The Waiting Glass

The Scratch

In the winter of 2007, Steve Jobs had a problem he couldn't buy his way out of.

For six weeks he'd been carrying an iPhone prototype in his pocket, on purpose, right alongside his keys. Not carelessness. A test. This was where the thing was actually going to live — in a pocket, against coins and keys and denim — and if it couldn't survive that, it wasn't a product. It was a demo.

One evening he pulled it out under a lamp and the face was ruined.

Not cracked. Worse than cracked, in a way, because cracked is honest. This was a fine gray web of scratches, so shallow you couldn't catch a thumbnail on them, completely invisible in a dark room. Then the screen woke up, light came through from behind, and every one of those scratches lit at once. The whole surface went soft and gray and cheap.

The screen was plastic. So was every phone screen in America — Motorola, Nokia, Palm, everything in every carrier store in the country — for one excellent reason: phones get dropped. Plastic survives being dropped. Glass is the thing a phone gets dropped onto.

But here's what nobody outside Cupertino had fully absorbed yet. Jobs had taken the keypad off. Not shrunk it, not hidden it — deleted it. There were no buttons on the front of the device at all. Which meant the glass wasn't a window onto the product. The glass was the product. It was the only part of the thing a human being would ever touch, backlit ten thousand times a year so that every imperfection in it would glow.

A scratch on a Razr is wear on a window. A scratch on an iPhone is a defect in the object.

And the clock was not his anymore. He'd stood on a stage in San Francisco on January 9 and told the world the iPhone would go on sale in June. Roughly five months. So he needed glass — thin, hard, optically perfect, in the millions — and he needed it from a company that had never made anything like it, on a timeline that didn't exist.

What Jobs didn't know, and what almost nobody knew, was that the material already existed. It had been invented in 1962. It had been sitting in a file in upstate New York for thirty-six years, inside a company that had spent nine of those years trying and failing to give it away.

A Company That Waits

Corning, New York is a small town in the Finger Lakes that is more or less identical with the company in it. The Houghton family had been making glass there since 1868, and they'd figured out something early that most of their competitors never did: in a business crowded with people making bottles and windowpanes, the way to win isn't to make more glass. It's to know more about glass than anyone alive.

So they hired scientists. In the 1850s. For a glassworks, this was a genuinely weird thing to do.

It paid off almost immediately, and it kept paying off in ways nobody planned. In 1879 an inventor showed up needing a very specific small glass envelope — thin, uniform, able to survive being sealed around a filament running white-hot for hours. Nobody made anything like it. Corning figured it out, and by 1880 Thomas Edison had named them his sole supplier. When electric light spread across the planet, it spread inside glass blown in upstate New York. By 1926 a Corning engineer named William Woods had built a machine that spat out bulb blanks in a continuous ribbon, hundreds per minute, faster than any human hand could get close to.

Then there's the story that tells you everything.

In the early 1900s two Corning scientists, William Churchill and George Hollister, developed a glass that could take violent temperature swings without cracking. The application was railroad signal lanterns: hot lamp, cold rainstorm, and a shattered lens meant a signal with no color in it, which meant a train running blind. Useful. Unglamorous. Life-and-death.

A few years later, somebody's wife wondered whether that same glass might work for a baking dish.

That's Pyrex. Corning invented it without meaning to.

The pattern held for a century. They cast the 200-inch mirror for the Palomar telescope in the 1930s, a disc so big it took most of a year just to cool. They made the glass for the cathode ray tube, and by the 1960s essentially every television screen on Earth was theirs. In 1970 they made the first optical fiber clear enough to carry a signal for miles — years before anyone had built a network that needed one.

Make the material. Wait for the world.

Sometimes the wait is short. Sometimes it isn't. And a company that operates this way accumulates a strange kind of inventory: not unsold product, but unused answers. Things that work beautifully and have absolutely nowhere to go.

Which brings us to a bad afternoon in 1952.

The Accident

Don Stookey was trying to make a photograph out of glass.

The material was called FotoForm, and the concept is still gorgeous: a photosensitive glass you could expose like film and then develop, so the image wasn't printed on the surface but suspended inside the sheet, permanently, like something frozen in ice.

Stookey slid a plate of it into a furnace to be heat-treated at 600 degrees, shut the door, and went off to do something else.

The controller failed.

Nobody caught it for hours. By the time anyone looked, the plate had been sitting at 900 degrees — 300 past where it belonged — which meant it wasn't a plate anymore. Glass at 900 degrees runs like syrup. It would have poured out of its holder, flooded the furnace floor, welded itself to the lining, and cost somebody a full day with a chisel.

Stookey opened the door expecting exactly that mess.

Inside was a milky white plate. Intact.

He reached in with tongs. The plate slipped. It hit the floor.

It didn't shatter. It bounced.

What he'd made by accident was the first synthetic glass-ceramic. Corning called it Pyroceram, sold it as CorningWare, and put it in more or less every kitchen in America. But the casserole dish was never the interesting part. The interesting part was the sound that plate made hitting the floor — proof that glass, the most reliably brittle material in industrial life, could be talked into behaving like something else.

Project Muscle

The obvious follow-up question: could you get that toughness without turning the glass milky? Could ordinary clear glass be made to refuse to break?

The program they set up to find out was called Project Muscle, which tells you a lot about the mood in that building.

The answer arrived in 1962, and it wasn't heat. It was chemistry.

Start with the thing most people get wrong. Glass doesn't break because it's weak — glass is extremely hard. Glass breaks because its surface is covered in microscopic flaws, and a flaw only ever has to be pulled open once. So you don't need to make it harder. You need to squeeze it.

Here's how. Put aluminum oxide in the melt. Then take the finished sheet and lower it into a bath of hot potassium salt.

The glass has sodium atoms parked in its structure. The bath is full of potassium atoms, which are identical in every way that matters except one: they're bigger. At temperature, the two trade places. Big atoms move into spaces built for small ones.

Then it cools, and the structure locks around a surface now carrying more than it has room for. Permanently compressed. Every one of those invisible flaws squeezed shut by the material around it.

Corning gave the composition an internal number: 0317.

It took 100,000 pounds per square inch of force — around fourteen times ordinary glass. They branded it Chemcor and spent nine years hunting for anybody who wanted it.

The sales list is a little heartbreaking. Phone booths. Prison windows. Safety eyeglasses, which were recalled — because a glass that holds and holds and finally lets go does not let go politely. It goes all at once, everywhere. A little of it made the road: some American Motors Javelins carried Chemcor windshields, and about a hundred drag-racing Darts and Barracudas took it, not because it was tough but because it could be made thin, and thin glass is light glass, and over a quarter mile weight is the only enemy there is.

Then they put it on the crash sled and found the wall they couldn't engineer around.

The windshields didn't break.

That sounds like a win. It's the opposite. A windshield that gives way slows a head down on its way through. A windshield that holds doesn't slow it down at all — it just stops it. Corning had built a windshield too strong to survive. Ford passed. Everybody passed.

In 1971 they shut Project Muscle down, filed the results, and 0317 went into the dark for the next thirty-four years.

The Drawer Opens

What finally woke it up was a flip phone.

In 2005 Motorola shipped the Razr V3, and the little external display on the lid — the window that showed you the time with the phone closed — sat behind glass instead of plastic. Somebody at Corning noticed.

It was nothing. An inch across. But it meant that for the first time in a generation, a consumer electronics company had decided a screen was worth putting glass in front of.

A small team went back to the old question: how thin could 0317 go? The surviving samples were about four millimeters. They were confident they could do much better. A market study said there might be real money in it. Somebody gave the project a name that stuck: Gorilla Glass.

Eighteen months later it had gone exactly nowhere, for exactly the reason it had gone nowhere in 1971. No handset maker was going to spec a glass nobody manufactured. Corning wasn't going to convert a production line for a glass nobody had spec'd. Everyone waited politely for someone else to move first. Samples went back on the shelf. The team went back to other work.

Then, in February 2007, the phone rang in the CEO's office.

The Phone Call

You need to know something about the man who picked up.

Wendell Weeks joined Corning in 1983, at twenty-four, in the finance department. In February 2007 he'd been chief executive for about eighteen months. And he had already watched this company come very close to dying, from near enough to have been holding part of it when it happened.

In September 2000, Corning traded at $113 a share. The world was going to be wired end to end in optical fiber. Corning made the best fiber on the planet. So Corning did what a confident company does, which is build — new plants, new capacity, thousands of people hired into a future every serious person in the industry agreed was already on the way.

By the fall of 2002 the stock was a dollar and change.

Weeks had been running the fiber business. Not observing it from somewhere safe. It was his. He went out to the plants and stood in rooms full of people who had moved their families for jobs he'd helped promise them, and told them the future they'd been hired for was canceled, and then drove home. Nobody who's done that ever fully stops doing it.

And here's the thing that mattered most: it hadn't happened because the fiber was bad. The fiber was magnificent. It happened because they'd built a road to a city that never got built.

He came out of those rooms with one rule. He would never again bet this company on a customer who wasn't standing in front of him.

The other thing he did, once he had the top job, was take the ceiling out. He liked to say the separation between himself and any bench scientist at Corning was nonexistent, and he seems to have meant it operationally — he'd made it possible for a researcher to start working on something without first sitting through a meeting about whether they were allowed to. He asked his people for what he called a degree of insubordination. It sounds like a line from a leadership book. It was about to be the only reason any of this worked.

So: February 2007, the phone rings, and the voice on the other end starts explaining to the chief executive of Corning how glass is made.

Weeks let him run for about a minute.

"Can you shut up," he said, "and let me teach you some science?"

Pause. Then: go ahead.

So Weeks walked him through ion exchange — the sodium, the potassium, the surface that ends up crushing itself shut. And then he mentioned, because it was relevant, that Corning had already built this material. In the sixties. And had spent nine years failing to find a single customer for it.

Jobs wanted it. He wanted it thinner — 1.3 millimeters — in quantity, shipping in June.

Nothing like that had ever been manufactured at that thickness, in that volume, anywhere on Earth. There was no line running the material and hadn't been in decades. To do this, Weeks would have to take a plant that was currently running, currently profitable, currently producing glass that customers were currently paying for, rip the process out of it, and re-point it at a product that had failed commercially before most of the people on that floor were born.

He said no.

Then Jobs told him he was afraid. Which was true, and Weeks said so — anybody who actually understood the question would be.

And then he thought about the rule.

Fiber had nearly killed them because they'd built for a customer who existed only in a forecast. This customer was on the phone. Right now. For five months. And then, almost certainly, never again.

He said yes.

Seven Variables

What came out of the file wasn't a formula. It was a formula for a different object — a four-millimeter sheet of automotive glazing, made by processes Corning no longer ran, designed in a decade when nobody had heard of a touchscreen.

Two composition scientists, Adam Ellison and Matt Dejneka, got handed it at Sullivan Park, the research campus on the hill above town. Make it a third as thick. Make it clear enough to stare at all day. And — this is the constraint that nearly ended the whole thing — make it on machinery that already exists.

Inventing a new way to form glass takes years. They had until spring. Which meant it had to run on the fusion draw.

The fusion draw is one of the great pieces of industrial elegance and it's worth understanding. Picture a long narrow trough, filled slightly past the brim with molten glass. It spills over both long edges at once and runs down the outside walls in two thin sheets. The bottom of the trough comes to a point, and right there the two sheets meet and fuse back into one, and that single ribbon gets drawn steadily downward. Pull faster, get thinner glass.

Why go to all that trouble? Because of what doesn't happen. The two outer faces of that finished ribbon are the faces that ran down the outside of the trough. Neither has touched anything but air. No roller pressed them. No belt carried them. Nothing polished them. That's how you get glass that thin and that flat — and it was the one part of this entire operation that was not up for negotiation.

The fusion draw is also picky about what you feed it. It wants glass that goes, in Ellison's phrase, extra stretchy — like chewing gum — at a fairly low temperature, so the ribbon holds itself together on the way down.

0317 would not do that. And the reason is the kind of problem that makes a room go quiet.

Everything that lets a glass swap ions deeply and fast — exactly what a phone screen needs — is the same thing that makes it stiff and hard to melt, which is exactly what the fusion line can't have. Improve one, wreck the other. There's no version that's simply better. There are only trades, and somebody has to pick them.

So Ellison's team went into the composition and moved seven things at once — adjusting oxide levels, adding a single ingredient Corning has never named publicly. Every candidate meant a melt, and a melt takes time nobody had. Dejneka spent that winter building samples and then destroying them: dragging grit across them, dropping them on things, bending sheets in a frame until they let go with a noise like a gunshot.

Nobody stopped to ask permission for any of it. That was the insubordination Weeks had been talking about, and it was the only thing that made the calendar survivable. Corning's normal path from research to product runs about two years. The heartbeat of a 156-year-old company that fully intends to see another 156 is slow on purpose, and usually there's a good argument for that.

By the end of March they had it. A glass that took deeper compression than the original, exchanged its ions faster, and would go over the trough like chewing gum.

One Tank

The plant sits in horse country an hour south of Lexington, Kentucky, among bluegrass and white fences, which is nobody's first guess for where the future of the telephone gets manufactured.

Harrodsburg opened in 1952 making optical glass — binocular lenses, periscope lenses, glass for aerial photography. When that work dried up in the eighties, Corning didn't close it. They turned it over to liquid-crystal display glass, and the engineers on that floor improved the fusion draw enough that the company still calls the place its center of excellence for melting and fusion. It's a plant that had already reinvented itself once and lived through it.

In early 2007, Harrodsburg was running seven tanks, fifteen feet each, producing more than a thousand pounds an hour of display glass for televisions. A real business. Customers, orders, delivery dates.

They took one of the seven.

You cannot ease a glass tank from one product to another. It runs at temperatures that take days to reach and weeks to come back from, and everything about it — melt chemistry, flow, draw rate, the people who can hear when something's starting to go wrong — has been tuned over years to make one thing extremely well. What that tank was being asked to do was keep the fire and give up everything else.

Most of April went into it. Run out the old melt, charge the new batch, then teach the line a glass it had never handled. The first ribbons came down wrong — a composition that behaves beautifully in a lab melt behaves like a stranger at a thousand pounds an hour — and every correction took hours to travel through the tank before anyone could see whether it helped. No second tank to practice on. No time to build one.

It began producing to spec in May.

What came down off the draw was a ribbon 1.3 millimeters thick that had never been touched by anything. It got cut into sheets, and the sheets went into potassium salt held at 400 degrees and sat there for hours while the small atoms climbed out and the big ones climbed in. What came out of that bath and cooled was a piece of glass locked in a permanent argument with itself: surface crushing inward, core pulling outward, the whole thing holding.

By the last week of June they'd made enough to cover seven football fields.

Seven Football Fields

Somewhere in there Weeks made a call west. There's no record of what was said, only that the answer was yes: tank's running, glass is good, it'll be there. Two companies that had each spent the spring betting everything on the other one's competence, finding out they'd both been right.

It shipped June 29, into the hands of people who'd stood in line overnight and had no idea that the surface under their thumbs was designed for a windshield in 1962, rejected by Ford, filed away in 1971, and pulled back out of the dark by a phone call in February.

Corning made $20 million from the material that year. Four years later it was making $700 million.

Some time after the launch a note reached Weeks in New York, from California. One sentence.

We couldn't have done it without you.

Here's what I keep coming back to. Every story we tell about this industry is a story about invention — the flash, the garage, the thing that didn't exist yesterday. But 0317 wasn't invented in 2007. It was invented in 1962 by people who were completely, painfully right and could not find a single human being who needed what they'd made.

The genius in 2007 wasn't creating something new. It was two men on a phone call recognizing, in about eleven minutes, that "nobody wanted this" and "this doesn't work" are entirely different sentences.

The drawer is always full. The hard part is knowing when to open it.