Stain etching method leads to formation of porous silicon and the main chemical reaction is results in a continus process of oxidation-followed-by-dissolution in which HNO3 is devoted to injecting holes and oxidizing silicon, and HF for the removal of the formed oxide [8, 9]. Figure 2 shows the typical SEM images of stain-etched black silicon in top view at 30°angle for different etching time. It’s clearly to see that etching appears first in defect and damage area caused by thinning step. As the reaction progress, physical scratches have been converted to dense bowl-shaped corrosion orifices in 5 minutes. And as the corrosion deepens, adjacent corrosion pits appear staggered overlap leading the formation of irregular bowl-shaped structures with a size of 0.8-2μm eventually.
The porosity of BSi rises with the increase of reaction time, but the effective numbers of reflections is greatly reduced due to the erosion in the sidewall of holes. As the results shows in Fig. 4, the average light absorptance of samples is promoted to 80% in visible range and in near-infrared band there is still an improvement of nearly 15% compared to untreated planar silicon. In the procedure of TiN nanoparticles growth, stain-etched BSi samples were deposited on ultrathin layer of titanium in nitrogen environment. The target material of experimental coating film is high purity titanium metal and deposition rate was controlled at 2 nm per second by adjusting the gas flow velocity and intensity of pressure. It can be known from the film kinetic growth process; the atoms of the incident material first form a stable nucleus on the substrate after absorbing or revaporization [24]. As the thickness of the surface layer increases, atoms coalesce at nucleation sites to form nanoislands and result in a continuous film finally. Since the uneven surface of stain-etched BSi, the original linear sedimentation path under the action of gravity in the vacuum chamber is disturbed and makes it more difficult to form a continuous membrane which called self-shadowing effects in obliquely deposited [25]. To ensure the formation of isolated TiN nanoparticles, deposition time was limited to 25 seconds and Fig. 3 shows the SEM images of TiN nanoparticles decorated BSi at different time points.
When depositing 5 seconds, the bottom and side walls of the corroded bowl hole are attached with TiN nanoparticles as shown in Fig. 3(a). And it is found that due to the shadowing effect of the black silicon microstructure during the deposition process, TiN after nucleation accumulation resulting in the formation of particles maximum size is not limited to 10 nm, but distributed in the range of 10 ~ 25 nm, and attachment location is relatively scattered. When the deposition time reaches 15 seconds, the metal particles become denser and appear partially continuous without significiantly change of particle size, nevertheless. In Fig. 3(c), granular deposits of TiN have been replaced by continuous TiN film with nonuniform projections. It also can be demonstrated by the absorption line in Fig. 4 that BSi/10 nm TiN basically retains the absorption characteristics of BSi in long wavelength ultraviolet band which means the coating materials does not form membrane structure. In contrast, BSi with 15 and 25 sec deposition samples shows higher absorption due to low reflectivity of TiN materials to ultraviolet light. And the absorption rate decreased as the deposition thickness increases cause of the reflectance effect of TiN in 600–1100 nm. However, in the near infrared waveband(1100–2500 nm), the LSPs effects of TiN nanoparticles shows extensive enhancement of spectral absorption from 15% to nearly 65% compared to separate BSi materials. As for these samples shows lower performance as exhibited, this is because the continuous TiN film increases reflection of near-infrared light.
To further study the effect of the introduction of LSPs-BSi composite absorption layer on detector performance, we ended up dealing with it in the form of back incidence (structure of PIN detector as shown in Fig. 1(b)) and test the device in terms of spectral response range, responsivity and dark current. Figure 6 gives the results of PIN photodetector with BSi/10 nm TiN absorber, the responsivity of two different devices are re-plotted and exhibited for comparison. Device 1 is a commercial Si-PIN detector(S1336-44BK) and the parameter is taken from the public Website of Hamamatsu Photonics Company [26]. Device 2 is also a BSi-PIN detector forming light-trapping structures on back surface from reference [27]. It can be clearly seen that our newly fabricated detector has higher response after the wavelength exceeds 700 nm and shows satisfactory performance in near infrared band. The response rate was 0.46 A/W at 1060 nm and 0.27 A/W even at 1100 nm. The LSPs-BSi composite detector peak wavelength of 980 nm, responsivity of 0.64 A/W, even above the silicon band gap width of 1100 ~ 1170 nm peak also has more than 10% of the highest response, this part of the responsivity from the impurity level absorb photons, and plasmons materials hot electron collection detection results.
At less than 700 nm in visible spectrum, Device 3 shows a relatively low responsivity compared to commercial PIN detector. This decline is caused by two factors. Firstly, device 1 used an antireflection layer for infrared band on the front of the detector which lowers the incidence of visible light. The second, unlike the normal incidence of device 1, design of back-incidence structure (refer to Fig. 2) enables the photon in the visible band mainly absorbed and completing generation and recombination of carries in the P layer which hardly any photocurrent is produced. This interpretation also applies to the performance of device 2 which has analogous structure. Above 700 nm and near infrared band, the photon energy is able of penetrating P layer and absorbed by LSPs-BSi layer, and then lots of generated carriers can be collected under the action of reverse bias. In the meanwhile, due to the localized surface plasmons resonance effects of TiN nanoparticles and light trapping effects, the original indirect bandgap structure of silicon-based materials has been transformed to quasi-direct band gap which leads the peak response of the detector was redshifted. And the dark current of the detector under working condition is less than 10nA measured in the aluminum black box. From what has been discussed above, the deposition of ultra-thin TiN film on rough black silicon surface can form nanoparticles with metal properties of random size and hybridizing LSP modes. The absorption layer is introduced into the PIN photodetector, which effectively improves the sensitivity of the near-infrared band.