Graphene (top layer) is a hexagonal arrangement of carbon atoms. Hexagonal boron nitride is a similar arrangement of boron and nitrogen atoms whose lattice constant is just 1.7% larger. Boron nitride’s attributes make it an excellent substrate for preserving graphene’s intrinsic properties. Image: Lawrence Berkeley National Laboratory |
Graphene
has intriguing electronic properties which include very high electron mobility
and very low resistivity. However, graphene is so sensitive to its environment
that these attributes can be wrecked by interference from nearby materials.
Finding the best substrate on which to mount graphene
is critical if graphene devices are ever to become practical.
Groups
led by Michael Crommie and Alex Zettl, scientists in the Materials Sciences
Division at the U.S. Department of Energy’s Lawrence Berkeley National
Laboratory (Berkeley Lab) and professors of physics at the Univ. of California at
Berkeley, have joined forces to examine the best substrate candidates for
preserving graphene’s intrinsic properties. Results of their research on
graphene’s interaction with a boron nitride substrate appeared in Nano
Letters.
“Any
substrate influences the properties of graphene, so if you want to study its
intrinsic properties the best way is to work with suspended graphene,” says
Régis Decker, a former postdoctoral fellow in the Crommie group, now at the
Univ. of Hamburg, Germany, and lead author of the Nano Letters report. “However, suspended graphene is quite unstable when investigated with scanning
probe techniques like scanning tunneling microscopy”—STM—”because the graphene
membrane can vibrate under the tip. So the idea is to find a substrate that
mimics the case of suspended graphene.”
A
group based at Columbia
Univ. reported, in
October 2010, that graphene supported on a boron nitride (BN) substrate had
dramatically better electron mobility than graphene mounted on the most common
semiconductor substrate, silicon dioxide (SiO2).
“The
Columbia group
showed that the mobility of electrons in graphene on boron nitride is much
better than graphene on silicon dioxide, but there were many questions that
their macroscopic measurements didn’t answer,” says the Crommie group’s Yang
Wang, co-lead author of the Nano Letters report. The Crommie and Zettl
groups compared the two systems side by side to find out just why boron nitride
works so well. “To investigate BN on the atomic scale we used STM to build up a
picture of the topography of the system and measure its local electronic
states.”
Searching for what makes boron nitride
special
Says Decker, “For a graphene-substrate system to be able to mimic suspended
graphene, the substrate needs a large electronic band gap and no dangling
bonds, so as to avoid any change in the electronic structure of graphene. The
substrate would also need to be very flat, as suspended graphene would be.
Boron nitride is a good candidate because it fills these requirements.”
The researchers deposited boron nitride flakes on a layer of silicon dioxide, grown on a layer of doped silicon. The doped silicon was used as a gate electrode for doping the graphene during scanning tunneling microscopy. Graphene was applied to both the boron nitride flakes (under the STM tip) and the bare silicon dioxide; the graphene (dark and light purple) was grounded by an electrode of gold/titanium (gold). The STM could scan across both substrate systems. Image: Lawrence Berkeley National Laboratory |
What
first attracted investigators to boron nitride’s potential as a graphene
substrate were its unusual structural properties. In its hexagonal structure
(h-BN), alternating nitrogen and boron atoms closely mimic the way carbon atoms
are arranged in graphene. Boron and nitrogen atoms in BN compounds are paired
equally, and together their valence electrons (three and five, respectively)
equal those of a pair of carbon atoms (four each). Although the h-BN lattice is
some 1.7% larger than graphene’s and not commensurate with it, the two
honeycombs laid one on the other can be aligned much more closely than graphene
on silicon dioxide. Unlike graphene, which normally has no band gap, h-BN has a
wide band gap, due to the alternating boron and nitrogen atoms in its lattice.
To
create graphene/BN devices, the Zettl group first reduced boron nitride
crystals to tiny flakes by the tried-and-true method of “exfoliating” them
between strips of Scotch Tape. The BN flakes were deposited on a layer of SiO2,
which was grown on a layer of doped silicon which, in turn, was used as a gate
electrode to tune the charge concentration—a way of “doping” the graphene layer
above—during scanning tunneling microscopy.
The
Crommie group’s Qiong Wu created graphene by means of chemical vapor deposition
on copper; on copper, carbon atoms self-assemble into a honeycomb lattice a
single atom thick. The graphene sheets were transferred from the copper to soft
plastic and then placed on top of the boron nitride flakes by pressing the plastic
onto the BN. The whole assembly was annealed at high heat.
The
graphene layer was grounded by depositing a titanium gold electrode on it.
Three graphene/BN systems were made this way, ready for direct STM comparisons
with graphene on silicon dioxide. The STM tip could scan across the graphene
layer, measuring topography and local charge concentrations at various doping
levels determined by the silicon-layer gate electrode.
Boron nitride versus silicon dioxide
“A couple of things were thought to interfere with electron mobility in
graphene on silicon dioxide,” says Victor Brar of the Crommie group. “One is
impurities that dope the graphene and locally alter the concentration of
charges.”
Results of measuring graphene on a boron nitride substrate are at left, graphene on silicon dioxide at right. The STM mapped both the topography of the systems (back) and the local charge densities (front). Graphene on boron nitride is extraordinarily flat, and inhomogeniety of local charge statesis is signicantly reduced compared to silicon dioxide. Image: Lawrence Berkeley National Laboratory |
A
sure way to shorten the mean free path of electrons is to strew the path with
obstacles known as charge puddles, which are fluctuations in local
concentrations of charge. In graphene on SiO2, charge puddles are
common.
“We
had previously studied the properties of graphene/silicon dioxide systems in
detail,” says Michael Crommie, “and showed that charge puddles aren’t caused by
ripples or corrugations in the graphene sheet, as had been suggested, but
rather by impurities below the graphene layer.”
One
source of those impurities could be foreign matter trapped between the graphene
and the substrate when the graphene layer is applied. Tiny air bubbles or water
molecules or other foreign matter could act as dopants.
“When
we made the graphene on boron-nitride devices we looked for atmospheric
impurities, but we didn’t see any evidence of their effects,” says Brar. “For
making practical graphene devices, that’s good news, because it means they
don’t have to be assembled in a vacuum.”
Another
source of graphene doping and subsequent charge concentrations is dangling
bonds in the substrate. A valence electron available for bonding with another
atom is a recipe for chemical reactivity, and silicon dioxide has a high
concentration of dangling bonds. Boron nitride, however, has no leftover
electrons to form dangling bonds.
STM
comparisons of the two systems vividly displayed the differences between them.
Topographically, graphene on boron nitride is much less rough than graphene on
silicon, with height differences on the scanned surfaces reaching only around
40 picometers. Height differences with the silicon dioxide substrate were up to
30 times greater.
Electronically,
variations in charge density were reduced dramatically in the BN substrate.
Compared to the nearly unvarying values of the boron nitride system, graphs of
the silicon dioxide systems resemble modern-art color-field paintings.
Finally,
Decker says, “because its lattice constant is very close to that of graphene,
theorists predicted that this induces a band gap in graphene, which would be interesting
for applications”—if not for maintaining graphene’s intrinsic properties. The
Crommie group investigated how electronic properties might vary according to
the orientation of the graphene sheet on the boron nitride substrate. The two,
not-quite-commensurate lattices betrayed their alignment by exhibiting changing
moiré patterns with different orientations.
Says
Wang, “We saw many different alignments, including alignments that were almost
perfect. But the graphene still showed no band gap.” In sum, how graphene is
oriented on a boron nitride substrate makes no detectable difference in its
excellent electronic properties.
Michael
Crommie says, “The graphene/BN system is really much nicer than any other
substrate for a range of applications. There are many fewer impurities, much
less charge inhomogeneity, much less bumpiness, and much more stability—in all,
a much cleaner environment for studying graphene’s intrinsic properties. Boron
nitride is a really fabulous system for practical graphene devices.”
