Misconceptions are common. Everyone has preconceived notions of what people are like; words like “jock”, “nerd”, “cheerleader” will bring certain stereotypes to mind. The same goes for professions— “doctor” and “politician” probably call up different images. Buzzwords, like “Genetically modified organisms”, “fracking”, and “NSA” also come with their own baggage. When people learn that I am a chemist working in nanotechnology their imaginations race and they ask me questions as if I were from some TV show that airs late at night on SyFy. I sometimes get asked, “Hey, are you making grey goo?” Grey Goo is an apocalyptic sci-fi term used to signify self-replicating nanomachines that will copy themselves indefinitely, until the world is a mass of undulating “grey goo” of nanorobots. Eric Drexler coined the term in his book, Engines of Creation, in 1986, which grasped the imagination of sci-fi writers over the years. So I thought about it. The more I thought about it (and read about it on conspiracy theory websites), the more I realized not only is it improbable, I began to believe that it didn’t make sense at all.
Abundance of Food
The first, and most apparent, fallacy that replicating machines can transform the world into more copies of itself is the abundance of starting materials, or “food” for the nanomachines. To put it simply, a nanorobot has to “eat” atoms or molecules that it can turn into itself. If it eats food made of elements that it is not composed of, it won’t be able replicate. For example, humans have to eat food with the elements oxygen, hydrogen, phosphorous, nitrogen, sulfur, and carbon for the most part (trace metals like iron, nickel, and cobalt, etc. are also necessary for many life forms). But if you were to go to the beach and stuff your face full of sand (silica and oxygen) your cells would not be able to replicate because there would not be the correct starting elements or compounds. So what are current nanomachines made of that scientist actually play with in a lab? Many are organic based, like DNA-based machines (carbon, oxygen, hydrogen, nitrogen, and phosphorus), while others have metals in them (gold, iron, cadmium, etc.). DNA is a good replicator (we’ll get to that later in the article) but it requires other molecules (such as proteins) and compartments (a cell) to replicate efficiently. Many nanomachines have metals in them. Metals are more scarce than most organic molecules, so the total amount of replication they can undergo is even less–especially if it is an uncommon metal, like platinum or palladium.
If I wanted to be rigorous about why replication is hard (which I do), the conversation would ultimately turn to thermodynamics. I’ll keep it as simple as I can and I’ll use pictures to keep it fun. Let’s say you want to go from Grey Goo to more Grey Goo. To do this you need food, and lets assume we have a lot of it. Now we need to turn food into Grey Goo, but without magic—that’s ok, because we can use something called energy. But energy can’t be created or destroyed—it is conserved. What if to go from food to Grey Goo, we need to input energy (breaking the bonds between atoms holding ‘Food’ together, and reorganizing them into ‘Grey Goo’)? In this case Grey Goo is in a higher energy state compared to food (Figure 2). Then Grey Goo couldn’t replicate effectively, and unless it had an infinite energy source, we would be relatively safe from mass extinction.
But let’s look at the flip side—what if Food and Grey Goo had similar relative energies, or even if Grey Goo was at a lower relative energy to Food (Figure 3). In this case, energy is released when going from Food to Grey Goo. This is actually what animals do when they eat. We convert energy from the bonds in glucose (a sugar) to energy (and water and carbon dioxide). With this energy we can do all sorts of activities, like covert chemical compounds in our food into other compounds useful to us, so we can grow.
In this case, the energy released by turning food into Grey Goo would help Grey Goo find more food, and turn it into more Grey Goo. Because Grey Goo is at lower energy than the starting material, food, chemists would say that food would “spontaneously” turn into Grey Goo. Counter-intuitively, “spontaneous” is a loaded word that is full of its own misconceptions, which I will get to in the next section. But which scenario is likely? What are the chances that Grey Goo is at a higher relative energy that of food, or at a lower one? A general answer is that some compounds will be at a higher relative energy to Grey Goo, and some lower. For instance, humans can’t eat graphite (pencil “lead”) to get carbon because it is one of the lowest energy states of carbon—we get our carbon from relatively higher energy sources like carbohydrates and sugars.
Let’s imagine, however, that we live in a universe where our Grey Goo has lots of food, and that the thermodynamics works out, and that it is conceivable that it is energetically favorable for the nanobots to replicate. Just because a process is thermodynamically possible, doesn’t mean that it will happen on any meaningful timescale. If you look closely at figures 2 and 3, you will see a funky red line connecting the black lines that represent the relative energies of Food and Grey Goo. This line represents the energy of the chemical compound at every step as it changes from Food to Grey Goo. At the top of the red arc, it is above the relative energy for both Food and Grey Goo. What does this physically mean? In short, as you are transitioning from Food to Grey Goo by breaking bonds and reorganizing the atoms, you are creating structures that don’t normally exist in everyday life because they are “high energy” and therefore “unstable” as a large amount of energy is needed to stabilize them. This transition state will tend to collapse into a lower energy state, in this case, Food or Grey Goo. However, the transition state exists for every reaction, and cannot be avoided. To perform a reaction, you need to input a certain amount of energy, called and activation energy to reach the transition state (Figure 4).
So if a reaction is called “spontaneous”, is there still an activation energy required? There is, which makes “spontaneous” a loaded word. A spontaneous reaction can happen in milliseconds, minutes, weeks, years, or millennia. An everyday example is the conversion of diamonds to graphite. Imagine the horror of buying a beautiful 2-carat diamond ring for your loved one, only to find that when you work up the nerve to pop the question it has turned into a black lump of graphite, and your romantic overtures are immediately rebuffed. Your initial reaction may be, “that would never happen,” but diamond spontaneously converts to graphite releasing 2,900 joules per 12 grams of carbon [McQuarrie and Simon, Physical Chemistry, pg. 1083]. The problem is that a “spontaneous” process may need an enormous activation energy to convert its crystal structure from diamond to graphite. This tranition can be observed at around 2000 Kelvin (1726.85 °C) where the activatio energy is only 188 kJ/mol [Butenko et al, J. Appl. Phys., 2000]. But at normal temperatures, (25°C), the activation energy is so high that this process takes millions of years [Zhumdahl, Chemical Principles, pg. 829]. It is likely that if you have a nanobot that can “spontaneously” convert raw material into new nanobots, you and possibly your greatgrandchildren will be staring at it for a long, long time before your petri dish or test tube turns to goo.
Actual Replicating Machines
In the end of the day, a lot of scientists look to nature for inspiration. The most successful replicating machine is DNA, which is the central molecule of life on earth. Nature has had about 3.5 billion years on to play with life since it started, and DNA has been the most robust replicator of them all. Nanoscience, on the other hand, has been going for a few decades. While it would be a great accomplishment for me to make something that could rival DNA in duplicating abilities, it would be hubris for me to think that my colleagues or I could do such a thing in our careers. DNA has had a remarkable run–it has turned brown landmasses green by over running them with plants, turned the atmosphere from mostly nitrogen and carbon dioxide to 21% oxygen, and filled a vast ocean with colorful rotifers and plankton and later sharks and whales. For the Grey Goo apocalypse vision to appear, what ever this nanorobot is, it will have to outcompete DNA, which is a tall order. Even scientists who work with DNA everyday would not be able to make a super DNA machine—DNA works best in a cell with hundreds of proteins and other molecules to help it thrive.
In spite of popular science and fantasy, there are many factors that make a Grey Goo doomsday unreal. The Goo would never have enough food, or a food source made of molecules that are easily digestible. And finally, even if God were to send manna from the heavens to feed these nano-replicators, there is no guarantee that the machines would replicate quickly, or at least more quickly than we could destroy them. David is a graduate student in chemistry in the Alivisatos Group. He studies DNA mediated self-assembly of nanoparticles, because it’s interesting, but also because it sounds awesome. He enjoys hiking, archery, and origami (both with paper and with DNA).