News Article
Exposing anti-phase profiles in GaP-on-silicon
Scrutinising samples with several microscopes reveals the presence of anti-phase boundaries in GaP that jump from one atomic plane to another
High angular dark-field imaging on GaP grown on silicon reveals the nature of anti-phase domains, which can lead to device-degrading free charges.
For many years, engineers have dreamed of the possibilities that would result from monolithic integration of III-V transistors and lasers on large, low cost silicon substrates. And GaP has been marked out as a compound with a big role to play, thanks to its very favourable lattice constant – but little is known about the boundaries between device-degrading anti-phase domains that can form in this binary when it is deposited on silicon.
To address this shortcoming, researchers from the Philipps University Marburg and Jülich Research Centre have carried out a theoretical and experimental study into the nature of the anti-phase boundaries found in GaN grown on silicon.
This team was surprised with what they found: “Theory predicted something different from what was experimentally observed,” explains team member Kerstin Volz from Philipps University Marburg. The anti-phase boundaries were not atomically abrupt, as expected, but have a finite thickness, due to random jumps between atomic planes.
One key finding by the team is that the jumping between planes can lead to a charge distribution that is macroscopically neutral along some planes, but leads to local charging along others.
“If there are free charges, even if macroscopically neutral, these charges can still have an impact on the functionality of devices that rely on charged carrier transport,” explains Volz. “Knowing the structure of these defects and correlating the structure to the growth conditions will allow us to control the charge distribution in these layers.”
Samples for the teams study were formed by using MOCVD to deposit a GaP layer on a silicon (100) substrate with an intentional miscut of 0.1° in the [110] direction. “Growth conditions are chosen in a way that no twins or stacking faults occur, and only a minimum of anti-phase domains remain,” reveals Volz. To realise this, the deposition process includes a pulsed nucleation step at 450 °C and continuous overgrowth at 675 °C.
High-angle, angular dark-field (HAARF) images of the electron-transparent samples that were prepared from the epiwafers were obtained with three different scanning tunnelling electron microscopes: a JOEL ARM 200F, a JEOL 2200 FS, and an FEI Titan 80-300. Spherical aberrations of specific lenses were corrected to realise the high level of resolution.
To complement these measurements, the team calculated HAARF-intensities using commercial software.
According to Volz, it is essential to carry out these simulations: “The often-used assumption that the HAADF intensity is proportional to the square of the atomic number is just a rough approximation. In reality, the contrast is more complex, and depends on imaging conditions, as well as sample properties, such as strain.” One example of this is that despite having a lower mean atomic number than GaAs, GaNAs produces a higher intensity.
Viewing along the [110] direction, Volz and his co-workers acquired images showing an anti-phase boundary with a thickness of one atomic layer in the GaP film. They explained this by arguing that the anti-phase boundary is not fixed on a {110}-plane, but jumps from one to another {110}-plane in the viewing direction.
To confirm the presence of these jumps, the team looked at cross-section samples in a projection. This revealed that the anti-phase boundary does not run straight along a charge-neutral [110]-plane, but is facetted on higher indexed planes.
Volz believes that these jumps maybe driven by temperature, so they could be supressed by lowering the deposition temperature. “However, a certain temperature is needed for decomposition of the precursors, which makes the jumps unavoidable.”
The jumps are not necessarily bad news, because they can provide a mechanism for self-annihilation of the anti-phase domains. The German team have done this, producing a GaP-on-silicon (001) template that is free from anti-phase domains after 50 nm of deposition of the binary compound.
Goals for the team are to now perform calculations to estimate the energy of the jumps, and to carry out additional HAADF measurements to determine the three-dimensional nature of the anti-phase domains.
A. Beyer et. al. Appl. Phys. Lett.103 032107 (2013)
