Copper Alloy Fragments : Corrosion Products

The copper alloy fragments were sent in for conservation because of the appearance of pale blue-green pits. The museum was worried that it could be bronze disease. Bronze disease is the active corrosion of the copper by chlorides, which can cause complete deterioration. I tested these spots of corrosion for chlorides using a silver nitrate test. If they were copper chloride corrosion products, then I would have to remove them.

I scraped off some of the corrosion products with a scalpel into a test tube and added deionized water. I dissolved a small amount of silver nitrate in deionized water and added a small amount to the test tube with the sample. I added a few drops of dilute nitric acid. The acid should have encouraged the dissolution of the copper chloride, which should react with the aqueous silver to form silver chloride, which precipitates out of the solution. Silver nitrate and copper chloride are both soluble in water, but silver chloride is insoluble. I ran this test on two pits on two different pieces, but a solid precipitate did not form.

There were a few possible reasons for this negative result. The sample could be too cohesive and thus it did not dissolve and react with the silver nitrate. This could possibly be remedied by soaking a entire fragment in water, then using that water in the reaction, but I do not want to immerse the copper fragments in water. Water is damaging to copper, and is often the root cause of active corrosion. The second option is that this is not a copper chloride. Without chlorides, the precipitate wont form.

I then decided to analyze the fragments using a scanning electron microscope (SEM). The SEM can reveal the elements in a sample through energy dispersion spectroscopy (EDS). This is a non-destructive analytical technique. An electron beam hits the fragment. The excited electrons that bounce back off are collected and create a back-scatter image, where light elements are dark and heavier elements are brighter. The fragment also generates x-ray emissions, which are gathered to create a spectrum. The more 'counts' of these emissions, the greater the amount of the element, and the higher the peak.

Top: The SEM with its three dedicated computers. Bottom left: the SEM stage, with the door open. Bottom right: looking through the view hole at the fragment on the stage in the SEM.

 I placed each fragment in the chamber and picked a place or two to test with the electron beam. I tested both the patina and the pit on two different fragments to compare. The fragments only have trace amounts of chlorine and sulphur, eliminating chlorides and sulphates as possible sources of corrosion. In fact, the SEM spectra and back-scatter images showed that the pit has a composition close to that of the patina, with no major differences in elements.

One of the tested fragments, showing the sampled sites. The top images are the EDS spectra, and the bottom images are the back-scatter images, with pink boxes around the specific sample sites.

Although the SEM-EDS did not tell me what the corrosion product was, I was able to determine the alloy composition of all the fragments. Seven of the fourteen are brass (copper and zinc), most with both zinc and tin. A couple had arsenic, and all of them contained lead. The other seven were bronze (copper and tin), with five of them as high-tin bronzes. These too had lead inclusions, and one had arsenic.

Knowing the composition, I was able to see a pattern. All the of pitting corrosion occurred on the bronzes, and the most severe corrosion was on the high-tin ones. The brass fragments had a rough patina, and a couple had a light green patch, but none seemed to have active corrosion. This made me think that it could have something to do with the tin content rather than trace elements.

I scraped out all the corrosion product from a pit. I analyzed this sample using Fourier transform infrared spectroscopy (FTIR). The easier method is to place the sample on a stage with a small crystal. A beam is bounced through the crystal, and the interaction with the sample causes adsorption peaks on a spectrum. The peaks can be attributed to various bonds in the material. The spectrum I got from this test was noisy and unhelpful.

The unassuming FTIR machine, which looks much like a printer.

To get better resolution on the spectrum, I made a KBr pellet to test. I ground the sample with potassium bromide for half an hour until it was a very fine powder. In theory, the potassium bromide is invisible on the spectrum. I then pressed the powder into a disk using 10 tons of force. It looked like a thin white plastic wafer, with a hint of green. The disk was placed in a slide mount, then the FTIR beam passed through it. The resulting spectrum was much more resolved and not very noisy, but it was nothing like anything in the database.

Left: grinding up the KBr and sample with a tiny mortar and pestle. Right: placing the slide mount with the inserted KBr pellet into the machine.

I went through and tried to assign all the peaks to possible bonds. I created a list of all the copper corrosion products that can be green, blue, or blue-green. I eliminated the ones that had elements that did not show up on the SEM-EDS spectrum. I looked at the peaks of the remaining corrosion products and compared them to mine. Most had too many peaks missing, so I eliminated them as possibilities. The closest product was copper carbonate, or malachite, which shared many of the peaks with my spectrum. I looked at other corrosion products of tin and lead to see if I could find anything to explain the other peaks. Hydrated tin oxide matched well, taking out the peaks assigned to malachite. Copper oxide and lead carbonate, or cerussite, were also possible. Therefore, my mixture seems to be oxides and carbonates, the similar to the patina.

If I am correct, than this is not active corrosion. Although damaging, no treatment can fix it but it will not continue or spread. The fragments had a hydrated tin oxide patina, a tin-rich area above the copper oxide layer. This dried out and cracked at some point, perhaps when placed in the museum environment where the relative humidity was lower than during burial. The cracks allowed the copper below to continue to corrode, and the lead inclusions in the alloy reacted with the atmosphere or carbonic acid. Lead carbonate formed, which increased in volume. Copper corrosion is is similar size and volume to the original material, but lead products expand significantly when lead corrodes. The lead carbonate pushed the patina up, forming a pustule. At some point, this pustule could have burst, or been brushed off, or somehow failed mechanically. This caused the top layer of patina to fall off and the lead carbonate could have flaked off as well. A pit with malachite and oxides was left, a damaged spot with the same corrosion products as the patina.

My interpretation of what is happening: cerussite is lead carbonate, malachite is copper carbonate, and cuprite is copper oxide.

These products are stable. The cracking has already occurred, the damage has been done. Removing the pale, damaged products in the pits will not effect the fragments. Therefore, I will not be cleaning off the pale blue-green corrosion surfaces.