Theoretical Calculation and Experimental Verification for Dislocation Reduction in Germanium Epitaxial Layers with Semicylindrical Voids on Silicon

Low-threading dislocation germanium is very important to realize high-performance silicon photonic chips. Voids at the germanium-silicone interface work as dislocation sinks to reduce the threading dislocation density. Demonstrating the procedure will be Mohammed Faiz, a master student from my laboratory.

To begin, define germanium growth areas by preparing a design file with line and space patterns and square shaped silicon window areas using commercial software. Then prepare a selective epitaxial growth mask by determining the window width and mask width, while drawing rectangles by clicking open file, then structure, and rectangle or polyline option using the software. To prepare boron-doped p-silicon substrates with the resistivity of one to 100 ohm centimeter, open the lid on the tube furnace and load the silicon substrates into the furnace using a glass rod.

Start to blow dry nitrogen gas into the furnace by opening the gas valve. Then set the gas flow rate to 0.5 liters per minute by controlling the valve. Set the annealing temperature by changing the program.

As the temperature reaches 900 degrees Celsius, close the dry nitrogen valve. Open the dry oxygen valve and keep it for two hours. Coat the oxidized silicone substrates with a surfactant using a spin coater and then bake it 110 degrees Celsius for 90 seconds on a hot plate.

After the surfactant coating, coat the silicone substrates with a photoresist using a spin coater as demonstrated previously. And then bake at 180 degrees Celsius for five minutes on a hot plate. After preparing a photoresist developer and a rinse for the developer in a draft chamber, dip the exposed silicone substrates into the developer for 60 seconds at room temperature.

Then put the developed silicone substrates on a hot plate to bake at 110 degrees Celsius for 90 seconds. Next, dip the silicone substrates into a buffered hydrofluoric acid for one minute in order to remove part of the silicon dioxide layers exposed to the air as the result of electron beam exposure and development. To remove the photoresist from the silicone substrates, dip into an organic photoresist remover for 15 minutes and then into 0.5%diluted hydrofluoric acid for four minutes to remove the thin native oxide in the window regions but to retain the silicone dioxide masks.

For epitaxial germanium growth, load the silicone with selective epitaxial growth masks into a load lock chamber. Set the buffer main growth temperature in the Recipe tab shown on the operation computer. After determining the durations for the main growth of germanium so that the selective epitaxial growth germanium layers coalesce with adjacent ones, click Start in the main window and the silicone substrate is automatically transferred into the growth chamber.

As the silicone substrate is automatically transferred from the growth chamber to the load lock chamber, vent the load lock chamber and unload the silicone substrate manually. For etch pit density measurements, dissolve 32 milligrams of iodine in 67 milliliters of acetic acid using an ultrasonic cleaning machine. Mix the iodine-dissolved acetic acid with 20 milliliters of nitric acid and 10 milliliters of hydrofluoric acid.

Dip the germanium grown silicone substrates into the acid cocktail solution for five to seven seconds to form etched pits. Observe the etched germanium surfaces with an optical microscope to ensure that etched pits are successfully formed. To count the etched pits, put the etched germanium sample on an AFM stage and then approach the probe by clicking auto approach.

Decide the observation area using an optical microscope integrated with an AFM and scan five different 10 by 10 micrometer areas. Threading dislocation densities in coalesce germanium originated from 113 faceted and round-shaped selective epitaxial growth germanium was calculated, demonstrating that threading dislocation generation occurs only at interfaces and dislocation densities should be reduced with aperture ratio. SEM images and distribution maps of coalesced or non-coalesced germanium layers were obtained, which shows that the coalescence took place when the window width is smaller than one micrometer.

Threading dislocation density for the coalesced and blanket germanium was studied by AFM, which shows that the thickness of germanium layers was reduced for those grown at 700 degrees Celsius. The interaction of threading dislocation with the surface was monitored by STEM and TEM images of coalesced germanium layers, demonstrating that strain accumulation occurs at the top of the semicylindrical voids and strain relaxation at the subsurface layer of the voids in order to minimize its energy during or after the growth. The TEM images of a coalesced and blanket germanium layer shows that the length of the defect lines in coalesced germanium are longer than those in a blanket.

TEM images of a small area with high threading dislocation density were obtained for incline dislocations, indicating that the screw dislocation disappeared when diffraction vector G was changed. While the mixed dislocation did not disappear whatever diffraction vector G was chosen. The most important protocol in this procedure is the substrate patterning by lithography, followed by a germanium epitaxial growth.

And unfortunately, due to the machine difference, we cannot directly show the protocol. Instead of using electron beam writer, i-line step is also one of the machines that can do the patterning and applied to germanium epitaxial on different type of second substrate.