colloids

colloidal color

Gold, copper, or silver nanoparticles grown inside the glass absorb light by plasmon resonance, giving the reds of gold ruby and cranberry glass. The color appears only when a controlled reheat strikes it.

Colloidal color in glass comes from metal nanoparticles (gold, copper, or silver) that are grown inside the glass after it is formed and then absorb light. The mechanism is physically distinct from dissolved-ion color: here the absorbing entity is a metallic particle tens of nanometers across, not a single ion. The phenomenon is called surface plasmon resonance, and it produces stronger colors than dissolved ions can: the red of gold ruby, the crimson of copper ruby, the yellow of silver stain.

A defining feature of the mechanism is that the glass has no color when it leaves the furnace. A gold ruby gather fresh from the melt is colorless or faintly tinted. The color appears only during a controlled reheat called striking, which nucleates and grows the colloids. A goblet unstruck in storage looks like clear glass; held over a torch for the right amount of time, it turns red. Because the color appears late and has to be judged during the reheat, gold ruby was one of the most valued and difficult products of Baroque glassmaking.

How it works

When a metal nanoparticle much smaller than the wavelength of light sits inside a transparent glass, an incoming light wave drives the particle’s conduction electrons into collective oscillation. At the resonance frequency, the particle absorbs light far more strongly per atom than any dissolved ion. For gold spheres a few to a few tens of nanometers across in silicate glass, the resonance sits near 530 nm (green; the size range is unverified): green is absorbed, and the transmitted light is the ruby red of gold glass.

Particle size tunes the color. As gold particles grow coarser, the resonance shifts toward longer wavelengths and broadens (the direction Mie theory gives for larger spheres). A correctly struck ruby is clear red. What an over-struck glass looks like is less settled than glassmaking lore suggests: in Haslbeck and colleagues’ gold ruby glasses, fewer and larger particles gave purple-red while tin-rich glasses with many small particles went brown. This site’s reheat stage and the striking slider on this page compute it with Mie theory from gold’s measured optical constants: as the spheres grow past a few tens of nanometers the band moves red and broadens and the particles begin to scatter, so the transmitted color goes purple and then blue while the light they scatter back is orange-brown. The brown of Haslbeck’s many small particles does not come out of that model, and we have found no measured spectrum of over-struck gold ruby to check either against.

Silver resonates at shorter wavelengths: Mie theory puts small silver spheres in glass at about 400–420 nm (blue absorbed, yellow transmitted), which is why silver in glass gives yellow rather than red. Copper’s resonance computes to about 580 nm for 40 nm particles, on top of copper’s own interband absorption of the blue; in particles under about 10 nm the surface damps the resonance until it is barely a shoulder. The result is an oxblood red similar to gold’s at far lower cost, and a dark one.

Striking. The metal enters the melt in ionic form (gold dissolved via aqua regia, copper as an oxide) together with a reducing agent. Historically, tin was the usual partner of gold, and the standard gold preparation for ruby glass was Purple of Cassius, a gold-tin precipitate named for Andreas Cassius of Hamburg, whose De Auro appeared in 1685. The Corning Museum credits a similar preparation to Johann Rudolf Glauber earlier still. Johann Kunckel had perfected gold ruby at Potsdam shortly before 1679, and by the mid-1680s was making it at the Elector of Brandenburg’s glassworks there with purple of Cassius in the batch (Corning Museum of Glass).

On the reheat (around 600 °C in Haslbeck and colleagues’ study, and in general a few hundred degrees below the melt), reduced metal atoms nucleate and aggregate into colloids. Time and temperature control the particle size, and particle size controls the color, so the striking schedule is part of the recipe.

Artifact stories

Gold ruby

Gold ruby glass contains actual gold, though very little: the laboratory gold rubies of Haslbeck and colleagues hold 200 parts per million (0.02% by weight), and the Lycurgus Cup’s glass as a whole about 40 ppm. Even so, the cost means it is often used as flashed glass for sheet and window applications, a thin ruby layer fused over a clear gather. Solid body color is used for vessels and tableware, where the thin walls keep gold content manageable. In the dilute form known as cranberry glass, the gold loading is lower still, producing pink rather than red. Cranberry became a popular Victorian tableware color (unverified): the pink shows well in thin-blown pieces, and the dilution kept costs down.

Roman glassmakers did get gold into glass, as the Lycurgus Cup below shows, but only a handful of other Roman dichroic pieces are known, and gold ruby made as a matter of routine starts in 17th-century Europe, usually credited to Kunckel (Freestone and colleagues, 2007). Potsdam ruby vessels date from about 1685 to 1700 and range from pale pink to deep garnet; how much of that range comes from striking rather than gold content is not established here.

Copper ruby follows the same striking logic and is the industrial and historical substitute when gold is too expensive. Medieval reds in stained glass windows are almost always copper ruby, flashed thin over clear glass because a full-thickness copper ruby sheet looks nearly black (general account, unverified here).

The Lycurgus Cup

The Lycurgus Cup, a Roman cage cup now in the British Museum dated to approximately the 4th century CE, is the best-known example of nanoparticle color in glass. Its glass holds about 300 ppm of silver and 40 ppm of gold, much of it as silver-gold alloy particles 50–100 nm across (roughly 7 parts silver to 3 of gold, with some copper) that both absorb and scatter light (Freestone and colleagues, 2007).

In reflected light (under normal room lighting) it appears opaque pea-green, because scattered light dominates. Lit from behind, it is translucent wine-red, because the eye receives light transmitted through the absorbing colloid. No pigment is involved; both colors come from the nanoparticles, and the craftsmen who made it did not know their colorant was metallic particles. Robert Brill’s analyses in the 1960s tied the color to the traces of gold and silver (R. H. Brill, “The chemistry of the Lycurgus Cup”, VII International Congress on Glass, Brussels, 1965, paper 223, as summarized by Freestone and colleagues); D. J. Barber and I. C. Freestone confirmed in 1990, by analytical transmission electron microscopy, that the metals sit as alloy nanoparticles.

The Lycurgus Cup is the standard example of ancient, unintended nanotechnology. It shows that empirical craft knowledge can control structure at the nanoscale centuries before there is scientific language to describe it.

← back to the start