Observaciones desde la filosofía de M.F Sciacca
C) Sociedad y escuela débil
Ion implantation of singly and doubly charged 63Cu was performed at a range of energies to achieve a nearly uniform concentration profile of Cu within the SiO2 film. To ensure the correct selection of doses and energies, TRIM simulations were performed prior to implantation. A total of four samples were implanted, see Table 5- 1 below and Figure 5-3 following page. The implanted Cu was located in excess of 1 mm away from the SiO2–Si interface to prevent any possible interaction with the underlying Si substrate during annealing.
Sample # Table 5-1: Concentration profile parameters. 1
(“1e16”) 2 (“3e16”) 3 (“1e17”) 4 (“3e17”) Implanted at energy E (keV) Dose (ions/cm2)
E = 800 6.37e15 1.91e16 N/A N/A
E = 1430 8.58e15 2.55e16 8.27e16 N/A
E = 2300 1.00e16 3.00e16 1.00e17 3.00e17
E = 3450 1.07e16 3.20e16 1.20e17 N/A
E = 5000 1.29e16 3.87e16 N/A N/A
Median dose (ions/cm2) 1e16 3e16 1e17 3e17 Median Cu conc. (atoms/cm3) 1.8e20 5.5e20 1.8e21 5.5e21 Depth of SiO2 covered by Cu (mm) 0.5 - 2.8 0.5 - 2.8 0.9 - 2.3 1.4 - 1.7
Areal density of Cu (atoms/cm2) 4.86e16 1.45e17 3.03e17 3.00e17 Si substrate
Si substrate 1
2
3
Step 1: Ion implantation at 0.5 - 2.8 mm below the surface of the 5 mm silica film.
Step 2: Annealing of the implanted samples to promote formation of nanocrystals.
Step 3: Selective removal of Si substrate by mechanical grinding and chemical etching for EXAFS experiments.
Note that the concentration of indigenous species in the SiO2 film is approximately 7¥1022 atoms/cm3. The Cu concentrations implanted are in the range of 1.8 – 55 ¥1020 atoms/cm3. Referring to Figure 3-4 this should ensure super-saturation prior to the annealing procedure. However, at high temperatures (1000 – 1100°C) the solubility of Cu approaches 1020 atoms/cm3 as shown in Figure 3-4.
The samples {1 through 4} will henceforth be referred to as 1e16, 3e16, 1e17 and
3e17 respectively, followed by the appropriate annealing temperature, i.e. as-imp,
500°C, 800°C or 1100°C. Also note that the term as-imp will naturally be interchangeable with 21°C (room temperature).
While implanting, the samples were maintained at LN2 temperature to limit diffusion/precipitation. A summary of the implant parameters is tabulated below, Table 5-2.
Figure 5-3: Concentration profiles as predicted by TRIM; a) 1e16 sample; b) 3e16 sample; c) 1e17 sample; d) 3e17 sample.
Sample Isotope Charge state (q) Ion energy (keV) Implant area (cm2) Dose (ions/cm3) Average beam current (mA) Total implant time (hrs) 1e16 63Cu 1 1 2 2 2 800 1430 2300 3450 5000 9.3 6.37e15 8.50e15 1.00e16 1.07e16 1.29e16 3.0 3.0 3.5 5.0 6.5 0.9 1.2 2.4 1.8 1.6 3e16 63Cu 1 1 2 2 2 800 1430 2300 3450 5000 9.3 1.91e16 2.55e16 3.00e16 3.20e16 3.87e16 5.0 4.0 8.0 4.0 8.0 1.6 2.6 3.1 6.6 4.0 1e17 63Cu 1 2 2 1430 2300 3450 6.3 8.27e16 1.00e17 1.20e17 3.0 4.0 5.0 7.7 14.0 13.4
3e17 63Cu 2 2300 6.8 3.00e17 9.0 Total: 20
5.1.2 Furnace annealing
Isochronal annealing was conducted at 500°C, 800°C and 1100°C for 1 hour. The annealing regime was selected based on previous reports, e.g. [30]. Samples were inserted into a quartz tube furnace on a non-implanted, clean Si wafer inside a silica boat. The annealing was conducted under a flowing forming gas (5%H2+95%N2) ambient to prevent oxidation of the nc-Cu.
5.2
Conventional characterisation techniques
5.2.1 X-ray diffraction (XRD)
All samples were scanned with x-rays emerging from a 40 kV, 30 mA Co tube with characteristic Ka1 wavelength of 1.788970 Å. Scans were performed at a glancing incident angle (2°) over a 2q range of [42.5°, 67.5°] covering the two main peaks of crystalline Cu, namely (111) and (200). With a step size of 0.005° (2q) and scan time
of 5.000 s/step, good spectral resolution was achieved. Prior to XRD a 300 nm thick Cu layer was evaporated onto a glass slide providing a bulk reference standard.
Following the XRD scans, the raw data were loaded as txt-files into the Microcal Origin 6.0 software for processing and analysis. This software allows a wide range of processing functions, including gaussian peak fitting.
5.2.2 Rutherford backscattering spectrometry (RBS)
To confirm the existence of Cu in the thin film silica the RBS method was used. This technique also provides the ability of determining the concentration profile as well as an estimate of dose implanted.
The incident beam consisted of 4.233 MeV He++. This incident energy was selected in order to separate the Cu profile from that of the underlying Si as far as possible. The beam current was kept constant at ~45 nA to an accumulated charge of 200 mC on target, with an ion beam spot size of ~1 mmØ. At the beginning of each run, a calibration sample was measured to correlate channel number to energy of the backscattered He ion. The full matrix of samples (16 in total) was characterised using RBS. The raw-data were subsequently analysed using the RUMP (Rutherford Universal Manipulation Program) software.
5.2.3 Transmission electron microscopy (TEM)
The sample preparation process for TEM is a rather longwinded and technical challenging procedure, on the borderline between art and science. In order to achieve good images, it is crucial to perforate the samples while leaving the section of interest very thin (<50 nm), i.e. electron transparent. A schematic outline of the procedure is given in Figure 5-4.
TEM imaging was performed with a Philips CM300 electron microscope using an accelerating voltage of 300 kV. The corresponding wavelength, l, and lens settings resulted in a minimum resolvable distance, rmin, in the micrographs of ~3.5 Å according to [49]:
where Cs is the spherical aberration coefficient of the objective lens.
Both TEM negatives and digital micrographs were recorded during operation.
5.3
Characterisation by synchrotron radiationA customised processing protocol was applied for the isolation of the thin film silica [15]. The objective of this step was to reduce the noise in EXAFS measurements considerably. By isolating the implanted 5 mm layer, and subsequently stacking the
1. Cut the samples into 3¥5 mm pieces. The implanted area is the cross-sectional interface of interest. 3 mm 5 mm Implanted face
2. Clean the samples with acetone and ethanol. Glue two samples together, face to face using Gatan G1 epoxy. Glue this couple to several dummy Si wafers of varying thickness (to identify the interface). Ensure interface is close to centre of sandwich. 5 mm 3 mm 4 mm Dummy wafers Interface
3. Cut the sandwich into 4¥0.7 mm pieces using a 0.006” thick diamond wheel saw. 4 mm
5 mm
4. Core each piece to produce five 3 mmØ discs using a ultrasonic disc cutter.
Interface 4 mm 3 mm
~0.7 mm
5. Mount each disc on a glass stub using
crystalbond.
Progressively grind and polish the first side, finishing with 0.1mm grit diamond paper.
Sample Glass stub
6. Flip the disc over and re- mount on glass stub. Grind sample down to 100mm thickness. Thinned sample Glass stub
7. Use a dimpler to grind a hemispherical dimple ~80mm deep into the surface. Polish the dimple with 0.3mm alumina paste. Remove the sample from the stub using acetone. Glass
stub
Dimple grind & polish, 80mm deep
8. Place the sample in an ion beam miller, which bombards the sample from both faces with 3keV Ar ions at a 5° angle until perforation occurs. Ar
Ar
Figure 5-4: Schematic outline of the TEM sample preparation procedure.
4 / 1 3 min 0.91(Csl ) r ª
silica in layers, the incident synchrotron x-ray beam could be directed through an effective layer thickness of ~6¥5 mm implanted silica.
The majority of the 520 mm Si substrate was removed by mechanical grinding, the remaining 40 mm was removed by a selective anisotropic potassium hydroxide (KOH) etch24. An outline of the procedure is shown schematically in Figure 5-5.
5.3.1 Extended x-ray absorption fine structure (EXAFS)
Samples listed in Table 5-3 were prepared for EXAFS measurements at the Photon Factory, Japan.
Cu Æ 5 mm SiO2/Si Bulk standards
SiO2 (unimplanted 5 mm layer) Cu (300 nm evaporated film) full matrix of
16 implanted samples Cu oxides:
CuO and Cu2O (powder mixed with BN)
The sample material was stacked in an aluminium sample holder with aperture size of 3 mm ¥ 6 mm. The aperture was covered on both sides with x-ray transparent Kapton tape, see Figure 5-6(a). The cryostick was pre-cooled in LN2 before insertion into a closed cycle He cryostat sample chamber for EXAFS measurements. All measurements were performed at liquid He temperature to cancel out thermal
24 See Appendix H for further details.
Step 4. Remove 40 mm Si: Anisotropic KOH etch under light, ~10 hrs
Step 3. Remove 40 mm Si: Manual disc grinding, 1200 grade (15 mm) wet sanding sheets
Step 2. Remove 40 mm Si: Manual disc grinding, 800 grade
(20 mm) wet sanding sheets
Step 1. Remove 400 mm Si: Manual disc grinding, 400 grade (40 mm) wet sanding sheets
Nanocrystals, ~0.5 - 2.8 mm deep from top surface
(100) plane Bulk (100) Si substrate, 520 mm thickness Implanted layer of SiO2, ~5 mm thick
Figure 5-5: Summary of the processing protocol (not to scale)
Table 5-3: Samples prepared for EXAFS analysis.
disorder, and single out structural disorder. By stacking the sample in ~6 layers, high quality EXAFS data were acquired.
Figure 5-6: a) EXAFS sample mounted and attached to the end of a cryostick b) EXAFS scan parameters
†depending upon fluorescence signal intensity from sample.
During insertion of the sample into the cryostat, the chamber was flushed with He and then evacuated to remove any gaseous contaminants. The sample chamber was subsequently cooled to < 15K. Using computer control, the initial energy was set just below the K absorption edge for Cu. Using the incident and transmitted beam ionisation chambers to monitor the normalised flux through the sample, the optical table was scanned in xy-direction to align the sample. The beam energy was then set just above the absorption edge to monitor the emission of fluorescence x-rays. Finally the distance between detector and sample was optimised in terms of count rate. The scan parameters are summarised in Figure 5-6(b).
In summary we have seen that ion implantation and furnace annealing and subsequent sample processing prior to XRD, RBS, TEM and EXAFS measurements is time consuming. However, for the first time, high resolution atomic-scale structural parameter measurements are consequently enabled.
b) a)
Absorber Cu
Edge K
Edge energy (keV) 8.979
Scan range (keV) 8.770 – 10.255
No. of scans 1-3†
25,000/sec
Typical countrate in Ge-detector