is a nichey opportunity. Different kinds of harvesting suit environments with different energy reservoirs subject to the constraints of the application and attached device (e.g., energy levels required/available, the conversion efficiency at that energy, and allowable area/volume/mass for the harvesting technology). If your device is buried inside a wall, in a closed shipping box, or embedded in the human body, for example, there is no light available, and then perhaps, if the expense is warranted, thermal conversion (on a warm/hot steampipe, for example) is used [13], or mechanical harvesting (if there’s a lot of vibration, ideally at a higher frequency as in an engine) [14], and RF harvesting of beamed or ambient radio with a backscatter tag can work for very low power applications [15]. Within the body, power is generally proximately inductively coupled from transmitting coils on the skin, but ongoing work looks to harvest small amounts through adaptive RF beamforming [16], vibrational scavenging via inertial reaction and induced strain (including on the heart—an idea with a long legacy [17] but only now nearing implementation [18]) and using the body’s own energy transport mechanisms via implantable biofuel cells [19]. But in most places where people spend time, there is light by default, as we need it to see, and here photovoltaics are often a top choice if energy harvesting is mandated.
Solar cells offer a harvesting mechanism that has been applied across many orders of magnitude of scale and plant capacity, from urban power systems to spacecraft to mm-scale “smart dust” [20] stacked IC dies gestating in research labs that glean power from a top silicon layer with a micro embedded solar array that also doubles as a coarse video sensor [21]. Photoelectric conversion technology has developed enormously since Einstein won the Nobel Prize in part for explaining it a century ago, thus playing his role (to his purported later chagrin) in ushering in the dawn of quantum mechanics. They have gone through decades of laborious development, enabled through (and likewise inspiring) advances in materials, power conversion and energy storage. Indeed, they will play a major role in renewable energy, together with safe, economical nuclear power and within a few decades fusion reactors designed either by humans or the AIs that will augment us by then. The market and societal need for large-scale solar power is now clear, and has driven the technology and manufacturing into a variety of performance achievements with different form factors while lowering the cost. Thanks to Elon Musk, one notable player out of many in this field, I can now tile my roof with aesthetically-appealing solar shingles and power my Tesla electric car significantly off-grid, even with the available solar illumination up here in New England. The world is now lacking only in political will to buck economic inertia and transition to an emission-free/renewable energy basis in which the capture and harnessing of incident solar photons will be a major part.
That, of course, is the topic of many other books. But the broad, recent expansion of photovoltaics and associated research has also driven the lower end, making light conversion increasingly feasible for low-power embedded applications. Already as a child growing up in the 1960s, I delighted in buying the cheap selenium solar cells they used to sell at the Radio Shack (the dominant electronics component chain store in the US that enabled generations of fellow DIY-ers before it recently went out of business) and hooking them up to small motors, etc., to see them work with a little bit of daylight. With decades of subsequent improvement, solar cells became cheaper and better—certainly at the microwatt level, if you could afford the surface area for some cm’s worth of cell area, they have powered simple LCD calculators with indoor light since the 1980s, and watches have even hidden tiny solar arrays in their face—we were proud back then to show off these versions of common gadgets that never needed batteries replaced. But they stayed nichey, as in those days there wasn’t much you could do when constrained to microwatts.
Moving to recent times, the long-anticipated explosion of wireless devices in our living environments has taken hold, and simultaneously the amount of function you can achieve at lower power has expanded. Processing, sensing, and even wireless communication have exploited advances in microelectronics fabrication/design, smart power management, and frameworks/protocols to do much more with fewer ergs, while energy conversion and dynamic optimal load matching has become much more efficient, even at the microwatt level and for a variety of different source impedances. This increased function per microwatt has boosted what can be done with a little bit of exposed PV cell, and applications have followed. Already in the mid-1990s we designed and deployed scores of smart IR beacons for indoor location at the MIT Media Lab that ran off light from proximate florescent bulbs [22]. During the last decade, we have run several of our embedded wireless sensing systems from small PV cells (e.g., our Tidmarsh environmental sensing node powers from a tiny solar cell atop its package; a few hours of diurnal daylight or a bit longer in overcast skies will serve to keep it powered [23]), and for low duty-cycle applications, we have built similar embedded light-powered sensors that work indoors.
This is no longer unusual, and various PV-powered wireless sensors have appeared for years now, some by commercial vendors, like the Schneider HOMES temperature/humidity/CO2 sensor developed by our former collaborators at Schneider Research in Grenoble a decade ago. But nonetheless, although the outdoor market for low-power wireless PV devices is robust (witness, for example, commonplace solar-powered outdoor lighting), few indoor PV-powered sensors have achieved widespread commercial viability. This is mainly due to the fact that as power decreases to the point where energy harvesting becomes an option, an embedded battery can last potentially many years, often approaching either the anticipated device lifetime, the battery’s shelf life, or just lasting long enough to make replacement a limited irritation (witness how we tolerated batteries in smoke alarms over the past decades), now often announcing over a wireless link when batteries need to be replaced. Batteries pose the ultimate dilemma to energy harvesting, as they are just so cheap and convenient. Plus, they don’t run the risk of failing in darkness, and can deliver brief pulses of high current (e.g., a commodity wireless IoT sensor bought today can draw a couple of amperes from a pair of AA cells for a few seconds when first joining an 802.11 network).
Things are changing now on many fronts, and this situation will evolve. For one, indoor photovoltaics are themselves improving. Although the efficiency of real high-end silicon solar modules in bright sunlight can significantly exceed 20%, recombination and spectral mismatch effects in the low light conditions we find indoors can cut efficiencies at least in half [24]. Indeed, the efficient world of LED and fluorescent lighting that we live in now emits at wavelengths less friendly to canonical silicon, and partly in response, ongoing research in photovoltaics has developed many approaches to converting light efficiently across broader spectral ranges. Exploiting these techniques to exhibit better performance at spectra now commonly exhibited indoors, as well as shifting away from straight silicon to new materials like perovskite, GaAs, and Cd compounds, hint at raising this efficiency by a factor of 2 to 3, albeit introducing some complexity in toxicity, manufacturing, etc.; the following chapters in this book explore this in detail. Similarly, needed power levels are still decreasing, not only for processing and sensing, but also for joining “heavier” mainstream networks like 801.11 or 5G, not to mention continued evolution in their low-power brethren like the now ubiquitous BLE and the multitude of other short-range contenders (e.g., ZigBee, Z-Wave, RF-backscatter, etc.).
The sheer number of wireless devices in the environments we live and work in will also continue increasing; at some point, this will cross a level at which battery replacement is less tolerable, together with a profusion of sensors in hard-to-access areas that still have low but usable levels of light (e.g., for environmental control, material integrity, leakage and structural assessment, safety, etc.). Also, as lighting has become more efficient and networked, and cheap renewable energy drops basic power costs, one possibility would be for areas of dwellings to illuminate for an interval when people aren’t present just to charge the sensors; they will tell the smart home or IoT servers when they are running low and need an illumination hit, or just raise the shades to let enough daylight in. Long-range inductive charging is a potential competitor here [25], but efficiency loss with range, orientation, and still unacceptably large EMI are show stoppers here, together with the lack of infrastructure; our homes and offices are amply lit, but don’t yet (and probably won’t) have large induction coils in the floors and walls (on the other hand, I see Helmholtz-coil-laden closets, draws, and shelves as potential ways to conveniently charge wireless electronics embedded in clothing).
