Reverse Engineering a Commercial LED Light

Siyuan Liu

University of Rochester, Department of Electrical and Computer Engineering

1. Introduction

The sample I explore is a LED component. It is obtained from a commercial LED lamp. The LED component is a 5 mm*5 mm thin chip. It emits white light under working condition. On the top of the chip, there are four individual LED crystal. They are covered by a thin layer of fluorescent resin and a hemi-sphere of transparent resin. They are used to transform the color of the emitted light and condense the emitted light from the four LED crystal. After removing the resin cover, the tiny LED crystal can be clearly seen with light microscope or the SEM. What I mainly investigate are the LED crystal and materials right beneath it. The overall image of the LED chip is obtained in the SEM. It is shown in Figure 1.1.

Figure 1.1: SE2 Image of the Overall LED Chip

Light Emitting Diode (LED) is a semiconductor light emitting component. The core of LED is the p-n junction. When a semiconductor material is doped with donors, this semiconductor will become a n-type semiconductor. On the other hand, when a semiconductor material is doped with acceptors, this semiconductor will become a p-type semiconductor. When these two types of semiconductor contact with each other, the p-n junction will be formed. LED is the p-n junction working under positive bias. When electrons from the n-type material recombine with the holes from the p-type material in the p-n junction, light is emitted.

The equipment used to explore the LED is the Scanning Electron Microscopy. Compared with light microscopy, the SEM has higher resolution and larger magnification. The secondary electron mode allows me to observe the micro-structure of the LED. Also, the backscatter mode in the SEM can help me distinguish different materials. With its EDS function, even the elemental composition of the LED and the distribution of each element can be easily determined.


2. Sample Preparation

In order to observe the multi-layers structure of the LED component, the cross section of the LED must be obtained. First, we used an aluminum sample holder to fix the LED chip and adjusted the position of the LED chip to make sure the cross section we wanted was slightly above the aluminum rings. Second, we put it into a round plastic box and filled it with acrylic resin. We cured it in the oven at 70 Celsius degree for about two hours. When the resin was dry, it became solid. The LED chip was tightly fixed within the resin. Third, we used the polisher to grind the LED chip with the resin. During this process, we observed the polished surface in the light microscope constantly. We used back light mode to find the crystal we wanted to explore and used front light mode to observe the cross section. Finally, when we observed the entire cross section of the crystal, we changed the polishing paper into finer one and kept polishing until the surface looked very flat and clean in the light microscope. The cross section of the LED crystal was obtained.

For scanning electron microscopy, the surface of the sample must be conductive. Otherwise, the image will be affected by charging. It will not only decrease the resolution but also erase many details of the image. When we observed the overall structure of the chip, the electrons hitting on the sample could be conducted to the bottom and released through the sample stub. But after we fixed the sample with the resin, the electrons hitting on its surface could not be released anymore. So, we sputter coated the sample with gold. First, we fixed the sample on the sample stub with carbon tape. Second, we placed the sample stub into the coating chamber and turned the system on. Third, we opened the valve of the argon gas to make sure argon gas could be filled into the system. Forth, we opened the sputter solenoid valve and closed the argon gas needle. When the system was pumped down to 100 mTorr, we opened the argon needle to backfill the system with argon gas. After several repetition, we pumped down the system to 100 mTorr. Finally, we started the operation. A high voltage was applied to the gold cathode. We controlled the current at about 15 mA for 60 seconds. Then, the sample was coated and ready to be observed in the SEM

3. Result of Microscopy and Discussion

At first, we used SE2 mode to get a sense about the structure of the LED component. The SE2 image is shown in Fig 3.1. We got the cross section of two paralleled LED crystal. Under the thick cladding crystal, there are a few thin layers. They are hard to distinguish in this magnification. We can tell that these layers connect with the metal pads on the circuit board. The metal pads are separated by a small trench. This is corresponding to the trenches shown in Fig 1.1. Based on this structure, we think the cathode and anode of the LED should locate at the left and right bottom layers of the crystal. It means we can both observe the cathode side layers and anode side layers on this cross section. Beside the crystals, we fortunately got the residue of the fluorescent material. Most of the residue is between those two crystals. The residue mainly consists of two parts, the particles and the glue material. Based on our understanding, those particles should be the real fluorescent material. The glue material was used to attach those particles to the crystals. Same area was captured in BSE mode. As is shown in Fig 3.2, the height information on the surface of this cross section is eliminated. But the difference between each material becomes more obvious, especially for those thin crystal layers.

Figure 3.1: SE2 Image of the Cross Section

Figure 3.2: BSD Image of the Cross Section

The most interesting area in this cross section is the layers between the electrode and the thick cladding crystal. We used SE2 mode and BSD mode to determine the distribution of different layer. As is shown in Fig 3.3 and Fig 3.4, there are two thin layers. The thickness of these layers is about 1 micron.

Figure 3.3: SE2 Image of the Layers Structure

Figure 3.4: BSD Image of the Layers Structure

We used x-ray analysis to determine the element composition of these layers. The x-ray spectrum of the cladding layer is shown in Fig 3.5. This layer contains carbon and silicon. Based on our understanding, this layer should be Silicon Carbide crystal. The lattice constant of Silicon Carbide is nearly the same as that of Gallium Nitride which is commonly used for white light LEDs. Also, the absorption coefficient is pretty high in UV and drops rapidly near the visible range. It is transparent for visible light. So, Silicon Carbide is commonly used for GaN LEDs substrate. The layer right under the cladding layer is GaN. It is determined by the spectrum shown in Fig 3.6. This layer should be the light emitting layer in this LED. However, we didn't observe p-n junction within this layer. Based on our understanding, the p-type layer and n-type layer should vary a little. The main composition should be GaN. But they are slightly doped with different atoms. For n-type layer, group II element should be doped. For p-type layer, group VI element should be doped. The dopant concentration usually is too small for x-ray to analyze. So, we couldn't distinguish these layers. Under GaN layer, it is the electrode layer. It consists of gold and platinum. As is shown in Fig 3.3, this electrode layer is separated by a thin insulative layer. It contacts with the bottom layer only at some specific location. We think this is designed to concentrate the electron flow and increase the lighting efficiency. The bottom layer is just copper pad on the circuit board. We used the element mapping function to obtain the distribution map of each element. They are shown in Fig 3.7.

Figure 3.5: X-ray Spectrum of the Cladding Layer

Figure 3.6: X-ray Spectrum of the Semiconductor Layer

Figure 3.7: X-rays Analysis Maps of the Layers Structure. The Maps Reprensent the Distribution of Au, Cu, Ga, Pt and Si respectively.

Beside the crystal layers, we also find the fluorescent residue covering on the cladding crystal. By analyzing its elemental composition, we find the particles shown in Fig 3.8 are compound of Aluminum and Lutetium. It is commonly used in white light LED. It can transform the harsh white light into warm day light color. The x-ray spectrum of the particles is shown in Fig 3.9.

Figure 3.8: SE2 Image of Flourescent Residue

Figure 3.9: X-rays Spectrum of the Flourescent Particles

4. Colorization:

False color image of the overall LED chip and overall LED cross section are shown in Fig 4.1 and Fig 4.2:

Figure 4.1: Colorized LED chip in SE2 mode. The yellow area represent the flourescent residue. The violet parts represent the LED crystal.

Figure 4.2: Colorized Cross Section Image. The yellow area represent the flourescent residue. The violet parts represent the LED crystal. The brown regions represent the copper pads.

5. Conclusions

Based on our analysis, this component is a GaN LED. The light emitting layer is fabricated on SiC substrate. Gold and Platinum are used as its electrode. The entire p-n junction should be fabricated based on GaN and doped with different elements. The amount of the dopants is so small to be detected by SEM. Also, those layers are usually deposited with thickness in nanometer scale. That might be the reason why we cannot observe the p-n junction distribution and detect the dopant elements

Acknowledgments

Sincerely, I would like to thank Brian McIntyre for the teaching of this course and the help for this entire project. I also would like to thank TA Ralph Wiegandt for the guidance of all laboratory exercises.

Please enter any comments, criticisms, questions, etc. below.

Your name:

Email address:

References

[1] H. Kong, J. Ibbetson and J. Edmond, "Status of GaN/SiC based LEDs and their application in solid state lighting," Physica Status Solidi (c), vol. 11, (3-4), pp. 621-623, 2014.

[2] D. A. Neamen, Semiconductor Physics and Devices: Basic Principles. (4th ed.) New York, NY: McGraw-Hill, 2012.

[3] Anonymous "X-Ray Spectrometry in Electron Beam Instruments," Analytica Chimica Acta, vol. 312, (3), pp. 353, 1995.