Image Download

An image sensor converts light waves into digital signals to form an image. It’s an essential technology for smartphones, in which cameras are considered crucial, and is also expected to play a key role for future technologies such as self-driving cars and robots. Image sensors are regarded as core categories in the semiconductor industry as it closely connects with other related industries, and can thus generate further value.

Currently, smartphones account for 70% of the total demand of image sensors, which is the largest. However, it is predicted that demands for image sensors will increase significantly in future industries such as self-driving cars.

As the importance of image sensors grow, global tech companies are taking a keen interest. Industry players are predicting that the market dominance of current leaders in the image sensor market could decrease considerably in the future. As a result, current leaders, such as Sony, are actively investing into research and development for the next generation of image sensors.

3D Image Sensors: The Form Factor to Lead the Image Sensor Market

Lately, 3D image sensors have been receiving much attention as the new form factor expected to lead the image sensor market. While existing image sensors can only realize 2D images, 3D image sensors can more precisely recognize objects and actions by measuring the distance (depth) to objects, visualizing them as 3D images.

Image Download

According to last year’s Yole Developpement report¹, the 3D image sensor market is expected to achieve an average annual growth rate of 20% from $5 billion in 2019 to $15 billion by 2025. During the same period, the smartphone segment is expected to grow at 26.2% and account for more than half of the 3D image sensor market share by 2025. The automotive segment comes second, as it is expected to record the highest annual growth rate of 27% during this time.

Image Download

3D image sensors have been used in smartphones since the early 2010s, in features such as AutoFocus² and Proximity Sensing³. In addition, after Apple heavily promoted the Face ID feature—a 3D facial recognition that was first introduced on the iPhone X in 2017—it garnered a lot of attention as a new form factor expected to lead smartphone innovations.
Now, more major smartphone manufacturers, such as Samsung, LG, Huawei, OPPO, and Vivo, are engaged in a tech race to come up with the best 3D image sensors.

How Do 3D Image Sensors Work?

The way 3D image sensors capture images is largely classified into three technologies: Stereo Vision⁴, Structured Light⁵, and Time of Flight (ToF). ToF is further divided into two techniques, including Indirect ToF (I-ToF) which measures phase differences and Direct ToF (D-ToF) which measures time differences.

Image Download

The I-ToF is a method that measures the distance of an object by using lasers modulated at a specific frequency to measure the phase difference using the reflected and bounced signal. Although it is comparatively easier to implement with existing Photodiode (PD) components, it is difficult to measure the distance from objects that are more than a few meters apart because of the low efficiency of photodetectors.
The D-ToF method detects the distance from objects by sending out short pulses of light and then measuring the time it takes for the emitted light to come back. It can measure objects that are ten or a hundred meters away, but a Single-Photon Avalanche Diode (SPAD) would be required.

D-ToF: The Key to Next-Gen 3D Image Sensors

Structured light was used for 3D image sensors in the front camera of Apple’s early iPhone X models, but since then, micro image sensors using I-ToF were developed and are now more widely adopted in smartphones. It is worth noting that Sony, a key player in the image sensor market, secured a high market share with I-ToF image sensors after acquiring Softkinetic Systems S.A. in 2015 to strengthen its image sensor-related offerings.

That said, the importance of D-ToF image sensors will grow even more. Although the number of 3D image sensors in smartphones is expected to increase significantly, they will mostly be installed on the backside of the phone as this would allow for more flexible application than the front. Since 3D image sensors installed in the back should be able to measure distances of more than 5 to 10 meters, it is important for SPAD based D-ToF technology to be researched competitively.

In fact, according to research⁶ by Markets and Markets regarding the ToF image sensor market in 2020, it is estimated that while I-ToF sensors will experience an average annual growth of 11% by 2025, D-ToF sensors will achieve 37.3% average annual growth—more than triple that of I-ToF sensors.

Last year, Apple was the first smartphone provider to add D-ToF sensors in the back—which came equipped on the iPad Pro and iPhone 12 Pro. Apple used SPAD elements and processing technologies from Sony to develop the sensors and referred to them as a LiDAR (Light Detection and Ranging) Scanner to differentiate the technology from existing sensors.

Apple’s LiDAR Scanner can measure up to 5 meters in distance and boasts better performance than I-ToF sensors. When iPhone 12 was launched, Apple emphasized various AR based apps and features by making the most of this technological advantage.

Yole Developpement predicted last year that the 3D image sensor market will grow significantly in 2021 as Apple launched LiDAR Scanner-based smartphones last year. It also predicted that LiDAR sensors for self-driving cars will lead the 3D image sensor market’s growth from 2024⁷.

Image Download

In fact, LiDAR sensors for self-driving cars that deliver precise 3D images with excellent resolution⁸ are in the spotlight as key elements of automated driving. However, current LiDAR sensors are considered inappropriate for mass production as they mostly adopt the mechanical scanning method with motors, making these sensors big and expensive. Thus, D-ToF LiDAR sensors are increasing in demand as they can measure mid to long distances while also being more cost-efficient and compact in size.

Image Download

Moreover, D-ToF sensors are expected to play an integral role in the progress of next-generation mobility industries such as robots and drones. A good example is Amazon’s logistics robots and drone delivery service. D-ToF sensors are also considered essential in the factory automation field.

SPAD, An Essential of D-ToF Sensors

Single-Photon Avalanche Diode (SPAD) is a photodetector for next-generation semiconductors that is extremely efficient enough to detect even single photons⁹ due to its very high gain.

Image Download

Image Download

When a higher voltage than the Breakdown Voltage¹⁰ is applied to SPAD, the impact ionization phenomenon occurs in which a huge electric field accelerates carriers which makes them clash with atoms, increasing the number of free carriers released from the atoms. This phenomenon is called Avalanche Multiplication and results in a large number of free carriers created by photons lit up by image sensors. This means that it can amplify photons and recognize them as a much larger number of photons even though only a small number of them were actually captured, due to dark surroundings or if light was emitted from a far distance.

Also, as SPAD arrays emit digital pulses when photons enter, it is easier to track flight time. Furthermore, it can also capture precise time differences, so it is able to determine accurate depth resolutions¹¹ even down to the range of mm and cm.

SPAD-based D-ToF sensors have been successfully verified and researched at Swiss Federal Institute of Technology Lausanne in Switzerland and University of Edinburgh in the UK. STMicroelectronics also adapted the technology to launch proximity sensors which have been applied to a wide range of smartphones.

Just as research and development was carried out using the Backside Illumination (BSI) method to improve performance in image sensors, 3D-stacked BSI SPAD array research has been carried out mainly in the field of SPAD-based D-ToF sensors.12) As mentioned earlier, Apple and Sony have recently collaborated to successfully develop better-performing 3D-stacked BSI SPAD-based D-ToF sensors to mount them on the rear camera of Apple smartphones.

In Korea, researchers from Korea Institute of Science and Technology (KIST) are currently leading the research of essential SPAD for D-ToF sensors to obtain the original technologies and drive the development of the next-generation of 3D stacked BSI D-ToF sensors.

Sony was able to dominate the I-ToF sensor field through its acquisition of Softkinetic Systems S.A., which had the original technologies. Similarly, suppose a key semiconductor company with state-of-the-art technologies and infrastructure, such as SK hynix can actively collaborate with local researchers—we can expect to lead the growing global market of D-ToF and LiDAR sensors.

¹‘CMOS Camera Module Industry for Consumer & Automotive 2020’ by Yole Developpement (2020)
²AutoFocus: A camera technology that automatically focuses on the object
³Proximity Sensing: A feature that detects light waves and recognizes the presence of objects or people. It is used in smartphones to detect user’s face that touches the screen to turn off the screen as to prevent unnecessary touch, and is also used for self-driving and robot technologies to measure the distance between cars or robots and inform the location or access of the objects.
⁴Stereo Vision: A method measuring the distance to objects with two image sensors and implement 3D images similar to that of human perception of perspective with both eyes. It has a fundamental disadvantage that it’s difficult to miniaturize.
⁵Structured Light: A method gaining 3D images by shooting certain pattern of light to objects and analyzing the distortion of light pattern returned to image sensor after being reflected from stereoscopic objects, through software. It is difficult to operate accurately under strong light outside, and the software can be costly.
⁶‘Time-of-Flight (ToF) Sensor Market – Global Forecast to 2025’ by Markets and Markets(2020)
⁷‘CMOS Camera Module Industry for Consumer & Automotive 2020’ by Yole Developpement (2020)
⁸resolution: An ability to distinguish two objects apart from each other
⁹Single Photon: A single quantized light particle contrasted with classical electromagnetic wave
¹⁰Breakdown Voltage: When the level of reverse voltage is beyond certain limit, it results in Avalanche making huge currents. The voltage in this called Breakdown Voltage.
¹¹depth resolution: An ability to distinguish two objects very slightly apart from each other
¹²Reference:
https://doi.org/10.1109/IEDM.2016.7838372
https://doi.org/10.1109/IEDM.2017.8268405
https://doi.org/10.1109/ISSCC.2018.8310201
https://doi.org/10.1109/ISSCC.2019.8662355
https://doi.org/10.1109/JSSC.2019.2938412
https://doi.org/10.1109/IEDM13553.2020.9371944
https://doi.org/10.1109/ISSCC42613.2021.9365961
https://doi.org/10.1109/ISSCC42613.2021.9366010

The post [Future Semiconductor Technologies] Next-Generation 3D Image Sensor Component- SPAD (Single-Photon Avalanche Diode) first appeared on SK hynix Newsroom.

According to this trend, various electronic products including mobile devices are evolving in the direction of smaller (“ultra-small”), less power consumption (“ultra-low-power”), and “faster (ultra-high-speed”) products. Furthermore, combined with artificial intelligence (AI) or IoT, they are developing into intelligent devices.

To accumulate an enormous amount of information in a small area, semiconductor devices for storing or processing information are also becoming more miniaturized in a several nanometer level with higher density. As a result, the manufacturing process has become more complicated, and power consumption due to heat generation is also rapidly increasing.

Accordingly, there is a growing voice about the necessity of improving the existing semiconductor operation method which uses electron charge³ in order to reduce power consumption drastically. To achieve this, many technological developments are currently being made. Among the next-generation low-power semiconductor technologies developed until today, a technology called “spintronics”⁴ that uses electron spins⁵ is evaluated to realize ultra-low power.

This technology simultaneously uses two physical quantities⁶ of electrons, which are charge and spin, and is implemented in a way that the resistance of a device changes according to the direction of the spin. Therefore, the processing speed is fast and heat is not generated in the device when a current flows, which makes its power consumption very low. For this reason, the magnetic random access memory (MRAM) technology using the spin is expected to lead the growth of next-generation ultra-low-power information storage devices in combination with semiconductor technology in the future.

Principles and Limitations of the MRAM Technology at Early Stage

The core part of an MRAM is a magnetic tunnel junction (MTJ) located at the intersection of a digit line⁷ and a bit line⁸ of the memory circuit. An MTJ basically has a three-layer structure consisting of a fixed (magnetic) layer, a dielectric tunnel barrier, and a free (magnetic) layer. Here, a magnetic layer that can easily change the direction of magnetization⁹ is called a free layer, while a magnetic layer that is pinned so that the magnetization direction cannot be easily changed is called a fixed layer.

One of the key features of an MTJ is that the resistance of the device varies depending on the relative magnetization directions of the two magnetic layers. When the magnetization directions of the two magnetic layers constituting the MTJ are aligned in a parallel direction, the device has a low electrical resistance value. On the contrary, when they are aligned in an anti-parallel direction, the device has a high resistance value.

If the high resistance value when not parallel is defined as RAP, low resistance value when parallel as RP and the difference between the two values as ΔR (ΔR=RAP–RP), the magneto-resistance (MR ratio) can be generally defined as ΔR/RP. The MR ratio is determined depending on which magnetic material and dielectric tunnel barrier are used, and is usually divided by tens to hundreds of percent.

Resistance changes in MTJ according to changes in the external magnetic field in an MRAM

Image Download

An MRAM stores information in “1” and “0” in the binary system by using the state of different RAP and RP when the magnetic field is 0. In the graph above, the x-axis represents an externally applied magnetic field and the y-axis represents the level of electrical resistance. When x value is 0 (x=0), RP and RAP show two different electrical resistances of 650 ohm¹⁰ and 1,400 ohm, respectively. By using this, RP is recognized as 0, and RAP is recognized as 1 to store information. Conversely, to read the information recorded in the MTJ, the resistance state of the MTJ should be measured.

To store information at this point, the magnetization direction of the free layer in the MTJ should be changed. In MRAMs at an early stage, a current was flowed through the digit line to induce a magnetic field to change the magnetization direction of the free layer. As the degree of integration gradually increased, however, an interference phenomenon occurred in which information of adjacent cells was read, and energy required for changing the direction increased. For this reason, this method is no longer used with a limit of 64 kilobytes (KB).

Principles, Advantages, and Disadvantages of Spin-Transfer Torque MRAM (STT-MRAM) Technology

What has overcome such fatal disadvantage is the spin-transfer torque magnetic random access memory (STT-MRAM) technology. Unlike the conventional MRAM method that changes the magnetization direction of the free layer through generating a magnetic field by making a current flow in the digit line, STT-MRAMs use a method of making a current directly flow in the MTJ to change the magnetization direction of the free layer. What makes this possible is the spin-transfer torque (STT).

When making a current flow in the MTJ from the fixed layer to the free layer, the spin direction of the conduction electrons¹¹ is aligned to the spin direction of the fixed layer (arrow direction in the figure above) due to the influence of the magnetic exchange coupling energy¹² in the process where the conduction electrons pass through the fixed layer. At this time, the current in which the spins are aligned in one direction is called a spin-polarized current.

When this spin-polarized current enters the free layer, the interaction between the spin electrons inside the free layer aligned with the magnetization direction (diagonal arrow in the figure above) of the free layer and the conduction electrons flowing in a state aligned in the magnetization direction (vertical arrow in the figure above) of the fixed layer changes the spin direction of the conduction electrons just before and after passing through the free layer.

The time taken when the spin direction changes like this is called “torque”. When spin-polarized conduction electrons are introduced in the free layer and torque of a certain level or above is applied, the magnetization direction of the free layer can be changed to the same direction as that of the fixed layer, through which the information is stored. When the magnetization directions of the two magnetic layers are aligned in a direction parallel to each other, the device can have a low resistance value.

To delete the stored information, the magnetization directions should be changed from parallel to antiparallel. To achieve this, electrons should flow from the free layer to the fixed layer, contrary to the previous process. As seen from the right of the figure above, when conduction electrons move from the free layer to the fixed layer, electrons which have spins in the same magnetization direction as the fixed layer pass through the fixed layer, while electrons which do not are reflected back to the free layer. These electrons apply a torque in a direction opposite to the magnetization of the free layer and reverse the magnetization direction of the free layer, which makes it possible for the magnetization directions of the two layers to be aligned in an antiparallel state. At this time, the device has a high resistance value.

The resistance changes of the MTJ when applying a current
(application: the act of applying power voltage between terminals of an electric circuit) in an STT-MRAM

Image Download

The figure above shows the magneto-resistance values measured while changing the magnetization direction of the free layer by making a current flow through the MTJ of an STT-MRAM. RP is 650 ohm and RAP is 1,400 ohm, which are the same as the measurement values in the conventional MRAM method that generates a magnetic field by making a current flow through the digit line.

To read the information stored in an STT-MRAM, what should be done is measuring the resistance of the MTJ, just like in MRAMs. The only difference between STT-MRAMs and MRAMs is the method of applying the current to record information. Through this, however, STT-MRAMs completely overcame the limitation of density and the interference phenomenon of adjacent cells, which were pointed out as disadvantages of MRAMs. For this reason, STT-MRAMs are recognized as a key technology to realize the next-generation non-volatile memory.

Principles, Advantages, and Disadvantage of Spin-Orbit Torque MRAM (SOT-MRAM) Technology

When a magnetic field is applied to a conductor through which a current flows, a voltage is induced in a vertical direction to the direction of the current and the magnetic field, and this effect is called the “Hall effect”. Also, the “spin Hall effect” refers to a phenomenon where the Hall effect appears in a material with a strong spin-orbit interaction (SOI) even without applying an external magnetic field.

Image Download

When a current is applied in the vertical direction to materials with a high SOI level such as tantalum (Ta), tungsten (W), and platinum (Pt), conduction electrons are separated into spin-up electrons with the upward spin direction (left in the figure above) and spin-down electrons with the downward spin direction (right in the figure above). At this time, a spin current in the horizontal direction is generated from the spin-up electrons to the spin-down electrons. At this stage, the spin current can be used to switch the magnetization direction of the free layer of the MTJ, and the memory which adopts this method is called spin-orbit torque MRAM (SOT-MRAM).

Comparison of circuit structure of MRAM, STT-MRAM, and SOT-MRAM

Image Download

Unlike STT-MRAMs, where a current is directly applied vertically to the MTJ, SOT-MRAMs utilize a method in which the spin current polarized in the vertical direction by the spin Hall effect changes the magnetization direction of the free layer when a current is injected in a horizontal direction into a material layer with a high SOI level at the bottom of the MTJ cell.

In particular, when comparing to STT-MRAMs, SOT-MRAMs can generate more conduction electrons with the same spin and inject them into magnetic layers. This suggests a stronger torque and an easier switch of the magnetization direction of the free layer, which makes it possible to achieve a faster processing speed and lower power consumption.

A table comparing energy consumption for programming 1 bit in each type of memory semiconductor

Image Download

Assuming that a 90 nm process is used, a large amount of 120 pJ¹³ is used in a typical MRAM, while only 0.4 pJ of energy is consumed in an STT-MRAM. When using a process with a line width narrower than 90 nm is used, the gap between the two technologies is increased. This low-power characteristic of STT-MRAMs is very significant in terms of energy efficiency. Also, the power consumption of SOT-MRAMs is about a tenth of that of STT-MRAMs, which is expected to further reduce power consumption.

In the modern von Neumann computing technology architecture, there is a gap between the operating speed of the processor (0.1 ns¹⁴) and the operating speed of the main memory (10-100 ns) and the storage memory (0.1-10 ms¹⁵). To ensure that the next-generation memory technology is able to cover a wide area from cache¹⁶ to main memory, an operation speed of 1-10 ns is required. Among the currently available technologies for the next-generation memory semiconductor, STT-MRAMs (10 ns) and SOT-MRAMs (1-10 ns) are the only technologies that satisfy this requirement. Among these two, SOT-MRAMs have the lowest power consumption level, drawing attention as the optimal technology for the next-generation memory.

¹ Zetta is a decimal unit denoting a factor of 1021 and a byte is a unit denoting the magnitude of data. The units increase in the orders of kilobyte (KB), megabyte (MB), gigabyte (GB), terabyte (TB), petabyte (PB), exabyte (EB), zettabyte (ZB), etc. Each time the unit is changed, the size of data increases by 1,000 times.
² This refers to a computing environment in which computers and the Internet can be used anytime, anywhere. Ubiquitous is a term derived from the Latin word “ubique” which means “everywhere”.
³ A property or physical quantity (a quantity representing the property or state of a substance) of a substance that causes an electric phenomenon; each of all particles is positive (+), negative (-), or neutral.
⁴ Spin is one of the physical quantities representing the basic properties of a particle. As a particle’s inherent angular momentum (intensity or momentum of a rotary motion), it has a momentum level and direction. In addition to the motion of the electrons orbiting the nucleus, it rotates around an axis that passes its center of gravity, and this rotational motion is called a “spin”.
⁵ An electronic engineering technology in which data is stored by dividing the phenomenon of electrons rotating in different directions into two digital signals of 0 and 1
⁶ A quantity representing the property or state of a substance, described by length, weight, viscosity, mass, temperature, capacity, etc.
⁷ An electrical circuit for entering information
⁸ An electrical circuit for reading stored information
⁹ To induce magnetic properties into an object with no magnetic properties; the magnetization direction of magnetic materials is determined with the direction of N pole to S pole.
¹⁰ The unit of electrical resistance; 1 ohm is defined as the resistance of a mercury column of 1 mm2 cross-section area and 106 cm long.
¹¹ Electrons that are not bound to any particular atom and can move freely
¹² The energy that describes the magnetic interaction between adjacent spins; when the spin direction is the same as the direction of adjacent spins, the energy is low.
¹³ J is a unit of energy and 1 J is equal to the energy transferred to an object when a force of 1 newton (N) acts on that object in the direction of the force’s motion through a distance of 1 meter. The pJ, abbreviation of picojoule, is a unit that means 1 trillionth of 1 J.
¹⁴ nanosecond; a unit representing one billionth of a second
¹⁵ microsecond; a unit representing one millionth of a second
¹⁶ A high-speed memory device in the form of a buffer that is installed between the main memory and the central processing unit (CPU) to temporarily store commands or programs read into the main memory

The post [Future Semiconductor Technology] The Present of the Next-Generation Ultra-Low-Power MRAM Technology first appeared on SK hynix Newsroom.

The industry and academia concentrated on the development of blue LEDs because white LEDs made of red, green and blue LEDs are 4 to 10 times brighter than existing lamps. Along with this excellent energy efficiency, the maximum usage period is 100,000 hours, which is 10 to 100 times longer than existing lamps.

Three Japanese scientists, including tenured Professor Akasaki Isamu, completed the research into blue LEDs. In 1992, they succeeded in commercializing blue LEDs for the first time by developing a technique to make high quality gallium nitride thin films, and a p-type doping technology that could utilize this technique. They went on to win the Nobel Prize in Physics in 2014 for this specific achievement, and since then, Japan has maintained their competitive advantage in the blue LED market.

Meanwhile, researchers from Korea Institute of Science and Technology (KIST) recently developed a new device technology to replace gallium nitride used in blue LED semiconductors. The researchers succeeded in developing a copper halide compound semiconductor technology by using copper iodide (CuI) semiconductor composed of copper (Cu) and iodine (I), securing a device technology that can emit blue lights with high efficiency.

Why does “Gallium Nitride”, a Next-Generation Semiconductor Material, Need a Replacement?

Gallium Nitride had been attracting attention as an ultra-low-power next-generation semiconductor material. Power consumption can be greatly reduced if electronic devices are based on gallium nitride instead of silicon, and can emit light not only in the blue region, but also in the ultraviolet region which has a shorter wavelength.

Figure 1. Structure of Gallium Nitride and main features of Gallium Nitride semiconductors

Image Download

Currently, gallium nitride is widely used in real life as a core material for smartphones, displays, electronics and high-frequency devices. In addition, since signal switching speed is fast and energy loss rate is low during signal switching, its range of utilization is rapidly expanding into high-frequency high-output communication systems, automotive power systems, and semiconductors for extreme environments.

As the price of gallium nitride wafers decrease and more advanced technologies for molding and growing gallium nitride thin films on silicon substrates are developed, the market size is expected to increase further. Previously, the interatomic distance between silicon and gallium nitride was so long that a gallium nitride thin film had to be grown on a sapphire substrate; however, technology has been recently developed to shorten the interatomic distance between silicon and gallium nitride, and now a gallium nitride thin film can be grown directly on silicon substrates without being damaged, allowing a lower substrate cost.

Figure 2. U.S. Gallium Nitride semiconductor devices market, by product¹

Image Download

Nevertheless, gallium nitride also has some considerable disadvantages. Gallium nitride is expensive, and difficult to build highly integrated circuits. Quantum efficiency² is also low due to the strong intensity of the internal electrostatic field between gallium (Ga) and nitrogen (N) atoms, and low exciton-binding energy between electrons and holes³. In addition, due to GaN’s large interatomic distance from either sapphire (Al₂O₃) or silicon carbide (SiC) substrates, numerous defects occur during the production of thin films, which impairs the lifespan and features of devices.

It is also difficult for GaN to conduct doping as a p-type semiconductor, which is required for blue LEDs. Doping modulates resistance properties by adding impurities into the semiconductor crystals. A p-type semiconductor has more remaining holes than electrons after doping, while an n-type semiconductor has more electrons than holes. Since gallium nitride shows the n-type semiconductor properties with excess electrons, doping it as a p-type semiconductor is difficult.

Development of a Source Technology Using Copper Iodide

Image Download

KIST researchers focused on copper iodide to overcome the weaknesses of gallium nitride and develop a new semiconductor technology that can replace it. The established theory in academia had been that substances of Group 1-7 elements of the periodic table, which includes iodine, are difficult to be used as semiconductors, since their strong electrical interactions led to high degree of interatomic bonding strength. This is because the current can flow only when the bonding strength between atoms is low enough.

However, the development of new technology opened a new horizon for semiconductor material technology research. Unlike gallium nitride, copper iodide has the weak intensity of the internal electrostatic field and high bond energy, resulting in high efficiency in photoelectric transformation – the conversion of light information into electrical signals.

Figure 3. Cross-section of a copper iodide thin film on a silicon substrate,
observed with a transmission electron microscope (TEM)

Image Download

Furthermore, since the crystal structure of copper iodide is the same as that of silicon and the interatomic distance is also similar, thin films can be grown on inexpensive silicon substrates with fewer defects. The temperature for growing copper iodide thin films is also similar to the temperature for silicon device processing (below 300 °C), so it can be applied to the silicon semiconductor process without degradation.

In particular, copper iodide thin films have p-type semiconductor properties. This is a great advantage compared to the n-type semiconductor gallium nitride, which can’t be easily doped as p-type. By focusing on this point, researchers have developed a hybrid LED where n-type aluminum gallium nitride (AlGaN) and p-type copper iodide are combined together, succeeding in emitting blue light.

Since a high-performance, high-reliability blue light-emitting device can be made with copper iodide thanks to its high quantum efficiency, it is considered that CuI can be utilized to develop high-efficiency photonic devices by replacing nitride semiconductors in the future. In addition, as it can be directly grown on a large sized silicon substrate without the need for high-temperature thin film growing process and equipment, it can be used for various applications like micro-displays, which were difficult to be realized with gallium nitride.

Researchers are currently working on developing an LED consisting of copper iodide and copper chloride to enhance efficiency such as increasing the wavelength of emission. For commercialization, some challenges like improving the quality of cooper iodide thin films, optimizing the process for mass production, and the development of related equipment must be solved in advance.

With the development of this technology, it is expected that the commercial production of CuI-based semiconductors will be possible within a few years as materials for both blue and ultraviolet light sources. In particular, it is expected to play an important role as a new material for the light-emitting semiconductor that replaces gallium nitride.

¹‘GaN semiconductor devices market size, share & trends analysis reports’ by Grand view research(2018)
²Efficiency levels when converting electricity into light
³A space for the electrons in semiconductors

※ This article is based on the subjective view of the contributor, and may differ from the official stance of SK hynix. Any unauthorized use of the contributor’s article may have legal responsibility.

The post The Newly Developed Blue Light Semiconductor Device Technology Set to Replace Gallium Nitride first appeared on SK hynix Newsroom.