QTC (Quantum Tunneling Composite) made an appearance on the flashlight scene several years back and was met with an initial flurry of interest. Since then, interest has dropped to a steady but much lower level. As a then-novel material, QTC appeared to offer many benefits, yet it quickly became apparent that it came with a unique set of challenges as well. In this post, I aim to coherently compile a body of information on using QTC in flashlights. The first sections are mostly background information; they can be skipped if you only care about QTC’s application in flashlights. The application sections will only concern QTC in “pill” form rather than as an FSR, and since I do not yet have any experience building drivers, I will limit this post to using QTC by itself as direct load control rather than as an integral part of a driver. Finally, I have flashlight build examples using QTC, which (once they are tweaked and cleaned up for best performance) I will post as updates to this thread in a few days time.
It appears that QTC is no longer particularly interesting in the flashlight community. After playing around with it and testing it, I can understand why, as it has some peculiar properties and is not necessarily the easiest material to work with. Furthermore, the effort put into flashlight drivers by people on this forum is nothing short of incredible, and thanks to them we now have readily available, highly customizable drivers for almost any project – therefore lessening the need to work with a material like QTC. I have a very personal interest in QTC however, for two reasons. First, flashlights with genuinely variable brightness fascinate me – the very first flashlight I purchased as an adult was a Jetbeam Jet-I Pro v3 which had Jetbeam’s IBS (Infinite brightness system). Second, QTC is just too cool of a material to not play around with 
On to the details (there’s a summary section at the top of the APPLICATION section, for those who just want the essentials). Brace yourselves, it’s gonna be a long one 
WHAT IS IT, AND WHY SHOULD I CARE?
For those who haven't seen it before, this is what QTC looks like (image courtesy of mindsetsonline.co.uk):
And here's what it does in a flashlight:
This is an L3 Illumination L10 that I modified. The Nichia 219a is replaced by a 219c on a copper sinkpad, the driver has been removed and replaced with a copper heatsink, and the QTC acts as the "driver." Twisting the head turns it on at a firefly mode and further twisting continues to ramp up the brightness -- up to an estimated 800 lumens OTF (using my integrating sphere). I plan on fine tuning it further (and replacing my 26awg wire with 20awg) and am confident it will reach >900 lumens, possibly over 1000, on a freshly charged 14500 
. A build log will follow later.
So what exactly is QTC?
Quantum Tunneling Composite is a metal/polymer composite which acts as a piezoresistor, changing its resistance with applied force. With no applied force, it acts as a near-perfect insulator, but its resistance drops exponentially as force is applied until it becomes a near-perfect (<1Ω) conductor. In “pill” form, QTC is 3-4mm square, 1mm in height. Since the most basic method of driving an LED is with a series resistor setting the current, QTC promises a potentially simple, robust mechanical method of controlling a flashlight’s output.
HISTORY
Quantum Tunneling Composite was discovered in 1996 by David Lussey. That same year, he founded the UK-based company Peratech, which holds the IP relating to QTC and continues to do active R&D, especially concerning QTC’s application in force sensing and HMI (human machine interfaces). Earlier this decade there was a spate of new uses for QTC as it became available in pill, cable, sheet, ink, and granule form. Of these, the pill form was by far the most widely available for the average consumer, and as near as I can tell remains the only elastomeric form of QTC still available today. However, elastomeric QTC and QTC pills are now considered legacy products and are no longer supported by Peratech. It would appear (someone can correct me if I’m wrong) that the QTC pills available today are manufactured by another company that licenses it from Peratech.
HOW IT WORKS
Traditional piezoresistors and conductive rubbers work by percolation. Conductive particles (often carbon) are suspended in an elastomeric polymer in a (more or less) random arrangement. As the ratio of carbon to filler increases in the manufacturing process, more and more particles find themselves in contact with one another, and they begin to form conductive “pathways” through the material. Because of this, piezoresistors that function by percolation have a resistance even when uncompressed of only around a few kOhms. As the material is compressed, the average distance between particles shrinks, causing more and more of them to come into contact with one another and further decreasing the resistance. Generally speaking, maximum compression results in a resistance of a few Ohms, and percolative piezoresistors exhibit a linear relationship between force and resistance. Finally, since conduction occurs via physical contact between particles, the metal to polymer ratio must remain fairly low to ensure an adequately resistive uncompressed state, which means also that conductive rubbers generally cannot carry a high current, making them less suitable for direct load control.
QTC stands in stark contrast to all this. Nickel particles with a spiky surface are mixed into an elastomeric filler. The single biggest difference between QTC and percolative conductive rubbers is that the nickel particles in QTC are fully wetted – i.e. each individual particle is completely covered by the rubber. This means that contact never occurs between the particles and percolation does not occur, since there exists an insulating barrier between each particle which is treated as insurmountable in classical physics Quantum mechanics, on the other hand, treats electrons like a wave, and describes a process (quantum tunneling) whereby it is possible for an electron to pass through the insulation.
In an uncompressed state, the lack of contact between nickel particles means that QTC is resistive on the order of >1012 Ohms – several orders of magnitude higher than conductive rubber operating via percolation. When placed in a circuit and compressed, highly localized electrostatic charges build up on the tips of the nickel “spikes” and reach significant voltages. Electron tunneling becomes increasingly possible as the electrostatic charges build up and the distance between nickel particles decreases. This is a type of field assisted emission called Fowler Nordheim tunneling – electrons tunneling through an insulating barrier because of the presence of extreme electrostatic fields. This mechanism means that the resistance of the material decreases not linearly but exponentially with force – resulting in a final resistivity of <1Ohm rather than several Ohms. Furthermore, operating by tunneling means that QTC’s resistance decreases not only with compression, but also extension and torsion. For our application, however, we are concerned primarily with compression.
PROPERTIES
This section will only concern the properties of QTC in pill form.
Basic Info from Datasheet
Mechanical Properties
Width: 3.6-4.0mm
Height: 1mm
Weight: 0.04g
Force Range: 0 N – 100 N
Lifetime: > 1,000,000 compressions
Operating Temp: -20°C to 120°C
Humidity Range: 0% – 100%
Electrical Properties
Unstrained Resistivity: > 7 x 1012 Ohm cm
Typical Resistance Range: > 1012 Ohms to < 1 Ohm
Operating Voltage: 0 - 40 V
Max Current: 10 A
Max Voltage: 40V
Unstrained Dielectric Constant: (1 KHz) 23.5
Okay, so here’s where the information relevant to application begins. I’m going to break this into further sections for easy reading.
Mechanical Durability
The mechanical durability of QTC is a common concern, but it need not be a barrier to its use in flashlights. There are two (three) significant sub-concerns that have to do with durability.
Torsion
This is the single biggest hurdle in using QTC. The obvious way to ensure smooth, linear axial compression of QTC is to make use of flashlight threads – either those connecting the body to head, or those connecting the tail to the body. Unfortunately, QTC has almost no mechanical resistance to torsion, and so implementations in which QTC is simply placed between a battery and a contact result in a drastically shortened lifetime on the order of only a few dozen compressions. Under torsion, QTC tears and breaks apart, resulting in inconsistent and eventually nonexistent ramping. This problem can be completely solved very easily, however, with any number of possible solutions that ensure the QTC only experiences axial loading and not torsional.
Over-compression
A secondary issue is that of over-compression. QTC can handle up to 80% compression without noticeable decay – beyond that however, it usually splits and is rendered non-viable. At 70% compression, QTC reaches a resistance around 0.17 Ohms, and provided that other sources of parasitic resistance in the flashlight are minimized or eliminated, this is plenty low enough to ensure performance almost identical to a direct drive setup. In my experience, most applications will not require more than about 50% compression. Over-compression can be prevented by building in a stop. It is also possible to integrate QTC into a flashlight in such a way that an over-compression stop also bypasses the QTC entirely and shorts the battery to the LED, which provides direct drive performance and avoids over-compression. Of course, in a setup like that, care must be taken to ensure that the battery and LED are well matched so that the LED isn’t burned out or the battery destroyed. A 14500 IMR is fine with most 3V LEDs provided there is adequate heatsinking, but an 18650 IMR runs the risk of destroying the LED.
(Pinching)
A tertiary issue is that of pinching. As QTC is compressed, its footprint also expands slightly. Some holder designs that ensure QTC does not experience torsional load or over-compression can run into the issue where this expansion gets caught between contacts and is pinched off – again resulting in the same performance issues as torsion and over-compression.
When these durability concerns are addressed, the lifetime of the QTC pill will far exceed the lifetime of the flashlight.
VOCs
QTC is fairly sensitive to VOCs (Volatile Organic Compounds). Interestingly, this property created an early spike of interest in using QTC as an “electronic nose.” For our applications, however, it means that using glue to anchor QTC is generally a poor idea. Apart from the fact that most glues adhere poorly to QTC, they also often change its properties. For instance, QTC will absorb the fumes from CA glue (superglue) and form a hard “skin” on its surface – greatly decreasing its conductivity as well as its compressibility and making it non-viable for flashlight applications. If CA glue (or any other glues that release fumes) are used in constructing the holder for QTC, it is advisable to wait 24 hours for the CA fumes to fully dissipate before adding the QTC to the apparatus.
Hysteresis & History Dependence
QTC displays considerable hysteresis. In other words, it has a tendency to respond slowly to certain changes. At a constant compression and resistivity, the I-V curve of QTC is non-Ohmic and displays a hysteresis loop. The shape of this curve depends heavily on the QTC’s resistance – but in general, increasing voltage results in a slow initial increase in current followed by an exponential increase and then a decay to zero. Following this state, decreasing voltage increases current very slowly until a large exponential spike again (at a much lower voltage than the current increase alongside increasing voltage) followed by a decrease to zero when voltage reaches zero.
For our applications, this hysteresis loop is not particularly important, since we are dealing with relatively constant (and generally low) voltages. More of concern is a second type of hysteresis which shows itself as variance in resistance with a constant voltage. If we start from an uncompressed state and compress QTC by 10% which results in initial resistance of say, 100 Ohms, we can expect the resistance to drop further over time even though no further force is applied. This property has been noted by flashlight enthusiasts and described as QTC “settling,” and it manifests itself as a slightly increasing brightness. It is suggested that this hysteresis occurs because the localized electrostatic fields generated at the nickel spike tips are close enough to interfere with one another, and it takes time for the optimal “pathways” through the material to appear, as current flowing on some pathways can close off or open up other pathways. Generally speaking, the resistance decreases very little after 30 seconds. This resistance change can be up to 10% of the initial resistance, and is usually only visible at a lower brightness. Edit 7/28/16: Further informal testing suggests that the change in resistance over time with constant compression may be larger than initially expected. I will run more tests to be sure and will update the OP again when I have more definitive conclusions. This mechanism also appears to be responsible for the phenomenon where QTC’s resistance slightly increases with compression before again decreasing. In practice, this hysteresis is only a major issue for moonlight modes and for runtime tests at a constant brightness. Edit 7/28/16: Some further testing indicated a higher level of hysteresis, which would be an issue for low and medium modes as well. Even so, it can be controlled so it is not a (significant) issue for moonlight modes.
QTC also displays mechanical hysteresis. It can take up to 20-30 seconds for it to return from >70% compression to an uncompressed state. What this means for flashlight applications is that when going from a high brightness level to a lower one, it can take additional time for the QTC to fully expand and consequently the flashlight will become noticeably brighter. It can also be an issue for turning off the flashlight – because QTC takes time to reach an uncompressed state, it is possible to go from a high mode to no light and then have the flashlight “turn back on” because the QTC re-initialized contact as it expanded to its uncompressed state. This issue does not need to be a significant one, however – good design can ensure that accidental turn-on does not occur.
All of this means that it is generally not possible to exactly match a set brightness with a set compression. The resistance of QTC at a given compression depends on a number of factors – most importantly, its previous state of compression (this is called history dependence). In practice, I have found this to be only a minor annoyance – the beauty of an infinitely variable system is precisely in the lack of modes and the ability to instantly increase or decrease brightness.
BEHAVIOR IN A CIRCUIT
As direct load control, QTC acts as a variable resistor. Placed in series with the battery and load (the LED), it burns off any voltage above the LED’s vF and delivers a constant current. Since it’s rated to 400W (10A at 40W), joule heating is insignificant. This means that QTC’s efficiency is below a good step-up/down switching regulator (boost/buck driver), but is essentially equivalent to linear drivers like those driven by 7135 chips. The difference is that since QTC does not rely on PWM for dimming like a 7135 based driver does, QTC is as efficient at any current draw as a 7135 is at 350mAh. It does not behave identically, however – since it’s essentially a resistor in series, current is not constant for as long as a 7135. As voltage begins to drop close to the vF of the LED, the current draw decreases. This can be compensated for quite easily by dropping the QTC’s resistance through further compression, however (incidentally, this is more or less what a 7135 chip does, only more precisely – change the value of an internal sense resistor to maintain constant current). Runtime graphs for QTC driven lights will resemble constant current lights, although they will begin to look more and more like the graph for a directly driven light at higher amperages. One concern here is low voltage protection, LVP. While it would be possible to add a simple driver with a voltage cutoff, in practice it is not needed – the vF of the LED serves as LVP. Using a Nichia 219C, for instance, provides a hard cutoff at 2.7 volts, but by 3.0 volts there is a noticeable decrease in available brightness at high levels of compression, and a significant decrease by 2.8 volts. So long as the LED vF remains conveniently close to the safe voltage cutoff for li-ion batteries, there is no risk of over-discharge.
Obviously, since it is a glorified resistor-in-series, QTC by itself cannot be used to drive an LED with a single NiMh battery. Likewise, it becomes less efficient with greater difference between the vF of the LED and the voltage of the battery. Using two li-ions in series with QTC to drive a 3V emitter would be horribly inefficient. With a single battery and single LED, however, efficiency is between 75%-95%, depending on the actual battery voltage as well as parasitic resistance in the circuit. With a single li-ion and a 3V emitter, QTC offers advantages over a 7135-based driver, insofar as it matches at all current draws the 7135’s efficiency at 100%, and does not rely on PWM for dimming. On the other hand, there are benefits to discrete modes over against an infinitely variable brightness system, and 7135s have the advantage of maintaining a better regulated current (although in practice the difference is not usually significant).
It seems to me that it would be possible to use QTC on the input side of a good boost/buck driver to achieve ideal efficiency and variable brightness. I’m not knowledgeable enough about circuits and circuit design to say that with any authority or confidence, though. I hope to run some tests soon on one of my flashlights with a boost/buck driver that is known to be very efficient (in the 90% range), and I’ll report back with results.
APPLICATION
If you’ve skipped the previous sections, here are the highlights:
Summary of Benefits
- Infinitely variable brightness from so-dim-you-don’t-even-know-it’s-on to direct-drive level performance
- Good efficiency (comparable to a current-controlled linear driver) at all brightness levels
- More efficient low modes than a 7135-controlled linear driver
- No PWM
- Can be combined in parallel to achieve amperages higher than 10A
- Can be combined in series to decrease the rate at which resistance changes with compression
- Can be cut in half (or punched through the center) to decrease the force required for compression
- Mechanically and electrically simple
- Has a lifetime of >1,000,000 compressions; very durable when not mistreated
Summary of Drawbacks & Design Challenges
- Not as efficient as a well-designed boost/buck driver
- Sensitive to torsion, over-compression, and pinching
- Noticeable hysteresis which makes obtaining an exact brightness difficult
- Direct load control is limited to applications where battery voltage and LED vF are well matched (i.e. single li-ion with 3V LED or 2S li-ion with 6V LED)
- Any significant parasitic resistance in the circuit lowers overall available brightness as well as circuit efficiency (n.b. also a problem for other driver types)
Addressing Design Challenges
Torsion
Let’s talk about the big issue first. There are probably dozens of different ways to prevent rotational forces from being applied to the QTC. What most (all?) of them have in common is that the QTC is sandwiched between two conductive surfaces which are prevented from rotating with respect to the QTC, although they may rotate with respect to other elements in the flashlight. I’ll mention here just a couple designs or design ideas that work. In larger flashlights with more space to work with, the best design seems to be one where two plates, composed of a conductive center and non-conductive ring, ride freely on two or more posts. In a design like this, the posts can be anchored to one of the two plates and the secondary plate can ride freely. The range of motion can be limited (addressing the over-compression issue) with stops on one or more of the posts. Another design relies on the same plate design, but instead of fixed posts, the plates are attached to one another with double-sided adhesive foam with a hole cut out in the center for the QTC. Depending on the foam, this design can be viable – but low-quality foam can still transfer some rotational force to the QTC and there are lifetime concerns as well. A third design utilizes a small key (not a household key; the machine element sort) and two plates with a keyseat. A fourth design makes use of a thicker non-conductive plate with a square bore that accepts the QTC and two conductive plates on either side – the height of the plate slightly less than the conductor-QTC-conductor sandwich so as to provide a set range of movement. This last design is pictured below (this was an earlier design for the L10).
Over-compression
From the same design pictured above, we can see one mechanism for preventing over-compression. In the below photo, you can see that one of the copper plates has a small circular disk soldered to it, and the QTC has a hole cut to match. This disk is approximately .45mm in height and the QTC 1mm, which means that no more than 65% compression can be achieved.
Pinching
Pinching can be eliminated simply by ensuring proper tolerances – the QTC should be given room to expand slightly (~0.1-0.2mm).
Hysteresis
There are two issues here that can easily be corrected – the first is the difficulty in obtaining a consistent firefly/moonlight mode due to brightness increasing as the QTC “settles”, and the second is the fact that the pill needs several seconds to recover from high levels of compression, making it difficult to obtain low modes immediately after a high mode. The first issue can be solved by designing a holder which slightly pre-compresses the QTC pill, providing a suitable firefly mode. This pre-compression can also be achieved by elements not part of the QTC holder – for instance, a small bit of foam or spring providing a small amount of force on the QTC before battery contact is made. The foam or spring needs to be weak or at a low compression, however, since only a small amount of force needs to be exerted.
The second issue, which concerns the mechanical hysteresis of QTC, is not solved without some compromise (fortunately, I do not find it a significant problem in practice). Rapid expansion of the QTC material can be ensured through the use of conductive copper foil tape. This tape consists of copper foil and a conductive acrylic adhesive. The easiest way to implement it is to solder squares of the tape to two copper plates on either side of the QTC and then attach the adhesive sides to QTC. The copper plates are then mounted in a holder with an integrated compression spring. So long as the soldering is done quickly, the acrylic adhesive remains strong and will adhere acceptably to the QTC. The problems with this design are significant for high power applications, however: the acrylic adhesive is not as conductive as copper, and (depending on the quality of the tape used) it has enough parasitic resistance to limit the achievable brightness to less than direct-drive performance. For instance, the tape I tried had a resistivity of .05 Ohm/in2, and it is necessary to use two layers of it. It is possible that 3M’s copper/nickel tape, which has a resistivity of only .004 Ohm/in2, would have a low enough parasitic resistance to avoid a significant performance hit (but it is very expensive). I may try this in the future when I have more available funds. An additional issue with this design is that because of the presence of the compression spring in the holder, it requires additional force to compress and is therefore not suitable for flashlights with smaller diameter bodies.
Thread Anodization & Accidental Turn-On
The simplest way of implementing QTC is in a flashlight with un-anodized threads – simply drop the QTC and its holder in the bottom of the battery tube and go. This is the design I went with for my L3 Illuminations L10. The drawbacks are that the threads are not as durable, and it can be difficult to ensure that the flashlight is actually off, since there is no lock-out. However, QTC can also easily be implemented in flashlights with anodized threads. In this case, it is usually easier to place the QTC and holder in the flashlight head rather than the tail. The contact plate of the holder that faces the battery tube simply needs a conductive ring on its edge in addition to the conductive center, and that ring needs to be connected to ground. A design like this allows for physical lock-out as well as the benefits associated with anodized threads. Additionally, it bypasses the issue of battery compression, which is discussed below.
Battery Compression
Many implementations of QTC, especially in flashlights with un-anodized threads, rely on transmitting the force provided by the flashlight threads through the body of the battery. Consequently, denting the negative end of the battery can be an issue. This is mitigated by making the negative battery contact larger – the design pictured above, for instance, has a large enough contact surface (16mm2) that pressure on the battery is minimized and I have had no issues with battery deformation. An 18650 battery may require a larger contact surface to avoid denting. There is also danger of deformation if the user overtightens the flashlight threads – this is prevented in the pictured design by matching the plate width to the battery width so that the force is distributed widely. Additionally, building in a hard stop to prevent QTC over-compression provides tactile (and usually also visual, in the form of a slight jump in brightness, if the stop doubles as a bypass) feedback that lets the user know further compression is superfluous. In practice, I have found that even without a hard stop, battery denting is a non-issue – the amount of force required at high levels of compression, especially on a small diameter flashlight, provides good feedback to the user that further compression is not advisable.
Of course, as noted above, battery compression can be avoided entirely by designing a QTC holder that relies only on the body of the flashlight for the transmission of force.
Battery Rattle
Battery rattle is a problem in many twisty lights. The L3 Illumination L10 solves this with a ring of foam around the positive contact – the same solution works to prevent battery rattle in a QTC enabled light. It is important, though, to choose a foam (or spring) that is easily compressible so as to not unnecessarily compound the force required to compress the QTC.
Drops
In a QTC light, sudden drops often cause the brightness to drop out or spike. A well designed holder minimizes this, but it cannot be avoided entirely. It occurs for two reasons – first, because the slop in the threads of the flashlight allows for some movement. This can never be completely eliminated, but can be improved by adding a single layer of copper or aluminum foil to the threads (this is not a particularly robust solution, however) or by choosing hosts that have well machined threads with little slop. Some slop is unavoidable, however. The second reason is that designs which rely on transmitting force through the battery are necessarily less stable insofar as the battery is a loose element, often loaded by springs, which can move independently from the body of the flashlight. In practice the brightness spike caused by drops is usually not significant with a well-designed system.
Starting on Low
Most implementations of QTC will start on low and ramp up. Since some users prefer their flashlights to start on high or medium, this is not ideal. By including an additional (side or tail) switch, the flashlight can be set to a preferred brightness via turning the head, and then turned off. QTC lights need not start on low but can start on any brightness the user requires.
WHERE TO BUY
Since QTC is manufactured in the UK, sourcing it outside of the EU can be difficult. Fortunately, it is not impossible. For large orders directly from the UK, QTC can be obtained here from Mindsets or here from Technobots. These two sites provide the lowest per-pill costs, but shipping outside the EU can be expensive. Even so, for larger orders over about 20 pills (such as a BLF group buy) this is the most cost effective solution. QTC can be found on Ebay as well for about $10 for three pills or $20 for ten pills – obviously, a much more expensive per-pill solution. The cheaper option for USA based customers is to order here from Versa Robotics. The per-pill price of $1.55 is still an almost 300% markup from the Mindsets price, but for smaller orders this is the most economical solution. I am sorry, I am not aware of any retailers that offer low per-pill pricing and/or inexpensive shipping options outside the US and EU. The good news is that the UK based suppliers appear to ship internationally for a flat fee for lightweight orders – so it should be no more expensive for someone from another country (for example, Australia) to place an order than it is for someone from the USA.
--------------
And that's pretty much it. If you've made it this far, you're either very interested in QTC or a brave soul
Hopefully this serves as a decent reference text for anyone looking to use QTC in a flashlight. In the next week or so, I will post builds of the L10 and Preon 1, which will hopefully pique the interest of some.
--FrontPorchCarver
Sources:
http://scitation.aip.org/content/aip/journal/apl/88/10/10.1063/1.2183359
http://www.nanolab.unimore.it/en/wp-content/uploads/2013/06/qtc_-background_reading_EN1.pdf