The Experiment

When I first arrived in Korea I didn’t know what my job would entail. I thought that I would be working on Prof. Lee’s project. On my first day I sat down with Prof. Lee and he told me that I would be working on my own little research project. He didn’t give me a specific topic and told me to do some research and think of something to do. He eventually gave me a little bit of guidance by suggesting that I look into the synthesis of metallic nanoparticles.

I was a little disappointed that I would not be working on Prof. Lee’s project because I was looking forward to contributing to a real project and I was concerned that my little research project would just serve to kill time before I returned to Canada.

For the next couple of days I read article after article on nanoparticles until I came across an article which described the synthesis of gold nanoparticles with a silica shell. Prof. Lee’s research uses gold nanoparticles as the charge trap elements of his memory devices and the structure of the particles gave me an idea.

Quick lesson in flash memory:

The basic idea behind flash memory devices is that a charge is trapped between two insulating layers. If a memory cell (one byte) holds a charge then it is in the programmed state, if it doesn’t hold a charge then it is in the erased state. The drawing on the left is the basic structure of a non-volatile memory device (doesn’t need energy to store information) using nanoparticles. The device is constructed on a semi-conducting wafer, generally p-type silicon, on top of the wafer a thin (4nm) blocking oxide layer (often silicon dioxide or hafnium oxide) is deposited followed by a dense monolayer of nanoparticles. A thick (15nm) blocking oxide layer (hafnium oxide or aluminum oxide) is deposited on top. Finally, metallic gates are patterned on top. Each gate represents one memory cell. To program a cell a voltage is applied to a gate, electrons from the semi-conductor flow towards the gate. However, there are two in barriers in the way, the tunneling and blocking oxide layers. The oxides are insulators, which means that a large amount of energy must be added in order for them to conduct a charge, compared to no energy for a conductor and a little energy for semi-conductors. The applied voltage supplies the electrons with sufficient energy to “tunnel” through the tunnelling oxide layer and reach the nanoparticles but not enough energy to get over the blocking oxide barrier. The voltage is applied as a pulse which allows the electrons to reach the nanoparticles but after the pulse, the electrons do not have sufficient energy to tunnel back through the tunnelling oxide layer to the semi-conductor, they are “trapped” in the nanoparticles. The memory cell can then be read by applying a small voltage to the cell and reading the capacitance. To erase the cell, a large negative voltage pulse is applied to push the electrons back through the tunnelling oxide layer to the semi-conductor.

My idea (on the right) was to coat the gold nanoparticles with a thin silicon dioxide coat (4nm) that would serve as the tunnelling oxide layer. The tunnelling and blocking oxide layers are generally deposited by a sputter machine which is quite expensive. My method would eliminate one of the sputter processes as the tunnelling oxide layer would be prepared directly on the gold nanoparticles through a couple simple chemical reactions. I presented this to Prof. Lee and he liked the idea and I was given two goals:

1. Synthesis of gold nanoparticles and the construction of high density monolayers.
2. Synthesis of gold-silica core-shell particles and the construction of a non-volatile memory cell.
   

My first job was to synthesise the gold nanoparticles (GNPs) which was not very difficult as the procedure I used is quite common and involves a couple of simple reactions. As expected, despite the simplicity of the procedure, my first attempt still managed to go horribly wrong. As I was heating my gold chloride solution (precursor of the GNPs) in a hot water bath I heard a loud POP! I spun around to see what had happened when I started to hear a fizzing sound coming from my experiment. The glass bowl containing the water bath had cracked a quarter way around, half way up the wall and right through the centre of the base and water was starting to seep out onto the hotplate. I immediately shut off the hot plate and unplugged it.

With time, and a new bowl, the synthesis of GNPs has become quite routine and I have become the GNP expert in the lab and have taught my colleagues the procedure so that they can continue making their own GNPs when I leave (I have been their supplier for the last couple of months).

The next step was to stick the GNPs to a silicon wafer. This proved to be easy enough. While it was easy to stick the particles, getting them to stick in large quantities and in an ordered fashion took a month to accomplish. Finally, after many experiments I prepared two samples with particle densities that are high enough for memory applications. This was the main goal of my work term as it is essential for my colleagues to have high density monolayers for their research.

Unfortunately, my research has moved quite slowly as it is often limited by the frequency with which I can use the SEM (scanning electron microscope). My particles range from 13-30nm in diameter and can only be seen with an electron microscope. There were many times where I couldn't move ahead with my research until I knew what my results were (what my particles looked like) but could use the SEM once every week or two. This slowed down my work considerably. The extra time allowed me to slowly work on my silica coated GNPs.

The synthesis of the gold-silica core-shell particles was fairly simple as the procedure was well explained in a couple of journals. After the synthesis of the particles I have been pretty much on my own. As far as I know this is a novel idea. I have done a lot of research and read upwards of 40 articles and have yet to find any hint that someone else has tried what I am doing. This is exciting because I'm doing something unique. It also makes it much more difficult as there is no literature on how to stick these particles to a silicon wafer, let alone construct a memory device with them.

Sticking the particles was the first hurdle to overcome. My first attempts were disastrous. It is no fun sitting in the SEM room with the technician as sample after sample comes up blank...not a particle in sight. After reading many articles on the surface properties of gold nanoparticles and silica nanoparticles I designed various methods and a few of them worked, but the densities were not nearly high enough. I am still working on increasing the density of the silica coated particles, but I fear the I will run out of time and will have to leave without constructing a memory device because the particle densities are too low. I have a couple of samples that are ready for SEM imaging and I am quite optimistic about their results, but the SEM operator will be on holidays and there isn't much chance that I'll see the results before the 11th.

In all I have been quite successful, though there have been problems along the way. I cleared the lab in my second week because my cleaning solution started emitting terrible fumes. We evacuated the lab for over an hour to let the place air out. I use a different cleaning procedure now. One of my experiments was delayed when our lab was flooded and we spent the day bailing water out of the lab. We're still not sure where the water came from. Of course there are all of the instances of human error, where the wrong sample was placed in the wrong vial or where vials were knocked over by a clumsy elbow. Though I haven't knocked anything over in a while. I was terribly clumsy around the lab for the first few weeks, but that isn't a problem anymore.

You can see more picture from my experiment here.