Tellurium, used in both photovoltaic and thermoelectric technologies, has become a recent topic of debate in cleantech and materials science because of its rarity. With massive recent commodity price increases in rare earths and precious metals, I attempt to make some sense of the tellurium demand picture and whether we might expect a similar rush for the inconspicuous chalcogen element.
Tellurium (Te) is an element, number 52 on the periodic table, whose rarity on earth rivals only that of a handful of other elements, including gold's. It is the namesake of the Telluride Film Festival and Telluride, Colorado, a mining town where gold telluride was thought to be found in the late 1800s.
The chart above shows the natural abundance of elements on Earth. For all practical purposes, this picture is static: it is the result only of world-formative events like the big bang and major asteroid impacts (despite any attempt at alchemy, black magic old or particle accelerator new). The fact that gold and tellurium have similar abundances on earth is merely a coincidence, but they are nonetheless found together in alloys of tellurium, or tellurides. Tellurides like calaverite were initially foolishly discarded during the first gold rush of 1849 and the subsequent discovery of gold in them spawned a second gold rush a few years later. Mining and metallurgy, an often high-tech profession at the time, was probably the only field that had tellurium in its vernacular.
Unlike tellurium, gold is the element of basilicas and twenty twos not because of its rarity, but because of its (anti) corrosive properties. Technically speaking, gold's chemical reduction potential is positive, meaning it requires extra energy to become oxidized - and therefore it never loses its luster while adorning our teeth and bathroom door handles. Tellurium, on the other hand, was not blessed with the pretty gene. Its current spot price on metals markets (~$200/kg, 200x cheaper than gold) reflects this.
Any engineer will tell you that looks aren't everything. Tellurium, in the throws of its own fifteen minutes, has the potential to become similarly priced in the long term due to its increasing use in the highest tech applications. Tellurium has a rich history, if anything, having played a role in every stage of mining and materials science since 1849. During the space race, mining and metallurgy eventually became materials science, an interdisciplinary field of science and engineering incorporating physics and chemistry. With the discovery/invention of quantum mechanics in the 1910s and '20s, solid state physics was born and so was the modern notion of a semiconductor material. It would be thirty years before a semiconductor was applied to computation through microchips, and it would be silicon that proved ideal for this application. Initially, however, an alloy of tellurium, bismuth telluride, was the star of the solid state physics community for its fantastic Peltier, or thermoelectric, properties. A Soviet physicist who published much of the seminal work on solid state physics named Abram Ioffe believed that the commercial application of semiconductors would come in refrigeration through something known as the Peltier effect. The Peltier effect is when a semiconductor material pumps heat when electricity is made to flow through it. Any material that does this efficiently is called a thermoelectric.
Clearly, bismuth telluride and thermoelectric technology lost to the common compressor refrigerator, but one can nonetheless find these semiconductor coolers in a few places. If you have a quiet wine fridge in your living room, a climate controlled seat that cools you down in your Ford F150, or night vision goggles, then you'll find a Peltier cooler inside.
Today tellurium is seeing its first real growth in use in a different application: solar cells. First Solar (FSLR), darling of all solar startups, uses a thin-film of cadmium telluride (CdTe, or "cad-tell") as one of the functional layers in its cells that helps collect sunlight. The company is increasing production quickly. Raw cadmium and tellurium demand is increasing and alas, tellurium could finally have its day. The question is whether Te, currently priced at ~$200/kg on the spot market and with the same supply constraints as gold, will ever have the type of demand that has gold currently trading two hundred times higher at ~$40,000/kg.
Tellurium supply has historically never been much of a concern to anyone. Today tellurium is produced from the refining of copper in the same way gold is produced, as the byproduct of an electrolytic refining process that manifests itself in nasty resultant anode sludges. Tellurium is produced primarily by a Canadian company named 5N Plus which extracts it from these sludges. According to the US Geological Survey, 200 metric tonnes of tellurium were mined in 2009 worldwide and the world can sustain 1,600 metric tonnes of production per year maximum (but these estimations are hard to make accurately - see Jack Lifton's piece on tellurium supply here). In comparison, there were about 2,500 tonnes of gold mined in 2009, and 165,000 tonnes of gold have been mined, ever. Gold production peaked in 1999 at 2,600 tonnes. Let's assume that tellurium could be produced at similar levels to gold going forward.
Gold demand currently comes from three areas: jewelry (~2,750 tonnes/year), reserve assets (~350 tonnes/year), and the electronics industry (~350 tonnes/year). Adding this up we get 3,450 tonnes/year demand, well over the amount produced. The difference is made up from both recycling of jewelry and the selling of reserve assets.
Gold, however, as an element that is also tied to the world economy through federal reserves and currencies, is not truly a commodity because its price is not generally close to its marginal cost of production. For that reason, let's consider an element that might be slightly more similar to gold (Au) in that sense: platinum (Pt) or palladium (Pd). Both of these elements currently trade at $60,000/kg and $25,000/kg, respectively. The order of magnitude of these spot prices are the same as that of gold's; while these elements tend to follow gold and are therefore somewhat subject to price swings that may not be concomitant with economic fundamentals, tellurium's price can still be expected to rise significantly, albeit perhaps not quite by 200x, if demand for it was also >2,000 tonnes/year.
So how could tellurium demand increase by a factor of ten? Should First Solar be worried? Should producers of bismuth telluride thermoelectric devices be worried?
First, let's examine how much tellurium First Solar uses, and what this costs as a fraction of their total cost to produce a CdTe photovoltaic cell.
The density of CdTe is 5.8 g/cc. This gives 3.08 g/cc of Te in CdTe.
The efficiency of FSLR's modules is ~11%. At a solar irradiance of ~1100 W/m^2, their cells will have a maximum power density of ~121 W/m^2.
At a CdTe film thickness of 3 um, and at a 2.7 GW target production in 2012, they will be using roughly 71 m^3 of tellurium per year in the cells alone.
This would mean using 218 metric tonnes of tellurium per year in their cells. As described earlier, global production is currently estimated at 200 metric tonnes/year and could go as high as 2,500 if we do a straight comparison to gold.
This means FSLR would have to be producing somewhere near 27 GW per year of solar panels to ever be truly supply constrained by tellurium. Considering 20 GW of new power plants were built in the US in 2009, and that 550 GW of new capacity is expected to be installed in China between 2010 and 2020, 27 GW of PV production per year is somewhat plausible many years out and by no means likely.
To understand whether First Solar is shielded from volatility in the price of tellurium, let's look at what the cost of tellurium is within their cells. From our analysis above it takes 80 metric tonnes of tellurium to manufacture a gigawatt of cell, assuming FSLR's CdTe deposition is 100% efficient in that no tellurium is wasted or lost (not the case, but we'll stick with this assumption). At $200/kg, 80 metric tonnes costs $16 million, or 1.6c/Watt. At an overall production cost of $1.00/Watt, the price of tellurium would have to increase by at least 10x before FSLR would feel significant pain. It is safe to say that they are not going to affect the tellurium market nor be sensitive to much volatility in it with business as usual.
Now let's look at thermoelectrics and their application for something converse to refrigeration: power generation. Bismuth telluride and a similar alloy, lead telluride, have been studied for a long time for their ability to generate electricity from an applied temperature gradient such as a waste heat source. The automotive industry, in particular, has big plans (pdf) to incorporate thermoelectric waste heat recovery technology into engine tailpipes to turn wasted heat in exhaust back into electricity. These systems require roughly 1 kg of bismuth or lead telluride per car typically, roughly half of which is tellurium.
Will adoption of automotive thermoelectric generators cause a tellurium shortage? About 60 million motor vehicles were produced in 2009, only a tiny fraction of them not with an internal combustion engine. If each had a thermoelectric generator on them with 0.5 kg of tellurium within, over 30,000 metric tonnes of tellurium would be required per year - over 10x more production than what is thought to be possible, and over 100x more than what is currently produced annually. Modestly, if only 7 million cars per year had thermoelectric generators (all of GM's and BMW's autos, for instance) we would expect tellurium demand to be 3,500 metric tonnes/year. Even if we assume this tellurium usage could come down by a factor of ten through going to more power dense configurations and by using thin film materials, the long term picture is still bleak. Surely a problem for anyone expecting to scale this technology - and for FSLR for that matter! (Note FSLR's relationship with the largest tellurium supplier 5N Plus.)
More importantly, however, is that it currently costs $100 for just the tellurium in an automotive thermoelectric waste heat recovery generator, and these systems typically produce no more than 500 W of power. In the low margin automotive industry, $0.20/Watt will never cut it; the question of whether the automotive industry will ever impose a tellurium shortage practically moot.
Gold, platinum, palladium, and rare earth elements have all seen their values skyrocket in recent months. While there is nothing immediately suggesting tellurium will follow suit, it will surely be an interesting metal to follow over the next decade - and an analysis like this hopefully helps guide technologists away from the use of telluride materials in all but the niche-est of applications.