The 120nm Miracle
Lisa Su hummed softly as she peered through the microscope at the ID Electronics semiconductor plant in Daejeon. It had been a long time since she had concentrated so completely on a single task. Fabricating semiconductors was her favorite work, and adjusting countless parameters to produce dozens of wafers in a single day was something she had rarely experienced even in the past. What made it even more remarkable was that this was not some makeshift laboratory experiment; she was using a full-scale production line as if it were her own workshop.
The semiconductor research team members who had joined the task force grew anxious every time a wafer emerged from the testing equipment, but Lisa Su remained calm. They treated the numbers provided by the chairman as the most important challenge in the world, so any significant deviation sent them into a state of near-panic. Lisa Su, however, understood that improving semiconductor processes was inherently tedious work. Even with a perfect recipe, yield could fluctuate due to minor factors. Hunting down those variables one by one gave her the satisfying feeling of being a detective in a mystery novel. Of course, if anyone else had known how much she enjoyed it, they would have called her a pervert, so she kept the thought to herself.
“Dr. Su, would you take a look at this?” someone called out while she was focused on the microscope. It was Professor Lee Jong-hyo.
“It’s a freshly processed J3 sample. Quite interesting.”
“If Professor Lee says so, I can’t help but be curious.”
Lisa Su accepted the wafer loaded in the transfer cartridge and swapped it in immediately. Professor Lee Jong-hyo had joined the ID Electronics semiconductor team about a month earlier with a strong recommendation from Chairman Yoo Jae-won. As expected, the chairman’s name carried weight; the professor had already made significant contributions to the development of the fine process. At the same time, Lisa Su felt a strong sense of kinship with him. His passion for semiconductors was something even she, who had only recently begun working closely with him, could acknowledge.
She glanced briefly at Professor Lee, who was absorbed in examining another J3 sample, then returned her attention to the microscope.
“Oh!”
The difference was immediately apparent. The circuits fell away with knife-like precision. People often assumed that because semiconductors were cutting-edge products, their circuit patterns would be perfectly sharp and rectangular. Under the microscope, however, the reality was far less clean. Every step of etching circuits onto the wafer was an analog process, much like taking photographs with an old film camera. Some areas came out sharp; others were blurry or incomplete. Only wafers where every section rendered correctly counted as good product. Any imperfection in the wiring or transistors would cause malfunctions, power leakage, or overheating.
This wafer was different. The circuit lines were cleanly defined. The transistors forming the logic circuits, as well as the L1 and L2 cache memory, were all uniformly rendered. With her accumulated experience, Lisa Su could tell at a glance that the chips from this wafer would deliver outstanding performance.
The wafer that passed through her hands moved on to the testing equipment. After receiving an all-green verdict, it proceeded to the packaging process. The chips mounted on the test board soon began operating under everyone’s expectant gaze. Performance metrics appeared on the monitor, and cheers erupted from the task force. At that moment, an urgent text message was sent to Yoo Jae-won.
The next day, Yoo Jae-won returned to the Daejeon plant. Though he had wanted to rush down immediately upon receiving the report, Tiffany had been waiting at their Seoul residence, so he had delayed his departure. She had other matters to attend to in Seoul, and he traveled to Daejeon with his entourage. Fresh snow had fallen overnight, delaying the journey, and they arrived at the plant just after lunchtime.
“So the actual device has finally been produced?”
“Yes, Mr. Chairman!”
The MAP 3 was packaged in black and roughly the size of a thumbnail. Although they had handled the packaging in-house, the quality was clean enough for retail sale. The ID Group logo was laser-etched prominently on the top surface, along with the model name containing the MAP 3 information. What pleased him most was the final line: “Made in Korea.”
“Show me its performance.”
“Of course, sir!”
At Yoo Jae-won’s request, Lisa Su guided him to the test board. The board was larger than a typical computer motherboard. Since the chip was destined for mobile phones, testing it on a phone board would have been ideal, but none had been prepared yet. They had therefore used a general-purpose test board for semiconductor evaluation. The operating system running on it, however, was the mobile Android OS. Because MAP 3 maintained full backward compatibility, it booted without any modifications to the OS.
Once the boot sequence finished, the benchmark application launched immediately. Brilliant graphics filled the screen. The test displayed floating-point operations for CPU performance and polygon and texture processing for the graphics module—visuals far more impressive than anything possible on the MAP 2 installed in the T-Touch Phone. That was only natural; MAP 2 supported only 2D, with no 3D acceleration module.
After several minutes of testing, a large score appeared on the screen.
“9,891 points!”
It was a massive success. He had already received the report via text the previous day, but seeing the result in person carried a completely different weight. The same application running on MAP 2 scored only 2,000 points—and even that was an artificial result obtained by briefly refrigerating the phone. Yet MAP 3 had produced more than four times that score on a mere test board. Although the inclusion of 3D rendering tests unavailable on MAP 2 contributed to the jump, the result far exceeded Yoo Jae-won’s expectations. He had optimistically hoped for roughly four times the MAP 2 score. On the test board, they had achieved 4.5 times. Once optimized for an actual phone board, the score would likely rise even higher.
“As expected, Japanese materials,” muttered a researcher behind him.
Everything was excellent except for one regret: the sample had been fabricated using Japanese raw materials. Every material composing the J3 sample—from the silicon wafer to the photoresist, the etching gas hydrofluoric acid, and the cleaning liquid hydrofluoric acid—had come from Japan. There had also been samples made with Korean materials, labeled with K designations, but the results were poor. Yield was the first problem. The best wafer achieved only about 40 percent yield. A single wafer could produce roughly 300 MAP 3 chips, of which only around 120 functioned. Even among the functioning chips, achieving full performance was another matter. The highest score recorded among the hundreds of K samples tested was in the high 6,000s—the very score Yoo Jae-won had conservatively anticipated. If the yield had been acceptable, they would have begun mass production with the K samples, but 40 percent fell far short of expectations.
The J3 sample, however, delivered overwhelming performance and an excellent 60 percent yield. With further optimization, it had a strong chance of reaching over 80 percent. The difference clearly stemmed from material quality. The purity of the etching gas hydrofluoric acid was 95 percent for the domestic product but 99.9 percent for the Japanese version. Though only a 4.9 percent difference, Yoo Jae-won had known such quality gaps existed and had assumed they would not pose a major issue at the current 120-nanometer process node. Ultra-high-purity hydrofluoric acid was expected to matter only at single-digit nanometer scales, yet even this relatively coarse process showed a significant disparity.
From a corporate standpoint, the origin of the materials did not matter, but the long-term implications were critical. Relations with Japan would deteriorate sharply, reaching their worst point around 2020, when a sudden export ban on raw materials would strike Korean companies. At that time, the victims had been Ilsung Electronics and Hynix; now, ID Electronics would take the direct hit. After all, ID Electronics was the company formed by merging Ilsung Electronics and Hynix. He had no desire to support Japan by using Japanese materials. That was why he had intended to base MAP 3 production on domestic materials and had designed the process accordingly. He had been fully aware that Korean materials were inferior to Japanese ones, but since this was not a 10-nanometer or 7-nanometer ultra-precise process, he had believed the difference would be manageable. Reality had proven otherwise. The most painful fact was that no domestic materials of higher quality than those already secured existed. They had gathered the best available, yet the results remained disappointing.
Yoo Jae-won fell into brief contemplation.
“Prepare for mass production based on the J3 sample for now.”
In the end, he chose the Japanese materials. The gaps in yield and performance were too large to force the use of domestic alternatives.
“Would it be acceptable to use the K samples for memory semiconductor production?”
“Yes! Compared to MAP 3, which has high circuit complexity, memory chips have a simpler structure, so domestic materials should be far more suitable.”
Lisa Su’s straightforward answer was reassuring. Japanese materials offered superior quality, but they carried a drawback: cost. Their high purity made them expensive to begin with, and as imported goods, transportation fees were substantial. Storage requirements were also stringent, pushing the total cost to more than five times that of domestic materials.
“And please continue running the test lines while pursuing further research on eliminating Japanese materials.”
This was not merely a matter of patriotism; economic viability also demanded the transition to domestic sources.
“Ah, and please gather the presidents of the semiconductor division’s partner companies as well.”
Yoo Jae-won would have preferred to handle everything internally, but ID Group had no chemical companies. At best, there was Sanyo Electric, which produced batteries. Chevron’s side only dealt with compounds such as naphtha and polyester derived from petroleum. Chemicals essential to semiconductor production, like hydrofluoric acid, had limited applications and were difficult to handle. Moreover, even if they produced such materials, securing designation as a partner company was as difficult as passing through the eye of a needle. That was precisely why a quality gap existed between domestic and Japanese materials. Yoo Jae-won intended to use Japanese materials only up to MAP 3 and switch to domestic sources thereafter. He had therefore arranged this meeting to request active development from the partner companies.
Although the idea had existed for a long time, this was the first time he had voiced it. The partner companies received the sudden summons, yet the very next day their presidents all gathered in Daejeon—an unmistakable demonstration of Yoo Jae-won’s presence. At the meeting, he announced the successful development of the mobile non-memory semiconductor and explained in detail the differences between Japanese and domestic materials. He also declared that once domestic quality reached a level comparable to Japanese materials, they would use only domestic sources without exception, and that ID Group would provide both technical support and active investment to the partner companies.
Despite their delight at the announcement, the partner company presidents could not hide their unease. It was not Yoo Jae-won’s fault. Too many large corporations had promised investments only to change their terms later or extract technology and then abandon their partners. When Yoo Jae-won signed a formal document guaranteeing that any company meeting the specified specifications would be contracted without fail, the presidents finally understood his sincerity, and thunderous applause filled the room.
Soon afterward, the contract between ID Electronics and its partner companies was reported in the press under the name “Semiconductor Korea 21.” It was an homage to Brain Korea 21, the ambitious plan to cultivate outstanding research talent under the People’s Government. The goal was to achieve 60 percent localization of semiconductor materials by 2008 and 100 percent by 2012.
Before long, January 2001 arrived. Secretary-General Kim Dae-seok, returning from year-end vacation, presented a newspaper during the morning media briefing with a remark that something interesting had appeared.
- ID Electronics Successfully Develops 120nm Mobile CPU In-House!
- A Ten-Year Miracle Achieved in Korea, Once a Semiconductor Wasteland!
- All Concerns Over ID Electronics’ Integration Completely Dispelled!
Usually, the newspaper clippings reported to Yoo Jae-won contained only articles related to ID Group. This time was different. From the front page to the society section, the papers were filled with stories about ID Group and Yoo Jae-won. There were so many articles that, at a glance, the newspaper resembled an IT industry magazine. Moreover, the tone of the coverage was so favorable that it could have passed for an ID Group press release. The newspaper’s name emblazoned at the top of the front page was Daehan Ilbo. Without that title, it could easily have been mistaken for ID Group’s internal newsletter. The excitement was extraordinary.
Although test production of MAP 3 had succeeded and Korea had indeed produced a non-memory semiconductor, mass production had not yet begun. Mass production required large-scale orders of raw materials, and those items were tightly controlled by the Japanese government, so obtaining permits took time. Even the political section, which contained no ID Group stories, was filled with praise for President Kim Dae-jung’s New Year’s press conference remarks on the internet, artificial intelligence, and cultural openness. At the same time, the paper did not hesitate to criticize the opposition party, introducing for the first time the framing that the opposition was merely obstructing without offering alternatives. The phrase “opposition foot-dragging” had frequently accompanied the democratic forces when they were in opposition, but had rarely appeared since the regime change—until now. Thanks to this, Yoo Jae-won could easily infer why Daehan Ilbo had produced its first newspaper of the new year in such a manner.