bottom of the heat dissipation module, the higher the temperature, the thicker the thermal boundary layer, and the greater the impact of heat dissipation, resulting in higher velocity upstream and lower velocity downstream. The downstream smoke forms a vortex retention area, and hot air flows back to the bottom of the cooling module, which reduces the heat dissipation capacity.

Airflow speed of 1.0 m/s
The results obtained under air convection with the airflow speed of 1.0 m/s are shown in Fig. 13(a). Compared with the case of airflow speed of 0.5 m/s, the effect of thermal buoyancy  decreases because of the higher airflow speed, and the circulating eddy current retention area also lengthens. Also, as shown in Fig. 13(b), the smoke in the circulating vortex detention area is still squeezed by the flow field. Although the backflow of smoke causes some interference on the heat dissipation capacity, the backflow of hot air decreases and the effect on the heat dissipation module gradually decreases owing to the lengthening of the circulating eddy current retention area.

Airflow speed of 1.5 m/s
The results obtained under air convection with the airflow speed of 1.5 m/s are shown in Fig. 14(a). Compared with the case of airflow speed of 1.0 m/s, the effect of thermal buoyancy decreases because of the higher airflow speed, and the circulating eddy current retention area gradually disappears. As shown in Fig. 14(b), the airflow is still turbulent. Compared with when the airflow speed is 1.0 m/s, the amount of smoke that returns to the recirculation vortex retention area decreases, that is, less hot air returns, and the heat gradually dissipates from the heat dissipation module area.

Airflow speed of 2.0 m/s
The results obtained under air convection with the airflow speed of 2.0 m/s are shown in Fig. 15(a). Compared with the case of airflow speed of 1.5 m/s, the effect of thermal buoyancy gradually disappears owing to the higher airflow speed and the circulating vortex detention zone is not obvious. As shown in Fig. 15(b), the state of flow is relatively smooth. Compared with when the airflow speed is 1.5 m/s, the circulating vortex flow retention zone also gradually dissipates, such that the hot air reflux decreases and the amount of heat that can be removed by the heat dissipation module gradually increases.

Airflow speed of 2.5 m/s
The results obtained under air convection with the airflow speed of 2.5 m/s are shown in Fig. 16(a). Compared with the case of airflow speed of 2.0 m/s, the effect of thermal buoyancy gradually disappears because of the higher airflow speed and the circulating vortex detention zone is not very chaotic. As shown in Fig. 16(b), the state of flow is relatively smooth. Compared with when the airflow speed is 2.0 m/s, the return of smoke to the circulating vortex retention zone is negligible, which means that less hot air returns and the amount of heat that can be removed by the heat dissipation module gradually increases.

Airflow speed of 3.0 m/s
The results obtained under air convection with the airflow speed of 3.0 m/s are shown in Fig. 17(a). Compared with the case of airflow speed of 2.5 m/s, the smoke at the rear end disappears because of the higher airflow speed and there is no chaotic disturbance in the circulating vortex detention area. As shown in Fig. 17(b), the smoke and state of flow are relatively smooth. Compared with when the airflow speed is 2.5 m/s, the recirculation of smoke in the circulating vortex retention zone almost disappears and the hot air reflux is negligible. The heat removal capacity of the cooling module greatly improves.
The flow visualization phenomena shown in Figs. 12-17 reveal that with increasing airflow speed, the thermal boundary layer between the cooling fins gradually deteriorates. Under low airflow speed, the thermal boundary layer is not easily damaged, resulting in the fluid being unable to pass through the fins smoothly and thus affecting the heat dissipation effect. When the airflow speed gradually increases, the air pressure gradually crushes the thermal boundary layer, so that the cold air can be closer to the surface of the heat sink fin, effectively accelerating  the removal of heat. The circulation eddy current retention zone gradually dissipates the eddy current and little reflux occurs, which greatly improves the heat removal ability of the heat dissipation module.

Infrared thermal imager temperature measurement
The temperature distributions obtained under different airflow speeds are examined. When the airflow speeds are 0.5, 1.0, 1.5, 2.0, 2.5, and 3.0 m/s, the infrared thermal image shows the maximum temperature, measured T f (T heat sink fin ), and emissivity ε, which are indicated in Table 1. Figures 18-23 show infrared thermal images that indicate the temperature distribution of cooling fins at different airflow speeds. The temperature at the heating position of the COB LED substrate is the highest, and heat is transmitted to the heat dissipation fin module. In each side view, it can be seen that at a lower airflow speed, the high-temperature distribution area (red) is larger. With increasing airflow speed, the red area gradually decreases, indicating that heat is removed smoothly under high airflow speeds. The actual measurement position is at the center of the middle fin module, the continuing study of which is valuable and of physical significance.

Luminance
The COB LED is placed in an environmental control room and T case is controlled by forced convection cooling (1.5-1.7 m/s). The improvement in luminance caused by heat dissipation is observed. The changes in voltage and current are recorded and the change in luminous efficiency is analyzed. The illumination is measured with a spectrophotometer. As shown in Fig. 24, T case is 58 °C, T j is 92 °C, the illumination is 9626 lux, the color temperature is 3261 K, the color rendering property is 82, the spectral peak is at 601 nm, and the input power is 101.4 W at the ambient temperature (T a ) of 25 °C. The input power is 100 W when T case is 50 ± 5 °C. The temperature of the environmental control room gradually decreases from 25 to 20, 15, 10, 5, 0, and −5 °C. The voltage, current, illuminance, power, T case , and T j are recorded every 5 °C decrease in temperature. As shown in Fig. 25, T case is 36 °C, T j is 86 °C, the illumination is 10433 lux, the color temperature is 3272 K, the color rendering property is 82, the spectral peak is at 601 nm, and the input power is 97.4 W at T a of −5 °C. The results show that the overall       efficiency of COB LED illumination is improved by 8.3%, and 4.0% of input power is saved. The overall efficiency of the COB LED is significantly improved, and the change in T case with illumination and input power can be seen in Fig. 26.

Conclusions
The design of the COB LED light-emitting module adopted in this study is the mainstream in multichip solid-state lighting. The use of convection heat dissipation is needed to strengthen the customized heat transfer mode. The use of an interface thermally conductive material is also important. Moreover, graphene has become the development option of the new generation of thermal conduction materials. Its heat conduction capacity is good and it is easy to deploy. The development of thermal graphene grease materials has been highly successful, and they will be widely used as thermal conduction interface materials in the future, mainly because if T case can be decreased, the power consumption of the LED module will decrease and its luminous lighting will increase, which means that the overall luminous efficiency will be improved with the improvement in heat dissipation capacity. For convection with low airflow speed, the larger the temperature difference, the less heat energy will be taken away. If the  temperature difference is small, much heat energy will be taken away. Between 0.5 and 1.0 m/s, the difference in T case is 4.2 °C. Between 2.5 and 3.0 m/s, the difference in T case is 2.6 °C. The high-speed convection helps to reduce T j , and the temperature easily reaches stability quickly. The measured data are obtained by thermocouple temperature sensing and airflow speed sensing under steady state. However, when the air convection wind speed reaches a certain level, the heat dissipation capacity decreases because the shorter the stay time of the air-cooled molecules on the surface of the fin, the less effective is the heat exchange and, therefore, heat energy cannot be reduced and the load and energy consumption increase. Furthermore, it is found that the surface of the heat sink fin is affected by the buoyancy of high-temperature air, resulting in a greater thermal boundary layer thickness when the temperature is high. As the temperature increases, the thermal boundary layer thickens, resulting in smaller gaps between the cooling fins. High-speed convection cannot destroy the thermal boundary layer owing to the wind pressure, but a retention area is formed at the tail end of the heat sink fin. The heat dissipation capacity is reduced by backflow around the fin, and the flow rate increases when such backflow occurs. Therefore, the development of the COB LED thermal conduction interface graphene grease material is important, and it is also necessary to introduce and strengthen material configurations in a timely manner.