Hello Professor Iyer,
Thank you for the question.
The short answer is yes. However, there certainly is a tradeoff between high vacuum (energy intensive) transparent conductive oxide production and solvent intensive silver nanowire (AgNW) synthesis. The synthesis requires (very roughly) 10ml of ethylene glycol (a petrochemical) for 1m^2 of nanowire coverage. After synthesis the nanowires are usually washed with a solvent such as acetone. This linking of the AgNW synthesis to oil prices is not ideal to say the least. However, there is a current research effort on the synthesis side (which I am not involved with) to make the solvents recyclable from batch to batch- this would be a game-changer. Furthermore, one may be wary of using silver a scarce and expensive metal, however very little is used on a per m^2 basis, and the synthesis precursors are silver salts which have very long shelf-lives. Some great analysis of the environmental and economical impacts of scaling up AgNWs was done by Professor Gao from HKUST, but much work is left to be done with this great question. Please let me know if I can provide any more information!
Regards,
Joshua

Hello Professor Clancy,
Thank you for the question!
As you presume, the nanowires are randomly oriented. Not ordering them allows us to use the simple deposition techniques such as spin coating, meyer rod coating, or doctor blading. In brief, I don’t believe ordering the nanowires gives us benefits which outweigh the difficulty of ordering the nanowires.
The long-winded answer: while not present in the video or the poster, I have studied random paths vs. orderly paths in percolating nanowire networks, and I have come to the conclusion that the benefits from ordering the nanowire arrangement are outweighed by the benefits of scalable, in atmosphere coating. In particular, I have extracted the mean behavior of a parameter I like to call ‘tortuosity’: this is the length of the network normalized to the length of the device (or a Euclidian straight line path vs. the ‘twisty-turny’ path in the nanowire web). Why I am talking about length is loss from ‘mobility’ issues in the mesoscopic regime will scale with the length of paths in the network. I have found added power loss in a device setting due to tortuosity is small relative to other power loss mechanisms (contact resistance, lateral resistance) and therefore efforts to order the nanowires should be secondary.
I would be remiss if I did not point out that there is current research to coat nanowire networks with some degree of ‘order’ in a manner that is scalable. For example, flowing a nanowire suspension over a substrate deposits the nanowires with some degree of alignment with the flow direction. This research is exciting but I expect that these orderly networks will show little improvement in a device setting over randomly ordered networks.
Josh

I agree. BTW, “tortuosity” is a perfectly appropriate term to use here.
Thanks for the thoughtful reply.

Hello Professor Koodali,
Thank you for the question,
The issue of residual organic layers on the nanowires (presumably the PVP from the solvothermal synthesis) is a hot topic right now. The silver nanowires I use in this study are from two sources: either purchased from bluenano, or homemade. The purchased nanowires are said to be washed multiple times to remove possible organic layers. This is also the approach I take after synthesis; I wash at least three times with plenty of acetone. However, I have not done a TEM study focusing on the presence and thickness of an organic layer. I will say that before the laser processing step the purchased nanowires are better (lower sheet resistance) than the homemade nanowires, I have always attributed this to better washing of the nanowires to remove this PVP layer. The saving grace is that after the laser processing step I see similar performance between homemade and purchased nanowires. I believe the post-processing step (whether my laser process or a popular oven annealing step other groups use) burns off the organic layer, as polyvinylpyrrolidone starts to degrade at about 150C.
The second issue, of oxidation, I have looked into a bit more. The main oxidative process in silver is a tarnishing with Sulfur which is familiar if we look at antique silverware or other objects. I have performed XPS on some samples (both with and without the laser treatment) aged one week in the lab environment to see how quickly this process may happen in nanowires, we found no evidence of surface chemistry change after this time. We think this suggests good stability of these electrodes. However, in a PV (especially an organic PV) setting we would expect the entire device to be hermetically sealed to prevent moisture- we posit this will more than sufficient to protect against tarnishing agents such as hydrogen sulfide and sulfur dioxide in the air. I will admit however that lifetime testing (months scale, and accelerated aging conditions) of these electrodes is still needed.
Let me know if I can follow up at all,
Josh

Thank you for your excellent response!

Hello Professor Harrison,
Thank you for the question,
The laser source I use is a frequency tripled vanadate (Nd:YVO4), with a roughly 20ns pulse duration. I am not sure I understand what you mean by explain the scientific mechanisms of the fluence data, but I get this number from two measurements. First is the pulse energy, for which I use an energy meter, and record a minimum of 300 pulses and take a mean. The second measurement is spot size, for this I ablate a thin layer of polyimide, and characterize the ablated region for the spot size of the laser beam. From these two measurements we can get the fluences (energy/area) on a per pulse basis. As you can see from the centerpiece plot (sorry it’s a very ‘busy’ plot) there is no turn on for the ‘welding’ process, it just seems to be optimized (for sheet resistance reduction that is) at 25-45mJ/cm^2. Perhaps the threshold you are looking at is in the red data points; these show the change in transparency (on the right vertical axes). We want the red data points to stay flat- as there is no dominant mechanism for transparency change. We see the red points start to rise after ~30mJ/cm^2 most likely because we are ejecting silver from the surface, thereby increasing the transparency. I have not characterized the shape of this increase in transparency with respect to fluence per pulse, mainly because this is a mechanism we want to avoid. Ejecting silver could cause gaps in the silver network that had been percolating, and thereby destroy is functionality as an electrode.
I hope this addressed the question adequately, but if not please let me know- even if it is after the deadline for the responses here as I would like to know what you think, you can email me directly at Spechler@princeton.edu.
Thanks,
Josh

Thanks Josh. Back to your plotted data, are you saying that your x-axis is the integrated fluence summed over a variable number of laser pulses (or time under the laser irradiation) or is your x-axis the integrated fluence over a single laser pulse? I had presumed the latter which for a fixed laser pulse duration would mean that you were scanning the laser intensity (e.g., 25 mJ cm^-2 (20 ns)^-1= 1.25 MW cm^-2). Normally, I would expect your welding process to be a strong function of the laser intensity. If your data is actually plotted just as a function of # of laser pulses, how did you pick the single laser pulse intensity?

Hello Professor Harrison,
Sorry – that was my confusion!
Firstly, the fluence is given per pulse. Therefore your calculation of instantaneous power is absolutely correct. However, as far as intensity I prefer to use the fluence multiplied by the rep-rate, in my case the rep rate is 1KHz.

An interesting comparison can be made when looking at intensity between my work to the work which came out of Prof. Brongersma’s group at Stanford about a year ago- this group used a 120W halogen lamp to achieve the same sheet resistance drop via plasmonic welding. We believe the laser is the way to go- as we can process the nanowire electrodes rapidly (20ns*1000 pulses) rather than >100s with the CW broadband lamp. Additionally, we can tune into specific resonances in the structure- with the hopes of making the process more efficient. Tuning the laser (using an OPO or other tunable system) and finding the most efficient nanowire welding wavelengths is one of my current research directions.
Again, even if we run into this midnight deadline, please feel free to email me as I would like to continue this discussion if you have more thoughts or questions.
Thanks
Josh

Hello Professor Yates,
Thank you for the question!
As you suspect, the nanowire network contact onto a device (let’s say for this discussion- a PV device) is very different than that of a transparent conducting oxide (TCO). The TCO allows separated charge carriers to enter the electrode from any point- eliminating the need for finger contacts- when a high quality TCO is used (like ITO) all resistive power loss before the busbar and downstream electronics is dominated by only the surface resistance of the TCO. Alternatively in a setting such as crystalline silicon one can use a buried finger contact and an n+ top layer. There some resistive power is lost as the carrier moves laterally through the n+ layer and into the finger electrode, but after being collected by the finger electrode resistive power loss is small as the metal chosen for the finger is of high conductance. The silver nanowire networks act more like this latter case. We cannot contact every point of the top of the device- as this would ‘shadow’ or block the light from coming in to liberate the charge. However, the size of the ‘holes’ in the network (where the light travels through in into the active layers) are small (micron scale), and allow minimal lateral resistance loss simply because there is not much distance to traverse- even in settings with high sheet resistance top layers (for example PEDOT:PSS may have ~500 Ohm/sq vs. n+ Si at ~40 Ohm/sq).
Additionally, making sure the contact from the nanowire to the top of the device is of high quality is of utmost importance, however, we have not seen major issues in the devices we have fabricated by spin coating on top electrodes. There is however current research into improving this contact by other groups in the field.
I hope that answered your question please let me know if I can clarify or follow up on any of this.
Thanks,
Josh