Self-assembly on graphene

Imagine a sculptor, stood inside his studio, a large block of marble in the centre of the floor. They want to create a statue. They approach the block and start removing pieces, discarding material until, after many hours, they have the finished article.

Now imagine an alternative reality. In this one, the studio floor starts empty. The sculptor is throwing tiny blocks of marble into the middle of the floor. The blocks start to assemble, and again, over many hours, a statue is constructed. In this reality, something has caused the blocks to assemble themselves into a grander structure.

The first of these realities is a an example of what we would call in nanotechnology a top-down approach. We start with a larger object and gradually remove material until we have the finished article. The second reality demonstrates the bottom-up approach, and, while it is unlikely to ever happen in the sculptor’s studio, it is very common when making nanomaterials.

In bottom-up synthesis the interactions between the individual blocks arrange each other into a grander structure. These structures can be used for many applications and their properties are determined by the arrangement of the blocks. To understand the final structure formed in bottom-up synthesis, it is important to understand the processes that make it.

There is a crucial moment in bottom-up synthesis. Imagine the blocks are forming on the floor, one next to the other. At some point, they will start to stack on top of each other. We would call this the 2D to 3D transition. It is flat when it is 2D, and at this transition point it becomes 3D.

This is interesting because lots of information is available about 2D structures formed through bottom-up synthesis. This is driven by the tools we have available for examining them. The best of these is scanning tunnelling microscopy (STM). In STM a sharp tip—so sharp the tip’s end is often a single atom—approaches the surface. We apply a voltage between the tip and the surface and then measure the electrons tunnelling between the two. This tunnelling current is very sensitive to the distance the electrons have to travel, and so STM gives us that important surface structure information. The difficulty arises when that film becomes thicker: that 2D to 3D transition. After a few layers, the information is all mixed and the information is muddled.

This is where other techniques would be used. In transmission electron microscopy (TEM), electrons pass through a material and are collected on the other side. These electrons are affected by their interactions with the material, and so can provide that structural information. TEM samples still need to be thin, but they can be up to 200 nm thick and still give useful information.

There is, however, a challenge to this approach, and its the reason TEM is not that popular for studying bottom-up synthesis. The electrons we use damage the material when they pass through because they have been accelerated to high speeds. This is particularly true for sensitive organic materials, the most commonly used materials for bottom-up synthesis. The damage can be so severe that all the information about the sample is lost almost immediately after being exposed to the energetic electrons.

We have been developing solutions to this problem. We use automation in the TEM to expose the sample to damaging electrons for the shortest possible time. More details of this are available in another post. Here’s how we used it to study bottom-up synthesis.

We were looking at two molecules TMA and TPA, and how they self-assemble on graphene. To start with we looked at how they self-assembled on graphene grown on copper. The graphene-copper surface is ideal for STM because the copper is flat and conducting. The STM showed how TMA and TPA arranged in their 2D structures. These patterns were expected and have been seen by other researchers before.

Scanning tunnelling microscopy of TMA (a) and TPA (c) on graphene. The chemical structures are shown in (b) and (d).

Now we have a look at them in the TEM. We deposited the TMA and TPA onto graphene that is now freely supported on a TEM grid. When we do this for a thin layer we see that the molecules have the same structure as those seen in the STM.

The surprising thing happens when we start to add more molecules onto the graphene. For TMA the diffraction patterns we take look no different: the molecules are arranging into layers just like the first one, and simply stacking on top of each other. The only change is that the patterns become easier to see as there is a greater signal from more molecules.

This is different for TPA: the patterns we measure start to change. Looking further we find that it is forming its bulk structure—that is the structure that large numbers of molecules form when they are allowed to arrange freely. Instead of adding more and more layers, a transition has occurred. The key point here is that TMA does not go through the same transition. It just keeps stacking molecules as before in their layers. It does not do the same transition to its bulk structure as TPA does.

Why does this happen for two molecules that are chemically very similar? We think that the difference lies within the different bulk structures of the two molecules. For TPA, the bulk structure is similar to the one seen for the single layers: the molecules are just tilted and packed closer together. However, for TMA, the bulk structure is distinctly different: it consists of molecules in planes that are interwoven. There is no easy way for the 2D layers to transform into this structure, and so it simply keeps packing layers onto each other.

This result implies that different 3D structures can be made by self-assembly by choosing the molecules in the beginning. The molecules are chemically similar but have a different transition.

The graphene production problem

Despite the many success and promises of graphene, it is yet to be widely used in mainstream devices. One of the big hurdles in this area is producing graphene. There are many methods to produce graphene, but they each have their problems: those that produce the highest-quality graphene cannot produce enough, and those that produce lots often give graphene that is too poor for most applications.

The original isolation method was the now-famous sticky tape method. Here, chunks of graphite are peeled away using sticky tape, and these are then placed onto a flat surface. More sticky tape is then pressed onto the chunks and peeled away again, giving thinner chunks. If this process is repeated, eventually there are flakes that are only a single atom thick. However, by this time the flakes are very small (only a few microns across) and they are buried within a crowd of thick flakes. This makes finding and investigating the flakes difficult. They are, however, of a very high quality and so this graphene is useful for early stage research. But it cannot make enough for any applications.

The question then becomes: is there a way to separate graphene sheets on a much bigger scale? The first steps in this direction came from chemical exfoliation. In this method, graphite is oxidised, which then allows water to move easily between sheets. With some stirring, the sheets then separate in water quite easily. Now there are many single layer sheets floating in solution. However, these sheets are not really graphene, they are graphene oxide. Removing the oxygen from graphene oxide is still providing many challenges and pristine graphene is yet to be recovered from graphene oxide.

Separating sheets without oxidising would be the obvious next step. Recent efforts have been made to do this using high shear mixing. Here, a mixer creates forces on the graphite layers that is strong enough to separate the sheets, and a surfactant (like soap) can coat the sheets to stop them from restacking. Again, this method often produces thin sheets, but the sizes are still too small (tens of microns) for many applications.

A route to large area graphene that has lots of promise is chemical vapour deposition. Here, metals are heated to 1000°C, and carbon-containing gases like methane are introduced. The metals break the gases down into carbon atoms, which then arrange onto the metal surface to form graphene. This method produces high quality graphene and has been scaled up to metre sizes. The downside here is that the graphene is attached to a metal surface, and efforts to transfer the graphene off have yet to be perfected. Further, growing graphene in this way on a non-metallic substrate are still in their infancy.

In summary, the current research efforts in graphene production are along these lines:

  • Can the oxygen on graphene oxide be removed completely, and yield perfect, high-quality graphene?
  • Can liquid exfoliation give bigger sheets, and more routinely give only single layer graphene?
  • Is there a way to transfer graphene perfectly, leaving no contaminants, wrinkles, or defects?
  • Can we find a way to grow perfect graphene on any surface that we want?

Using micro-ARPES on heterostructures

Some of the most interesting research into 2D materials involves their electronic properties. The electronic properties determine how charge carriers (like electrons) behave within them, and this then dictates how the materials will perform in electronic devices. The details of the electronic properties can be seen in the electronic structure of the material, and studying this structure attracts significant research efforts.

The electronic structure of many of the new 2D materials has been investigated, and attention now turns to how heterostructures behave. Heterostructures are the result of stacking (to form a structure) two or more different materials (hence hetero-). It will be heterostructures that make devices, not single materials on their own. Therefore, understanding how they interact, particularly how their electronic structures interact, will be essential for electronic devices.

The big challenge of looking at heterostructures is that they are very small. The most reliable way of making them currently involves mechanical exfoliation. This is where tape is used to peel off single flakes that are then place on top of each other. But the flakes are only a few microns across usually, and so the stacked areas end up even smaller. Techniques that look at areas this small are still under rapid development.

Schematic of a heterostructure. A single layer of WSe2 is placed on top of MoSe2. The heterostructure is then the overlapping region. In this experiment, the heterostructure was placed on graphite because it is flat and helps stop the heterostructure from charging.

This is particularly the case for investigations of the electronic structure. The most robust technique to give accurate, detailed information on the electronic structure is angle-resolved photoemission (ARPES). Here light is shined onto the surface, which causes electrons to be photoemitted. Measuring the properties of these electrons after emission gives information about their properties in the solid. But the light shining on the surface normally covers about a millimetre. This will not work with a heterostructure that is 1000 times smaller.

Recent developments have enabled the use of a focused beam of light down to a spot only microns across. Using this technique, called microARPES, we can put the light on the different regions of the heterostructure.

Panel A is an optical image of the heterostructure, which is about 5 µm long, near the blue H. The other panels show details of the electronic structure around this heterostructure. A key result is shown in panels F and G. In F, there are two lines labelled as W and M, whereas in G, it looks like there is only one line in the same place. This demonstrates that the heterostructure interacts more strongly when they are aligned to each other.

In our recent paper we used microARPES to study the electronic structure of a heterostructure made from MoSe2 and WSe2. The ARPES beam could then be placed onto the different regions of the sample so the electronic structure of the individual layers can be measured, as well as the heterostructure. With this we found that the two layers did interact with each other and changes in their electronic structure were observed.

These results show that it could be possible to tune the electronic band structure to give the specific properties required to fit an application. This band engineering will help design heterostructures to start to fabricate ultrathin transistors or LEDs.

Image: graphene coated copper

Scanning electron microscopy (SEM) is an essential tool for studying graphene. It is particularly useful for graphene that has been grown on copper, as it gives a direct visualisation of the copper surface. The image below is an SEM image from a graphene coated surface. SEM images are normally presented as greyscale, but this one has been false coloured. The surface shows wavy lines of steps in the surface, just like terraced rice fields. The steps are caused by the interaction of graphene with the copper, and tell us that graphene is likely present in this area; without this, it would be challenging to infer whether graphene was there or not.


A scanning electron microscopy (SEM) image of a graphene coated copper surface. The graphene causes the copper to arrange into a stepped structure.

The challenge of mixing TEM with organic molecules

Some organic molecules form crystals that have semiconducting behaviour. These crystals could become the foundations of the next generation of electronics, replacing the current silicon-based versions. But organic electronics do not yet perform at the level of the silicon-based counterparts. One route to improving their performance is through control of their crystal structure. To do this, we need accurate and fast methods to measure the structure of organic molecular crystals.

Direct measurement of the crytsal structure can be obtained by microscopical techniques. Transmission electron microscopy (TEM) allows the direct visualisation of individual atoms in a crystal structure. This has been used to improve our understanding of silicon-based materials and has benefitted traditional electronics enormously. Now TEM is being used to understand the structure of organic molecular crystals.

However, there are challenges to overcome. Principle amongst these is the damage caused by the electron beam. This beam is made from electrons that are accelerated to almost the speed of light, and millions of these electrons pass through the sample every second. The electrons interact with the molecules and disrupt its structure, rendering any TEM images useless.

There are ways to get around this problem. One way is to reduce the number of electrons passing through the sample, called the dose. Traditional dose-minimisation involves searching for the sample with a weak beam, where only 100 electrons pass through a square nanometre per second (100 e/nm2/s). Once an interesting region has been found, the microscope can be focussed on a nearby region with a greater dose of around 10 000 e/nm2/s. Finally, the beam is moved onto the region of interest, and the images captured.traditional_low

This has produced many successful images of beam-sensitive materials in the TEM. However, for some of the thin molecular crystals, even the searching dose of 10 e/nm2/s is enough to destroy the crystals.

Therefore, another approach is needed. This method uses TEM supports consisting of regularly spaced holes and a molecular film that is present uniformly across all the holes. First the microscope is focussed using an already destroyed part of the film. Then the beam is pointed away from the sample and the microscope programmed to move to another hole. After that, once the sample has stopped moving, the beam can be brought back and images instantly captured using the very first electrons that have passed through this part of the sample.


Using this new low-dose TEM imaging we were able to study how the organic molecule vanadyl phthalocyanine (VOPc) grew on graphene.

Spectromicroscopy at Elettra in Italy

An essential area of materials research is the electronic properties of a solid. Fundamentally, it is the answer to the question: how do electrons (and so electricity) behave in that solid? You can measure these properties by making a device and testing it, but this can be hard to interpret.

An alternative is to measure the energy and momentum of electrons in the solid, called the electronic band structure. The most powerful tool for this is angle-resolved photoelectron spectroscopy (ARPES). With ARPES, you shine a light (a photon source) on the surface of the material and measure the electrons that are emitted. The electron’s energy after leaving the surface can be related directly back their energy in the material. Further, the ‘angle-resolved’ part of ARPES means you collect the number of electrons at different angles off the surface, which tells you what momentum the electrons had in the solid. With both the electron’s energy and momentum, you get the electronic band structure.

synchrotron diagram

A futher improvement to ARPES is to focus the light to a microscopic point, called scanning photoemission microscopy (SPEM). While ARPES is now a relatively common laboratory technique, SPEM is not. This is because you need a lot of photons to be able to collect meaningful spectra at each microscopic point, and even more so because a lot are lost during focusing. This is where a synchrotron comes in. They can produce the many photons needed for SPEM.

microscopy synchrotron diagram
We have used SPEM for two main purposes. The first is to make a map of the property that you are interested in. For example, we can use this to see how the orientation of graphene differs across a surface and see if its electronic properties change with this orientation.

The other way we have used the microscope is to find a small (1 µm) area of interest that can then have its band structure mapped. This has prooved quite successful in our investigations of small flakes of exfoliated materials where the sample is only 1 µm across.

Spectromicroscopy is a beamline at Elettra synchrotron near Trieste in Italy, run by Alexei Barinov and his PhD student, Victor. At Spectromicroscopy we have used SPEM to measure and map the band structures of many different 2D materials.


Me, Neil, Alexei, Victor and Natalie in front of Spectromicroscopy at Elletra.

VOPc on Graphene

The majority of the electronics that we use are made from inorganic materials that are often hard to recycle, include dangerous elements like lead, and sometimes include rare elements that are expensive. This combines to increase both the financial and environmental cost of the vast electronics industry.

One alternative is to use organic molecules. These are molecules that are mostly made from carbon, and, in contrast to inorganics, they are often cheap, disposable, and easy to process. Organic molecules, like one named vanadyl phthalocyanine (or VOPc), could be used to make cheaper, more environmentally friendly solar cells and transistors. But the best attempts of using organic materials have yet to surpass their inorganic counterparts. The performance of organics lies in how the molecules stick together to form semiconducting crystals.

In this paper, we looked at the growth of VOPc crystals on graphene. Graphene is chosen as a growth substrate for two key reasons. One, is that it is an excellent conductor that is almost completely transparent to visible light, and so could be used in solar cells and LEDs. The second reason is that it is also virtually transparent to electrons. This allows us to deposit VOPc onto freely suspended graphene, and use the latest transmission electron microscopes to study how the VOPc crystals grow. This is something you cannot do with traditional substrates like glass.


We made three key findings. The first is that heating the graphene substrate during the VOPc deposition causes large crystals to grow. This is likely because molecules move around faster on a hot substrate and so are more likely to find and stick to an already-existing VOPc island than to make their own. This allows crystals to grow larger, and larger crystals generally perform better than smaller ones in devices.


Atomic force microscopy (AFM) from graphene on copper after VOPc deposition at a) room temperature, b) 75 °C, c) 125 °C, d) 155 °C, and e) 175 °C. The graph in f) shows how the islands get bigger but the coverage drops as the graphene gets hotter during deposition.

The second finding is that the VOPc crystals do not necessarily align with the graphene crystal structure underneath. This lack of epitaxy (a term used to describe how one layer can influence the orientation of the layer above) is quite important, as epitaxy is often used to increase crystal size. These results show, then, that it is not essential to rely on epitaxy to grow large crystals.

The third result is that the large-crystal behaviour seen for graphene is not seen in graphene oxide (GO). GO is a much cheaper, solution-based form of graphene that is easier to process, and so is also being carefully considered for large-scale applications. Its structure is a graphene sheet with oxygen groups attached, and it is these oxygen groups that disrupt the large crystal growth: the VOPc molecules are more likely to nucleate at these groups than reach the already-existing islands.

Overall, the results show that graphene could be an ideal substrate for organic molecular crystal growth because it enables large crystal growth, without the need to rely on epitaxy. However, the graphene should be clean and flat; any disruption (like the oxygen groups in graphene oxide) can hinder the growth.

For more information, please head here to read more.


Dark field TEM – From December 2012

When graphene was first produced in 2004, it was using the famous “sticky-tape” method. This simple method produces high-quality, single sheets of graphene, but it doesn’t produce very much of it; you end up with flakes that are (at the larger end of the scale) hundredths of a millimetre across. In 2009, researchers developed a technique that could produce large areas of graphene, and this was quickly expanded up to 30 inch pieces. The technique is called chemical vapour deposition, and involves breaking down carbon-containing molecules on a metal surface.

Here we are investigating the most common CVD route: using methane and copper. Although it can produce huge sheets of graphene, the quality of the material is much lower. CVD graphene did not quite have the same amazing properties that had been measured before.

It didn’t take too long to work out why. CVD-graphene is actually formed of lots of smaller graphene sheets stitched together, like a patchwork quilt, with all the different patches (or grains, as they are actually called) rotated a bit compared to those nearby.

Understanding why graphene grows with different orientations is key to improving the quality of CVD-graphene.

It is possible to visualise the graphene grains using a technique called “dark-field” imaging in a transmission electron microscope (TEM). In a normal TEM image, you cannot see very much of graphene because it is so thin: it often looks like the grey image below (the darker strands hold the graphene in place). But in the dark field image – seen by hovering your mouse over the image – we have managed to separate the electrons from each orientation, and used these to make another image. Each colour corresponds to a different orientation.

The image has only two colours, and we interpret this to mean that this small area of graphene is formed of grains that have only two orientations. This is an interesting result for growing graphene on copper, and is discussed more in the paper we recently published here.


Friction Force Microscopy – From June 2013

Studying 2D materials brings obvious challenges. It is especially challenging for those on rough substrates. This is the case for graphene grown on copper foils by CVD. Detecting graphene on copper is quite difficult and there are many different microscopes that have certain advantages. Here I want to introduce friction force microscopy in an atomic force microscope.
The key parts of an atomic force microscope (AFM) are shown in the schematic below. A laser light is reflected off a beam onto a photodiode, which collects the light in each quadrant and measures how the beam bends. A sharp tip on the end of the beam pushes it up when it moves over a lump on the surface; we measure the lump with the movement of the laser.


We can measure the graphene on copper surface like this, but this height image is governed by the rough, wavy copper surface, and seeing a single atomic step of graphene becomes impossible.


However, AFM can probe many surface forces, and we can exploit a different property of graphene to image it: graphene has a very low friction coefficient, much smaller than copper. We can map the friction at each point on a surface and use this to identify graphene.

To measure the friction we have to see how the beam twists instead of how it bends. We measure a lateral signal when the beam twists as the spot moves left or right on the photodiode.


The schematic below shows what we record as a tip slides over two surfaces with different friction values. The copper colour represents high friction copper and the grey represents the low friction graphene.


What we are most interested in is the difference between the trace and retrace lateral signals; the larger the difference at that point on the surface, the greater the friction. When we map the same area as above in this way we are left with the image below (part e). The graphene is now clearly visible from the copper in the gaps.


More information can be found in the paper:front