The Ion-Stuffed Surface

Donald Stookey's Chemcor Bath and Compressive Stress Armor

Calculate the theoretical strength of ordinary window glass from the covalent bond energy between silicon and oxygen atoms ($\text{Si}-\text{O}$), and a pane of glass should withstand more than $10,000\text{ megapascals}$ of pulling stress. That theoretical limit is roughly ten times stronger than structural bridge steel. In everyday life, however, an annealed glass tumbler shatters at a tensile stress of barely $50\text{ megapascals}$, less than one percent of its atomic bond strength.

In 1920, British aeronautical engineer Alan Arnold Griffith at the Royal Aircraft Establishment in Farnborough solved this two-hundred-fold discrepancy. Drawing pristine silica glass fibers directly from a flame and testing them before dust or fingers could touch their skin, Griffith proved that freshly drawn glass threads reached strengths approaching the multi-gigapascal limit. Touch the fiber lightly with a glass rod, however, and its strength immediately collapsed back to $50\text{ megapascals}$. Every ordinary sheet of glass is covered in millions of invisible, sub-micrometer surface scratches called Griffith flaws.

Unlike a ductile metal such as steel or aluminum, where atomic planes slip past one another at a crack tip to blunt the sharp edge, amorphous glass has no crystal slip planes. When a glass sheet bends, the outer convex surface is pulled into tension. The fracture stress $\sigma_f$ at which a surface flaw of depth $a$ catastrophically unzips obeys Griffith's fracture mechanics equation:

Here $K_{Ic}$ is the material's plane-strain fracture toughness ($\sim 0.75\text{ MPa}\cdot\text{m}^{1/2}$ for soda-lime glass) and $a$ is the depth of the deepest microscopic surface scratch. Because the radius of curvature at the atomic tip of a Griffith crack is less than a nanometer, an applied surface tension of $50\text{ megapascals}$ multiplies two-hundred-fold right at the crack tip. Silicon-oxygen bonds snap one by one at over $1,500\text{ meters per second}$. Crucially, however, Griffith's law works in only one direction: glass breaks solely under tensile stress. Squeeze a Griffith crack under pure compression, and its faces press tightly shut so the crack cannot grow.

For decades, glassmakers exploited that asymmetry through thermal tempering—heating a thick glass plate near its $600^\circ\text{C}$ softening point and blasting both surfaces with cold air jets. The outer skin solidifies while the hot core is still expanded. When the core cools and contracts later, it pulls the rigid outer skin into permanent compression. Thermal tempering works well for thick car side windows, but it fails below two millimeters of thickness because heat conducts out of a thin sheet too rapidly to sustain a temperature gradient between skin and core.

In 1962, following chemist Samuel Kistler's discovery of alkali diffusion in glass, Corning Glass Works research director S. Donald Stookey (working with Martin Nordberg and Ellen Mochel in Corning, New York) invented a chemical way to armor thin glass under Project Muscle, marketed as Chemcor. Stookey formulated a sodium aluminosilicate glass where aluminum ions opened wide diffusion channels through the silica network. He then submerged the solid glass sheet into a molten bath of potassium nitrate ($\text{KNO}_3$) salt at $400^\circ\text{C}$ for several hours.

At $400^\circ\text{C}$, the silicate network is rigid and cannot flow. Driven by concentration gradients, small sodium ions ($\text{Na}^+$, ionic radius $102\text{ picometers}$) diffuse out of the outer forty-five micrometers of the glass into the salt bath. Larger potassium ions ($\text{K}^+$, ionic radius $138\text{ picometers}$) diffuse in to take their exact sites. Because each potassium ion has a thirty-five-percent larger radius than the sodium ion it replaces, stuffing millions of oversized $\text{K}^+$ ions into the rigid silicate cages packs the outer skin of the glass under $700\text{ to }900\text{ megapascals}$ of compressive stress.

Before a drop onto pavement can open a surface Griffith scratch $a$, the bending impact must first overcome that entire $700\text{-megapascal}$ compressive shield just to bring the surface to zero net tension. Corning shelved Chemcor in 1971 after testing it in laboratory glassware and AMC Javelin windshields. Thirty-six years later, in early 2007, Apple CEO Steve Jobs realized plastic phone screens scratched in pockets and asked Corning CEO Wendell Weeks for an unbreakable $1.3\text{-millimeter}$ glass face. Within five months of Apple's February 2007 request, Corning's Harrodsburg plant in Kentucky produced enough 1.3-millimeter ion-exchanged aluminosilicate glass to armor the screens of the first million iPhones launched in June 2007.