4.2.1 Reactor setup
The AK400M batch type PECVD reactor from Roth&Rau (Fig. 4-2) features a loadlock helping to achieve good vacuum conditions inside the reactor in successive deposition processes. Inside the reactor chamber, power can be coupled into the plasma either by RF or by microwave excitation or using both sources at the same time. In the case of RF operation, the graphite plate with the substrate is powered and the second electrode is formed by the grounded reactor walls. The additional micro-wave excitation allows for an increase in plasma density and hence in decomposition of the precursor gases without increasing surface damage. This is especially useful for the breaking up of CH4 molecules for the deposition of C-rich films. Two different gas inlet positions (see Fig. 4-2) allow for a further tuning of the deposition conditions.
Amorphous silicon carbide by PECVD 45
Although reactive etch gases such as nitrogen trifluoride (NF3) and oxygen (O2) are connected to the plasma reactor, throughout this work the cleaning of the deposition chamber was mainly performed mechanically by opening the reactor in regular inter-vals. Particularly for the prevention of cross-contamination, the mechanical cleaning turned out to be mandatory when switching from phosphine (PH3) (n-type doping) to diborane (B2H6) (p-type doping) containing processes and vice versa.
4.2.2 Substrate temperature
The deposition temperature for a certain plasma process commonly refers to the Fig. 4-2: Schema of the AK400M PECVD batch-type reactor from Roth&Rau used for the
majority of the a-Si1-xCx depositions performed throughout this work (draft from Roth&Rau company).
Fig. 4-3: Tracking of the substrate temperature (Treal) during a typical deposition process of Si-rich a-Si1-xCx (without plasma).
46 Amorphous silicon carbide by PECVD
nominal temperature measured within the heating system close to the substrate holder (in our case the graphite plate). This temperature can vary drastically from the “real”
temperature that the substrate assumes throughout the process. As the evolution of the a-Si1-xCx film properties with post-deposition annealing as a function of the real depo-sition temperature were of central interest for this work (see chapter 5), direct tempera-ture measurements of shiny etched silicon wafers were carried out for different nomi-nal deposition temperatures at various pressures. The measurement was performed by gluing (ceramic glue) thermocouples to the surface of the wafer. The processes were therefore conducted without actual plasma discharge. The measured substrate tempera-tures are nevertheless assumed to fairly accurately reflect the prevailing temperatempera-tures during the real deposition processes. The results of the temperature measurements are summarized in Fig. 4-3 and Fig. 4-4. The former shows the evolution of the wafer temperature during a typical deposition process of Si-rich a-Si1-xCx and reveals the thermal inertia of the heating process. Fig. 4-4 gives the relation between nominal and substrate temperature, the latter assuming values 130-180°C lower as compared to the set parameter. The measurements furthermore disclose the substrate temperature de-pendence on the chamber pressure. As a consequence the pre-heating period of the substrate in the reactor was henceforth performed in hydrogen ambient at a pressure of 0.3 mbar. The deposition temperatures cited throughout this work refer to the meas-ured “real” substrate temperatures.
4.2.3 C-, B- and P-incorporation in passivating a-Si1-xCx films
In accordance with literature [67], our investigations on the quality and thermal
Fig. 4-4: Measurement of the real temperature (Treal) for a silicon substrate in the AK400M reactor. Left: tuning of nominal temperature (Tnom) at various hydrogen pressures.
Right: nominal temperatures of the AK400M reactor versus measured silicon sub-strate temperatures at various hydrogen pressures.
Amorphous silicon carbide by PECVD 47 stability of the surface passivation properties of amorphous silicon carbide show a sensitive dependence on the C-fraction of the matrix (see chapter 5). In general, for passivation purposes, it is necessary to assure Si-rich a-Si1-xCx films (x < 10 %). For films prepared in the AK400M reactor, a low RF power deposition regime was chosen (15-30 mW/cm2). The power density is sufficiently high to allow for breaking of SiH4 bonds contrary to the more stable CH4 bonds (Table 5-3). The incorporated C-atoms therefore predominantly stem from secondary reactions between SiH4 radicals and CH4
molecules [68]. No microwave excitation was used for the deposition of passivating a-Si1-xCx films. For a quantification of the incorporated carbon, boron and phosphorous atoms for our standard passivation process, SIMS (secondary ion mass spectroscopy) measurements were performed. For this purpose multiple layer stacks with varying precursor gas flows (CH4, B2H6 or PH3) were deposited onto polished silicon sub-strates keeping the remaining process parameters constant. The standard deposition conditions for this process (without doping gases) are: p=0.3 mbar, CH4=30 sccm, SiH4=30 sccm, H2=100 sccm. The growth rate under the given deposition conditions is in the range 3-5 nm/min. The doping gases diborane (B2H6) and phophine (PH3) are highly diluted in hydrogen exhibiting a concentration of 2600-2800 ppm.
Fig. 4-5 through Fig. 4-7 display the SIMS measurement results. The specific atom concentration vs. measurement depth are shown in the graphs on the left side, the correlation with respective gas flows are shown on the right. For the investigated gas flow range, the C-density increases fairly linearly with the methane gas flow, assuming a value of 3-4 % for 30 sccm of CH4. Under the given conditions, the incorporation of phosphorous in the a-Si1-xCx matrix is more effective as compared to boron. For a typically applied doping gas flow of 100 sccm, the measured phosphorous density amounts to 1×1021 cm-3 and the boron density to 5×1020 cm-3.
The conductivity of the as-deposited films with 100 sccm doping gas flow was de-termined on the bases of the TLM (transmission line measurement) principle [115].
For n- as well as for p-type doping the conductivity was found to be in the order of 10-3 Ω-1cm-1.
48 Amorphous silicon carbide by PECVD
Fig. 4-5: Left: SIMS profile (carbon density vs. measurement depth) of a Si-rich a-Si1-xCx stack with tuned C-fraction x. Right: dependence of incorporated carbon concentration on methane gas flow.
Fig. 4-6: Left: SIMS profile (boron density vs. measurement depth) of a Si-rich a-Si1-xCx stack with tuned B-concentration. Right: dependence of incorporated boron concentration on diborane gas flow.
Fig. 4-7: Left: SIMS profile (phosphorus density vs. measurement depth) of a Si-rich a-Si1-xCx
stack with tuned P-concentration. Right: dependence of incorporated phosphorus con-centration on phosphine gas flow.
Amorphous silicon carbide by PECVD 49