Counterfeit semiconductor devices entering the Department of Defense (DOD) supply chain continue to pose risks to many of our country’s most sophisticated aerospace and defense systems. However, considerable progress is being made by the DOD, industry and academia in developing approaches to help eliminate these devices before they are put into critical systems.

Supply and Demand

Electrical testing of a counterfeit IC. One of the surest ways to detect counterfeit IC’s is to perform full electrical testing.
A short review of how we have arrived at this point in the counterfeit battle will help put these new approaches in perspective. As with any market in the world, there must be both supply and a demand for counterfeit devices. In this case the majority of the supply comes from an unintended source – electronic waste. An unfortunate side effect of the effort to recycle the millions of tons of obsolete e-waste generated each year is that it has created a nearly unlimited supply of very inexpensive semiconductors that can be easily removed from e-waste printed circuit boards and cosmetically refurbished, remarked and resold as authentic material.

The demand side of this market exists because many of the DOD’s most sophisticated systems have product lifetimes that far exceed the lifetime of a typical commercial semiconductor device. Defense systems can have useful lifetimes of 30 years or more and can be prohibitively expensive to redesign and requalify. Meanwhile, commercial semiconductor production is dominated by devices intended for the consumer electronics market, where lifetimes can be as short as 3-5 years. With a nearly unlimited and inexpensive supply of obsolete semiconductor devices in the world and a DOD long term demand for these same obsolete semiconductors, the counterfeiters will find a way to bring that supply and demand into balance.

Table 1. Initial Counterfeit Mechanical Screening Techniques
At this stage in the battle with counterfeiters, the defense industry, their suppliers and leading test labs have become adept at screening out the relatively unsophisticated counterfeit devices that have made up the majority of the initial wave of counterfeits washing over our supply chain. These initial screening techniques tend to be mechanically oriented tests but have been effective at identifying these relatively unsophisticated counterfeits (Table 1).

As good as these techniques have been at catching the initial waves of counterfeits, the counterfeiters continue to develop more refined and difficult to detect counterfeits. In particular, the counterfeiter’s ability to remark devices has become so good that it can be impossible to differentiate them from an authentic device.

Today’s more sophisticated counterfeits are taking advantage of the fact that many semiconductor devices come from families of devices derived from the identical die design. For instance, a single microprocessor design will typically yield a range of performance characteristics as a natural consequence of the variability of the underlying wafer fabrication process upon which it is built. The manufacturer will test these devices and “grade” them according to their performance characteristics such as speed and functionality over temperature.

The parts are marked according to this grading process and sold into different applications for different prices. For instance, a slower speed device could be sold to a consumer electronics manufacturer for five dollars, while a high speed device may be sold to an Air Force contractor for a high performance application for twenty-five dollars. All that is needed to create a very difficult to detect counterfeit is to re-mark the poorer performing device as the higher performing one. Re-marking, as stated above, is one process the counterfeiters are getting very good at. Since both of these parts have the exact same die in them and come in the same package, the traditional counterfeit detection techniques of Table 1 will not be able detect them.

The only way to reliably screen these more sophisticated counterfeits is to perform full electrical testing across the device’s entire specified temperature range. This is the same type of testing the original manufacturer performed and will be able to verify the “grading” marked on the device. Table 2 lists some of these more sophisticated counterfeit types that require full electrical testing to detect.

Table 2. Counterfeit Types Requiring Full Electrical Test Screening
The second entry in Table 2, where a commercial temperature range device is re-marked as a military temperature device, is particularly problematic. In this circumstance, the device will perform all of its intended functions at full rated speed across the narrower commercial temperature ranges – just not at the wider military temperature range. Making detection of these counterfeit types even more difficult is that there is normally only a one-character difference in the way the device is marked. Unless full electrical testing across the entire military temperature range is performed, this counterfeit type might not be discovered until it’s in the field attempting to operate at full load and/or in extreme temperature conditions.

Clones and Trojans

Another particularly difficult to detect counterfeit device type from Table 2 is the “Clone.” A cloned device is one that is actually redesigned and remanufactured using todays semiconductor technology in order to meet the original manufacturer’s performance specifications. By using current semiconductor technology, these cloned devices often perform much faster and with more device operating margin than the original device. The only way to catch these devices is to perform full electrical testing over the entire operating temperature range, as discussed previously, but in this case take the testing one step further. Instead of only testing to make sure it at least meets its performance specs, the actual device performance must be measured to see if it statistically exceeds its performance specification. This actual performance testing will determine if the device is ‘too good” and is, therefore, a potential counterfeit.

When discussing clones, the question is inevitably asked that if a cloned device has even better electrical performance than the original device, why can’t it just be used in place of the original? The answer is that while it is relatively easy to redesign a simple device to electrically function like an old one, it is much more difficult to make that new device with the same quality and reliability as the original manufacturer. Remember that the counterfeiters are out to make money by doing as little as possible to get someone to buy parts and are not likely to employ the same rigorous military qualification processes over the entire operating temperature range that the original manufacturer did.

Counterfeit Example. Identically marked parts contain different die revisions. They both function the same, but the counterfeit consumes much more power.
Another important category of counterfeit devices is the “Trojan.” A Trojan is generally defined as the introduction of malicious hardware or software into a device such that it can be activated by some external event or after a predetermined amount of time.

Trojans, if designed well, can be extremely difficult to detect. Today, there are no commercially available ways to assure your device doesn’t contain a Trojan. On the other hand, a good Trojan is even more difficult to engineer than a Clone, so it is less likely that the older, common industry devices (that make up the bulk of today’s counterfeit devices) would include a Trojan.

But as the counterfeit device supply issue moves from the hands of relatively unsophisticated e-waste recyclers and into the hands of establishments or governments with malicious intent, the resources and desire to create Trojans increases significantly. This is compounded by the fact that most new semiconductor fabrication is done outside the United States where designs and mask sets could be more easily compromised to introduce Trojans.

One solution to limiting the Trojan threat is to use the DOD Defense Microelectronics Activity (DMEA) “Trusted” source program. This program accredits US based semiconductor manufacturing locations that have proven high level security processes and procedures in place. Numerous other solutions are being studied by the DOD and academia, but so far none of the published approaches can be considered a universally viable solution.

There is some good news in the industry when it comes to identifying and eliminating counterfeit semiconductor threats. There is a growing awareness in the industry with an abundance of technical conferences and papers proposing solutions. Academic institutions are studying techniques to guarantee newly fabricated devices can be authenticated. Finally the DOD is sponsoring defense industry research programs in numerous counterfeit related areas.

One such program is sponsored by the Missile Defense Agency (MDA) and targets the development of new and more effective counterfeit screening techniques for Field Programmable Gate Arrays (FPGA’s). FPGA’s are used extensively in the defense industry because of their flexibility and wide range of available functions. They are often the “brains” of the system they are designed into, where failures caused by counterfeit devices could lead to catastrophic consequences.

Another is DARPA’s Supply Chain Hardware Integrity for Electronics Defense (SHIELD) program. The goal of DARPA’s SHIELD program is to eliminate counterfeit integrated circuits from the electronics supply chain by making counterfeiting too complex and time-consuming to be cost-effective. SHIELD aims to combine NSA-level encryption, sensors, near-field power and communications into a microscopic-scale chip capable of being inserted into the packaging of an integrated circuit.

Although the industry is doing much more to thwart the incursion of counterfeit semiconductors into the DOD supply chain than ever before, there is no realistic end in sight. Continued diligence is required in employing the initially developed mechanical counterfeit detection techniques as well as the even more effective comprehensive electrical test techniques. The counterfeiters will be relentless – it will take the combined efforts of the DOD, academia, the original semiconductor manufacturers and independent test labs to ultimately bring this threat under control.

This article was written by Joseph L Holt, Vice President, Integra Technologies LLC (Wichita, KS). For more information, Click Here .

Aerospace & Defense Technology Magazine

This article first appeared in the December, 2015 issue of Aerospace & Defense Technology Magazine.

Read more articles from this issue here.

Read more articles from the archives here.