A significant fraction of global silver production is used for photovoltaic modules, and this accounts for something like 10% of their cost, and something like 10% of global silver production, at present, a number which is likely to grow. The USGS’s silver report doesn’t mention photovoltaics, but of course photovoltaics aren’t being produced in the US now. It says apparent consumption in the US was 6500 tonnes in 02019 out of 27000 tonnes of global world production.
(As a side note, it says photography use in the US is down to 3% of total silver consumption, down from 28% in 01999, which is still almost 200 tonnes.)
In PV panels, as of 02018, conductive silver paste is used to make electrical connections to the photocells at a loading of 130 mg Ag per 4.7-watt cell, down from 400 mg in 02007. The Silver Institute anticipates that cells will grow to 6 watts by 02030 and reduce silver usage to 65 mg per cell by 02028. (I’m assuming these numbers are peak watts because that’s how cells are normally sold; the capacity factor varies wildly depending on where you install them, making it not only anticommercial but impractical to derate them for an expected capacity factor.)
In 02020 about 140 GWp of new PV capacity was installed, doubling about every three years, bringing the total to about 770 GWp. 130 mg / 4.7 Wp is 28 mg/Wp, so those 140 GWp amount to 3.9 billion grams or 3900 tonnes of silver consumption, 14% of world silver production. Perhaps some of the anticipated reduction from 02018 has already happened; apparently from 02007 to 02018 the trend of silver intensity was about a 4.3% decline per year, so if that had continued for two more years, in 02020 it would be down to 119 mg per cell.
This 3900 tonnes per year is almost double the 70-million-ounces number given in the article cited above: “CRU experts forecast silver demand for the PV industry of around 70 to 80 million ounces per year until a decline to between 50 and 55 million ounces in the mid-2020s. Only by 2030 is demand expected to recover, to approximately 66 million ounces per year.” 70 million troy ounces is almost 2200 tonnes. They also cite historical figures:
According to a report published by the Silver Institute in April, global industrial demand for silver grew around 4%, from 5,768 million ounces in 2016 to 5,990 million last year. This spurt was mainly due to the record growth of the PV industry, which pushed demand for silver as a component of silver pastes for solar cells, from 79.3 million ounces in 2016, to 94.1 million ounces in 2017 – year-on-year growth of around 19%.
In SI units, 5768 million troy ounces is 179400 tonnes, almost seven times the number the USGS gives for global annual silver production, including that sold to investors and that used for non-industrial uses such as jewelry; 5990 million troy ounces is 186300 tonnes; 79.3 million troy ounces is 2470 tonnes; and 94.1 million troy ounces is 2930 tonnes. (If we were to go further into SI units, 2930 tonnes per year is 92.7 grams per second.)
Clearly the article is in error about global industrial demand for silver, both because it’s an order of magnitude too high and because an increase of 222 million ounces per year cannot be “mainly due to” an increase of 15 million ounces per year.
2930 tonnes ÷ (130 mg / 4.7 Wp) gives an estimate of 106 GWp of PV cells manufactured in 02016, which is of the right order of magnitude but about 38% higher than the Wikipedia estimate cited above of 76.8 GWp of new PV capacity installed in 02016. One or the other number seems likely to be in error; there’s no way that 28% of the year’s global PV manufacturing product was stored in warehouses or in transit at the end of the year. (Or, one supposes, on jobsites.)
Though, as I write this, silver prices have spiked to about US$26/troy ounce (US$0.85/g), the average price for several years has been closer to US$16/troy ounce (≈US$0.50/g). As of 02021-01-27 a polycrystalline solar module in China costs US$0.167/Wp on the spot market (one-week average). (PVXchange cites €0.16/Wp for “low cost” modules; at €1.20/US$ that’s US$0.19/Wp, about 15% higher.) Multiplying out US$16/troyounce × 119 mg / 4.7 Wp gives US$0.013/Wp of silver, which is about 7.8% of the spot price of the entire module. The cost of mono PERC modules and poly PERC modules are given as US$0.183/Wp and US$0.190/Wp, while the corresponding cell prices are given as US$0.086/Wp and US$0.116/Wp, suggesting that the rest of the module costs in the range of 7¢-10¢ per watt, including the silver paste.
There will be continued growth in solar PV will increase global silver demand; the expected 25% or so growth in PV installations in 02021 (continuing the trend of doubling every three years) would increase total global silver demand by about 4% this year, but if doubling were to continue at this rate, PV would consume all of current world silver consumption by the early 02030s.
For a few years investors may cushion the price effects of such demand increases, but the price will probably go higher.
Because, like indium and gallium, most silver is not mined from silver mines, but rather as a byproduct of other mining, it’s likely to have a fairly inelastic supply; the price would have to go extremely high before the minimal amounts of silver produced as a byproduct in, for example, a zinc mine, would justify increasing the mine’s production. So silver prices can probably go quite high before increased mining limits their rise.
Solar cells that can accept the lower efficiencies that accompany the use of copper-filled conductive paste rather than silver-filled paste will do so, due to copper’s much lower price and much more elastic supply. More efficient solar cells using silver will increase in price and experience continued pressure to reduce silver usage.
Old solar cells with larger amounts of silver will become increasingly attractive recycling targets thanks to their high content of now-more-valuable silver. Even at today’s US$26/troy ounce price, the 400 mg in a solar cell from 02007 is worth US$0.33, almost half the price of a replacement module (assuming both use 4.7-watt cells). If prices shoot up to US$50 per troy ounce, the silver in the module is probably worth more than the module itself, even aside from the additional possible profits from recycling the silicon; at US$100 per troy ounce there will be bandits roaming the countryside to mine silver from solar cells.
Ultimately silver will probably be abandoned in favor of more sophisticated patterning techniques than screen printing, which will permit the use of copper or aluminum with no loss in efficiency. Fractal-like branching patterns and conductors with high aspect ratios (height divided by shadow width) require more precise control over manufacturing, but can reduce the shadow and surface recombination losses to almost arbitrarily low levels.