Chapter 109: The Third Light
Location: ISMC Semiconductor Facility, Gorakhpur
Date: 27 January 1973 — 14:15 Hours
The photographer had been rearranging the same lighting stand for eleven minutes.
Karan had counted.
"Little to the left, Mr. Shergill. Good. Now if you could look toward the facility with — yes. Thoughtful. Forward-thinking."
Karan looked thoughtful and forward-thinking while mentally reviewing the morning's production reports. Steel division, furnace number two: temperature control still inconsistent. Twenty-three tonnes behind schedule. The procurement note from Balachandran needed a response before Thursday or the man would send three more follow-up memos, each one slightly more passive-aggressive than the last.
The flash went off. Then again. Then three more times because apparently modern photojournalism required multiple identical photographs of the same person standing in the same spot.
"Excellent, Mr. Shergill. Now for the cover shot — something more dynamic? Perhaps at one of the LED production lines? The innovation, the technology, the future of Indian semiconductor manufacturing?"
"The LED line is in operation," Karan said.
"Clean room protocols—"
[LED — light-emitting diode — was a semiconductor device that converted electricity directly into light, with none of the fragile filaments or glass envelopes of traditional bulbs. When electrical current passed through certain crystalline materials, electrons dropped between energy levels in the atomic structure and released the difference as photons — particles of light. The color of that light depended on the material: gallium arsenide produced red, gallium phosphide produced green. The facility's clean room ran two active lines producing these red and green variants that were now appearing in consumer electronics and control panels across the country.]
"Just a few shots through the viewing window," the photographer said. "Equipment in the background, you in the foreground. Perfect visual metaphor. Five minutes, maximum."
Karan looked at his watch. Meeting with Lev at three about the new transistor specifications. Aerospace division at four about component delivery schedules. Japanese Investors (for access to japan electronic industries not for selling shares of ISMC )at seven, which would require him to spend two hours pretending to enjoy sake while discussing market penetration strategies he'd already decided on six weeks ago.
"Five minutes," he said.
The LED production floor was visible through a large viewing window — Lev's idea, installed specifically so that investors could see the technology operating without having to suit up and go inside. It had turned out to be one of the better decisions in the facility's layout. There was something immediately convincing about watching manufacturing happen in real time.
Inside the clean room, technicians in full bunny suits moved between the epitaxial growth reactors.
Epitaxial growth reactors were specialized furnaces that built semiconductor crystals one atomic layer at a time — a process called epitaxy, from the Greek words meaning "arranged upon." Inside these reactors, substrates — thin wafers of crystalline material — were heated to high temperatures while carefully controlled mixtures of gases flowed over them. The gases decomposed on the hot surface, depositing atoms in precise crystalline arrangements. For gallium phosphide LEDs, the gases contained gallium and phosphorus compounds. For gallium arsenide LEDs, gallium and arsenic compounds. Layer by layer, atom by atom, the crystal structures grew until they reached the thickness needed for functional devices.
The yields were good. Production volume was climbing month over month. Red LEDs, green LEDs, finding their way into consumer electronics panels, automotive dashboards, industrial control systems, the computing division's indicator arrays.
"This is perfect. The precision, the workers, the scale — India's technological future in one frame. Now, Mr. Shergill, if you could stand here and look just slightly—"
The assistant, a young man fresh enough out of college that he still looked surprised to be employed, reached for the battery pack of the secondary flash array to adjust something.
He adjusted the wrong thing.
All eight xenon flash units fired simultaneously.
Six thousand watts of white-blue light detonated against the viewing window.
Xenon flash tubes worked by ionizing xenon gas with a high-voltage electrical discharge, producing an intensely bright flash lasting only milliseconds. Unlike incandescent bulbs which burned warm and yellow like miniature suns, xenon discharge light was cold-white and loaded with blue and ultraviolet wavelengths. The spectral difference was dramatic — incandescent light was heavy in red and yellow, weak in blue; xenon flash was the opposite.
For half a second, the production floor was lit with that cold-white surge. Under it, phosphors glowed with unusual intensity, materials responded in ways their designers had calculated for, and some in ways they hadn't.
"Sorry! Sorry, sir, I'm so sorry, the switch was — I didn't mean to—"
Karan wasn't listening.
He was looking through the window.
The production floor had returned to normal in the same half-second it had taken to illuminate.
But in that half-second, something had registered.
He stood completely still while the assistant continued apologizing behind him and the photographer made sounds of professional embarrassment and two ISMC employees appeared from somewhere down the corridor to see what the noise was about.
Karan was looking at the control panel of the nearest epitaxial reactor.
It had three indicator positions. Red. Green. And a third socket — installed with a temporary white incandescent bulb, the standard workaround, the universal admission that no one had solved the problem yet.
An incandescent bulb worked by heating a thin wire filament — usually tungsten — until it glowed white-hot. It was the same principle as a candle, just faster and hotter. But it was profoundly inefficient: roughly 95% of the input energy became heat, only 5% became light. And the light it produced was never true white — it was always yellowish, the color temperature of hot metal, skewed heavily toward the red end of the spectrum with very little blue content.
Under the xenon flash, the red LED had looked red. The green LED had looked green.
The white incandescent had looked wrong. Dim. Yellowish. A different category of light entirely. Not a substitute — a placeholder, and a poor one, and under that flash it had been impossible to pretend otherwise.
He looked at the gap between the green light and the empty space where the third color should have been, and his mind did the thing it occasionally did — the thing he'd never fully explained to anyone, the way knowledge from a life he no longer lived surfaced not as memory but as certainty, arriving fully formed and immediate.
Red and green made yellow.
Red and blue made purple.
Green and blue made cyan.
Red and green and blue made white — actual white, clean white, the kind that didn't require burning a filament and wasting most of its energy as heat.
This was additive color mixing — the principle by which colored lights combined rather than subtracted. When you mixed paint, you were subtracting wavelengths: blue paint absorbed red and green wavelengths, yellow paint absorbed blue, and mixing them gave you green because that was the only wavelength both paints reflected. But when you mixed light, you were adding wavelengths: shine red light and green light on the same spot and you saw yellow, because your eye received both wavelengths simultaneously. This was how television picture tubes worked, painting phosphor dots in red, green, and blue and relying on the human eye to blend them into a full-color image. It was how computer monitors would work. It was how every display technology that mattered would work.
Every screen. Every display. Every indicator panel in every facility he'd ever built or ever would build. All of it depended on three primary colors of light. Two of those primary colors existed as manufactured solid-state devices. The third — blue — was still a theoretical promise.
All of it waiting on the third color.
The year was 1973. In the timeline he remembered — the life before this one, the future that had already happened in a world he no longer occupied — the problem wouldn't be solved until the 1990s. Three Japanese researchers Isamu Akasaki, Hiroshi Amano, and Shuji Nakamura would receive the Nobel Prize in Physics for exactly this. He could see the headline, a Sunday afternoon in 2013, reading on his phone in a flat in Delhi: Nobel Prize awarded for blue LED breakthrough that enabled white LED lighting revolution.
Gallium nitride. The material that everyone had tried and most had given up on because it was difficult in ways that took decades to fully understand.
Gallium nitride — GaN — was a compound of the metal gallium and nitrogen. In theory, when properly structured as a semiconductor junction and subjected to electrical current, it could release photons at the blue end of the visible spectrum. The bandgap — the energy difference between electron states in the crystal structure — was approximately 3.4 electron volts, corresponding to ultraviolet light. By carefully incorporating indium into the structure to create indium gallium nitride (InGaN), you could tune that bandgap down into the blue range at 450-470 nanometers wavelength.
In practice, it had defeated every attempt. The crystals grew wrong. The lattice mismatch between GaN and available substrate materials was severe, creating enormous densities of defects — dislocations and grain boundaries that ruined electrical properties. Even when decent crystal quality was achieved, getting p-type conductivity — necessary for a functional LED junction — was impossible. Magnesium doping had been tried. It didn't work. No one knew why.
He knew it was gallium nitride.
He knew it was the right answer and that the researchers currently working on blue light emission didn't know that yet, were pursuing silicon carbide and zinc selenide and other materials, were years away from the realization.
What he didn't know — couldn't know, not without the work — was how to actually make it function. The theory was one thing. The process, the specific temperatures and pressures and doping concentrations and crystal growth conditions that made the theory real, was another thing entirely. That part had taken the Japanese team years of incremental, often failing experimentation.
But he could get that knowledge.
The system existed for exactly this.
He stood at the window for another moment.
Then he turned.
"The session is over," he said.
"But we haven't finished the—"
"Thank you for your time," Karan said, and walked away.
15:00 Hours
Lev Abrahamov was already in the meeting room when Karan arrived, which was normal — Lev was always early, on the grounds that being late was a form of intellectual disrespect and being on time was merely acceptable.
He was reviewing a printed spec sheet and had the expression of a man who had found something in it that he disagreed with and was deciding how to raise it.
Karan sat down. "Before the transistor specs — I want your honest assessment on something."
Lev looked up. This was apparently more interesting than the spec sheet. "Go on."
"Blue LEDs," Karan said.
A pause.
"That's a short sentence for a very large problem," Lev said.
"I know. What's the current state of research?"
Lev set down the spec sheet entirely, which meant this had his full attention.
"Blue LEDs are theoretically possible," he said. "Multiple materials have been proposed. Silicon carbide, zinc selenide, gallium nitride. All have problems. SiC can produce blue light but the efficiency is terrible — most of the input energy becomes heat, not light. ZnSe degrades rapidly under operating conditions. GaN has the right bandgap but nobody can grow high-quality crystals, and even when they do, p-type doping doesn't work."
"Why not?"
"Unknown," Lev said. "Magnesium should work as a p-type dopant based on the crystal chemistry. But experimentally, it doesn't activate. The material stays n-type or insulating. People have been trying for a decade. No success."
"What if," Karan said carefully, "the magnesium is being passivated during growth? Hydrogen from the ammonia precursor binding to it, electrically neutralizing it."
Lev was quiet for a moment.
"That's a theory," he said. "I haven't seen it in the literature."
"It's my theory," Karan said. "And if it's correct, thermal annealing after growth might activate the magnesium. Heat in an inert atmosphere, drive off the hydrogen, restore electrical activity."
Lev looked at him with the expression of someone evaluating whether a statement was brilliant or completely wrong and unable to determine which.
"That is either an excellent insight or wishful thinking," he said. "The only way to know is to try it."
"I want to try it," Karan said.
Lev absorbed this.
"You want to start a blue LED research program," he said.
"Yes."
"Based on gallium nitride."
"Yes."
"With a novel theoretical approach to p-type doping that hasn't been demonstrated."
"Yes."
"This will be expensive," Lev said. "Custom MOCVD reactor, high-purity materials, characterization equipment, dedicated clean room space. Minimum two crore rupees over eighteen months to reach a meaningful result."
"I know," Karan said.
"And it might not work," Lev said.
"It might not," Karan agreed. "But if it does work, we'll have blue LEDs fifteen to twenty years before anyone else. White LEDs follow from that. Solid-state lighting follows from that. The entire global lighting market follows from that."
Lev was quiet for a long moment.
"You are very confident about this," he said finally. "More confident than the theoretical justification alone would seem to warrant."
Karan met his gaze. "I am."
"Why?"
"Because I've thought about this extensively," Karan said. "And I believe the path exists."
Lev studied him. "You have an unusual way of arriving at conclusions. I have noticed this before. You synthesize information in ways that produce results ahead of where the normal trajectory would place them."
"Some people call that intuition," Karan said.
"Some people call it genius," Lev said. "I don't know which you are. But your track record suggests your intuitions are worth pursuing." He paused. "Who would lead this program?"
"Dr. Ramesh Chandra as overall technical lead," Karan said. "Dr. Suresh Iyer for the doping and diffusion work. We'll need to hire a crystal growth specialist — someone with MOCVD experience, preferably GaN experience even if their results were negative."
"And you?" Lev asked.
"I'll be involved," Karan said. "Not just funding from a distance. I want to understand the experimental work directly."
Lev nodded slowly. "Then I support it. With the caveat that failure is possible and acceptable as long as we learn from it."
"Agreed," Karan said.
"When do you want to start?"
"Equipment procurement this month," Karan said. "First experiments by May."
Lev made a note. "I'll coordinate with Ramesh on the reactor specifications." He paused. "This is ambitious even by your standards."
"Good," Karan said. "We should be ambitious."
That night, alone in his study with the door locked and the curtains drawn, Karan accessed the system.
The interface appeared in his mind — not visual, not audible, but present in the way memory was present. He thought: Gallium nitride blue LED. Complete research package.
The system responded with structured knowledge:
[GaN Blue LED Technology - Complete Research Package]
Cost: 500 Gamer Points
He confirmed the purchase.
The knowledge transferred over approximately two minutes — not instant but gradual, information structuring itself into his memory as if he'd spent years learning it. The crystal physics. The MOCVD process parameters. The p-type doping solution via thermal annealing. The device architecture. The failure modes and their solutions.
When it finished, he opened his eyes.
His point balance had decreased by 5,000. The knowledge was now resident in his mind with the clarity of expertise.
He pulled a notebook toward him and began writing — not because he needed notes for himself, but because tomorrow he would need to present this to scientists who didn't have system access.
29 January 1973 — 10:00 Hours
ISMC Conference Room
Sixteen people sat around the conference table. Karan had called the meeting with twelve hours notice: "Blue LED Research Initiative — Technical Discussion Required."
The attendees included the ISMC leadership — Lev, Sato Hiroshi, Tanaka Kenji, Dr. Ramesh Chandra, Dr. Suresh Iyer, Vikram Malhotra — plus specialists Karan had specifically requested: Dr. Harish Mehta from materials science (background in III-V semiconductors from his years at Bell Labs), Dr. Ashok Kumar from crystal growth (expert in oxide crystal growth including sapphire), Rajiv Sharma (recent IIT Kanpur PhD, six months at ISMC), and several senior process engineers.
Karan stood at the whiteboard.
"Blue LEDs," he said without preamble. "I want to develop them. Gallium nitride as the base material."
The room absorbed this.
Dr. Harish Mehta — forty-one years old, fifteen years of experience including three years at Bell Labs before returning to India — spoke first.
"Blue LEDs using GaN," he said carefully, "have been attempted by multiple research groups. The material science challenges are significant. Poor crystal quality, high defect density, inability to achieve p-type doping. The general consensus is that practical blue LEDs are ten to fifteen years away."
"I disagree with that timeline," Karan said.
"Based on what?" Mehta asked. Not hostile — genuinely wanting to know.
Karan turned to the whiteboard and began writing.
"Gallium nitride. Direct bandgap at 3.4 electron volts. Incorporate indium to create InGaN, tune the bandgap into the blue range. Grow on sapphire substrates using MOCVD."
He sketched the layer structure.
"The lattice mismatch between GaN and sapphire is sixteen percent, which is large. But it's manageable with an aluminum nitride buffer layer deposited first. The AlN acts as a structural transition."
Dr. Ashok Kumar, the crystal growth specialist, spoke. "Sixteen percent mismatch creates enormous defect density even with a buffer layer. This is why GaN devices have consistently failed."
"The defect density can be reduced with optimized growth conditions," Karan said. He wrote temperature and pressure specifications on the board. "MOCVD growth at 1050 Celsius, pressure 200 Torr, V-III ratio around 30."
V-III ratio referred to the ratio of Group V elements (nitrogen, in this case) to Group III elements (gallium) in the gas phase during crystal growth. Getting this ratio right was critical — too little nitrogen and you got gallium-rich material with metallic inclusions, too much nitrogen and you got nitrogen-rich material with poor crystallinity. The ratio had to be optimized through experimentation.
Sato Hiroshi, who had been watching quietly, said: "These are very specific parameters. GaN MOCVD is not well-established. How have you determined these conditions?"
This was the question Karan had prepared for.
"Theoretical modeling based on thermodynamic equilibrium, surface kinetics, and extrapolation from related III-V semiconductor systems," he said. "The fundamental physics is similar to other MOCVD processes we already understand. The parameters can be predicted."
It was a good answer — technically accurate, theoretically sound, and completely insufficient to justify the level of specificity he was claiming. But it was the best answer he could give without revealing the system.
Sato looked at him for a long moment. "That is a confident prediction."
"I'm aware it's unconventional," Karan said. "But I believe the approach is sound."
Vikram Malhotra spoke up. "Even if crystal growth works, there's still the p-type doping problem. Nobody has achieved p-type GaN reliably."
"Because the magnesium is being passivated during growth," Karan said. "Hydrogen from the ammonia precursor bonds to it electrically. The solution is thermal annealing after growth. Heat to 750 Celsius in nitrogen atmosphere, drive off the hydrogen, activate the magnesium."
The room went silent.
"That's a theory," Mehta said.
"My theory," Karan confirmed. "And we'll test it experimentally."
Dr. Ramesh Chandra, who had been listening carefully, spoke. "You're proposing a complete research program. Custom MOCVD reactor, specific growth conditions, novel doping activation approach. This isn't exploration — this is a directed development path."
"Yes," Karan said.
"Why?" Ramesh asked.
Karan turned from the whiteboard.
"Because if it works, we develop blue LEDs ahead of everyone else in the world. Blue LEDs enable white LEDs — blue light plus yellow phosphor coating produces white light. White LEDs enable solid-state lighting that uses one-tenth the power of incandescent bulbs. For India, with chronic power shortages, that's transformative. Globally, the lighting market is hundreds of billions of dollars. First-mover advantage is worth far more than the development cost."
He paused.
"The budget is two crore rupees over eighteen months. I'm authorizing it personally."
Mehta looked at him. "That's a significant investment in an unproven approach."
"All research is unproven until you do it," Karan said.
"And if it fails?" Mehta asked.
"Then we learn why it failed and adjust," Karan said. "But I believe it will work."
Lev had been watching this exchange. He spoke now. "The question isn't whether to try — Karan has already decided that. The question is who joins the program."
He looked around the table.
"This is high-risk research," Lev continued. "It might not work. But if it does work, it's breakthrough-level science. Who wants to be part of it?"
Vikram raised his hand immediately. "I do."
Rajiv Sharma nodded. "I want to work on this."
Dr. Kumar looked at the whiteboard, at the reactor specifications Karan had sketched. "The approach is unconventional but the reactor design addresses real problems. I want to see if it works."
Dr. Mehta was quiet for longer.
Then: "I think the probability of success is low. But the potential impact if it succeeds is high enough to justify the attempt." He looked at Karan. "I'm in. With the understanding that failure is a possible outcome."
"Agreed," Karan said.
Sato spoke. "This will be interesting. Either you have solved a problem that has defeated researchers for years, or you have made expensive mistakes. We will find out which."
Tanaka added quietly: "The work will be difficult. But difficult work is often the most worthwhile."
Karan looked at the group.
"Then we begin," he said. "Equipment procurement starts this week. First growth attempts in May. Target: working blue LED by year end."
That Evening — 19:30 Hours
Shergill House
Sakshi was reading in the drawing room when Karan arrived home. She looked up, then looked at him more carefully in the way she did when she was reading the version of him that didn't correspond to the news he'd just walked through the door with.
"You look like you started something today," she said.
He sat down. "A research program. Blue light sources."
She waited.
"The indicator panels at the facility have three positions — red, green, and a third that doesn't exist yet as a solid-state device," he said. "We're going to develop it."
"How long will it take?"
"Years," he said honestly. "This isn't production engineering. This is the research that creates the technology."
She was quiet for a moment. "Is it important enough for years?"
"Yes," he said. "If it works, it changes lighting entirely. And India does it first instead of watching someone else succeed and then licensing their technology."
Sakshi smiled and nodded. "Then do it and don't second-guess halfway through."
She picked her book back up, and that was the end of the conversation. This was what she did — not the technical detail, never that, but the human accuracy. Knowing when to ask more and when to simply accept that he'd made a decision.
Later, after dinner, he sat in his study for a while.
The facility would be quiet at this hour. The production floor running on night crew, the reactors maintaining their temperatures, the red and green LEDs being manufactured in steady volume.
And the empty third sockets, still waiting for the color that didn't exist yet.
He pulled a sheet of paper toward him and wrote:
Blue LED Research Program
Lead: Dr. Ramesh Chandra
Start: February 1973
Target: Working prototype by December 1973
He looked at it for a moment, then wrote one more line:
This will take time. That's acceptable.
Outside, Gorakhpur was settling into its cold January night. The streets quieter, the air carrying the particular edge that passed for winter in eastern UP.
Some things were straightforward to build — you understood the target, you assembled the materials, you executed the plan.
And some things required you to first prove the path existed.
This was the second kind.
The hard part wasn't starting. The hard part was staying patient through all the failures that would come before the breakthrough — the crystals that cracked, the layers that wouldn't grow properly, the devices that produced nothing when you applied current and waited and the test chamber stayed dark.
You had to believe the light would eventually come.
He did.