 Let's we start with the demonstrations and For that we will have some slides After that we will organize groups This you are ready. You can start. Thank you All right, so this is going to be the hands-on part of the talk. I have some slides I'm going to go through them quickly just so that Everyone can have as much time as they want to look at the actual devices But I want to explain just briefly what each of them does so The first device I have is a plasmonic reader. I could talk forever about this because it's my project And I have some slides coming up that have this in more detail. It reads plasmonic sensors Localized surface plasmon resonance. This is something I think there's going to be a Lecture in a lab later this week about it's a very exciting project We have a normal transmission based microscope. This one is the most fun to play with I have a bunch of samples people can try out different samples. It's a very simple design I'll speak to the design in a second the third device is a mobile phone based Eliza reader This is one of my favorite projects in the lab because Eliza is a test that's done all over the world it's very common in clinics and labs everywhere and This is a device. I think can really have a lot of impact So we'd love to talk with you about that and I have some slides detailing this more and then The fourth and fifth device are the giardia analyzer and the DNA analyzer That I talked about in the presentation Before the break so the first I'd like to just briefly go over what the plasmonic reader is So that's that's this device with the dome on top and this reads plasmonic sensors Plasmonic sensors are A emerging type of sensor for detecting and sensing protein concentrations There's work recently that's been published that actually can get a virus count in a given blood sample an unprocessed blood sample So the way localized surface plasmon resonance works is it create there on this sensor is a Nanostructured grading there's nano hole array on the sensor and these nanostructures support plasmon resonances Which are collective electron oscillations When a molecule or something of some sort of substantial weight Something larger than an ion but something around the size of a protein molecule kill adult in size is adsorbed onto the surface Specifically through biochemistry It can react with a near field of the sensor and change the location of the plasmon resonance So where the Spectrally where the electron oscillation is occurring? So if you were to look at the transmission spectrum of a localized surface plasmon resonance sensor It would look something like this in the bottom here where the blue curve represents the transmission before Where the peak is the peak of the plasmon resonance and after molecules are absorbed that Spectra will undergo a red shift by three or four or five nanometers And you can read them out with the spectrometer or in the case of this project. I've created a device that uses LEDs and an image sensor to read out the plasmon resonance Look why surface plasmon resonance sensors are also very cool because they're very cutting-edge But they're now becoming very cheap to produce Using soft lithography techniques, which is basically a molding process with a UV glue or an optical glue You can actually replicate nanostructures And then coat them with gold in an e-beam evaporator Which does need a clean room and an expensive piece of equipment, but it's an incredibly high throughput process Meaning that these sensors can be disposable. They can be flexible They the ones depicted in this image are printed on a overhead transparency So we're looking at a future where maybe plasmonic sensors are Embedded in existing medical technology or are printed on your books or things like this Things of diagnostic value or maybe even wearable sensors So this this device uses four LEDs in a light cone Which guides Guides the light down to a single aperture Such that these LEDs can turn on one at a time and illuminate the sample at normal incidence Which is very important for reading local ice surface plasmon resonance All this is 3d printed like the other materials I've presented so far and I can talk to you more about this This design is not necessarily novel, but the framework behind how it was designed is very novel I won't go into all the details, but basically we used machine learning techniques as I've discussed before to Select the optimal LEDs with which to use in this design from a larger library online So there are thousands and thousands of LEDs online that you can order and it's hard when you're engineering a system like this to know Exactly which LEDs would be the most optimal given a specific plasmonic sensor design and an inherent fabrication Variability so this framework takes into account the fabrication variability and the specific plasmon resonance To actually select the optimal LEDs that can be embedded in a modular way into the sensor The second device I have is a transmission microscope. It's very simple This is the one we take to elementary schools and middle schools It has two white LEDs and a diffuser and then a simple sample tray and only a z directional stage You can mechanical you can manually move the sample on the XY by pushing or pulling And you can focus by adjusting the the z-stage with the focusing knob I have a bunch of samples here. You can try out in the microscope and take a look at I have blood smears and different animal samples that are on a glass line So this is a really Fun device that's actually currently being used for sickle cell imaging at UCLA As a resolution of point eight seven microns tested with the Air Force target as we discussed earlier This is an example of blood cells We also took this device In collaboration with a group from Ghana several years ago to actually look at system isis Which is a parasitic worm? tropical disease prevalent in many parts of Africa It Causes it has a bunch of negative symptoms rash itchy skin fevers coughs and chill and it's very treatable with medicine But one of the issues is it's hard to get the medicine to places that need it the most without having some sort of point-of-care diagnostic system so we actually It affects the 250 people were million people worldwide 35% of the population in Ghana is Has been said to have System isis at some point in time. Unfortunately, most of these people are school children 40% so we actually were able to take this transmission based microscope and look at urine samples and assess and validate the efficacy of Determining system isin in a urine sample and found that it had a hundred percent sensitivity for high infection rates Lower lower sensitivity for low intensity infection rates More specs about the device, but you can see it for yourself in one minute And then the third the third device that I haven't talked about yet is this Eliza well-play reader this reads an Eliza well-play which is a plastic Device that is very commonly manufactured. It is 96 wells for doing testing So it's very high throughput biological testing Typically, these are color-based Tests, so you place a sample in here with other reagents and the color in the well will change based off of the Concentration of the biomarker you're wishing to detect Eliza well-play readers are benchtop machines that are maybe this big and they're obviously very effective at what they do As they're everywhere in the world in well-funded clinics They're designed to be modular for many different Eliza tests however We've designed a device that is for one specific Eliza test so it uses one specific illumination and Can be read with a mobile phone So I can show that to you as well It has a really interesting design where it employs 96 plastic optical fibers, which are very cheap and more robust than glass fibers And it actually couples the fibers directly to the well and Condenses them onto the image sensor of the camera phone So this is an example of an Eliza test. It's high throughput. You get 96 tests at a time The left is an example of an image we can take where the brightness of the circle Corresponds to the concentration of the biomarker in the Eliza. Well, and then the fourth device I have is the Giardi analyzer, which I've already talked a lot about And the fifth device is the fluorescent microscope for DNA imaging So that's it So if you want to come up and see the devices I invite you to do it You know relatively small groups for five or six and I will be willing to spend as much time with you as you'd like So ask any questions whatsoever. I'll try my best to open up the devices and give you Simple demos and things like this. So thank you very much and you can just kind of come up whenever So I propose start to from group number one. Yes everybody Six Okay, so group number six can come first because they should come to M lab immediately after finishing this this seminar So group number six, please. Yeah, so this is here. Let me turn it on This is the transmission based microscope There's an example of the blood cells. Let me go to the live camera So there's this focusing knob which you can turn it should be relatively well focused already Loaded in here is a blood smear just a drop of blood smeared across the side. I have other samples here There's a switch to turn on the These so it's a bit overexposed now the camera settings don't Do an amazing job? We can turn the white balance up Yes, so these are the settings Sorry You have to close the You can if you'd like just play around and you can insert any samples you want There's frog tongues and other embryos and vertebrae in here Now one thing with this the application doesn't allow you to zoom in beyond like this But if you take a picture you can zoom in further if you go to the photos and the phone is limited Those are blood cells Oh This is live coming in right now So the idea here is that and the idea here is that the resolution Maybe is not as good, but we can input this image is still a little bit out of focus But we can input this image send it to a server and it can get back a sickle cell count a malaria count Yeah, so let's open this up. Yeah, all of these are made All these are made with a 3d printer I can all creating and one is a square green they have very different Oh, I have the mic here Yeah, so this takes into account the Specific nanostructure as well as the fabrication variability, which is actually a pretty significant with this soft lithography type of thing a Variation by one nanometer of the plasma resonance can make LED based sensing very unreliable And so this type of framework mitigates that error by selecting the most optimal LEDs So for a specific plasmonic sensor you need a specific LEDs These are interchangeable though. This is just a cap I can a cap I can take it off And it plugs in here So I can I can just take this off and put on another cap But actually the specific work what makes plaza plasmonic sensing so challenging is the fact that if you want to detect a specific protein for instance Lysozyme which is a protein in your tear fluid You have to functionalize the surface of this gold sensor so that it captures only Lysozyme The biochemistry to do that is very complicated and it's not incredibly robust at the moment So there's a lot of research into making that chemistry robust or for instance There's a paper published recently they captured whole HIV viruses from a blood unprocessed blood sample and Adhered those to the surface and used this exact type of sensor to measure the presence of those You can nothing Only bulk refractive index So yeah refractive index that's something yeah, that's something so without functionalization You can't detect any specific protein But for instance you can put a microfluidic channel on it like I've done here and I can flow water Glucose solution or whatever and in this this device would output a refractive index with point zero zero zero five refractive index unit accuracy five times ten the negative four The grating on these lines or not the cell the cell It's fixed everything's fixed The so each square here is a different design But the ones that we use mainly this square and this square the holes are about 380 nanometers Wide and about 300 nanometers in depth and they're the periodicity is They're what shapes their cylinders so their circles that recess down Into a they're not they don't go all the way through so it has a bottom and then it's a repeated structure Yeah, yeah, and they're very easy to fabricate You have to start with a silicon master that has the nano hole array in it already But that only needs to be fabricated once and then you can mold that into glue as many times as you want for what A label yeah And we're looking into using commutational techniques as well to detect the malaria infected red blood cells And there's a really interesting project to with gaming actually For detecting malaria, which I can talk to you about later, right? That's another big field. Yeah Well, I'm definitely interested in talking later Yeah, and the the comb in here. Let's see if I can and then this is really simple. It's a light comb It's just a foil And it just tries to keep as much light in there as possible and then the piece of paper which should be glued Is it just a diffuser? So it's actually a very specific piece of paper It's not any piece of paper and so it gets a really uniform elimination So this ensures that each LED is illuminating the sensor at the same Angle which is very important for plasma on resonance because it's momentum dependent You can couple into different resonances based on the angle of illumination So in order to create a system that's calibrated you every time need to be illuminating that the sample at the same angle Which in this case is normal illumination and normal illumination is also when the plasma on resonances are most Our most sensitive for the localized surface No, it doesn't need to be that there's an image sensor there So this one is an example of you could you could build it around a phone But I didn't there's no polar polarizers. It's unpolarized. Yeah, it's just random It doesn't need to be polarized for surface plasma on resonance on a gold film They sometimes do use certain polarizations Because only certain polarizations couple at a given angle, but for localized surface plasma on resonance when you're coupling into Structures nanostructures. There's not so much required. Okay students from group one two three That was not in the preparatory Please come View the experiment Group one two three that was not in the preparatory school only for winter college Please come here. So here. This is the cartridge. Here's the excitation filters here One two three you was in the preparatory or not. Oh So So, let's see, let's see if We're connected to the Internet I can show you the app So I don't have Giardia samples here Because you're really bad to bring on the plane, but I can show you the way that works with a previously taken image so So I'm gonna find an image on the phone Okay, group number six. We are waiting for you in the lobby for MLAP now and you can continue to talk tomorrow Professor will be available till first day in the actual image I had to transfer that onto the phone So I can't I can't find it in the database, but right now it's uploading it to the server in LA Los Angeles and Once it's uploaded it does the uploading does take sometimes a couple minutes because it's a raw image with 40 million pixels So once the image that is done uploading it will then process the image and then return the result So No, no, no, this is a Eliza well Say me that if you look this camera and this camera and we will try no student to be here They to be over here. Okay, or in this angle Okay, this angle to be free and this because okay, these three yes, okay take care about the microphone to be working Okay, so this one's on. Yes So this this is in here and so when you load in the Eliza well play This can be battery powered but we don't operate it as battery powered because the power of the LEDs Here So we actually use red illumination here and now what you're seeing here are these ends of the fibers each fiber is coupled to one of the wells and the Eliza well play to mess with the settings but here so this is the image we process and If there was a bio target you were looking for malaria something like this one of those would be very dim and that would correspond to very high absorbance because of the Immunolinked absorbent assake going on color richie. Yeah, and we actually do have a fluorescent based reader as well No, so There are let's see 20 LEDs or something like this here red LEDs So making those modular is very difficult. So this this device is for a specific Eliza test Which if you were doing a lot, this is a Plasmonic reader I developed and this has an image sensor at the bottom and A cone here and then this is the cone that guides the light down to an aperture Which then shines the light on to the image sensor. You then would take a plasmonic sensor Oh, which is Yes, the sample would be dropped on here with a cover slip and the services Functionalized to selectively capture an analyte and you would put this in here And attach the USB to the your computer Align it and then and then turn this on and These LEDs will sequentially turn on it triggers the camera to take an image and from these images You can then get a Concentration of the analyte or in this case The experiments I did for the paper was just bulk refractive index So I had a micro channel here and I just slowly increased the refractive index over time and allowed This to sequentially take measurements to show that the plasmon resonance was indeed So that's the DNA imager. So Unless you have Labeled DNA on you it's not going to do anything, but you can turn it on. It's laser diode. You can see here I think this one is the red Nice So here is where the sample would go. You can see the emission filter there We have this hole opened up to do bright field imaging with just ambient light So you can actually do it like this hold it up and take an image just to make sure your sample is aligned And then you can put it on the table like this it covers up the hole And then you can do your DNA the DNA imaging or size Yes No, no the labeling is the labeling is done before I am We won't leave The die is called yo-yo one Yo-yo one it's a common. There's common dies for DNA. I've never done a DNA labeling process, but it's very common in labs So as long as you have the reagents, it's yeah, you go ahead Yeah To Yeah, yes, that is the conventional way to read gold nanoparticle assays plasmonic assays Spectrometers high-resolution spectrometers, especially are expensive $1,000 maybe for a high-resolution spectrometer So this tries to make spectral measurements, but with limited spectral bandwidth from a collection of LEDs So I have four LEDs with very different No, it doesn't cover it takes the most important Sections of the visible range and creates actually a linear program to output the Refractive index normally what you would do is you would measure with a spectrometer the whole optical range You would find the plasmonic resonance and track the peak Which is the most effective way to do it? But it's it requires a spectrometer or and a broadband light source both of which are very expensive But you can do it with just LEDs and a image sensor, but selecting the bands of the LEDs is a challenge Because it changes given the nanostructure or the size of the nanoparticles and it also Some LED choices that would be intuitive Are not the best because of fabrication variability. Yeah, not at the same time. Yeah, they have to it has to be done It's a scan so here they're turning on individually it triggers the camera Automatically to take an image and if it was hooked up to my computer it would save the image so you get an image stack It's like a hyperspectral stack a multi spectral But you can imagine you have four images because there's four LEDs and you bin the image you take an average pixel intensity and you have essentially the Overlap integral of the spectrum with the LED spectrum and so this is spectral information It's incomplete spectral information But we showed in this paper recently that you can use this incomplete spectral information To still recover a very accurate bulk refractive index measurement. Yes, but it's more complicated than that Narrow band LEDs don't net aren't necessarily going to get you a better result because of fabrication variability So the work that was published recently in ACS nano on this device is all about the framework for selecting the optimal bandwidths with which to probe the plasmonic assay right Because this is dependent on like I said fabrication variability, you know You can imagine you have a solution of gold nanoparticles right hundred nanometers. They're not all exactly a hundred nanometers, right? So you can't make a model of a hundred nanometer gold particles and expect for your LEDs to Work perfectly with that model because there's there's inherent variations And there's even more so variations with these nanostructures which provide which provide a More sensitive assay than nanoparticles, but also provide additional challenges Yes That's that's the whole challenge of doing it low-cost It's very easy to deduce a spectral shift when you have complete spectral information If you're using a broadband source and a spectrometer, you just track the peak and your metric is change in nanometers per refractive index units if you're doing bulk refractive index measurements However spectrometers and broadband light sources, especially those that are stabilized are expensive So we aim to try to build a low-cost system that just has LEDs Which gives us incomplete spectral information and an image sensor to just capture the intensity As it is transmitted through the plasma and extensor So we actually all we do is we take images of this square under different illumination conditions And we we take the average pixel intensity and we actually have a linear model that outputs the refractive index No, the the difference here is wavelength and the design is such that this is actually a cone That it acts as a light guide So that four LEDs can illuminate the sensor at approximately normal illumination Yeah, the the cone is the cone here group number one and two you're invited to join this experiment But please leave free space here for recording by video camera And that's really important because the plasma resonance is a couple to different resonances depending on your angle of illumination So if you're creating a linear model you want to be able to want to be able to have I invite you for envelope We want to take a sensor that hasn't been calibrated before and put it into the machine and output a result We don't have to calibrate it clean it and then run it again So we need to always be illuminating it at the same angle, and this is the easiest way to achieve that Yeah, yeah, yes You're you're very right. Yeah, I Photoshopped it the I showed one that was absorption from another paper But for simplicity because all this is transmission. I just I said it's transmission. You're exactly right But you can sometimes get spectral features that are peaks in transmission. I've seen it before But these are dips In transmission, you're exactly right. You're exactly right. It's No, it's normal. No, you're exactly right. That's this one. Unfortunately, there's no DNA here but We have the laser diode that's here at a high angle to the First three group that was not in M lap. So you are invited. You can see the Emission filters. There's that square and this hole is so that You can do bright field imaging by by holding it up to the light and aligning your sample And then when you're ready to do a dark field measurement You just put it on the table and it covers the hole and so this this is the system for DNA It's battery powered it eats batteries pretty quickly though because it's a pretty high power laser diode And then we have a knob here for focusing the Z in the Z direction That's this this system here Excitation filters around here LEDs are in pairs of two. This is the cartridge where you would drop your water sample The contents of the cartridge are disposable So once you put your water sample here, you connect the cartridge to the device And I think I uploaded an image a second ago to the server. So let's see if Let's see if it gave us a result. Yeah job ICTP So then this is the image analysis result from the server So the green circles are the Giardia and it gives you a Giardia assist count here And you know how much volume of liquid you've dispersed so you can make the calculation of sys per milliliter to determine How contaminated the water supply That's the application right there Yeah, that's So technically not but that's how we do make measurements when we're doing Functionalization for Biological samples so we'll functionalize half of it to capture a biological sample and leave half of it Unfunctionalized or blocked This is actually a continual measurement and in this fluid fluidic channel I actually vary the refractive index over time So the reference is actually the spectrum of this one sensor in water at the beginning and I just use that beginning measurement I just subtract I normalize But you're exactly right this system because it's using an image sensor that has an area of two or three millimeters squared There's enough room to do a reference in one snapshot Which is unlike a spectrometer where if you're coupled to a spectrometer with an optical fiber That's 400 microns the only way to do a reference is to physically move The optical fiber underneath the sensor and this this can lead to problems If you don't get the movement just right or if you miss the reference if you're dealing with very small so this has So this does it it's multi pixel information So you can have the reference and active area side by side in one image and you just take pixel averages of both Yeah, there's a USB here That goes to the computer and the images are taken automatically and these are these are plasmonic sensors that are fabricated on the overhead transparency Yes, so this is made with a molding process. So we have one The overhead transparency Plastic it's nothing special And there's a gold layer on it. That's what gives it its reflective properties The gold layer is 50 nanometers 50 50 It's 50 nanometers of gold continuous 50 nanometers of gold in the electron beam evaporator is about is a electron beam evaporation so that is is an expensive bottleneck of producing these but but each sensor is Five millimeters by five millimeters and in a conventional e-beam evaporator you can load about seven Four-inch wafers so you can pack a thousand of these easily into one electron beam evaporation Most e-beam evaporators I've worked with three or four over the years Have a seven wafer dome Some of them are four way for dome, but even then it's a very high throughput process What's not high throughput is when people in the clean room make one sample and put it in and waste all the other space in the Dome, but the lithography soft lithography of molding can be done I can make I think I make sometimes 60 in a day Before I go and do an e-beam run and then I can go and deposit gold and that's expensive It does require a clean room But it's high throughput very high throughput and the material cost of the gold is very minimal Yeah, it depends on the system for the water-borne pathogen detection As we're using a learning algorithm. It is processed on a server and a GPU With my lab you can do it in C++ or CUDA to make it faster You can get Hundreds of times speed-ups doing that actually but most of these that have any significant image processing. Yes, it's done remotely You could in principle do it on the phone. It would just take hours For instance, I was able to send this image From here to LA and I got back to us within about two minutes and it gave us a cyst count Doing it on the mobile phone here There's no point because we're in a Wi-Fi space and you can do it with a network to a cell network The sample in this device. Yeah, so this it's a liquid. It's a water sample 10 milliliters And you put it in here in this cartridge the contents of this cartridge are Just absorbent paper so they're disposable So once you're done with this test you would throw these away and you would have more absorbent pads for the next test But yes, you put the absorbent pads in here And you just with a syringe the water on here. The DNA is done on just a glass slide On a glass slide and it slides into this device here. The cover slip goes It's just a conventional glass slide and then you put the drop of the DNA Take a cover slip and push it that stretches the DNA spreads it around. It's too cramped. Oh Yes, hold on It's I don't know why it's so low and then yeah, that's it And then there's a switch here for the LED or the laser diode laser diode sits back here has a hole Here so the sample would be here. I don't bring DNA It's here and you see the excitation filters that square that's reflective This allows this hole allows for bright field imaging with the ambient light So you can hold it up and make sure your sample is aligned With markers on the side and then when you want to image dark field you can just put it on the table. I Was really worried about that. I have a letter from UCLA saying that they are not bombs But they didn't even search my bag Now in Los Angeles nor in Paris did they search my bank I just put it on the belt and I looked at it through the x-ray machine and it looked very suspicious But they didn't say anything. They just let me go. I guess plastic. Yeah The batteries and the laser diodes that I wonder Yes, well, that's the way we're doing it now There could be clever designs. There are commercialized cell phone microscopes that are compatible with different types of phones They're not as large field of view and not as high quality images as this But yeah with the clever mechanical design you could adapt it to other phones and have it be So these Nokia Lumia phones are good because the camera's in the middle of the phone. Okay. I phones. It's on the corner That would lead to some awkward design. Okay, possibly but you can still design a system around is no problem But another another tricky thing not so much with the fluorescent microscope. So with the on-chip imaging You can't really do that on a cell phone because you need to de-cap the lens you need to actually take the lens on Which is something that people aren't going to do Right exactly so in service posmon resonance Yeah, yeah, you have a present crutchman crutchman coefficient you've got the light So this is your So this is all the rest people join join experiment come but left one window on the front for the That's for nanostructures, so that is very easy. We just have a nanostructure And we just shine yeah, and we just shine light down at a normal and we just capture it with a camera Yes, absolutely Transmission seems to be more Sensitive according to the literature, but this device is just transmission. You have LEDs you have a sensor here No, you do not need a Crutchman coefficient a couple into localized service posmon resonance Periodicity is 500 nanometers So these are like 380 nanometers wide the whole there's a nano hole, right? 300 nanometers Yeah, it's a dip exactly Red shifts And this device measure shifts without needing to use a spectrometer or a broadband light source Normally people watch the shift and they track the peak as moves This is the best way to read plasma sensors But it requires a spectrometer and a broadband light source which can be very expensive So this just uses LEDs Uses incomplete spectral information and it uses actually the pixel intensity of The transmission of each of these LEDs and it creates a linear model that outputs their factor index So it's a bit different than tracking the peak, but it does the same thing needs to be calibrated But the paper that I wrote recently on this device. It's not necessarily on this device It's on the framework that calibrates it that creates a universal calibration. It's nano paper Yeah, so I've created a universal calibration for 500 nanometer hexagonal and square gradings And you can do that using machine learning such that I can give anybody One of these fabricated through soft lithography and they can just use it There's no need for them to calibrate an individual sensor as long as it's the same structure and it's fabricated in the same manner Yeah, I mean you can do a syringe if it's if you're just interested in Having a biological sample in there. I had syringe pumps that were programmed for the purposes of the paper. Oh Someone wait. Hold on. I want to take a picture of this Well, wait, hold on. My professor will love this because it's not I'm a Nokia Lumia Yeah, so it works with any any so fun So this this I brought this for demonstration because I can't exactly open this up, but this is contained inside here So these here are coupled to the Eliza Wells in here So this sits on top of this and this is in here and then this is what you're seeing in the image You're seeing the red light coming out of all these optical fibers And we build we build the system for fluorescent Eliza well assays as well This isn't the DNA imager the DNA imager is there But I think that there might be some DNA Eliza assay kits again. This doesn't image Samples it just records the absorption from all of the wells so you don't get any Pictures of the things you just get dots It's a posmonic reader the these Are you familiar with the posmonic sensors? Okay, so these are gratings It looks like this the periodicity the distance between the center of those 500 nanometers and these are holes like a hole for a golf ball and There's a gold layer on top You shine light even right now There's electrons that are collectively oscillating here and when you look at the transmission There's a dip around some wavelength in the case of one of these that dip occurs around 580 nanometers You can track this dip using a spectrometer and it will tell you what the refractive index of the solution is or if you've Functionalized the surface to capture a specific protein or molecule you can then detect the Concentration of that protein based on this shift and so this device does those measurements But without the use of a spectrometer it does it with just LEDs and an image sensor Yeah, and there's the flexible ones. I think over there. I have some I've made on a flexible substrate So the idea the dream here is that you can make these You can put them on a band-aid and put them on someone's arm and they can measure an analyte and sweat Or something like this The physics is there for it to happen The the part that's currently being researched is how to Functionalize the surface to capture specifically what you want to look for That's the hard part and there's a lot of research that has demonstrated this but making it robust so that it's Commercializable is what's very difficult biochemistry biochemistry problem Antibodies you're having to work with antibodies Proteins Molecules they have to form a very dense densely packed monolayer that process of self-assembly is very complicated and hard to Get consistent from sample to sample storage is an issue. Yes to get a large signal This is the cap Yes, I think it looks cooler like this, but yes measures transmission. It's very simple Again the work here. That's novel is is selecting what LEDs so normally plasmonic sensors are measured by Using a spectrometer and you get complete transmission information spectrally speaking and There's a dip that corresponds to the plasma to absorption You then track how this dip moves in spectral space towards the red In response to bulk refractive index or molecules being adsorbed out in the surface That's the best way to read these but a spectrometer is pretty expensive and a broadband light source, especially if it's Stabilized is very expensive. So we tried to do that by using LEDs and an image sensor So in this system we have incomplete spectral information and this project the paper that I recently published is about how to select which subsections of the spectral of the spectral band as a whole are best for creating a calibration for these sensors It's not obvious what LEDs you would pick because of fabrication variability and differences in different nanostructures LED well I started with measuring with a spectrometer and You have so much more data than you need and you begin thinking about well How can I use this data and you can use it to inform you how to design low-cost devices? uses four LEDs What every links the battery is dying so only a couple of them are illuminating now, but it uses four LEDs at different wavelengths and the wavelengths are selected through a machine learning algorithm that Takes into account the specific nanostructure as well as the fabrication variability All of these so they're oriented at angles But they're in a light cone that actually outputs at a single single aperture Here they all so they're all illuminating the plasmonic sensor at approximately normal information, which is really important Yeah, it's very low. You lose a lot of light But you can turn the so again. This isn't a very high frame rate system I mean a frame is a Spectral stack of the four LEDs transmitted and you can see this is about how fast they switch and they switch slowly because the integration time on the sensor is really high because the power that eventually gets the sensors low also these are Not that transparent transparent You can hold it up to light and see through them But they're about maybe on average 20 15 to 20 percent transmission So you lose a lot of light even through this EBL is used to make a initial master one time which has the desired nanostructure on it Once we've made that nanostructure we can make as many of these Sensors that the one like you're holding now by just molding that nanostructure into a UV curable polymer This process is very fast. It's very easy and it's incredibly low cost because the material cost is just the UV curable polymer and obstacle glue. It's not very expensive and anybody can do it I've trained lots of people to do this You mold it into the polymer and then you you release the mask from the polymer and it's clean and the polymer is cured And now a rigid nanostructure and then you do have to take it to the clean room to deposit gold But this is very high throughput process. So we believe that these sensors are scalable enough to be disposable If they were implemented for some assay Plasmonic sensors are not Localized surface plasmon sensors are not commercially available yet. No built for this research project There are lots of commercialized plasmonic assays and there is obviously commercialized surface plasmon resonance with plain gold Gold films it's been commercialized for decades now, but there aren't commercialized nanostructures because People typically in research labs don't use a molding process because it's not as doesn't have as much control So graduate students will go and spend a day to make one of these and that's not scalable. So it hasn't been commercialized yet And then there's lots of hurdles to overcome with the biochemistry to functionalize them to specifically target a bioanalyte The substrate the backing layer here is glass and on top of the glass is a UV curable polymer Just a UV glue and then there's a 15 nanometer gold layer on top of that No, we haven't yet I suppose we could We're very interested in kind of open source type of work. So that would be a really easy way to do that the issue is a low-cost 3d printer People could just try but The 3d printer that makes all these is a nice one. It's an expensive one. I think it costs like $60,000 We purchased it like five years ago. So that price has probably come down a lot but if you were to make one of these with a Maker bot type printer, which we do have in the lab. There's a lot of problems that come up the plastic warps There's not as high resolution But maybe for the transmission based microscope and maybe it would be fine and it's great if people are trying Even if it's not, you know with the absolute best materials There there are commercial phone microscopes There's dozens They've come out in the past three or four years They clip on to all sorts of phones Lots of different lens attachments They're not as high quality as the one we have here Not as large field of view, but you can buy them online for like 30 or 40 euros And they're fun to play around with I should have brought one. I have one I bought for Christmas. They're all the same depth So actually here these three are line-gradings They're line-gradings These two columns are hexagonal nano-hole arrays with all of the same depth The only thing that changes is the periodicity, but the aspect ratio is locked. It's the same so The periodicity varies. So we go from 500 to 600 700 800 900 and 1000 and then over here a square So for the for the work published in the paper We stuck with the 500 nanometer Periodicity and gratings and we validated the framework that we proposed in the paper for the hexagonal grating and the square grating I probably read it Yeah, yeah, and that that's a large challenge, too How do you know what is the best nanostructure in my literature search? There's Dozens of different types of nanostructures nano-hole arrays Nano diamonds nano triangles all sorts of stuff and it's it's not obvious What's the best because people model it and get very good results and then fabricated and don't Experiments it's a function of that and There's just so many Factors it's a function of fabrication soft lithography does not produce as high quality in terms of the sensitivity and figure Merit of plasmonic sensors as if you were to make these structures from scratch in the clean room using traditional ebm lithography but It's a fine price to pay when you're making these in very high throughput manner and for filters. Yeah Yes, were you the one who asked this question in the talk? Okay, great. So yeah, I'm glad you came This is a this is a plasmonic reader what I've just purchased an image sensor You can build it on a phone but I chose not to because it's a little bit of a hassle So yes, we also do just purchase the image sensors from Sony and build systems just around us Sorry, so you're asking about decapping the lenses from the suffer these are designed to incorporate the Lens that's already in the phone into the design. So we did not decap the lenses here This works with any phone someone just put their Samsung phone on the color metric reader and the image you looked exactly the same as the Yeah, oh it's a compound lens system. Yeah, so we have external lenses and then we refer to the internal Right, so that's why we have focusing knobs So we do have a Z focusing knob here and on the transmission microscope We have a Z focusing knob. Yeah, because it's it's not good to have a completely fixed system You need some degrees of freedom exactly Oh, you're saying just working with the existing lens. Yeah. Yeah, so the The problem with that is a lot of cameras the iPhone 7 is different But they don't they can't focus very close you can only get a focused image maybe about Two or three centimeters away Oftentimes we need to put the sample immediately over it creates it and make some more compact device It also allows us, you know a little more in the on top of it to put other optics. Oh, can you? Or you just text yourself the image So the Giardia one we did I did a demo just a couple minutes ago I sent the image through the Wi-Fi here to Los Angeles to our server to process the image Sent it back the results, but yeah, you can do that with a local connection, too 580 nanometer LED. I've got the batteries dying. So the higher power LEDs are not turning on but we have they're very specific for Gold and for our plasmonic sensor design. We chose these LEDs because they were outputted by a machine learning algorithm Indicating that they would give the most reliable reading. That's the paper about this device is about that framework So the wavelength is very important and the bandwidth is very important So we have a 580 nanometer five 25 527 and 611 and what the what the paper shows is basically The intuition says to put the LEDs as close to the peak as possible Because this gets you the largest contrast because the peak is shifting, right? So that's going to be the largest signal But it gives a very bad calibration the universal calibration because of fabrication And you actually want to pick LEDs that are a bit removed from the peak These give a much better accuracy when you're using a universal calibration What I mean by that is I can take a sensor that I fabricated Independently and stick it in and just get a good result. I don't calibrate it for every sensor Right, so that's the really cool thing about local surface plasmon resonance So it's really important to distinguish those two surface plasmon resonance refers to a planar gold substrate on a dielectric layer Or a planar, you know negative permittivity material But if you have a nanostructure You can couple from many different illumination angles the k vector Requirement is not as strict. In fact, you couple in the different resonances if I were to measure the transmission With a light source a broadband light source And if I were to change the angle of illumination the spectrum would completely change But you're still coupling into plasmon resonances There's not one given angle because there's 3d topology LSPR sensors are like nanoparticles, right gold nanoparticles in solution also support plasmon resonances But they're completely symmetrical in three dimensions So people use nanoparticle assays all the time and they don't care at all about what angle the illumination gives it Why would it matter so? local surface surface plasmon resonance is Emerging as a more practical Technique for measuring molecular absorption. There's a lot of problems with it SPR surface plasmon resonance by itself has been very successful in the laboratory Biacore has had its device on the market for 15 20 years now and They can make incredibly sensitive measures that more sensitive than localized surface plasmon resonance, but Localized surface plasmon resonance is far more practical You do not need, you know, very precise optomechanical Equipment you don't need a crushman configuration No, no, it's a nanostructure It's like this and it's fixed there's not nanoparticles No No That's why that's why on here. There's actually lots of different squares because we kind of just fabricated a bunch of different designs periodicity pitches and Then experimentally found which one would be the best we modeled them before so we had an idea But it's very difficult to optimize that application Because there's so many differences between the modeling step and when your final fabricated product Especially when you're using soft lithography and high throughput techniques So you're using a laser source for instance So if you were on the plasmon resonance 532 is I think there are nanoparticles around that then they would absorb heavily because they're supporting the plasmon resonances and you would Throughout the beam Throughout the whole beam. Yeah, there's no like threshold or anything like that So you would with the detector just measure the dip Same thing or yes when nanostructures the reflection is the transverse of the Sobs so when you measure a dip and transmission you measure a peak in reflectance Because they're re-radiating up the plasma resonance Optical tweezers good luck Let's say if you have just a DVD DVD player Yeah, from what I understand though optical tweezers uses radiation pressure And I don't know what sort of enhancements you would get from service plasma resonance Oh just just in general Okay, yeah, that's difficult it's really hard because you need you need a really tight focusing Right so the technology is there on that smart right because it's fabricated a large so yeah, if you can if you can find that technology The way to do is just to go that route yeah to find Technology that exists and that's cheap because it exists in mass Well it depends on the size right For what I understand the smaller the particle the easier it is to track One micron That's pushing it. That's pushing it, but There's no technological barrier to achieving one micron resolution We have under one micron resolution in the transmission microscope, but it would be you know a pixel So underwhelming demonstration, but if you use, you know, maybe you can play a little the other way and use bigger particles There's a sweet spot Yeah So yeah, I don't So CMOS is higher noise Larger dark noise dark current CCDs are more Cool CCDs is you that's what you find on an optical microscope. They're better In terms of getting a signal, but the pixel size is bigger. So that's the reason I don't know the specifics You can in principle use a CMOS. I don't know why you could especially if you were using, you know computational techniques to Allow, you know overcome some of the noise problems and find you know the Gaussian beam profile Yeah, we have we have one whole lab dedicated to this is the Giardia app so so all Now I haven't taken any images of Giardia samples with this phone So I have to find find the image in the camera roll that I've uploaded But but this one is an example of a Giardia assist Doesn't matter with the preview And I'm gonna upload this Yeah, now now the app is sending the image to our server and the server will run the Yeah, it's very frustrating because When writing your own applications, you're actually with windows with Android with Apple. They are very restrictive and controlling the camera parameters and Nokia only allows us to take is the only one that allows us to take raw images So we actually use the basic camera application to do the image acquisition and the apps are for taking the images It's already store on the phone and sending it and receiving a diagnostic Diagnosis and diagnosis and adding whatever sort of information we want It would be great if we could do everything in that Absolutely, you can have a lot of stuff you could crop an image and send it and it would go like that Because here the algorithm on the server is fixed If I wanted to change it I would have to VPN and you know remote desktop there and change some stuff But if you had on the phone, it would be much more modular Yeah, it's a result. So here we actually So the the computer is gone through and actually circled the GRD assists and you see how it's differentiated it says this one's not Just or something and then it gives you a sys count a time Yeah, I'm glad you guys enjoyed Welcome, yeah