Program Manager
Prof David Jamieson - University of Melbourne

Ion Beam Researchers
Staff
Prof Steven Prawer
Dr Changyi Yang
Dr Sergey Rubanov
Dr Chris Pakes
Dr Grigori Tamanyan
Dr Alberto Cimmino
Dr Sean Hearne
Mr Robert Short
Mr Roland Szymanski

Students
Mr Toby Hopf (PhD)

Collaborating Centre Researchers
Prof Andrew Dzurak - University of New South Wales
Dr Eric Gauja - University of New South Wales
Dr Fay Hudson - University of New South Wales
Mr Victor Chan - University of New South Wales
Mr Mladen Mitic - University of New South Wales

OTHER COLLABORATORS
Prof Robert Elliman - Australian National University
Prof Peter Johnston - RMIT University
Prof John O’Connor - University of Newcastle

Program Description
The ion beam program is part of the topdown strategy. We focus on the implantation of single phosphorus atoms and clusters of phosphorus atoms into silicon. We have developed a precise single ion implantation method for counting implanted atoms and can now mass-produce devices containing pairs of atoms or clusters of atoms in the Colutron system. We are also making progress in the migration of the single ion implantation method from the established Colutron system into our Focused Ion Beam (FIB) system for the construction of arrays of single donors in a silicon substrate.

Figure 1 Cross-sectional and plan view schematic diagrams of the process steps to fabricate an island aligned with implanted buried metallic leads by ion implantation through a surface mask fabricated by electron beam lithography.

1. Cluster and Counted Atom Devices Fabricated by Ion Implantation
Throughout 2005 the ion beam program has been working in close collaboration with the Integrated Quantum Computer Device Program on the fabrication of ion implanted devices using a versatile strategy. This can be applied to fabricate devices that have been used to verify the integrity of the charge transfer and measurement process on the scale required for the charge qubit device with the possibility of scaling the fabrication process to single atoms. Our strategy employs a pure silicon substrate (>104 Ωcm) masked with a polymethylmethacrylate (PMMA) film patterned with Electron Beam Lithography (EBL).

This process allows the fabrication of sub-20 nm scale masks in 100 nm thick PMMA films that can be used as implantation masks as shown in Figure 1. Implantation of 31P through such masks allows a self-aligned structure to be fabricated incorporating buried islands aligned with leads using a single implantation step. For our devices the beam energy was 14 keV which produces islands located 20 nm (from SRIM [J.F. Ziegler, J.P. Biersack and U. Littmark SRIM – The Stopping and Range of Ions in Solids, Pergamon Press, New York, (1985)]) below the surface of the Si substrate which is protected by a 5 nm SiO2 passivation layer.


The 31P doping of silicon with this energy produces metallic layers at the end of range if the fluence is greater than 5x1013 P/cm2. Conventional charge integration, which may be used down to 1x1011 P/cm2, is adequate for measurement of the fluence in this case. Following implantation, the lattice damage is repaired with a rapid thermal anneal, 10 s at 1,000°C, chosen to limit diffusion of the implanted atoms.

The size of the buried islands produced by this process is limited in the inherent precision of ion implantation where the spatial accuracy of the implanted dopants is restricted by the ion straggle. This is evident in the Scanning Electron Microscope (SEM) images shown by the composite image in Figure 2 where the straggle of the implanted ions under the EBL mask is evident. The original position of the mask is determined to sub-2 nm precision by the metallisation through the mask. The width of 10 nm seen in Figure 2 is consistent with the standard deviation of the implant determined by SRIM. The lower limit imposed by straggling on the size of a gap between the leads and the island that can be produced with our parameters is shown in Figure 3. This shows SRIM simulations for 10,000 14 keV 31P ion trajectories for the fabrication of leads with narrow gaps from 10 to 50 nm. These
simulations show that sub-20 nm gaps are not electrically isolated which is also confirmed by experiment. Despite the fact that even for 50 nm separation there is still a strong probability of finding straggled atoms in the gap, electrical tests indicate the leads are electrically isolated and results from a series of devices fabricated with and without islands reveal an interesting range of charge transport phenomena through the leads. Phenomena include, in order of increasing gap width: ohmic conduction for narrow gaps, then wider gaps display Coulomb blockade due to the quantum tunnelling of single electrons from the metallic leads onto real or virtual metallic islands. These results are presented in the report from the Quantum Measurement Program.

Figure 2 Composite image showing the SEM image of the as-implanted device made with the mask of Figure 1 overlaid with another SEM image of the metallisation produced by the same mask in a different device. The metallisation is confined to the original boundaries of the mask and the implanted ion halo can be seen to extend outside the boundaries as expected from ion straggling.

Figure 3 Simulations, based on SRIM calculations, of pairs of buried leads fabricated by ion implantation with a gap from 10 to 50 nm. In each case the ion implantation direction is vertically downwards and the ions are 14 keV 31P implanted into a Si substrate with a 5 nm SiO2 surface oxide. The width of each simulation is 100 nm and the horizontal black bars represent the opaque area of the mask. The profiles represent the ion concentration as a function of position with the maximum concentration in the experimental devices chosen to be > 5x1013 P/cm2.

2. Development of theCounted Atom DevicesFabricated by IonImplantation
In order to scale this process down to the implantation, for example, of exactly two atoms aligned with independently fabricated metal gates, conventional charge integration for the ion dose measurement is insufficiently accurate. We have developed a method to count the individual ion impacts where integrated detector electrodes in the substrate register ion impacts as previously described. Briefly: the detectors are fabricated on the ultra-pure silicon substrate to create a reverse-biased PiN architecture, by utilizing a large back contact with n+ (1020 cm-3) doping positively biased against two top electrode fingers each making contact with a p+ well (1020 cm-3). In the area between these electrodes a central implantation zone is defined, with precision placement of the ions guaranteed by the fabrication of 15 nm diameter apertures in a PMMA mask using e-beam lithography and standard development processes.
Over the past two years we have directed intensive efforts to improvement in the method for detection of the extremely low signal generated by the ion impacts. It is essential that all of the charge from the ionization created by the ion impact be collected by the detector electrodes and that the fabrication process be reliable and reproducible. The modifications to the previous architecture in 2004 are shown in Figure 4. We have improved the stability and reproducibility of our detector structure by reduction of the area of the surface metallisation as shown by the central schematic in Figure 4. In this case the electric field in the central construction zone is sustained by an extended p+ well formed by boron diffusion. We are also in the process of investigating a new NiP device (inverted PiN structure) shown by the right schematic in Figure 4. The design of the NiP device is intended to minimise the effect of local electric fields in the surface of the device which cause electric breakdown in the PiN architecture. In the new scheme of the NiP device, the surface electrodes will be grounded and the back-contact electrode will be biased.

Last year we reported an Ion Beam Induced Charge (IBIC) investigation of the charge collection efficiency of these devices as a function of substrate doping concentration. In 2005 we used IBIC to map the charge collection efficiency as a function of detector electrode spacing and have shown that the electrode spacing may be set back to more than 50 μm from the implant zone and still retain high charge collection efficiency (see Figure 5). The present IBIC measurements were performed using a focused microprobe of 2 MeV alpha particles, produced by the 5U Pelletron accelerator at the University of Melbourne and the MP2 nuclear microprobe system. A large separation (> 50 μm) between the surface electrodes is highly desirable for qubit fabrication, as it allows sufficient space for subsequent nanofabrication steps near the implantation zone.

Figure 4 The modifications to the previous architecture have improved the stability and reproducibility of the detector fabrication by reduction of the area of the surface metallisation as shown by the central schematic. The electric field in the central construction zone is sustained by an extended p+ well formed by boron diffusion. The NiP structure is being studied for potential future applications.

Figure 5 Results of IBIC measurements showing the charge collection efficiency of the detector regions for increasing electrode spacing showing no loss in efficiency up to 45 μm electrodes spacing. The ion beam for these images was 2 MeV He+.

Figure 6 Single atom implantation event charge pulses for the counted implanted devices on the DP3 chip.

Figure 7 The first successful pulse height spectrum from the Focused Ion Beam system recorded with a sub-100 nm 60 keV Si++ ion implantation events. The ionization arising from interaction of the Si ions with the substrate is similar to that expected from P ions into the same substrate. Left: the single ion implantation module for the FIB; centre: the substrate after Si implantation; right: the corresponding charge pulse height spectrum from the 60 keV Si++ ion implantation events.