[00:00:05]
>> So it's a real pleasure to have you with us today Ollie got his bachelor's degree in electrical engineering from Shiraz University then came to Georgia Tech for his master's in doubly and then did his Ph D. in a post-doc at Cal Tech before coming back to Atlanta as a faculty member in 2000 where he is currently professor and Joseph and Pettit chair in Electrical and Computer Engineering.

[00:00:35]
He is editor in chief of The Journal Nano for tonics he's won a number of research awards including presidential early career award a Packard fellowship and a career grant from the National Science Foundation as well as several teaching awards from Georgia Tech He's also a fellow of the USA the S.P.I. E. and the AAA S. And so it's a pleasure to have with us today.

[00:01:01]
Thank you David for the introduction Good afternoon thank you for participating in this seminar and I'm really thankful to the organizers for inviting me for having the opportunity to talk to our campus people about some of our research in the area of hybrid nano conic materials and devices for integrated photonics.

[00:01:24]
As always. Representing a big group of the students and postdocs who did these works and I. Just manage them doing this work. The field of silicon photonics is. Getting into a new reality these days because of lots of advantages a solid common listed here silicon is. Abandoned Metairie all and it is available in platforms that are friendly for photonics it's reconfigurable in the sense that it's in the software fraction optical properties could be modified.

[00:02:05]
There is a large I.C. fabrication backbone that supports for tonics as well and that's one of the main motivations that silicon got into the field of photonics it can be integrated with electronics on the same chip and then same to Gratian because of the confinement of life lots of applications interconnection is a big deal these days it has been big deal for.

[00:02:30]
A long time information processing the field of R.F. photonics milimeter of a foot on it analyzing wireless signals in the optical domain because of the large band itself optics sensing there are a number of companies that came out of Georgia Tech Research in doing such sensing and then you know these days the end of law and the new paradigms for computing has brought back the concept of optical computing and when ever you have a chip you try to offer a solution and of course LIDAR which is very important for autonomous cars among others.

[00:03:11]
There has been a lot of advances in silicon photonics in the last 2 decades that we are really you know riding on those achievements ability to talk to non optical signal like our signals millimeter wave signal to high speed modulators the faster they are the faster we can put the data in optics and therefore the faster would be the processing.

[00:03:38]
Sources and detector silicon is not a direct semiconductor so we cannot make the lasers and detectors at somebody violence in solar can so when we make them in other materials we need to somehow seamlessly bond them and that has happened reduction of loss has been the major driver in.

[00:04:01]
This technology and then of course when you have good waveguides and resonators we can form a variety of functional devices that then they add up together to form a complete system on a chip and then you know there are. Adjacent technologies like photonic Crystal technologies where they try to somehow fill in the gaps of device technology that conventional devices cannot achieve.

[00:04:33]
What we do is primarily resonant space integrated photonics and this is an example of a photonic resonator which is the micro dest the. Radius could be somewhere between 1.5 micron 200 micron But you know miniaturization will push us to use a smaller and smaller and this is a big guy that brings light into the system and if the very bling matches the resonators wavelength and then if the couple and just proper The signal goes into the resonate or as you see the dip in the transmission corresponds the mode of the resonate or we have a lot of molds and then when the zoom in one of these lines we get the bandwidth and then from there we define that quality factor which is a measure of the sharpness of these resonance S. we have achieved quality factor of 3000000 for this and then you know it's just coming up of course a resonator like this has some problems you have a lot of high order modes and the distance between adjacent moments of the same nature for your spectral wrench called is very a small and then you know in order to critically couple it to the resonate or they always have a little bit of issue because the big guys has to be very close since that the large resonators confines a lot of their energy inside and don't talk to of a guide outside very well so in order to solve these problems we develop this mini A Cherise technology here this is the same mark read this great senator but reduced in size all the way to 1.4 my current by reducing the size we have the fundamental mode as shown here.

[00:06:25]
Is still confined but then the higher order modes get leaky and then they get out and then if the. Punch a hole. Quote unquote or Etch a hole inside the resonator of the effect the high order molds take them out because as you can see the fundamental mode is mostly in the Perry ferry of the resonator So this way and by optimizing the fabrication process thanks to our I.E.E.E. and solid East lever able to show the 1st devices at 1.5 micron with quality factors of 100000 now you can see that over 70 nanometer I only have just one moment just to.

[00:07:13]
Emphasize alike to compare this with what I showed a few seconds ago this is a big resonator and this is a small resonator So now they had Billy D. off having only one load in a large bandwidth allows me to do a lot of clever things now one idea is to have a so-called add drop a structure of a variety of very lengthy interview structure.

[00:07:42]
And then the resonator selects one of those and then send it into another Red Guard separate wavelengths in the wavelength division multiplexing system now the real thing is that if I mix some of these resonate tours then I will be able to somehow shape this spectral response because each resonator is responsible only for a tiny band this and doesn't talk to the rest of the frequencies so this is the idea that if I have let's say 6 resonators here you'll see that you know the drop strength is now at 6 each corresponding to one of the resonator and then for each one of the resonators I can I can develop architectures like couple the structures in order to shape this resonance and have either a little more flat band or or or an old Chevy shaft type and so on so now little by little I can shape a spectral response using these resonators each taking care of the business in a small band bit and natural thinking of that would be a spectrometer there I have an array of these resonators and in order to save a space we just wrap the waveguide around them as shown over here and these resonate tours each has a little different they've.

[00:09:07]
Decided by the size and then they cover and wideband the trench Each one felt like one variable length and sends it out of the of this structure. Using this we can have a 90 channel a spectrometer in a very small area and we can have down 2 point one nanometer wavelength Resolution 16 nanometer band data analyst than half a D.B. insertion loss I mean this is just for a simple chip that can be manufacture and doesn't need sophisticated detector cooling you know.

[00:09:44]
Large sized gratings and so on that's a very very small miniature eyes to for spectroscopy Now this is the actual device to be fabricated So the key thing here is we have an array of a god that then the light is selected by the resonator it just couples it to that very God and the big guy brings it out so the resonators are half here and half there and I kind of interleaved them in order for the adjacent channel not to be close to $12345.00 and then I can repeat is a structure in 2 dimension if I wish and then I D. end of each wave guide I have scattered I just sent the light out and then I will put the camera on top in order to monitor what happened in the structure.

[00:10:35]
The. Measuring system is here what I do as I bring a tune up a laser and then cueing the Babeland can expect that by tuning the blank light should just switch between resonators back and forth as the wavelength changes it goes to each one of the resonators and that's what you see.

[00:10:57]
So it clearly works in that ranch now there is a little bit of apparent crosstalk that just because of the saturation of the detector and the exact crosstalk issue. So. One of the advantages of silicone over some of the other material is the ability to reconfigure eat. You can use the name tuning modulation somehow changing its properties by an electronic or another optical signal.

[00:11:29]
And there are different means that we can use. The 1st one is by injecting carriers into solid con electrons and holes which in the index of refraction in the 2nd approach thermal optic effect which is heated up and then that changes the index of refraction in the 3rd after protests we.

[00:11:52]
Make a slot in it and then fill it up with polymers and in the 4th we use up to mechanical tuning which is more along the lines of up to called memes then you can have 2 structures that one of them is kind of moving and by applying a signal you can bring it down or up and change the effective resonance effective index of refraction these are the properties of any tuning mechanism it's not surprising to expect very fast operation at very low POV or very low loss so everything should be pushing the limits in order to make these devices practical.

[00:12:33]
This is the example that the did for making a reconfigurable filter by changing the index of refraction of silicon so in this a structure we have a unit cell that. Order one order to filter and then we just repeat that and then by repeating it the same way you multiply polynomials you just increase the degree of the filter which is identified by the number of pulls in the transfer function and of course the number of zeros so that's the technology we developed for DARPA and that's the actual micrographs of that.

[00:13:12]
We have 4 resonators this is a kind of a ORDER TO ORDER TO So that's order for and then a filter like that can be modified by applying electric signal to D's heaters that be put on the structure and that filter can be and not filter the tune ability or the same filter by a series of different control signals can turn into a bandpass filter with controllable center of a blend and controllable band that so you can have one a structure and by applying electronic signal you can get different functionalities out of that you know a speeds that are for the heaters in the ballpark of 100 nanosecond time.

[00:14:03]
So if silicon is so good the question is why do we look for other materials or hybrid materials Well part of the problems is that silicon cannot handle higher powers and that's always a problem. At high power silicon on linearity kicks in if your signal is analog it's dead the same way non-linearity and transistors are bad the same way non-linearity in photonic devices back the quality factor even though we've got a few millions but when you're pushing the limit to tens of millions you will stop because Silicon has to photon absorption and free carrier absorption.

[00:14:46]
Also there's a problem that occurs in all integrated photonics not silicon this is an example of micro disk resonate tours and the theater in the center of it when there is no electromagnetic field. When you design it you design it up to any significant digit that you want in nanometer size but when you fabricated is not going to be that size there's a little bit of fabrication imperfection that changes the resonance very well for publishing a paper you don't care because you have a tunable laser and then you tune into that very length and then you report but you want but if you want to narrate of these each one of the modulating frequency or wavelength for the W.D.M. then you do care so you always have to trim your restructures and that has to be low power this is not as I said this is not just silicon this any material and this technique has to be developed for that to be low POV or.

[00:15:43]
And fast and then you know if we use thermal Of course like this. The speed is going to be low now when you are starting out the power as here is shown or Originally you have a very nice resonance the Blue Care. But the power goes up the heat generation and the freak area loss start to compete and then at some point you'll see that your resonance has a bias to build behavior that's just the problem of high power and silicon that cannot handle so let's try to solve one problem at a time the goal eventually is to have a hybrid material platform that has different layers in this example silicon silicon I tried at Port graphene and Metallica structure and so on the idea is I have a substrate at different locations I want different devices that may require different functionalities that suit content not have you want to have a nonlinear device silicon only or if you might be very a small you have the better non-linear material you want to have a very high quality factor so they cannot drive as we will see in this talk is very good material but you just wanted there and then silicon has some properties that Thorp as everybody else US So you want to combination of these in a hybrid material there are different locations can have different properties so this is what we call it a double layer silicon that we developed the concept is simple we have to silicon on insulate or silicon oxide very 1st grow very thin layer of oxide about 10 and a meter 2 to $13.00 and a meter depending on application on each and then I bond them together.

[00:17:38]
Although I say that in probably 30 seconds it took us about a year to optimize this platform to get quality results but then this is what we get this is a very vague I silicon silicon in the middle oxide is probably visible even from the students and then the resonator of.

[00:17:57]
Now using this material be very able to demonstrate resonate tours with quality factors that are in the ballpark of the single layer Solich and that means the quality of the material is not sacrificed both small resonate source and large resonators a comparison of this work with other works.

[00:18:20]
The quality factor with size. Our case was bonding their position bonding their position and their position and now you can just compare the numbers to see the qualities really improved this is that transfers electric mode and this is transfers magnetic some have most of their energy in the in the middle region and the other one have the the T. M. and T. T.M. has most of the energy in the oxide between and then.

[00:18:52]
Most of the energy in solid com So this is an excellent platform for modulation kuning venue one silicon property to keep to and this is an excellent material if you want to oxide and replace it with the material to sense for example there most of the energies there we use T. Now the reason we are started this is because now I can think about this as a capacitor as shown over here silicon silicon oxide and by applying the voltage I can accumulate carriers there without requiring a current in contrast to conventional ways of using a P.N. junction and push the carriers in or take the carrier out by a current I just apply a voltage and capacitively I accumulate the carriers now I apply a voltage as shown here and clearly I can get a shift so that's the structure one electrode is shown is placed at the center of this silicon the other one is placed at the Perry ferry now the advantage of this a structure is multifold number one I can clearly see that by not requiring a P. and junction and a current I can reduce power comes from 2 I will have a tradeoff between the capacitance.

[00:20:13]
Which tells me that what voltage can be a plot so I want some charge large capacitance allow me to work with very low voltages and a small capacitance allows me to work with large will that just large capacitance has a larger R. C. lower speed small capacitance has a smaller R.C. higher speed so there is a tradeoff that I can play with using this a structure and most importantly if there is any problem in the size of the resonator and its resonance wavelength all I need to do is to apply a D.C. voltage to put some D.C. carriers in and then fix that problem without keeping a current in all the time or keeping feet in all the time so that allows me to reduce the power consumption of the device.

[00:21:04]
This is the design the structure as shown over here adding that contact is always a challenge because you have to add them in the place then electromagnetic field is minimum and then this is the R.C.S. structure that you can see there are some resistances in the structure. And depending on the resistances then the capacitor says that's are involved you will get an R.C. time constant the defines your speed of operation.

[00:21:33]
The polarity of the voltage of course makes a difference if your charge is. Provided by holes and electrons like in this case P. and then then you're going to get more index variation because only mobile carriers electrons and whole chain jury next if in this case you're also relying on acceptors and honors to some degree then you're not going to get that much faith shift or index shift and you can see that for good clarity you get a larger resonance shift because that all the charges come from electrons on health a complete system for characterization.

[00:22:17]
Which was done in electrons. George Allen tronic Design Center G D C Professor routes center. For highest speed characterization. Aside from all these. Elements here what you do is you control your polarisation you send the like into have a guy you monitor the lights out now these are the systems that allow you to monitor its frequency response if this error rate so on and so forth but at the end of the day you just send a light into it of a guide from a fiber out of the waveguide from a fiber This is their response the frequency response and you can see that you will easily get 3 D. bandwidth you know we believe it's going to be more than 15 gigahertz is limited by the characterization tool there I diagrammed is shown over here for 15 gigabit per 2nd and these are 2 zooms are diagrams this is the performance of our modulator about 30 you got it per 2nd.

[00:23:27]
$55.00 M. to Jewel per bit that's the energy us then to convert one bit from $0.00 to $1.00 can the be total modulation depth an insertion loss of 5 the B. by 2 eating and really optimizing easily a factor of 2 in both a speed and power consumption can be achieved.

[00:23:50]
So. This technique solves a couple of problems the trimmings need and power consumption but they don't address the high performance as high. As high powers and low quality factors meaning a few 1000000 those are inherent properties so we need another material and silicon nitrite has 20 times less effective non-linearity it doesn't have 3 carrier absorption it's technically an insulator so we could get in principle ultra high Q. resonators this is some examples of the fabricated devices silicon nitride on a silicon wafer you can see that we have got Q 16 millions and now by adding chemical mechanical polishing or C M P 2 our system through the new addition to I.E.E.E. and we're expecting to easily get about 30000000 which is a state of the art so that's like an order of magnitude better than sell it count.

[00:24:58]
We developed a very. Sophisticated optimized silicon nitride process again it took us probably about a year to optimize it and now with that they are able to get quality factors. That we feel you would be reaching to you know more than $30000000.00 this is the example 5 balls and the material on the top is the flowable oxide which can be removed that just like a mask which will be removed but you know very good qualities have been have been achieved.

[00:25:37]
That for the resonator that I showed quality factor of. 10000000 for 16 Micra and this is without any C M P and without C M P disses among the best results published to date with C M P It should get to as I said about 30000000 now of silicon nitride is so good why don't we use silicon nitride only the problem is that we cannot reconfigure or tune silicon light for an easily it's a dielectric so I cannot inject carriers carriers are a question it's term optic effect is weak so I have to apply a lot of heat to changing.

[00:26:17]
So now we have a material like silicon that can be modified more easily we have a very last material silicon nitrite So now the marriage of the 2 should give us. A much better material platform and that's what we did so here is a silicon on insulator very very again here is an oxide on solar con and then we grow Silicon Knights right here at C.N.N. using LP CD technique to have a high quality silicone nitrite and then be put oxide on the top oxide on the top the bonded to and then we remove the substrate and the oxide and they get silicon on nitrite Now these trenches are important they are placed far from each other like 500 micron from each other which in integrated photonic language is a lot because you know your device is tens of microns so now we put it there just because there are air bubbles that will develop at the bonding time if we don't put that so we can easily see that vents wrenches are there there is no air bubbles and there is no difference as there is their mobile and again this took us about a year to optimize to the quality that is competitive and you know not sacrificing either silicon or silicon nitride property after bonding.

[00:27:47]
This is the fabrication process so we 1st fabricate our devices in silicon which are smaller because of the higher index and then we come back and fabricate devices in tract So the key advantage of our technique is that the material development is done all prior to any fabrication process because like the position of nitrite is always at high temperature it will affect silicone so this way we just develop the material using this technique 1st and then we will fabricate our devices and now this is an example of couple ing light from solar con the using a grating and then there is a silicon waveguide that be tapered down this is the zoom version of the taper down to 15 nanometer when you make your structure various small you expand the feel it automatically goes into the nitrite guide on their meat so not much alignment is really necessary I mean reasonable and then it goes into night Friday of God The couple of into resonate or in the 9 that's the reason we wanted nitrite to begin it gives us very high quality factors and then the output signal will come into solid come true another taper and then by a grating it will come out so this shows the response and and it's very interesting to see that over white band they're more than 15 nanometer we don't get much loss because of going between one layer to the other and coming back our assessment experimental is that they get better than 97 percent efficiency going down and up between the 2 layers and then you can see that resonators in nitrite still get very good quality factor.

[00:29:40]
Now using this material we developed a new modulator architecture that you there's an existing technique called coupling modulator and let me compare the 2 philosophies in convents in on modulators like the one that I showed before with double layer silicon we have a very big garden and we have a resume to the Apply carriers either by peer injunctions or by a capacitor to the silicon Mon resonator that changes the index of refraction that changes the resonance wavelength so if the light from God is coupled to the resonant or at its correct resonance wavelength I get the royal If I shift the resonance of the resonate or it doesn't coupled to them I've got it so the very good life will go out on a fact that beat one this way I do the digital modulation Now this is very effective and I can use a very high Q. resonate over which is a very sharp prison and so I if I tweak it a little bit well I'll go from 0 to one I don't need to apply much power but the problem here is that when I am switching between resonance and non resonance the resonator doesn't respond immediately there is some like inside the resonate or that needs to scatter out and the higher the Q. the longer this time is so the way I like to bring our minds closer is to think of it as a dam and this is a border channel.

[00:31:20]
And I want to send some water from this dam to this channel that has much smaller capacity than the dam itself like a hike to resonate or has tens of thousands of times more photons than the one that brave God wants so now in this example what I do is that I will just empty the dam and field the dam up when I want to get any change in this channel this simply means that it takes time to fill it up and then give some into the resonator and then emptied up that's why even the highest quality fact to resonators are coupled to the very eve so that their effective quality factor is like 10000 because you need to reduce your cue in order to get the band with that you want this band with poverty trade off is one of the problems but what I do in here is that we say that OK what if I can change the couple and of the 2.

[00:32:23]
Just for the sake of argument bring the got closer or farther from the resonator Now this simply means that I will bring the wall of the dam down a little bit just an hour to get the water into the channel and then bring it back up this is going to be much faster than this case because here I'm not changing the resonance they have length of the resonator is always at the same result as always keeps a stronger electromagnetic field it just tweaks a little bit so this will be much faster because I don't need to empty the up and then it keeps the quality factor into so that's somehow a means of breaking that idea.

[00:33:06]
This shows some simulation results as a function of the couple in question coupling the strength of the 2 of course the stronger the couple ing that means I can get faster signal into the resonator into the heart so when the when the quality factor of the resonator is large I need to tweak the couple in question much smaller to go from 100 percent to 0 percent and the quality factor is a smaller of course I need to change it a lot so now the idea is is that how do I move that they've got it's fabricated Well we don't what we do is we form a resonator in the nitrite layer and then it is coupled into 2 wave God in the silicon layer which is technically one very big guy that goes through a round trip and come back now I modified the face of that wave guide by Apple eyeing the P. and junction in this example so the very God is and silicon the resonator is in the light right now this way I'm effectively changing the couple in because I do too place coupling and by modifying the phase between them I can make them constructive or destructive so this way I need only a small.

[00:34:34]
Amount of shift in here in order to get complete bits of ones and zeros the reason I do it in silicon silicon nitride is because I can get a very high quality factor in the nitride resonator than the silicon only resonate so this is an example that you can see if I go between one and 22 different couplings you will go from.

[00:35:02]
A low transmission into a high transmission. This is the actual view graph of the of the structure silicon nitride resonator and silicone and then the couple of between them is done by the tapers as I mentioned before this is a tapered signal in silicone that allows the signal to go into the nitrite they've got and then resonate or and then come back up.

[00:35:30]
To test it for a quick and dirty test the fabricated the device and use heat to change the phase of soccer by shifting the temperature by about 5 degrees you will see that you go from almost 0 minus 20 D.B. all the way to a very large transmission so that's what happens when you change the temperature of course this is not done by thermal because thermal is a slow this is done by applying a voltage to a P. and junction as shown over here.

[00:36:09]
So all the modulation is done in silicon and here are modulating only a waveguide not a resonator So it's not limited by the lifetime of the resonator and then I have my resonator at a fixed wavelength there this is the response of my peon junction that actually shows the diode response this is the performance of the system this is the.

[00:36:32]
Transfer function as a function of frequency the transmitted signal divided by the incident signal as a function of frequency and you can see that the 3 D.B. banded for this example was 12.5 gigahertz for this resonate or if I wanted to use the resonator only the conventional technique I would be limited to one gig or because of that feeling and seeing the structure.

[00:37:00]
Insertion loss is 3 to 5 D.B.. Some other numbers a rate of 9 times 10 to the minus 4 will allow me to have 10 gigabit per 2nd. And I know this is just the 1st try. It can easily get into you know 3040 yards operation. The next the step is now to.

[00:37:32]
Start with the silicon silicon nitride but then add up the double layer silicon technology the 1st technology that I mention to reduce the power consumption avoids having P. and junctions so that requires a 3 layer material silicon double layer silicon on silicon silicon and then silicon nitride So this is the true art for the bonding process there 1st we do the double layer silicon as before and then we do we grow nitride and then we bonded together and we get this 3 layer material.

[00:38:11]
Some examples of the devices fabricated again silicon silicon and nitrite couple ing is done using tape 1st couple hours signal goes into you know this is structure and that is again their response very good tapered couple laying and the quality factors are also in the 1000000 ranch 1200000 for decides ample So it's not an easy material to develop but no thanks to extensive development and I am facilities we were able to get into this place now this isn't recent results on published yet so that's what we are working on now to demonstrate the material in this platform.

[00:38:55]
I guess there is just one thought in here that we have looked into it and one idea if we can etched silicon oxide in this platform and fill it up with polymers electro optic polymer then I can have an a slot resonator or a slot a structure and then have you know a very popular officiant high for a high speed system because the electron take effect is very fast and electro optics modulators in the slots with 100 gigabit per 2nd have been demonstrated these are at the infancy yet so we have to develop the process for them.

[00:39:35]
In the remaining time I want to shift the gear and talk about another Metairie all that is very important I'm beyond silicon and silicon nitride and that's solid carbide Now a lot of things have been listed in here and I'm not going to go through all of them those who are interested in implantable devices know that it's biologically safe so one idea is for making brain probes and so on silicon carbide does the job.

[00:40:04]
And that's something that we are looking into with a colleague of mine at Carnegie Mellon University it's it has very good 2nd and 3rd ordinal in the area days and then by making carbon vacancies and if you can make quantum dots now it's also very good material at Visible same as silicon nitride of course less developed but with a difference silicon are 2 It is an insulator silicon carbide is a semiconductor so you can still inject carriers and get faster modulation and tuning performance out of it so there are lots of things and the reason we got into this is primarily the non-linear and quantum aspects that silicon carbide can bring potentially can generate single photon sources single photon detectors and that's the area that we are collaborating with a couple of colleagues in Italy on that who are expert in full account carbide Metairie So what did they do is a look at the silicon carbide this usually grown on top of silicone but you cannot make devices in it because index of refraction of silicon carbide is lower than silicon it's about 2.12.2 silicon is 3 point.

[00:41:23]
3. So light will go all into silicon so you have to remove silicon somehow and people on their cup or 2 from other things. And then via started this research the quality factor of resonators in carbide was very low part of the reason is that they're stressed at the silicon silicon carbide will result in low quality of the film in here and a lot of a scattering so you get a lot of loss.

[00:41:52]
So what we did was we used our bonding technology the put oxide on top of a dummy solid convey for they put oxide and then V. bonded dismissed Terry all upside down and then we removed the silk So now when I remove the solid cond the culprit the bottom layer will just come on the top and I can do whatever I want with it definitely polishing so then we did Chemical mechanical polishing to the surface to make the quality better a very simple idea that use of or knowledge of bonding in order to bring the quality factor to the record type.

[00:42:36]
Show in here so this is the origin I'll sample the top surface of it when we get it is reasonably good that's the A.F.M. results are a mess their reaction is about 1.4 angstroms because they polish the top surface the problem is the bottom surface this is when I bonded and bring the bottom surface up and monitor it A.F.M. you can see that and this is after chemical mechanical polishing of course we did not do at Rest of lean this if I we do that at least we should get one point for micro stripes but that by itself shows that our simple technique can reduce the scattering from that by a factor of 3 we also developed you know electron beam lithography an etching process to fabricate the structures then the Macquarie all is not perfect you see these voids in here but even through that we were able to fabricate resonators in it so these are and their radar figures that I try to explain this are a couple of different resonators with different sizes and different resonance of the key figure here is this one that shows other observation of quality factor as a function of the radius.

[00:43:54]
Now for this we actually have quality factor of 242000 which was 3 times you know the existing world record now the unpublished results this is published. But the unpublished result that we have by making this a structure better is now in the 250000 which is sold just by a simple thing we're bringing a material that was thought to be too lossy into a regime that now we can get a structures out of it we can get you know for tonic performance out of it because quality factor is getting into the ballpark of the numbers that are attractive.

[00:44:36]
And I just flush out a couple of slides that show other materials that. That we are working on this is double layer graphene on silicon nitride silicon nitride grafting grafting and this is similar to silicon nitride So it comes sole income the reason we use graphene is because on paper at least graphene should have very low resistance now the speed is limited by our resistance and c capacitance of this device by reducing the resistance we should get 100 gigabit per 2nd on paper now we are still working on this because of the sheath resistance of graphene is higher than what was expected but that's another material platform that they have developed being able to transfer graphene by vet transfer technique put oxide in between and then combine it with Silicon Knights where technology.

[00:45:36]
These are some other hybrid material platforms this is to the materials money Benyam sulfide money then you I'm selling are these are the metals the child could generate which combine a transition metal like money Danny on with a childhood Gen like sold for and sell a neon and they have active properties that are direct semiconductors one a farming core molecular layer thin and we are working in making quantum dots all those the paper just came in A.C.S. photonics where we are.

[00:46:10]
I'm sorry A.C.S. nano there we are showing lateral had to restructure in it I mentioned silicon carbide This is our polymer Warrick where we have a silicon waveguide Etch a slot in it it's proven that your mold can be completely or mostly confined in this lot and then putting an electro optical polymer and changing that electro optics polymer through electro optical effect will allow to develop very fast modulators and that's the double layer silicon and nitrite.

[00:46:43]
And let me conclude that silicon photonics has done this for with a lot of success when there's investment from U.S. government and industry to make foundries to mass manufacture silicon photonics chips aim for tonics is an example of such investments and Europe is investing so I would expect soon we will see people sending their tape out to these foundries instead of fabricating it themselves but all this capability can be argument by combining silicon with materials that brings properties that silicon does not have and then that's the vision of this work and then the results that I showed either in double layer silicon or silicon Silicon Knights suggest that for sophisticated system that push the limits of the speed and power consumption and size and density would eventually require some kind of marriage between different materials that could be silicon silicon nitride $35.00 semiconductors silicon carbide polymers or other materials and that's the approach that hopefully can push silicon photonics into higher applications and and hopefully into consumer and non consumer markets thank you very much for your attention.

[00:48:26]
Well the. The question was. Explain a little bit more on the metric you over V. Well when you look at the resonator let me and by the way our policy just for not mentioning the the funding people but these are the sources of funding for us so if I go to.

[00:48:55]
One of these are slides Well the key thing is that quality factor by itself shows how you can confine your mode inside the structure and therefore build up energy inside that but then the question is that is your energy bill topple over hundreds of microns or over tiny nanometers the key thing at the end of the day is energy density which is you know number of photons per unit volume now 2 properties have 2 different aspects of this story quality factor is about the confinement of energy putting it you know more and more in the resonate.

[00:49:42]
And then more volume which is in essence the volume that your mode occupies normalized by the index of refraction let's forget about that so now more volume is technically volume very little more is concentrated not just the resonator because the more these mostly are on one micron of this and then 10100 nanometer up and down of that structure the smaller the more volume is the better the structure is the larger the Q. is the better the structure is the ratio Q. over re is volume Yes Yes And there are cases that for example in plasma Onyx you have resonators that have quality factorial of 100 compared to 10000000 but then they are competitive because they confine your energy at the surface in a very small multiple and at the end of the day that's the parent or that defines one a structure over the other Thank you.