Thank you Dr. Xu. This is a very good question, because, admittedly, we are not entirely certain how the light couples to the performance of the device and our goal is to utilize the tunability of our system to find this out.
The shape does not have to be a crescent for plasmonic enhancement. There has been much research with plasmonic enhanced photovoltaics using other shapes, in particular nanoparticle spheres. These are often studied because they can be synthesized and fabricated in solution even within the active layer or interfacial layers. However, these spherical particles need to be very small (<100nm) otherwise light is scattered rather than inducing a localized plasmon. These resonances are often very narrow, meaning almost a single wavelength is enhanced. The tunability of these nanoparticles is also limited by their size, so much of the solar spectrum cannot be utilized. Nanorods and wires also have been studied for this application with slightly better broadband response, but still lack the controlled tunability across all wavelengths that is needed to really optimize the performance of such devices.
We use the crescent shape because of their broadband light harvesting properties and tunability. The two resonant modes in the crescent structure are the long axis (electron cloud oscillates from tip to tip) and short axis (electron cloud oscillates from tip to backbone) resonances. These resonances are both stimulated in unpolarized light or can actually be isolated through polarization of the incident light as seen in the graphs on the poster under Nanocrescent Fabrication and Analysis. From the graph in the Preliminary Results block of my poster, you can see the placement of the resonances in relation to the absorption of the active layer. It can be seen that the short axis resonance lies within the absorption band of the active layer and the long axis lies outside of the absorption band. In terms of light coupling into the active layer, we have seen absorption enhancement of the active layer with unpolarized light but we cannot yet conclude the exact mechanism of this absorption enhancement. More precise polarized optical experiments and polarization dependent performance tests are now underway to study what is exactly happening between the metal nanostructures and the active layer. With the wide range of wavelengths that we can access with the crescent structures, we can look further into how these structures interact with materials.
An additional reason for our interest in the crescent is being studied is their relatively complex structure (compared to spheres or rods/wires) can be fabricated using nanosphere template lithography (which is simple, large scale, and highly tunable). This enables more complex and broadband resonances without the complex fabrication techniques usually needed to produce such structures. The crescents can also be dispersed randomly (essentially isolating individual structures) or in an ordered fashion (where coupling among other crescent structures might introduce more interesting and potentially useful properties).
I hope this answered your question and helped clarify some of my presentation.

Hello Dr. Baskaran, thank you for the question. Plasmonic photovoltaics refer to photovoltaic devices that utilize the light harvesting properties of plasmonic nanostructures to enhance the performance of solar cell devices. Plasmonic nanostructures are metal structures which exhibit unique interactions with light. If the structure is much smaller than the incident light, their free electron cloud can couple to the electromagnetic field and, at resonance, create a high intensity electromagnetic near-field. Essentially, this acts as an antenna for visible light to trap it near the surface, typically a few hundred nanometers from the surface. If you place these structures in or near the active layer, you can ideally trap the high intensity light into the active layer to increase light absorption. Their role is to enhance the performance of the device, but do not necessarily affect the energy conversion mechanisms. The mechanism of converting light into electricity in the device I presented here is generally the same as for traditional organic solar cells: light is absorbed by the active layer material to create an electron hole pair, these charges then separate and are injected into the electrodes to run through an external circuit.

Our results have shown that introducing nanocrescent structures can increase the absorption of the active layer and increase the current extracted from the device. Because much of the resonance peaks for the structures lies within the absorption band of the active layer (see the results block of our poster), it can be assumed that the current enhancement is directly related to the increased absorption. The more photons you absorb, the more electrons you create and the higher your current will be. However, with our current results we cannot conclude this is the definite case. We have begun more precise optical and electrical characterization tests to try and prove the mechanism of the enhancement.

While the traditional idea of plasmonic photovoltaics focuses on enhanced light absorption, there have been studies that suggest the plasmonic nearfields can couple to other photovoltaic mechanisms. For example, plasmonic nearfields have been shown to couple to excitons (bound electron hole pairs which are a limiting step in organic photovoltaics) and increase their dissociation into free charges. There are also other studies, which claim it is possible to directly extract current from the plasmonic structures through emission of an enhanced electron into the embedding media. While these processes are much less studied, they open the door for what a plasmonic photovoltaic device is. We have taken these potential processes into mind when designing the long-term goals for this project. Our unique structures and fabrication techniques can easily study how different wavelengths, especially into the infrared, can affect mechanisms such as exciton generation, exciton dissociation, and charge transport. The overall goal is to push the role of plasmonic structures in photovoltaics.

I hope this answered your question and helped clarify my presentation. If you have any other questions/concerns/comments I would be happy to answer them.

Thank you for the comment Dr. Kong. Here Ref# refers to a reference device (complete device with no nanocrescents), and NC# refers to a nanocrescent device (a complete device with nanocrescents). The number refers to a particular device from that batch of samples, so the fact that they are all different numbers does not mean anything. I guess my eyes looked over that during my final edits. You can decipher which line corresponds to which device by looking at their short circuit current (Jsc) and open circuit voltage (Voc). NC2 refers to the left red line while NC1 refers to the right red line. Ref 3 has a larger Jsc (more negative) than Ref 1 so that means Ref 3 is the lowest dotted blue line and Ref 1 is the upper dotted blue line. I apologize for that confusion.
The as seen in the schematic at the bottom of my poster, the nanocrescents are added between the ITO and the hole extracting layer, PEDOT:PSS. That is the only difference between the NC and Ref devices. The material compositions of the active layers between all of the devices are identical (or at least they should be). They were fabricated on the same day from the same batch of active layer solution. This should leave the presence of the NC on the ITO the only difference between the NC devices and the references devices.
In terms of the NC devices, both NC samples were fabricated on the same patterned ITO glass slide. SEM of the structures in several areas of the device on the ITO surface showed uniform dispersion, so we assume the surface concentration of the structures on every device to be about the same. UV-Vis extinction spectra at each device location also show nearly identical peaks. From this, we assume that the devices should be essentially identical. This, however, does not explain the drastic difference seen in the device performances. The only difference between the two NC samples is placement on the patterned ITO glass substrate, where 6 devices are fabricated on a single glass slide. Our ITO pattern used in these results was designed for traditional solar cell devices, where only the thin film layers were being deposited. I have since redesigned the ITO pattern to take into account the fabrication of the crescents (polystyrene sphere deposition and angled gold deposition), where the ITO edges (~150nm) might be a problem. Irregular patterns of the crescents were seen on some samples, but were not consistent across samples. If the edges are the cause of irregular buildup and performance variations, then this will help to explain the drastic differences across the NC samples while the reference samples were much more consistent.
I hope this helped to clarify my presentation. If you have any further questions/comments/concerns I will be happy to answer them.

Thank you Dr. Noginova. You have some very good points. I will first address the cost of the gold for nanocrescent fabrication. The idea of using gold for “cost effective” solar devices sounds a bit contradicting. However, when you take into the account how much gold we are using, the price becomes negligible. The devices fabricated in this presentation used 20nm of gold and were fabricated on a 1”x1” ITO covered slide. The volume of gold that was used for the entire slide, including the area that is not even part of the devices, was only 1.25 x10^(-5)cm^3, or 0.125 microliters, of gold. With the current gold prices found online today (goldprice.org) of about $44/gram, this correlates to about 1 penny worth of gold. Compare this to the price of the active layer. PCPDTBT and PC71BM are two common active layer materials that were use in this presentation. The price of each is on the order of about $3/mg. With an active layer of 100nm and an assumed density of 1.5g/cm^3 this comes to about $0.28 or 28 times more expensive. Furthermore, a 1”x1” ITO coated slide costs typically more than $3. Granted, these are lab scale pricing so they do not represent the actual cost of large-scale production of the devices, but the relative pricing should remain the same after scale up. The real cost increase with the addition of nanocrescents will come in the fabrication. Current lab scale methods require two vacuum processes; vacuum deposition of gold and reactive ion etching of the gold. However, the crescents are fabricated on ITO, which itself requires some sort of vacuum deposition, such as sputtering, to deposit. The time and price of the NC fabrication might be reduced with an inline vacuum process after ITO deposition.
Second, you bring up a very good question about stability. Organic semiconductors are very susceptible to oxygen and humidity because of their high LUMO level. This requires fabrication in inert atmosphere and encapsulation to achieve high performance devices. It is important to note, though, that the devices for this presentation were all fabricated and tested in atmospheric conditions. As of right now, we are unsure how the presence of the NC structures will affect the stability of the device. No long-term irradiation studies have been explored in this device or on the crescent structures themselves. However, gold crescents stored in the dark show excellent stability for up to years.
Previous research studies on plasmonic particles have shown that plasmonic near-fields can create local heating effects depending on the size, shape and embedding medium. Theoretically, the local heating can either increase or decrease organic solar cell stability. Depending on the morphology of the active layer, thermal annealing has shown to enhance performance by improving crystallinity and optimizing phase separation thereby improving free charge generation and separation. On the other hand, polymers can degrade under extreme heat. Local heating of these structures have yet to be explored but will need to be done for stability studies in the future. Besides local heating, other potential stability effects caused by the NC structures are the high intensity EM field and how they affect the electronic characteristics of the materials, especially in the presence of oxygen.
I hope this answered your question and helped clarify my presentation. If you have any other questions or comments I will be happy to answer them.

Thank you very much for your excellent questions Dr. Wei. I will start by responding to the first question. I agree that is was quite interesting to see such a decrease in the Voc accompanying the current increase. However, we are not entirely sure what the cause is. This trend was seen in almost all samples to varying degrees. The devices shown in this presentation are representative of the large changes of Voc seen in the NC devices on the same ITO glass slide. Because these changes are even seen on the same ITO glass slide, we suspect that much of the problem could be systematic fabrication error. The devices used for this presentation were designed for traditional organic solar cell fabrication; meaning the polystyrene template sphere deposition and angled gold deposition were not taken into account. We realized that there was potential for the spheres (81nm) and the gold (20nm) to buildup at the ITO edges (150nm) causing non-uniform features across the slide. SEM images at the ITO edge did showed irregular NC formations and assemblies that were not consistent with the structures on the rest of the ITO surface. If there was significant buildup of spheres or gold, this might cause non-uniform deposition of the active layers possibly creating pinholes or allow for impurities to diffuse in, which are both known to affect the open circuit voltage. Admittedly, these are just speculations as AFM studies have not found such irregularities as of yet. Regardless, The ITO pattern has since been redesigned to hopefully reduce irregular deposition during NC fabrication. Due to an inoperable thermal evaporator, performance of complete devices with this new pattern have not been tested, but SEM images of the NCs on the newly patterned ITO show better uniformity around the edges. Other possible Voc degradation mechanisms could be unexpected coupling with the active layers or energy level shifting of the ITO or PEDOT:PSS hole extracting layer (although the work function of gold, PEDOT:PSS and ITO are all about the same so this should not be a problem). Optimization of the fabrication and device performance for consistent semiconducting behavior and stable Voc are currently underway.
Your second question about creating a broadband resonance through a mixture of different sized spheres is an excellent question. First, this can be done to create broadband resonances at shorter wavelengths. However, when you increase the size of the particles, you increase its scattering and decrease the plasmonic resonance. To our knowledge, spherical nanoparticles cannot show resonances much higher than 600nm. Coupling across particles can extend this slightly, but this would require ordering and is not easily controllable. This limits how far into the NIR and IR the resonances can be extended. The solar photon flux spectrum (found in the introduction block of my poster) has a peak at around 750nm, which is about as far into the NIR as most organic semiconductors can go, but there is active research to extend that, seeing that there is a significant amount of energy that is not being absorbed. Theoretical studies on the ideal bandgap, known as the Shockley-Queisser limit, calculate that a solar device should have a bandgap of about 1.1eV (1127nm) to maximize the Voc and Jsc. Most novel low bandgap polymer semiconductors still have bandgaps at around 1.5eV (826nm). Being able to harness this lower energy radiation will be very useful as the bandgaps of organic semiconductor materials continue to decrease.
Another very important factor to consider is the relative size of other layers in the device. For efficient charge extraction, the active layer of the device needs to be around 100nm. The PEDOT:PSS layer should be even thinner, about 30-50nm. To extend the resonance of solution-based particles, you need to extend their diameter. If the nanoparticle diameter becomes on the order of the active layer thickness, there will be difficulty creating a smooth and uniform active layer needed for good contact to the cathode. This showcases another advantage of our system, because the resonance of the nanocresents is tuned by changing the in plane x-y dimensions, the z-height can remain very thin, 20-50nm (we used 20nm). This enables us to go much further into the NIR without disrupting the active layer morphology.
I hope this answered your questions and if you have any further questions or comments we would be happy to answer them.

Thank you Dr. Zang!

Hello Myisha, and thank you for the question. Fabrication of the nanostructures are done under vacuum, first to deposit the gold and then to etch the gold not protected by the spherical templates (refer to the fabrication pane in our poster). The actual time to fabricate the structures is limited mainly on the pump down time of the vacuum instruments. Roughly speaking, from beginning to end, the crescents take about 2 hours to fabricate.
The crescents have a very good lifespan. Gold is very inert and gold crescents have shown stable optical properties for years when stored in dark. Silver will oxidize over time, but when stored under nitrogen can have a similar lifespan.
The nanocrescent performance lifespan in the device has yet to be tested. Because our experiments are done in air rather than a nitrogen glovebox, the lifespan of the organic active layer is very short making it difficult to identify the lifespan of the crescents in the device. However, we are using gold which should not be reactive in the device, so we do not see a significant potential for nanocrescent degradation. This is a good point though, and we need to look into this when we plan stability tests in the future.
Thank you for the question and I hope my answer helped clarify my presentation. If you have any other questions or comments, we will be happy to answer them.