The State Council has set a goal of meeting 70 percent of China's demand for integrated circuit chips by 2025. China Daily writer Zhang Zhouxiang explains how ICs are made from sand and the bottlenecks China faces in making them
Workers labor at an integrated circuit chips assembly line in Suining city, Southwest China's Sichuan province, in August. (LIU CHANGSONG / FOR CHINA DAILY)
If one were to list the greatest inventions of the past century, integrated circuit chips, commonly called ICs or silicon chips, would surely find mention. After all, it is the "brain" in almost every electronic device one can think of. Be it the computer, the smartphone, an Apple watch, or a smart TV set, they all have it.
Even electronic gadgets that do not look very smart, otherwise, cannot do without these chips. It is required in the modern-day refrigerator, for example, to smartly adjust cooling, an advancement over the simple on-off thermostat method. The microwave oven has a chip controlling its heating and timing functions, after taking inputs on the control panel and various sensors. Why, even the electronic lock needs a chip to identify which fingerprints should be allowed access and which ones denied.
Chips are rather small. If you break a smartphone and disassemble all parts, you will find that it is generally less than 1 square centimeter in size. However, inside that 1 sq cm space there could be as many as 10 billion microtriodes and PN junctions, with current, or an instruction, passing from one junction to another.
"No wonder some technological experts liken a chip to a city, even calling it the most complicated of cities in the smallest of spaces," said Dai Guoqiang, former director of the Institute of Scientific and Technological Information of China, explaining its manufacturing process. Like all cities, even the chip-city is made up of sand, literally, but not the sand that one finds at the nearest beach. Chip makers need silica sand, also called silicon dioxide, which is got from quarrying. And to extract silicon from it, the oxygen is removed by mixing it with carbon and heating in an electric arc furnace to temperatures beyond 2,000 degrees C.
This metallurgical grade silicon is 99 percent pure, which is not enough to meet the requirements of the chip maker. According to an article in www.techradar.com, this silicon is refined further, by grinding it into a fine powder, adding hydrogen chloride and heating it in a fluidised bed reactor at 300 degrees C. This creates a liquid silicon compound called trichlorosilane, and chlorides of unwanted elements such as iron, aluminium, boron and phosphorus, which are removed by fractional distillation. The trichlorosilane is vaporised in hydrogen at 1,000 degrees C. An electrically heated, ultra-pure silicon rod collects what is electronic-grade silicon that is 99.999999 percent pure.
That still won't do, because of its polycrystalline structure, meaning it is composed of lots of small silicon crystals, and the joints between those crystals can suffer from defects known as grain boundaries. Those boundaries can interfere with electronic signals. So, the structure of the silicon is changed, through what is called the Czochralski Process. It involves melting the silicon crystal in a quartz crucible at just over the melting point of 1,414 degrees C. A tiny silicon crystal is then dipped into the molten silicon and drawn out while rotating it in a direction opposite to the crucible's rotation. This attracts silicon from the crucible, creating what's known as a boule, which is a rod made from a single silicon crystal. A typical boule will be around 300 mm across.
The circular silicon rod is now cut using a device that works like an egg slicer, into multiple slices simultaneously to create wafers 0.775 mm thick. The sharp edges of the wafers are then smoothened to prevent them from chipping.
The wafers' surfaces are then polished until the wafers are flat. After that, the wafer is etched with a mixture of nitric, hydrofluoric and acetic acids to create an even smoother surface.
A staff member checks out chips at a plant in Hangzhou city, East China's Zhejiang province, in August. (LONG WEI / FOR CHINA DAILY
Time to build the chip city
The silicon wafers are then put into the working platform of a lithography machine, where light passes through a series of lenses before falling on the wafer to carve circuits on it. The circuits would form microtriodes, the most basic units of the chip. Each is a computing unit and can contribute to the whole computing capability. The smaller each circuit is in size, the more microtriodes a chip can fit in, meaning higher speed.
The smaller size ensures two additional advantages, as Chen Jing, deputy head of the Fengyun Institute, a think tank on technology based in Beijing, explained. First, the electric current travels a shorter distance, which means higher efficiency and computing speed. Second, it will consume less energy, which is essential for computers and smartphones, which need to cool off.
However, the accuracy is much easier said than done because there are too many procedures involved, each of them fraught with the risk of error.
The light being focused on the silicon wafers will first bounce off three light adjusters whose function is to adjust its angles and focus. The light then passes through an energy controller, shape setup unit, and a switch before it reaches the exposure platform, where the silicon wafer is fixed.
In order to ensure high accuracy, the lens system must be extremely accurate, being fixed via electromagnetic suspension. The silicon wafer must be accurately placed on the exposure platform. A minor error in the lens system might get amplified by 10 or even 100 times in the final product.
That's why the accuracy of chips is a major yardstick for measuring a company or nation's capability in the chip industry. By far, TSMC based in Taiwan, with machines from Advanced Semiconductor Material Lithography based in the Netherlands, can carve circuits that are 5 nanometers wide, or one 12,000th the diameter of a human hair, on chips. That means 100 million microtriodes can be placed together in every square millimeter of the 5-nanometer chip.
Sadly, by far even SMIC, the most advanced chip enterprise in China, can only produce chips with circuits that are 14 nanometers wide. That means only 12.7 million microtriodes can be placed in a square millimeter of the chip.
In terms of engineering, it means while a processor containing ASML chips can rally 100 million microtriodes to compute together, a processor containing domestic chips can rally only 12.7 million microtriodes, making it lag far behind in speed.
Even that speed has been achieved by SMIC using machines and technology from the US. By far, there is no chips assembly line that is totally domestically produced using domestic technology.
Clearly, there is a gap, which is making Chinese high-tech enterprises rely heavily on Western chip suppliers. Even in 2020, China was meeting 80 percent of its chip demand through imports.
The demand is particularly acute in the smartphone industry. A smartphone with a 28-nm chip will run much slower than one with a 5-nm chip. Also, it will produce more heat and will be larger in size.
Enter US, the spoilsport
Following Sino-US trade frictions in 2018, then US president Donald Trump threatened to issue administrative orders forbidding US enterprises from exporting smartphone chips to China. The order came into effect on Sept 15, 2020, a little more than 100 days before Trump left the White House.
Not just US companies such as Qualcomm, but even companies that do businesses with it, such as Samsung and TSMC, have cut chip supply to Huawei, one of the largest high-tech companies in China. The company was able to buy 100 million chips in August 2020, but it must now address the problem of chip shortage.
The Joe Biden administration has been trying to put pressure on ASML to not sell to China the extreme ultraviolet lithography system, a $150-million machine that is essential to making advanced chips in everything from cutting-edge smartphones and 5G cellular equipment to computers used for artificial intelligence.
While ASML chief executive Peter Wennink said that Washington's anti-China tech blockade is a bad idea that will backfire, China has realized that self-sufficiency in core technologies such as semiconductors will be critical going forward.
In August 2020, the State Council, China's Cabinet, vowed to further support the domestic chip technology in a guiding document. It set a goal of meeting 70 percent of the domestic demands with domestic production by 2025.
That will be a tough road ahead with very limited time to spend.
Contact the writer at zhangzhouxiang@chinadaily.com.cn