The composition of the anodic oxide layers used in stripping was measured by backscattering analysis. Thus, any crystal axis or plane of the sample can be aligned with the direction of the incident analytical beam. This system allowed the analysis to be performed at temperatures lower than the boron ion implantation temperature.
NORMALIZED RANDOM
CHANNEL NUMBER
The incident beam is well collimated with the beam angular spread less than 0.1°. This part of the beam will have an angle of incidence with the direction of crystal symmetry greater than The angular distribution of the analyzing beam is calculated from the number of scattering centers/cm in the amorphous layers.
BACKSCATTERING SPECTRA 1000
ANALYSIS 1.8 MeV He+
The energy corresponding to silicon surface atoms is the midpoint of the rise to the random level. For such a sharp interface, the trailing edge of the amorphous layer corresponds to the midpoint of the drop in the adjusted spectrum from the random spectrum to the minimum of the adjusted spectrum (see Fig. 6). The quantity m depends on the exact behavior of do/dn for small values of the argument 8.
The procedure used is an extension of the method used in the investigation of the minimum yields behind amorphous layers. All particles scattered outside the critical angle are dechanneled from the aligned component of the beam to the random component. A particle in the random component of the beam never becomes part of the aligned component of the beam.
Also, the backscattering or forwardscattering probability of atoms for the aligned component of the analytical beam is not known as a function of the distance of these atoms from the lattice. The fifth assumption requires that the probability of particles in the random component of the beam scattering into the aligned component of the beam is small. The value of N used in the calculation is the total number of lattice atoms corresponding to the sample thickness corresponding to one channel.
The problem of determining the disorder distribution from measurements of the aligned and random backscattered energy spectra.
MATE DETECTOR
RESOLUTION
ALIGNED SPECTRA
RANDOM
SPECTRUM
Relationships of amount of disorder, substrate temperature, and dose are discussed in If this is an indication that the choice of the critical angle for the transition from line beam to random beam was wrong, but this. However, for such a factor, the restored silicon atoms will probably do little to change the shape of the disorder distribution.
The doses are chosen to give similar disturbance densities at the peak of the disturbance for the two implant temperatures. An experimentally measured value (32. ) of the electronic braking power for boron in silicon is used. Assuming that the distribution of the projected drilling range is Gaussian with a standard deviation of 6Rp', the FWHM of the distribution is 2 o/2~n26Rp.
For 200 and 300 keV implants, the interference peak depth for cold implants is -85% of the calculated boron area and for room temperature implants -80% of the calculated boron area. Values for the depths and widths of the damage peaks as calculated by Brice(S2) are included in Table III. However, the measured FWHM values of the interference distribution for implants at room temperature are 60% to 70%.
To achieve the measured widths, these annealing processes must be more effective in the more lightly disturbed wings of the disorder distribution. Next, the effect of the analyzing beam itself on the lattice disorder will be described. A more accurate comparison of the disturbance caused by a given dose is not appropriate at this time, as recent experiments have shown a substantial difference in the amount of disturbance induced at room temperature.
50 °C SPECTRA
CHANNEL
NUMBER
18 is assigned RT because an annealing occurs between implantation temperature and room temperature (RT). Isochronal annealing sequences from -150°C to 300°C were obtained from analyzes performed at the North American Rockwell Science Center and at Caltech. To continue the baking sequence, it was then necessary to remove the sample and continue the process with.
The data for the two parts of the anneal curve for the 200 keV implants were placed in the region from RT to 100°C and are shown in Figs. The shift of the boron disorder annealing curve to higher temperatures than the implantation curve indicates the non-equivalence of the dynamic annealing during implantation and the annealing treatments after the disordered regions have formed. Annealing characteristics (dash–dot lines) for divacancies (1.8 µ absorption band) in samples at room temperature during 400 keV oxygen ion implantation are from Ref.
Growth (dashed line) of divacancies (1.8µ absorption band) in samples at -190°C under 400 keV boron implantation is from Ref.
DIVACANCY ANNEAL
DI VACANCY "'"
GROWTH \
DISORDER \
ANNEAL IMPLANTATION TEMPERATURE (°C)
MeV Hl IONS
5 x 10 ions/cm • However, the interference-reducing effect of the helium or proton analysis beam is not observed. A thermocouple was attached to one of the samples so that the helium and proton analysis beams hit positions that both abutted. This carbon causes energy loss in the incident beam that shifts the leading edge of the silicon spectra to lower energies.
The calculation of the disorder distribution described in (II.E) requires the determination of the sample thickness corresponding to one channel in the backscatter spectrum. This quantity was then used in (111.C) to determine the depth of the disorder peaks and the widths of the scattering center distributions. Next, the results of the direct measurement of the depth scale for the spectra of each alignment will be described.
The area of the anodic oxide layer was determined by a hole in the vinyl coating placed on the sample to be anodized. 2 was simply due to the smaller size of the implanted regions on the samples on which these oxide layers were to be grown.). Since this index was not known for the implanted part of the sample, the control oxides were.
This was the average of the measurement of 21 films and the calculated standard deviation was 35A.
This was the average of the measurements of 21 films, and the calculated standard deviation was 35 A. 0 For oxide T. 2, the surface of the anode layer has energy Ecbs = kSi Ein for silicon and oxygen atoms, respectively, where Ein is. The scale factor f is the ratio of the minimum efficiency determined as an average over ten channels in the aligned spectrum for silicon in the oxide layer, i.e. about 1830 in channel 146 in Figure 2.
22, to the value of the random spectrum (not shown in Fig. 22) in the same energy range. 0 is the total number of counts in the oxygen peak (shaded in Fig. 22), RSi is the total number of counts in the engineered yield of silicon in the oxide layer only (shaded in Fig. 22), and. A direct comparison of the spectra can be made and the level of recovery from silicon in the thermal oxide matches that of silicon in both anodic oxides only after heat treatment.
This result is now consistent with the interferometric measurement of the silicon removal value. The special case of the depth rate expression Eq. 50) to be discussed now is that for the single stretching type of spectrum with 1.8 MeV 4. The He+ beam was aligned with the <110> axis occurring in the implanted and bare region of the sample and a spectrum was recorded.
The energy shift 6E of the disturbance peak was measured based on data similar to that shown in Fig.
The depth scale just presented is for aligned spectra from slightly damaged samples, because the aligned yield from the region of the crystal removed by stripping is -10% of the random yield. The error when using this depth scale throughout the spectrum, even in the region of the clutter peak, is nevertheless small. As the amount of disorder in a region of the specimen increases, the stopping power of that portion of the specimen for the aligned beam increases from the aligned value SA(E.
If we consider this depth dependence of the depth scale, only the calculated maximum depth of it changes. Normalized to the corresponding random values, the peak height ratio is 140 in. the ratio of the smallest values of the aligned spectra behind the corresponding peaks is 110. Annealing was necessary to rearrange the sample lattice sufficiently to allow reproducible anodic oxides to be grown on it.
B~gh of the Institute of Physics, Aarhus University, Aarhus, Denmark, kindly supplied the 400 keV antimony implant. 2 were grown and removed from the sample by the layer removal technique already described with a spectrum recorded after each layer was removed as shown in Fig. The experiment was repeated on another section of the same sample for further removal of two layers.
The energy loss of the helium atoms when they are backscattered from antimony (Sb) is small since for 8.
RAN DOM BACKSCATTER I NG SPECTRA Sb PORTION OF SPECTRUM
NONE ONE
300 CHANNEL NUMBER
The behavior of the analyzing beam was first investigated by investigating the minimum yield of monocrystalline substrates. The insight into the scattering processes obtained from the amorphous layer results was applied to an iterative analysis of the backscatter spectra of discrete disorder peaks. Although the treatment was essentially phenomenological, it provided an adequate description of the desired disorder distributions.
Therefore, the mechanism governing dechanneling depended on the detailed structure of the disordered layer. This effect is attributed to disorder annealing, or perhaps the same nonlinear processes seen for antimony in silicon. The effects of temperature were described with particular attention paid to the drastic dependence of the amount of disorder produced on temperature.
In conclusion, this thesis described a method for obtaining consistent disorder distributions from helium and proton backscattering analyzes of disordered regions in boron-implanted silicon. The application of the backscattering analysis technique to study the composition of thin layers was presented. Sattler, in Proceedings of the Santa Fe Conference on Radiation Effects in Semiconductors, 1967, Radiation Effects in Semiconductors edited by F.L.
Watkins, in Proceedings of the Santa Fe Conference on Radia- tion Effects in Semiconductors, 1967, Radiation Effects in Semi- conductors, geredigeer deur F.L.
