What Makes a Good Synthetics Detector?

RAPAPORT... The massive project by the Diamond Producers Association(DPA) and Signet Jewelers to test and review 18 synthetic-diamond detectors hasraised an important question: What should you look for in such a machine? Accuracy is, of course, a major part of the equation(alongside speed, price and usability), but it isn't a clear-cut issue. The firstset of results from the DPA and Signet's Assure Program, which came out lastweek, gave up to nine different performance scores for each scanner, reflectingtheir varied abilities. The criteria don't all apply to every machine, as only fiveof them can conclusively detect lab-grown diamonds, while six merely claim torefer questionable stones for further checks. The other seven are stillundergoing testing. Definitely natural The DPA believes the most important factor is the ability toverify that a diamond is natural, rather than being able to say that a stone isdefinitely synthetic or definitely a simulant. It calls this key metric the "diamondfalse-positive rate." The good news is that all but two devices scored 0%,indicating they didn't mistakenly label any man-made stones as natural. (See thelist at the end of this article.) "It's a reflection of the instrument's ability to ensurethat only natural diamonds are verified as natural diamonds," said Jean-MarcLieberherr, CEO of the DPA. "From a consumer perspective, this is the primaryfunction of a diamond-verification instrument." Consumers expect retailers to give formal reassurances thattheir products are what they say they are, and those are only as reliable asthe detection machines the retailer or supplier used, Lieberherr explained. Outof all the associated risks, the main one is that a synthetic diamond shouldmistakenly be sold as natural, he added. Refer: A friend? That measure is far from the only important one. A machinethat's overly strict, wrongly referring some natural diamonds for furthertesting, might be good if you want to be sure you're left with only naturalstones. But those additional tests may be expensive. You might need to buy morethan one machine. Plus, a buyer who receives a batch of natural diamonds and finds that someof them can't be verified may grow unnecessarily suspicious of his or hersupplier. That's why some traders may also look at the "diamondreferral rate," which is the percentage of natural stones that a machinecategorizes as undetermined ("refer"). Again, 0% is the best, with only oneinstrument - De Beers' DiamondView - achieving that. Some machines didn'tget a score, as they don't classify stones as "refer." Looking for lab-grown? But what if you're a synthetics company looking to ensureyour inventory comprises entirely lab-grown diamonds? The relevant measure mightbe the "synthetic-diamond false-positive rate," which assesses how oftenmachines wrongly labeled stones as lab-grown, when they were in fact natural diamondsor simulants. "The relative importance of the other metrics is up to theindividual device buyer, as there may be different motivations," Lieberherrnoted. "For example, some synthetic-diamond dealers may want to pull out allthe synthetics, and would care less about the referral rate." Sure thing Other measures give an idea of machines' more categoricalfindings. Only a few were perfect at identifying natural or synthetic diamondsconclusively. DiamondView was the only one that correctly spotted all thenatural diamonds, without mislabeling or referring any of them. The De Beersmachine was also one of only two that categorically identified all thelab-growns, the other being the Sherlock Holmes from Yehuda. (Presidium's SyntheticDiamond Screener II didn't get a score for that measure because it only refersstones, though the Assure Program's initial report wrongly gave it 100%, a DPAspokesperson confirmed. Rapaport News has corrected an earlierstory to reflect this.) However, most people who buy synthetics detectors are mainlyinterested in being sure that the stones they think are natural are indeed so.By that measure, most of the devices gave a good showing. How the testing worksThe Assure Program tested 18 machines from 11 manufacturersagainst a standard developed by UL, a third-party testing agency, and by atechnical committee of gemologists. Some of the committee members were from devicemanufacturers, including De Beers, the Gemological Institute of America (GIA),and the Antwerp-based Scientific and Technical Research Center for Diamond(WTOCD), which developed the M-Screen+ machine that HRD Antwerp sells. For thatreason, an independent adviser supervised the creation of that standard toensure impartiality, Lieberherr noted. All committee members approved thestandard and the sample of stones, and UL carried out the tests independently, headded. The sample comprised 1,000 natural diamonds and 200synthetics, including stones created by chemical vapor deposition (CVD) andHigh Pressure-High Temperature (HPHT), while some of the man-made specimens hadundergone post-growth treatments. All were round brilliant cuts, larger than 2millimeters in girdle diameter, D to J color, and of mixed clarity. Machinesthat claim to detect simulants such as cubic zirconia also underwent testing ona sample of 200 of those. The batch was intentionally challenging, and includedcustom-made synthetics that are not yet available commercially. That ensuredthe test was "future-proof" and able to differentiate effectively between theinstruments, the DPA explained. The DPA initially tested each stone with several gemologicallaboratories to confirm their origins. The natural and lab-grown diamonds alsohad slightly different sizes, enabling UL to verify machines' answers using sieves. "This system, proposed by the technical committee, has provento be 100% reliable," Lieberherr said. The DPA will release results of further tests in the comingmonths, covering new samples with additional sizes and colors, as well asstones mounted in jewelry. Machines that scored full marks: Diamond false-positive rate of 0%:(the percentage of synthetic diamonds wronglyclassified as natural) AMS2 (manufactured by De Beers) DiamondDect 3 (Taidiam Technology) DiamondSure (De Beers) DiamondView (De Beers) GIA iD100 (GIA) M-Screen+ (HRD Antwerp) Sherlock Holmes (Yehuda) SYNTHdetect (De Beers) Synthetic Diamond Screener II (Presidium) Diamond referral rate of 0%:(the percentage of natural diamonds referred forfurther testing) DiamondView (De Beers) Diamond accuracy rate of 100%:(the percentage of natural diamonds correctlycategorized as natural) DiamondView (De Beers) Image: Taidiam Technology's DiamondDect 5 - one of 18 machines undergoingtesting under the Assure Program. (DPA)