nanoHUB-U Principles of Nanobiosensors/Lecture 5.4: Concluding Thoughts ======================================== >> [Slide 1] Welcome back. This is the last lecture for the course. In this course, we have covered a lot of things and we will quickly review them. But I also wanted to take this opportunity, first of all, to say what are the emerging directions of research for nano biosensor and then also acknowledge a number of people who have helped me put together the course. So let me get started. Let's see what we have learned this semester. Actually, if you have been following along, we have covered a lot of material and we will see. [ Slide 2 ] So I will start with the summary of what we have learned and then we'll be talking about looking ahead that what are the applications of biosensors. In some way, we have just an analogy with the transistor. We are in the 1950's where the transistor had just been invented and integrated circuit has yet to come. We are just beginning to think about integrated circuits. So the potential is remarkable and we say significant. It's going to be a significant event. And then I will conclude but my main point of the conclusion is geometry is physical and I'll explain what I mean and the importance of geometry in nanotechnology in general. [ Slide 3 ] All right. Do you remember why the motivation we had and understand why in thinking about nanobiosensors, the idea was that if you take a human being then by the time the disease is detected, most of the time, this disease show up in certain organs and people say they have, somebody has lung cancer or a colon cancer or something like that associated with an organ. But of course, before this got expressed in the organ itself and often very difficult to treat this especially for lung cancer, let's say, long before that something got broken, either in the intracellular level or in the molecular level or in the genetic level, at some level something got broken. And if we had a highly-sensitive sensor that could work with very small and a light volume, well, that would be great. And here is that one day, we'll be able to essentially map out, just like a Google map of the streets in high resolution the map of the protein network associated with our body for every -- and everybody is different. Everybody's system is slightly different and as a result, biosensors, nanobiosensors promise that future when one would be able to have a complete map of the protein network associated with the human being. Now, on that -- towards that goal, we have already sort of beginning to make progress. I already told you that small molecule detection, glucose, nitric oxide, this type of thing, were already significant development. DNA, genome sequencing, a lot of beautiful work especially if you heard the last two lectures on genome sequencing. That's really amazing, that amount of progress that has been made over the last 10 years. >From a billion-dollar genome sequencing, now you can do in a couple of hundred dollars. That's nothing, no technology has evolved faster. And then we are beginning to understand the protein network for many diseases. For example, obesity and others, people are beginning to unravel the corresponding protein networks using biosensors. And of course, the foreign invaders like virus and bacteria, it is always an important thing to check for especially when new strains of virus or bacteria come around that for which we may not have effective antibiotic. So significant applications, there are lots of applications and that inspired us to develop this course and to think about nanobiosensors. [ Slide 4 ] Now, I gave you a short history. I told you that just like in biology, people started first by learning about virus and bacteria because these are relatively big structures. Bacteria, you can essentially see under a microscope. Virus, you cannot see, of course. And then, of course, people, as more and more sophisticated microscope came along, people could see protein and DNA structure and of course, could amplify them. PCR is polymerase chain reaction. By now, you know, right? In the selectivity lecture, we discussed the polymerase chain reaction and also in the genome sequence. And so, bottom line is the technology has evolved. Biotechnology has evolved very, very rapidly. And at the same time, I explained that electronics have evolved rapidly also. Vacuum tube at the same time as the virus was detected. Around the same time CMOS integrated circuit, they have come along and en route, there had been this merging where people have used different types of-- of various types of sensors based on electronics to detect the corresponding type of molecules of interest from biology. So this has a long history and so the nanobiosensors that we just talked about in the last five weeks or so, is a culmination of a century-long effort. And of course, the work goes on. This potentiometric based, MOSFET based sensors, ion-sensitive field effect transistors we told you that how to make super Nerstian response sensors, how to use droplet to beat the diffusion limit and how to use Flexure-FET to enhance sensitivity. So there's a continuous work on the sensor technology itself. [ Slide 5 ] Okay, so the point I wanted to make in this course and I hope that I have done so repeatedly is that no matter how complex the sensor is, what the method of transaction is, you can always -- you should always go back to the fundamentals and think about three things. Settling time, how long does it take for the molecules to arrive at the sensor surface. No matter what technology, this is defined, the fundamental limit that if there is no molecule, you cannot detect it no matter how sensitive your sensor is. In the settling time, we emphasize the role of geometry. That was the key point. Once the molecule has landed on the sensor surface, then of course, it matters where there's a potentiometric sensor, whether amperometric or a cantilever-based sensor. And in that case, we discussed the physics in detail and of course, in that case, you want as few biomolecules as possible to initiate the response, robust response that you can detect. And as I told you, the third thing one has to worry about is selectivity and I'll give you some examples. But the bottom line is that unless your sensor is selective, this doesn't matter. Technology doesn't matter. Sensitivity doesn't matter. Selectivity is the first and foremost thing and we talked about energetic-based selectivity, selectivity arising from space, and selectivity associated with noise. Noise associated in the transducer. [ Slide 6 ] So let's talk very briefly just as a reminder that what is the essential point of the settling time and I said the settling time, if you think about how molecules diffuse in the solvent or in the fluid volume then you can arrange all the sensors that you know starting from the planar sensors all the way to the bio-barcode sensors. You can arrange them all in a very natural way and that is sort of like a mendeleev principle. It allows you to arrange the sensors in a way that it can analyze them at a later point. And the formula turned out to be remarkably simple, that the minimum amount of analyte that you can detect at a given time ts is given by the fractal dimension associated with the sensor about how the biomolecule swims towards the surface and the diffusion coefficient associated with the biomolecule. Larger molecules go slow, smaller molecule move faster. There's nothing complicated. And the only time that was sort of specific to particular sensors was this N naught? How many molecules do you need in order to initiate a robust response? [ Slide 7 ] Now, we've very quickly found out that although sensitivity is significant using nanobiosensors but often at ultra-low concentration, classical biosensors may not really work completely because it's very difficult to go below a few tenths of femtomolar. And so we say that there are various very simple strategies. One is that this tau, the time for the biomolecule to arrive is square goes to the square of the dimension of the fluidic volume divided by the diffusion coefficient. So one approach is to make L small by fragmenting the space so you catch the biomolecule by throwing a lot of sensors at it. You catch a thief by putting a lot of policemen in the city and thereby sort of trap the element of interest in a smaller box. Inefficient, yes. It costs a lot of money. That's also true but if you want fast response, it would be very, very fast response. The other approach was to reduce the space, make L small and the droplet's evaporation-based technique that we talked about, that sort of bring everything down to the sensor, pull everything down to the sensor and the median fold decrease in the volume within 10 minutes, it's possible. And that allowed us to detect by-- the diffusion limit, detect the biomolecules relatively quickly. And the third one is the genome sequencers I talked about, you know, in the last lecture where on the bid, you have the local generation and it is generated so close to the sensor that L is very small and moreover, here the proton essentially come in and diffuses to the sensor surface. As a result, D is extremely small also because D is inversely proportional to A. A is the size of the molecule and proton is very small, of course. So therefore, high D and very small tau. Milliseconds, you will get the signature on whether there is a DNA is present or not. So all this beat the fundamental limit and I'm sure, if you think about it carefully, you yourself will come up with different approaches how you can beat the diffusion limit. All right, the second topic that we discussed was -- [ Slide 8 ] these different types of sensors, potentiometric, amperometric and cantilever. I tried to explain the physics clearly. These things are not always so easy to explain in the sense because you need some background associated with semiconductor device physics or in this case, let's say, the chemistry, electrochemistry. So but if you assume, you have to accept some results [inaudible] given but once you accept that, I think the rest of the things follow relatively quickly. So I tried to explain how individual sensors work, how many molecules n naught that you need before a response is initiated. [ Slide 9 ] Now, over here again, we saw that there are fundamental limits. For the potentiometric sensors, what is the fundamental limit? I hope you would say that salt is both necessary and also is a very bad thing because salt screens. So this is like a Mafia which sort of gives you protection but then takes a little, a lot of money out of you. So in that case, this is necessary, both necessary but essentially it takes a lot of charge away leaving very small amount behind for the sensors. So what are the approaches we discussed to beat this detrimental effect of salt and pH? Well, one was to use high frequency so that the salt molecules cannot move and cannot screen the charges. And it's like a fast break in basketball so therefore you can quickly sort of sense the charge before the other molecules have a chance to rearrange themselves. And the other approach I said that use a large sensor for biomolecule detection in parallel with the small sensor and this nano and nanoplate combination under certain circumstances can also give you significantly enhanced sensitivity. Okay, so those are about potentiometric sensors, how to make potentiometric sensors better. What about amperometric sensor? Amperometric sensors again, a couple of different things I told you about how to make it better. First, one is instead of having a planar, why not use a spherical sensor? Here, for example, these small cubes act as the sensor surface and that allows molecules to come from all sides, ultra fast diffusion and therefore by nanotechnology, you can make an amperometric sensor better. What was the other approach? The other approach is essentially to cycle back and forth that the molecule comes in, there is a -- it gets oxidized and then there is an enabling electrode where it gets reduced. And every time it sort of cycles back and forth, back and forth, this produces one electron. So previously, we're in classical structure, every atom biomolecule would produce one electron and one electron only. Here, every biomolecule may produce 400 to 500 electrons. So that's the amplification and only nanotechnology because this gap is about 50 nanometer, it allows you this type of response to this space of one figuration. And so this was another corresponding figure where you had two electrodes there. And finally, what about cantilever sensors? How do you improve the sensitivity of cantilever sensors? Well, first of all, you make them small. Look at this. One micron in the width, two or three micron, three or four micron, let's say, on the length, 25 nanometer in thickness. This is what's called a nanoscale sensor. And therefore, you can now weigh a virus molecule. Never before in history, people could actually weigh a virus and here it is. You can not only see them but you can also measure them with ease. What is the other thing, the other way you can make it even better? Well, the other approach is this Flexure-FET where you couple the transistor technology, borrow things from here, potentiometric sensor and then couple with the cantilever to get the highest performance sensor that has been reported today. So lots of different ways, no matter which branch of sensor you are coming from. If you are a chemist, you may be working here. A physicist may be working on the potentiometric sensors. Depending on where you are coming from, no matter, significant work is being done. And hopefully, after this course is over, you will keep track of the recent literature towards this goal. [ Slide 10 ] Topic three, we talked about selectivity, three types of selectivity. You see, when the two DNA molecule comes in, comes together and this is something we'll also discuss in the homework problems that they are supposed to bind perfectly, right, [inaudible]. But many times, even when they are not perfect bind, an incorrect molecule can still bind locally, right? And so therefore, although there are some broken bonds, a potentiometric sensor doesn't really know that there is a broken bond, right? It just measures the amount of charge and the amount of charge is the same in every case. So therefore, this incorrect binding is a false-positive and we say that that's a big selectivity problem. The other selectivity problem was that there is no matter what you do, 54% of the space is covered by volume of molecules and remaining 46% is empty where parasitic molecule can come in, unless you cover it with something else. And this is again, a big selectivity problem. And finally, the noise I told you about, white noise, pink noise, those noise are also -- give rise to selectivity concerns. And so once you know the fundamentals of this, you can calculate these quantities alpha and beta. Quantify the false-positive, false-negative associated with the sensor technology and can decide then whether you want to go for -- go forward with the technology or not. But those people, either in the startup companies or in the earlier stage of research, those who don't ask these questions will almost certainly, will eventually not be successful because this is such a fundamental physics issue that is very difficult to get around it unless you start things, start thinking about it from the very beginning. Bottom line is that you say the noise, we can get larger than nanoscale, many of these noises, if it just gets larger than nanoscale and therefore nanoscale really sort of the selectivity issues get more accentuated at nanoscale, something to think about. [ Slide 11 ] So how do you improve selectivity? Once again, I told you about remarkable research that are going on. For example, you can have a new type of DNA. This is a polynucleic acid which doesn't have the backbone charge. You can have linked nucleic acid where the backbone has been stiffened and therefore the selection, the capture is more selective. We talked about how to incubate longer so that you have less amount of space that are unfilled and then you passivate the rest of the structure. It's like putting carpet on or grass on so that to prevent sprouting of various types of parasites. So in that case, you want to cover the surface, otherwise, there will be big selectivity issue. And the most important way of how I said that improve the signal-to-noise ratio is to average over the noise. And this is something that you cannot do in television or in radio because instantaneous signal-to-noise ratio is important. In biosensors, it's unique because signal changes in the second [inaudible] and yet the electronics allows you to sample a thousand times before the signal changes. And this repeated sampling once you average it out, signal goes up and down, noise stays flat. Signal-to-noise ratio can become significantly better. This is an advantage that nanosensors have that classical electronics may not. [ Slide 12 ] but the point I just want to make and those important class we didn't discuss is the optical sensors. There are various types. One type is called ELISA method then there is a SERS-based approach where it wants the biomolecule, surface-enhanced Raman scattering. So once the biomolecule comes in then if you bounce off a wave, electromagnetic wave then once the molecule comes in, it increases the thickness. It changes the optical path and from there, you can essentially see whether you have captured a biomolecule or not. So it's the optical way of doing it. People have been also been doing this like in nanosphere, in little sphere-based sensor and the biomolecules are coming in so you bounce a laser through that little sphere and you see what will happen, that there will be a whispering gallery mode. And once the biomolecule comes in, the same wave will repeatedly sample the biomolecule so you can have significant enhancement in the signal. And then there is, of course, this ring-based or [inaudible] based approach. Once again, the signal comes in, goes out and when the biomolecule comes in and gets attached, it changes the associated phase and can be very sensitive. This is a resonator-based approach. This may look very new to you. But you see, all you have to do is to go and look at the corresponding physics associated only for the sensitivity. For selectivity and for settling time, you don't have to do a single thing. Whatever we have discussed in this course, that is the only thing necessary for all future sensors that will be coming in contact with. For example, let me show you how it works. [ Slide 13 ] For example, I just took this picture, took this figure from a paper where once the biomolecule has landed, there is a corresponding phase shift and therefore change in the wavelength associated with the input, incoming signal index and it turns out it depends on what are the index was before the target molecule came and what was it after. And then since it's getting thicker and thicker, there's a corresponding change in the thickness of the molecule. But you see, it is getting thicker because biomolecules are gradually coming and landing on the sensor surface. We know how to think about that, right? That's settling time. So all you have to do is to express D of T in terms of the number of molecules they have. But the number of molecules I know, if it is two-dimensional sensor or a three-dimensional sensor, it should go as T linearly with time, put that thing here and you will immediately see the result is beautifully reproduced. So therefore, although we have no idea about the sensor, it's just there is something I just introduced, yet the essential feature of it, you can now explain in a few minutes. [ Slide 14 ] All right, so what should we be thinking ahead or looking ahead, what should you be expecting down the road? [ Slide 15 ] First of all, genome sequencing as I said is one of the sort of the seminal achievement of the last few years. It is as if that we have made a hundred years' worth of progress in computing, essentially in 10 years of similar progress in genome sequencing, very impressive. And the whole reason this could be done is because there were millions, hundreds of millions of sensors working in parallel, taking advantage of nanotechnology. Without nanobiosensors, there would be no more genome sequencing, it is as simple as that. [ Slide 16 ] So this is something that will continue. The other thing that will continue is what we will call the lab on a phone which is to use embed sensors on the cell phone platform technology, maybe one would have diabetes sensor and then if blood pressure sensors all embedded communicating directly with the doctor. So in that case, the nanotechnology, biotechnology and wireless technologies, they will converge. It is already beginning to happen. You can buy sort of inserts from other various companies which will do this. It's expensive but the price hopefully will come down. And once it is networked, all these sensors are networked and networked with the doctor then I think, healthcare is going to change fundamentally. So this is a promise, very near-time promise. I'd be very surprised if it doesn't happen within next five years in the large scale. [ Slide 17 ] And -- but further down, these sensors, they are getting smaller, people are going to put them on very small electronics. They said these are smart bandages type of structures that you can put down on the skin and after a while, it can automatically sort of wash away, go away once this is no longer necessary or if there is an infection, there is a continuous monitoring of the infection associated with the underlying wound, for example. So these things involves flexible electronics in the sensor platform, again integrated completely with the human body. [ Slide 18 ] Finally, I'll just give you one example before I move on. This is a type of stent. These are [inaudible] stents. Where previously the stent would go in, it was a mechanical, simply a mechanical thing. Now it is loaded with electronics so when the doctor goes in, if there is any extra blood or extra type of damage, the doctor would immediately know and then it will have a localized response to any damage that is done. So these electronic, these sensors are all embedded and we expect more of this type of embedding down the road especially for implanted electronics. [ Slide 19 ] So let me end with one of the most interesting direction that is going on. You know, many people let's say has epilepsy or various types of diseases. Sensor is one thing. Sensor simply sees something. But seeing and doing is not the same thing. Sensor simply says, it can tell you where there is some neural activity. There is something going on and it can measure the pH. It can measure the protein and whatever. It can measure various types of markers. But what is most interesting that once you have the information, what do you do with it? And that is the next wave in the sensor technology. Of course, you have to interpret it. You can compute it by computers. You can interpret what you are seeing but the most important is that then you can go back and actuate it. So therefore, you can provide a new signal that will hold the seizure or epilepsy and that type of thing. And that is, this is for the brain but that could happen for any of the other organs also so this would be a new direction for sensor technology that would be going on. [ Slide 20 ] And the most interesting thing that has been happening recently is called optogenetics regarding this sensing interpreting and actuation then the optogenetics, what it does essentially is that it embeds various types of proteins within the neurons and these proteins are photosensitive and therefore once there is a signal of something unusual happening, a laser can photo excite the corresponding neurons and thereby control the brain response to it. And this has been shown to have quite a bit of promise. So closing this loop between sensing and actuation will bring in a whole set of new promise for this field. [ Slide 21 ] All right, finally, on a very big level or a macro level, one thing I want to emphasize is that what we learned in this course, fundamentally is geometry is important. [ Slide 22 ] For example, we said that -- we started from planer sensors, the nano sensors, but you can immediately realize that fundamentally nature has been using this type of sensing technology for a long time. For example, the jellyfish collects it's food that way. It sits still and the food comes to it. And by branching properly it can essentially catch more food than if it ran around searching for the food. So, therefore, essentially it has the same approach, and similarly the way blood essentially picks up oxygen in the shape of the blood molecules essentially is exactly similar physics associated with a nanosphere biosensors. So in all these cases, sensing is fundamentally enabled by a new geometry. The origin and material is the same, right, but this geometry allows this new -- new things to -- new capabilities to emerge. Now Berg in 1963 originally pointed this out on this biomimetic importance of the biomimetic devices. [ Slide 23 ] But of course the history is much older from 1970 and onward. People had been noticing that how nature uses geometry fundamental to solve very complex transport problem. It doesn't use drift, nature most of the time doesn't have a big magnet to work with, or it doesn't have a big pump to work with. What it works with at the end of the day is just a -- different types of geometry to solve various transport problems in its environment. [ Slide 24 ] So finally, what I told you about biosensors is-- are very general as other technologies are emerging. Large scale, like solar cells, displays, flexible electronics. So I focused on biosensors, I told you the physics of it. But whether you go to energy harvesting, or you go to flexible electronics, in all those cases people are now using geometry, new types of materials which are geometrically complex to solve a broad range of problems. So things that you have learned in biosensors, you will find it very useful if you go in other areas and want to explore these other devices, you will see many of the things that you have learned in this course. You'll be able to carry it to -- forward to understanding other devices. [ Slide 25 ] All right, so let me conclude then. So we discussed the physics of biosensors. Hopefully you understood that nanobiosensors is a excusably the sensitive analytical tool, but this is not new. This really -- nano is new, that's like 15, 20 years old, but the physics that led to this sensors is at least a hundred years old. The second is that many of the sensors, the next wave, is not necessarily in the sensor technology. Many of the sensors are already super sensitive. Their problem is selectivity and their problem is then being able to integrate it with, for example, things like smartphone platform, how do you integrate them, and make sure that this is a consumer electronics not a FDA product, because then it will require sort of approval from a different agencies, which may make things more complicated. So embedding them in a smartphone platform may have a broad technological implications. Now I wanted to emphasize that nano is different, everybody knows that. Nano -- everybody is interested in nano technologies, but not because of any quantum effect or anything, but primarily because of the geometrical effect. Geometry is fundamental in this enhanced operation of these devices. Then I wanted to emphasize that biology is all about biosensing. You see, when the two proteins come about, recognize each other, then they essentially enable some process actuates some other molecules. This is essentially all about biosensing. So we have a lot of insights from the biosensors from this course, and hopefully in the process we'll be able to give something back to biology itself because we have sort of learned a lot from the -- from biological concepts. So before I conclude, I want to acknowledge a few people. [ Slide 26 ] For example, the Settling Time work that I discussed in this course comes from Pradeep Nair. We worked many years together, Aida and Piyush Dak, about beating the diffusion limit, you know, this droplet based sensing. Cantilever-based sensing is based on the work by Ankit Jain. Potentiometric sensors, Jonghyun Go, spent many years on this, and also the genome sequencing. So it's a lot of work. And homework and biosensorlab, Xin helped me out and Sambit. But my colleagues, I've been working many years with my colleagues, they are -- have excellent experimental insights into biosensing. And of course none of the work you could do without the funding from a broad range of -- a broad range of sources. The bottom line is that hopefully in this course you have learned a set of new concepts. Many of them, some of them at least, may not be completely clear, but it's possible you go back, take a look, and then I hope that you will enjoy the learning process, because then you can use it for a broad range of other technologies and other concepts.