Skip to content Skip to navigation

Research & Ideas

Search this site

Crystal paradox: Almost perfect structures made useful by tiny flaws

​A crystal is the repeated arrangement of identical atoms or molecules. But their value, as precious gems or silicon chips, often owes to tiny structural defects or impurities.

drawing of crystals

Crystals, which have fascinated humans for millennia, are common today in smartphones, laptops and televisions. | Illustration by Stefani Billings

Crystals have fascinated humans for millennia. Sodium chloride, the crystal commonly known as salt, is essential for life. Metaphysicists believed that crystals held supernatural powers. Rare crystals such as diamonds, rubies and sapphires are prized as gems.

These days, however, crystals are most commonly found not on the dinner table or fingers, but inside smartphones, laptops and televisions where silicon, gallium arsenide and various oxide crystals form the working foundations of modern electronic devices.

Bob Feigelson, a professor emeritus of materials science and engineering at Stanford, is one of the world’s leading growers of synthetic crystals. A prolific author whose writings include a chapter in Elsevier’s Handbook of Crystal Growth from ancient times to the present, Feigelson recently took time out to share his fascination with the field.

Robert Feigelson

Professor Bob Feigelson: “If you could ... walk inside a perfect single crystal, everything would look more or less the same ... ." | Photo by Tom Abate


“If you could shrink yourself to the atomic scale and walk inside a perfect single crystal, everything would look, more or less, the same in every direction,” he says.

A naturally occurring crystal is the repeated arrangement of identical atoms or molecules that create a structure, known as a lattice, not unlike the steel frame of a skyscraper. Each constituent atom or molecule in the lattice is bonded to the next in an orderly, recurring fashion, layer upon layer. Crystals can have a few different types of lattice structures. The lattice structure of salt is cubic, yielding the classic faceted grains with their perfect corners.

Paradoxically, this uniformity means that even slight impurities in a crystalline mineral can have dramatic effects. Under normal conditions, crystals contain tiny defects or impurities, including missing atoms. These imperfections strongly influence the optical, electronic, mechanical and magnetic properties of industrial crystals. The same is often true of gems. Rubies and sapphires, for instance, are both crystals of aluminum oxide. Both would be transparent if their lattices were perfect.

“A slight impurity here or there, a bit of chromium or titanium perhaps, can change the color of a crystal,” Feigelson says. “Chromium in aluminum oxide produces a blood-red ruby. Titanium yields the bluish star sapphire.”

In the 17th and 18th centuries, alchemists, those pre-scientific tinkerers best known for trying to turn lead into gold, occasionally set their sights on creating gems, a breakthrough which finally occurred around 1900, when the Frenchman Auguste Verneuil figured out how to grow large, industrially useful crystals of rubies and sapphires synthetically.

“Suddenly there was a lot of interest and money in trying to make crystals for the gem industry,” Feigelson recounts. “That’s when the field really started to grow – no pun intended.”

Dramatic shift

The crystal growth field took another dramatic leap forward in the late 1940s, when Bell Labs introduced the world’s first transistor. That original transistor was demonstrated using polycrystalline germanium, a material in which an assortment of small crystals is bonded together in random orientations. However, despite some resistance within the labs, the idea that improved and controllable properties could be better achieved with single crystals of silicon was proven by the work of Bell’s G. K. Teal and J. B. Little. The prospect of developing crystals for electronic applications and, soon thereafter, for optical applications such as lasers, led to the dramatic expansion of crystal growth studies and the development of new and more advanced commercial processing methods.

“It was also at this time that scientists started to study in earnest the important mechanisms behind the different growth processes so that crystal properties, influenced by defects, dopants and impurities could be controlled,” Feigelson says.

“Most of those early single crystals, even silicon, had to be continually improved to match new device constraints,” Feigelson says. “If you wanted to achieve the best electronic properties in your silicon crystal, for instance, you had to add something to it – you had to dope it. To make transistors, for example, requires two differently doped silicon crystals bonded to each other; you need to dope one silicon wafer with boron and the other with phosphorous. This alters the nature of the conductivity of each crystal and makes the transistor work.” In other words, one had to add an impurity to make it perform as desired.

“The crystal growth scientist’s job is to find out how to achieve a set of desired physical properties by controlling various imperfections during the crystal growth process,” Feigelson says. This work is aided by theoretical scientists who study growth behavior on the atomic or molecular scale; for example, how dopants segregate in a crystal during growth. This in turn provides solutions to achieve compositional uniformity.

The crucible

Feigelson came to Stanford amid the heyday of the crystal growth field, late in 1963 when Silicon Valley first began laying claim to its name. Faculty members at Stanford from various departments were clamoring for crystals and it was Feigelson’s job to figure out how to grow them.

“When asked whether we could grow a material, my answer was always, ‘Yes!’” he said. “If anybody else could do it, we could do it because our laboratory in the McCullough Building was very well equipped and we were financially well supported.”

Crystal growers would control the balance of perfection and impurity concentration and give prototypes to physicists. They in turn would make the measurements and publish joint scientific papers. In the process, these early scientists and engineers helped lay the foundations for a billion-dollar industry. Today, says Feigelson, commercial crystal growth companies can produce silicon crystals about a foot in diameter and 6 feet long, with high crystalline perfection throughout. Thin discs, known as wafers, are later cut off like slices of bread to be manufactured into the myriad integrated circuit chips that populate our electronic devices.

New directions

In the decades since the transistor first emerged, the crystal growth field has made itself useful in fields other than electronics and optics, most notably perhaps in biology. Here, X-ray crystallographers had to find ways to take fragile, supple biological molecules, such as DNA and proteins, and capture their basic structure in a crystalline state so that they could be more easily and more precisely studied. This branch of the field requires small but very perfect crystals for structural analysis and has provided new ways to understand how life works at the molecular level. The famous double-helix structure of DNA, for instance, while correctly theorized in the 1950s, was not actually proven until the 1980s, when X-ray crystallographers first created a static lattice of this important molecule. So far, X-ray crystallographers have revealed the structures of more than 100,000 proteins, nucleic acids and other biological molecules, in the process contributing to the development of numerous life-saving medicines.

Regardless of the application or challenge, the thrill of producing some new crystal continues to intrigue Feigelson and inspire the next generation of researchers. “I suppose it’s sort of like doing a crossword puzzle,” he says. “It’s more like a hobby than a job. It’s fascinating. It’s challenging. And it’s great when you can really pull something out of a hat. Currently we are working in my laboratory to develop low-cost methods of producing some highly efficient scintillator crystals for radiation-detection monitoring devices for use at various points of entry into the U.S.”

Get Updates from Stanford Engineering