Boron-doped a-Si_H_c-Si interface passivation_Degradation m.pdfSolar cells featuring a narrow-band-gap semiconductor sandwiched between two wider-band-gap films have in the past been suggested to allow for maximal device performance.1 In this respect, the study of crystalline silicon (c-Si) surfaces terminated by hydrogenated amorphous silicon (a-Si:H) films has attracted considerable attention in recent years. This is related to the fact that such films can be doped relatively easily either n or p type, allowing for the fabrication of abrupt heterostructure emitters and back surface fields.2,3 Direct deposition of doped a-Si:H films on c-Si surfaces may result in poor interface properties. For this reason, typically, a few nanometer thin intrinsic a-Si:H(i) buffer layer is inserted between the doped films and the substrate.4 Provided no material is epitaxially grown during deposition such intrinsic films can yield outstanding surface passivation, which is controlled by their dangling bond density,5 and may benefit further from low temperature ( ⩽ 260 °C) postdeposition annealing.6 High-efficiency solar cells have been demonstrated by using c-Si(p)/a-Si:H(i)/a-Si:H(p+)/μc-Si(p+) structures (μc-Si denotes microcrystalline silicon) for back electrode formation.7 Nevertheless, for sufficiently thin (few nano-meters) a-Si:H(i) films, the interface passivation quality can deteriorate significantly by subsequent deposition of a boron-doped a-Si:H(p+) overlayer.8
In this letter we show that the presence of a boron-doped a-Si:H(p+) film on top of c-Si/a-Si:H(i) structures may cause H2 effusion from the intrinsic buffer layer at lower temperatures, compared to the case without such over-layer. This suggests generation of Si dangling bond defects by Si–H bond rupture in the buffer layer of the c-Si/a-Si:H(i)/a-Si:H(p+) structure already at moderate temperatures ( ⩽ 260 °C). Consequently, we argue that these defects are the origin of the experienced electronic passivation degradation for such stacked films.
For the experiments, 320 μm thick bifacially mirror po-lished 0.7 Ω cm phosphorus-doped high quality float zone (100) (FZ)-Si wafers are used. For predeposition surface cleaning, the samples are first immersed in a (H2SO4:H2O2) (4:1) solution for 10 min to grow a chemical oxide. This is followed by rinsing in de-ionized water. The oxide is then stripped off in a dilute HF solution (5%) for 30 s. For a-Si:H film deposition, a clustered multichamber parallel plate direct-plasma enhanced chemical vapor deposition (PECVD) system consisting of separate chambers for i- and p+-layer depositions is used. After transfer of the samples to the re-levant deposition chambers and mounting at the top electrodes, the wafer surfaces are exposed to a 200 SCCM (SCCM denotes cubic centimeter per minute at STP) H2 flow for 20 min at a pressure of 0.5 Torr for temperature stabilization. During film deposition, all chambers are operated at radio frequency (rf) (13.56 MHz) power. For soft film deposition, the used power is consistently the minimum required to maintain stable plasmas. To assure that no material is epitaxially grown during deposition, for all films the deposition temperature Tdepo equals 155 °C. The deposition conditions used for all mentioned films are summarized in Table 1, unless otherwise stated. The B2H6 concentration is 4660 ppm in H2. No additional H2 dilution is used during deposition. To evaluate the surface passivation quality of such thin doped a-Si:H stacks, identical structures are deposited on both wafer surfaces. After this, the samples are consecutively stepwise annealed in a vacuum furnace (20 °C increment per step of 30 min, with annealing temperatures Tannstep ranging from 120 to 260 °C). In between these annealing steps, the value for effective carrier lifetime τeff of the samples is measured with a Sinton Consulting WCT-100 quasi-steady-state photoconductance system,9 operated in the so-called generalized mode.10 All reported values for τeff are evaluated at a constant minority carrier injection density of 1.0×1015 cm−3. Since high quality FZ-Si wafers are used, these values can be regarded as a measure for the surface passivation quality. The deposited film thickness dbulk is determined by measuring ellipsometry spectra (ψ,Δ) using a variable angle Woollam M-2000 rotating-compensator instrument. These data are then fitted to a two-layer model to take the 50% void surface roughness thickness drough of the deposited material into account.11 For bulk characterization of the films, thermal desorption spectroscopy (TDS) measurements are taken. For this an ESCO EMD-WA1000S system ope-rated at ultrahigh vacuum (<1.0×10−9 Torr) is used in which the samples are lamp heated up to 1000 °C, with a linear temperature ramp of 20 K min−1. During the annealing, a Balzers AG QMG 421 quadrupole mass spectrometer is used to determine the H2 effusion rate from the a-Si:H films. Boron outdiffusion from such films is determined by secondary ion mass spectroscopy (SIMS). Throughout this letter the following shorthand notations are used (see also insets in Fig. 1): i, p+, and i/p+ for respectively, the c-Si/a-Si:H(i), c-Si/a-Si:H(p+), and c-Si/a-Si:H(i)/a-Si:H(p+) structures.
Fig 1.
(Color online) Influence of stepwise annealing Tannstep on the electronic passivation quality, expressed by the effective surface recombination velo-city Seff of thin intrinsic and doped a-Si:H layers deposited on mirror po-lished (100) FZ-Si(n) surfaces. For such high quality wafers, approximately Seff = d*(2τeff)−1, with d being the wafer thickness. For reference, values for as-deposited films are given as well (labeled a.d. in the abscissa). The lines are guides for the eye. The table in the inset gives the values of dbulk and drough for these films, determined from SE measurements.
Figure 1 shows how the electronic passivation quality of the i, p+, and i/p+ structures changes by stepwise low temperature ( ⩽ 260 °C) postdeposition annealing. The table in the inset gives the values of dbulk and drough for the respective layers. For the i case, the applied annealing treatment is seen to have a beneficial influence on the passivation. This is different for the p+ case: here, annealing rapidly leads to passivation degradation. The latter situation is seen to be only slightly improved when an intrinsic buffer layer (of similar thickness as in the i case) has been inserted underneath the p+ film (i/p+ case): starting with values similar as for the p+ case, initially the passivation quality benefits from annealing. Nevertheless, from about Tannstep = 220 °C onwards degradation sets in. Such trends have been found to be irrespective of the dopant type of low resistivity wafers.12 Figures 2b,2c give TDS data of the same structures, as displayed in Fig. 1. For reference, Fig. 2a gives H2 effusion rate data for a bare c-Si(100) surface directly after HF dip. Similar profiles have been measured in the past (see, e.g., Ref. 13). The low temperature peak β likely originates from simultaneous rupture of two Si–H bonds and formation of H2,14,15 during which also surface structure changes may occur.13 The high temperature peak α is attributed to the desorption of hydrogen related to monohydride rupture.16 Internal reflection spectroscopy data similarly suggest such high temperature peak to be a signature of monohydride.17 Figure 2b shows that for (a few nanometers) thin a-Si:H(i) films similar signals are detected. In this situation, the β peak is considered to be a signature of (interconnected) internal voids and surfaces in the film.15,16 After such low temperature desorption, the films become more compact so that the remaining effusion is li-mited by diffusion.15 Here, a dominating α peak is a signature for dense material. During the initial deposition stages of a-Si:H(i) on c-Si surfaces, a dominant signature (at 2088 cm−1) has been shown to be present by real-time infrared (IR) attenuated total reflection spectroscopy data.18 This signature has been assigned to hydrogen terminated voids in a-Si:H.19 The p+ case shown in Fig. 2b reveals that, compared to the i case, H2 effusion occurs at significant lower temperatures. This phenomenon has been reported earlier in literature for thicker films.20 The same figure also shows similar data for the doped film case after the stepwise annealing cycle, as described in Fig. 1 (label pann+). The crosshatched area in this figure [labeled Δ(p+,pann+)] represents the difference between these two signals and clearly demonstrates that during the latter cycle already significant H2 effusion takes place. Figure 2c compares H2 effusion rate data of the i/p+ structure with that of the superposed i and p+ cases [labeled ∑(i,p+)], under the assumption that the effusion is not limited by H diffusion. It is seen that at low temperatures more hydrogen effuses out for the i/p+ case than for the combined i and p+ cases. Note that the combined i- and p+-layer thicknesses practically equals that of the stacked i/p+ structure (see inset of Fig. 1). Finally, in Fig. 3 the boron profile of a thin a-Si:H(p+) film (again, tdepo = 60 s), sandwiched between two a-Si:H(i) layers, is given. For all films, Tdepo = 155 °C. Results before and after lowtemperature annealing are shown. The asymmetry of the boron depth profile likely is due to crater edge effects which often limit the decay of a SIMS signal. As the changes by annealing are small, the displayed results do not indicate boron diffusion at such low temperatures.
Fig 2.
(Color online) Influence of a linear ramp annealing Tannlin on H2 effusion rate of a-Si:H films as given in Fig. 1: (a) data for a bare mirror polished c-Si(100) surface terminated by a dilute HF dip, (b) data for few nanometer thin single layer a-Si:H films, as deposited and after the low-temperature annealing cycle described in Fig. 1, and (c) data for as-deposited stacked doped films. All data are normalized to the low temperature peak β in panel (a).
Fig 3.
(Color online) SIMS profile of the boron concentration of the structure, as shown in the inset. For the a-Si:H(p+) layer, tdepo = 60 s. For the a-Si:H(i) layers, tdepo = 240 s. For all films Tdepo = 155 °C. The used annealing cycle corresponds to that given in Fig. 1.
The improvement of the passivation quality by low temperature annealing for the i case likely originates from defect reduction of the film close to the interface, as discussed in a previous letter.6 For a-Si:H(i) material annealed at higher temperatures (>300 °C), a correspondence between the H2 effusion rate and defect generation in the film has been de-monstrated in the past by comparing TDS, IR absorption, and electron spin resonance measurements.14 Consequently, for boron-doped a-Si:H(p+) material, the effusion data suggest that hydrogen likely is already transferred at much lower temperatures from a Si–H to a H2 state, creating defects in the material. The Si–H bond rupture energy has been argued to depend on the Fermi energy (rather than on the actual dopants) to explain hydrogen diffusion phenomena in doped a-Si:H material.21 The same mechanism has been attributed as well to doping dependent hydrogen desorption in such material.22 As discussed, in the present case (Fig. 1) postdeposition annealing rapidly results in electronic passivation losses for the p+ case. Again, the origin of this phenomenon likely is Fermi-level dependent defect generation, also occurring close to the a-Si:H/c-Si interface.
For the i/p+ structure, the initial improvement in passivation most probably is due the same phenomenon as experienced for the i case: out annealing of defects in the intrinsic layer. At higher values for Tannstep, the passivation degradation of this stacked structure could be caused, in principle, by boron diffusion into this intrinsic buffer layer. In this respect, for intrinsic PECVD μc-Si growth, it has been argued by photoluminescence measurements that such diffusion from an underlying boron-doped μc-Si(p+) layer may generate harmful defects.23 Nevertheless, the results of Fig. 3 suggest no such diffusion to occur in the present a-Si:H case. Hence, a more likely explanation for the annealing induced passivation losses of the stacked structure may be that due to the presence of the p+ layer on top of the i layer, also in the latter layer the Fermi level will be shifted towards the valence band of the material. In such a case, already at moderate annealing temperatures, Si–H bond rupture can be expected to take place in this buffer layer too. This is evidenced in Fig. 2c, where it can be seen that at lower temperatures more H2 effuses out of the stack than for the two films measured separately. Conversely, at higher temperatures a smaller amount effuses out. Consequently, it must be concluded that the presence of a p+-type overlayer likely enhances Si–H bond rupture in the intrinsic buffer layer, in agreement with the Fermi level dependence of the H stability. As a result, also for the c-Si/a-Si:H(i)/a-Si:H(p+) structure, high Si dangling bond defect densities may be generated close to the c-Si/a-Si:H interface, already at relatively low annealing temperatures. It is also worth noting that the total amount of H2 that effuses out of the stacked i/p+ structure appears to be smaller than for the case of the two films measured separately. This result suggests that already during the deposition of the p+ layer, hydrogen may effuse out from the intrinsic film underneath. In conclusion, the described phenomena demonstrate the need for a careful assessment of temperature treatments during c-Si/a-Si:H heterostructure device fabrication.
To summarize, it was demonstrated in this letter that whereas postdeposition annealing has a beneficial effect on c-Si/a-Si:H(i) interface passivation, this may not be the case for (a few nanometers) thin a-Si:H(i)/a-Si:H(p+) stacked structures deposited on c-Si surfaces. The losses in the latter case likely are related to the lowered Si–H bond rupture energies in a-Si:H(p+) material. In addition, it was shown that the presence of such doped layers may result in lowered Si–H bond rupture energies in underlying a-Si:H(i) films as well. Both effects are attributed to the particular energetic position of the Fermi level in the amorphous material.