Discussion on Refiner Methods

Shawna Kulpa takes a look into the assaying methods refiners use to determine the value of a lot. This article is a discussion on refiner methods from the journey a piece of metal goes through to how much they are worth.

It seems so mysterious: You gather up your bits of dirty precious metal scraps, send them to a refiner, and just a short time later get either a check in the mail or some new shiny metal just waiting to be transformed into jewelry. But do you know what happens to that metal once you ship it off, or how your refiner calculates exactly for how much precious metal you should be paid?

Prepare for that mystery to be solved.

Behind the Curtain

Despite their ability to transform dirty bits of scrap into clean, fresh precious metals, refiners are not magicians trying to keep their tricks out of view. Not only are they more than willing to discuss what happens to a lot as it's assayed and refined, they openly invite customers in to witness the process.

"We want our customers to feel confident that we're not hiding behind some screen or picking some numbers out of the sky or manipulating things," says Larry Fell, CEO of David H. Fell in City of Commerce, California. "If you have the ability to watch [your lot being] melted and take a sample with you, that's the best way to do it."

To understand the various assaying methods refiners use, you first need to know what initially happens when you send in your scrap for refining. Regardless of the type of analysis that will be done; your scrap will be melted down to form a bar, with the goal of creating the most homogenous mix possible to ensure the most accurate results.

"The accuracy of any sample depends on how well the scrap has been melted," advises Ron Kelly, refining technical supervisor of Hoover & Strong in North Chesterfield, Virginia. "If it's not homogenous, you won't get an accurate sample. It all starts with the smelter."

Once the scrap is melted, there are several methods refiners can use to assay the lot. The best method for your lot will depend on its contents, which is why you should separate your scrap. Gold and silver should be kept separate from platinum and palladium, and scrap metal needs to be separated from general shop sweeps. (The sweeps may contain combustible materials that will need to be burned off before any precious metals can be recovered.) While refiners can and will use multiple assaying methods to process mixed lots, it will cost a customer more in terms of the time needed to run the tests as well as the additional fees involved. Different assaying methods are best suited (and offer the best accuracy) for specific metals, so while a refiner could rely on multiple methods to process a mixed lot, the time and fee savings would have to be weighed against the loss of accuracy.

"The accuracy of XRF now is incredible, as long as you know what you're looking for." - Larry Fell

"There is a cost difference between tests," says Daniel Ballard, national sales manager of Precious Metals West/Fine Gold in Los Angeles. "Chemical analysis, which offers the best accuracy, can be expensive. There's a cost benefit analysis that people have to figure out for themselves." Many refiners have a preferred way to assay the majority of the lots they receive. In that case, the only charge a customer receives for the service is a percentage of the precious metal recovered. In instances where a customer insists on a different test, refiners will often charge an additional fee for that test in addition to a percentage of the metal recovered. While the most complicated chemical analysis methods may have better accuracy, you'll also be paying extra for that accuracy.

X-ray Fluorescence (XRF)

One of the easiest and fastest ways refiners can use to quickly determine the value of a lot is X-ray fluorescence (XRF). This technique uses X-rays to excite the molecules in the various elements of a metal sample. Every element, when bombarded in this way, emits a unique set of secondary fluorescent X-rays. An XRF machine will analyze the emission of those secondary X-rays to determine the presence of specific elements. In 30 seconds, it can determine not only which elements are present, but also the amount of each individual element, giving a refiner an accurate snapshot of the contents of a scrap lot.

While this technology has been around for a while, it has advanced a great deal over the last few years, yielding newer machines with an accuracy of +/- 0.1 percent. But even with increased accuracy, machine calibration and operator knowledge are still key.

"The accuracy of XRF now is incredible, as long as you know what you're looking for," says Fell. "One of the important facts to understand about XRF is that it normalizes the result to 100 percent. What that means is, if the material has 50 percent gold, 25 percent silver, 15 percent copper, and 10 percent zinc, and you don't select zinc as an element to look for, that 10 percent will be spread out to the other elements, resulting in higher readings and an overpayment to the customer. Conversely, if elements are entered that are not there, the XRF may assign a value, albeit a very small one. We have to use our knowledge of the material and those readings to determine if the numbers generated are in fact the results of those elements being present or if it is electronic noise.

"XRF is only as good as the technician doing it, and the calibration put into it," says Fell. "Deal with a refiner that you trust."

Although the process involves melting down the customer's lot before analysis, it is considered a non-destructive method because the sample that is used for testing is not destroyed. Refiners can test the entire metal bar either by taking readings of several locations on the bar or of metal shavings taken from several spots on the bar. It's important to run multiple tests a variety of locations because the from X-rays don't penetrate deeply into metal. "It only analyzes the surface," says Kelly, which makes getting a homogenous mix when melting down the lot critical.

"XRF is only as good as the technician doing it, and the calibration put into it. Deal with a refiner that you trust." - Larry Fell

"If the top of the bar is 50 percent gold, and the bottom of the bar is 62 percent gold, we know there's a problem" with the melt, explains Fell. "Until the top and bottom come out the same, we can't rely on the sample. We may need to oxidize something out, add copper, or do something else to get a homogenous bar. Our whole goal is to get a homogenous mix."

Another shortcoming of this method is that it shouldn't be used on sweeps. "Sweeps are low-grade metal bearing trash," explains Kelly. "The combustibles are slowly incinerated, then ball milled to a fine powder." That powder will feature a combination of metallic and non-metallic material that will make it difficult to get a homogenous mix. "In settling, metallic (heavier) material tends to migrate toward the bottom of the sample, whereas nonmetallic material will tend to stay near the surface. XRF measures only the surface area, so the settling is a problem." Instead, the powder should be analyzed with a different assaying method, such as Inductively Coupled Plasma Spectrometry (ICP) or gravimetric wet chemical analysis.

Depending on the refiner, many customer lots are settled based on the XRF results, with no other analysis done. This allows refiners to settle with their customers quickly, without the added costs to run additional tests.

"We do XRF on small lots to keep costs down," says Fell. "We like to use XRF for anything under 15 ounces. If we don't get good correlation or we see elements that we know may cause problems, we will back up the XRF with a fire assay." He notes that it can be expensive to do fire assay in both time and materials, citing an example of a 15 ounce lot they recently tested using both methods. The difference between the amount of metal detected by XRF and fire assay amounted to $5.

"As an engineer," continues Fell, "I want to get the most accurate result I can, but it is not worth spending $75-$150 to get to a parts per billion reading on a lot that is worth $5,000 and the difference in accuracy will only vary the payment by $5 one way or the other. When we have a 1,000 ounce lot, we will obviously do more extensive testing.

"Also, in California, we're under an obligation to reduce hazardous wastes [such as some of the chemicals used in analysis]. If we can eliminate the waste, we're doing our duty. We do a lot of testing of XRF to compare the results to fire assay to make sure we're getting good numbers. If we do only XRF, we will melt batches of the lots together and then fire assay to verify that the total content matches the total of the individual contents."

Fire Assay

The oldest of the assaying methods, fire assay is widely regarded as the most accurate technique for assaying gold (although refiners also use it for assaying silver). While it can be used to assay platinum, it's an expensive process, and many refiners have a minimum amount of platinum required for testing using this method. "The rule of thumb is that if there's less than an ounce of platinum, you're probably not going to see it because refiners typically can't efficiently remove low amounts of platinum from gold," says Ballard. "This is why you should always keep your platinum scrap separate."

To begin, a lot is turned to molten metal, and a glass tube is inserted into the blend; the metal is sucked up to create the pin sample used for testing. As with XRF, a homogenous mix of the materials is necessary to ensure that the sample contains a representative mix of the metals. (Also as with XRF, some refiners may instead take shavings of the final melt rather than pin samples of the metal blend.)

The pin sample is then cut into several pieces to ensure it is solid. "Sometimes pins can have hollow spots, which could hold slag, resulting in inaccurate results," says Fell. A sample is weighed and then a known amount of fine silver may be added to aid in digestion. The metal is wrapped in lead foil and placed in a cupel (a small cup made from bone ash) before being placed in a furnace to burn out any of the base metals. "The lead acts as a carrier for the cupel to absorb the base metals," says Kelly. "This leaves only precious metals."

The oldest of the assaying methods, fire assay is widely regarded as the most accurate technique for assaying gold.

At this point, refiners will weigh the resulting metal bead or wafer before repeating the cupellation process to help remove any residual base metals that may be left. After another turn in the furnace, the metal is weighed again.

Refiners will then take that bead to a rolling mill to roll it down into a thin strip before placing the metal into a flask with nitric acid. "The thickness is critical," explains Mykhailo Guch, refining manager of Umicore Precious Metals in Toronto. He notes that although the acid will dissolve away all of the silver (and a small amount of platinum group metals), it can't go deep into the metal. "Nitric acid cannot dissolve something in the middle of a cornet. If you have silver stuck in the gold, it will stay behind. That's why we roll the sample super thin—it gives the acid as much metal surface as possible."

After the metal is digested by the acid, the acid is decanted and the gold is annealed before being weighed again and compared to the weight of the original sample. "If we started with a 100 mg sample, and the final weight is 50 mg, than the sample is 50 percent gold," explains Kelly.

To determine what metals comprise the rest of the original sample, the leftover fluid is then precipitated down first for platinum and then for palladium. "Once you get your gold content, platinum content, and palladium content, you subtract them from your total and the remaining amount is silver," says Kelly. (Alternatively, some refiners will also use ICP to test the liquid for other precious metals.) After subtracting out the silver that they initially added to the sample, refiners will then pay out for the remaining precious metals.

As a means of double-checking their procedures, many refiners will also run a known control sample alongside a test of a customer's lot. Using a sample of gold whose content has been certified, the refiner will test the sample at the same time as a customer's lot, following the same procedures for both to verify that the results are accurate. If a problem appears with the known sample, it's assumed the same problem will affect the customer sample. "If the standard sample is off, the customer sample is off, too," explains Guch. "For example, say the control sample is off by 0.02 percent. In order to be perfect, you would have to adjust the customer sample by the same percentage."

Because fire assay requires first burning out and then dissolving metals, it comes with a longer timetable than XRF. Whereas XRF may take two minutes to complete, fire assay can typically take around seven hours. However, fire assay offers better accuracy than XRF.

"With XRF, we probably get 0.1 percent, which is good," says Fell. "But fire assay is still the go-to when we're talking about gold and silver analysis. We look for gold numbers down to 0.02 percent."

Kelly confirms: "XRF is good for two nines [99.0]. Fire assay is good for three nines [99.9] when analyzing for pure gold, though some refiners may claim 99.99. If you want four or five nines [99.999], you have to go to ICP."

"XRF is good for two nines [99.0]. Fire assay is good for three nines [99.9]. If you want four or five nines [99.999], you have to go to ICP." - Ron Kelly

Inductively Coupled Plasma Spectrometry (ICP)

If fire assay is the go-to method for assaying gold and silver scrap lots, ICP is the preferred method for testing platinum and palladium lots. That's because "it's designed more to detect trace elements," says Fell. "For example, if you have some 95/5 platinum scrap, 5 percent of it will consist of impurities (iridium, cobalt, etc.). We can use ICP to determine what those impurities are, and then calculate the platinum by subtraction."

That benefit is also the reason why ICP is not particularly suited for mixed alloy scrap other than low-grade sweeps. "The more concentrated the sample is of mixed precious metals, the less accurate the ICP is," says Kelly. "You can tell it to analyze gold, silver, platinum, or palladium if you want it to, but the accuracy would be suspect because it's not designed for high concentrations in a mixed alloy."

Because ICP works only on solutions rather than solids, the pin or shaving samples taken from a customer's scrap are placed in a flask with aqua regia (which is a mix of nitric and hydrochloric acid) to dissolve. The dissolution process is time consuming and can take several hours, depending on the alloy. Once the sample has dissolved, the solution is put into a spectrometer, where a nebulizer sucks up the solution and discharges it as a mist into an argon-filled furnace chamber of the ICP. The chamber contains a plasma field (created by high-energy radio frequency), and as the mist disperses into the plasma, the spectrometer detects and measures the wavelength of the energy. The resulting spectrum reveals all of the elements present.

As with XRF, calibration of these machines is critical. To verify their results, some refiners will add a small amount of an internal standard to their flasks. "We add yttrium to every flask to see if there is a drift during analysis," says Guch. "You're giving the ICP a solution with a known content to analyze. It [the yttrium] doesn't affect the reading, it just stays there and you know the value." If the machine detects a different amount of yttrium than what was added, it will automatically make adjustments to the reading of the other elements based on that correction.

Taking between three to four hours for analysis, ICP offers an accuracy of +/- 0.001 percent, but it's limited in its scope of materials to analyze, since it is best suited for detecting trace elements.

Gravimetric Wet Chemical Analysis

Another method that is ideal for platinum group metal lots, gravimetric wet chemical analysis begins similarly to ICP in that the pin samples taken from the lot melt are dissolved in a beaker with aqua

regia. After the elements are in solution, the refiner will then precipitate each element one after the other, starting with gold before moving on to platinum and then palladium, using specific chemicals to selectively precipitate each element.

 "Even if gold isn't in the lot, we still do the step to be on the safe side," says Guch. As each element is precipitated, it takes the form of a metallic salt. The salt is filtered from the solution and placed into a furnace for an hour to burn down to a pure metal. That metal is weighed and then compared to the weight of the original sample to determine how much of the content it composed.

After all of the precious metals have been precipitated, they're placed back into solution, which undergoes ICP testing to detect other elements that may have been in the sample. If the ICP results show the presence of other metals, those amounts are deducted from the original sample weight and then a final weight of the precious metal content is calculated. If it all sounds very complicated, it's because it is, and it's one of the reasons why it is not the go-to method for refiners to assay most customers' lots.

"It takes weeks," says Guch. "You need to make sure the chemicals precipitate out all of the metal, and that you don't lose any of the metallic salts. We try to stay away from this method as much as we can."

"The more concentrated the sample is of mixed precious metals, the less accurate the ICP is." - Ron Kelly

Atomic Absorption Spectrometry (AAS)

Another rarely used assaying method is atomic absorption spectrometry (AAS). Similar to ICP, AAS relies on a spectrometer to analyze the energy of an element to detect its presence. However, whereas with ICP multiple elements can be detected at one time, AAS can test and run only one element at a time. That's because the machines operate using a series of bulbs, each one correlating to a different element.

There are now sophisticated systems available that have a setup featuring several bulbs, which can be rotated so elements are tested successively, speeding up the process. But these machines also come with heftier price tags, which limit their use. In addition, many refiners agree that the results from AAS tests are not accurate enough to be used for official assays so it is unlikely that they will be used to determine the value of a customer's lot.

As you can see, how refiners determine the value of a customer's lot isn't so mysterious— no tricks, just a whole lot of science and some technical details. But it's important to be able to discuss those details, so you can have a better understanding of what to expect. Then all you have to do is sit back and wait for the refining "magic" to work.