Aging effects of porous silicon (PS) and the origin of blue shift are investigated. Photoluminescence (PL) measurements of the PS prepared with HF-EtOH solution showed a 210 meV blue shift after 1.5 months. It is found from deconvolution of the PL spectra that this shift is not fully related to the quantum confinement (QC) effect. For stable PS formation, a HF-EtOH-H2O2 solution is used. A stable luminescence at 2.01 eV with a Gaussian distribution is obtained when the samples are kept in H2O2 for 45 min after the anodization.
© 2003 Optical Society of America
In the past several decades, silicon-based process technology has advanced rapidly and has reached a mature level. Porous silicon (PS), first developed unintentionally by Uhlir while he performed surface cleaning and polishing of silicon (Si) , has become an attractive material. Following this invention, formation of PS has attracted much attention and has been studied widely. PS is used commercially for thin layer transfer because of its advantageous physical structure  and is used in IC technology because of its insulation properties . Thus PS has entered the arena of optoelectronics, particularly after an emission in the visible region from PS was reported in 1990 . Following the 1990s, studies have focused on finding the source of this emission and in the meantime to find application fields in optoelectronics by use of this feature. Even today the reason for the emission has not been fully explained. The most spectacular proposal is the model of the quantum confinement (QC) effect [4–5]. According to this model, PS emits light in the visible region at room temperature, owing to crystal in nanometer dimensions, which is different from bulk Si crystal whose band gap is in the far-infrared region. The emission of PS at room temperature makes the QC model more valid than others . However, surface passivation effects also have to be considered together with the QC model because the model alone cannot explain the aging effect.
In general PS is formed with either electrochemical anodic etching in a HF-based solution or with chemical etching in a HF+oxidized solution. The formation of PS with electrochemical anodic etching in a HF-based solution is the technique most widely used by researchers. In this method, PS is generally formed by use water-diluted HF or HF+EtOH solution as an electrolyte. Then PS is left in ambient air and is then dried with one of the drying methods. While the PL peak intensity of the PS has been observed to increase with time in some studies, others have reported a decrease in this signal. However, most researchers have commonly observed that the position of the PL peak has shifted to the blue region with the passage of time . The change of the PL features with time is caused by the interaction of the PS with oxygen in the ambient . It was also observed that the broadening of the PL spectra was caused by Si-O-Si vibration . Wolkin et al. showed that Si-O-Si bonds were formed from the contact of the PS surface with air even after a short exposure. They also demonstrated that these bonds caused more effective luminescence at 2.1 eV, which is dominant over the original luminescence at 3 eV. This situation has been reported as a red shift . As a result of weakening of these bonds by time and developing of SiO2 on the surface, the PL peak shifts from the red to the more blue region. To remove this drawback, the surface should be passivated. A simple technique for accomplishing this is to oxidize the PS chemically, anodically, or thermally . Yamani et al. reported that the red/orange emission became stable with the passage of time by use of electrochemical anodic etching in a H2O2-HF-based solution; however, the green/blue emission is unstable. It was observed that this was caused by a tiny crystallite formation .
In this study, we investigate aging of PS and the source of blue shift by using two different types of etchants. The first type of PS sample is formed by use of a traditional HF+EtOH solution. It is observed that the shift of the PL peak position is not fully related to the QC because of aging of PS in time. We also demonstrate for the first time to our knowledge that much more stable PS can be fabricated by use of HF+H2O2+EtOH (second type of solution) and by keeping the samples in H2O2 for 45 min after anodization and rinsing in methanol.
The samples used here were n-type Si oriented (100) with 7 Ω cm resistivity. Au/%2.5 Sb was evaporated to the rough back side of the Si, and then annealing at 400 °C for 5 min was carried out for ohmic contact formation. Two types of solution were prepared as electrolyte. The first solution was a mixture of HF (%40) and ethanol (%95) with a 1:1 ratio, and the second solution was a HF (%40):EtOH (%95):H2O2 mixture with a ratio of 6:4:1. The process of electrochemical anodic etching was carried out in a Teflon cell where the sample and platinum wire were selected to be anode and cathode, respectively. A schematic diagram of the cell is shown in Fig. 1.
The samples were illuminated by a 15 mW He-Ne laser at 633 nm, the output of which was expanded to 1 cm diameter, since free holes required for the PS formation in n-type samples can be obtained by a means of illumination with an energy above bandgap. A constant current density of 10 mA/cm2 was applied to all samples for 10 min. Both the current and the illumination were stopped for 1 s after every 3 s to prevent accumulation of gas, which is released during the PS formation, where the sample was in contact with the cell. We aimed not only to prevent the gas accumulation but also to make the freshened electrolyte reach the surface of the sample by use of a mixer in the solution when the formation of the PS was stopped for 1 s. Following the rinsing of the PS sample prepared by the first solution in ethanol, it was left in ambient. Two groups of PS samples were prepared with the second solution. After these samples were rinsed in ethanol, the first group was left in ambient and the others were kept in H2O2 for 45 min. Photoluminescence (PL) was used in the optical characterization of the PS samples. The PL spectra were obtained at room temperature by use of the emission line of Hg at 365 nm. A PMT (Hamamatsu R955) was used as a detector. Luminescence from PS is dispersed with a 50 cm monochromator and a signal from a PMT detected with an SR850 amplifier. The PL measurements were repeated for different periods of times to investigate the aging effect.
3. Results and discussion
PL spectra of the PS samples are shown in Fig. 2 where the first type of solution has been used.
As the time passed, a shift to the blue region in the peak position and a decrease in the PL intensity were observed, which is in agreement with the results reported in the literature. After 1.5 months, the total shift was ~210 meV. To investigate the source of the shift, a Gaussian deconvolution of all the PL spectra in Fig. 2 is done, as illustrated in Fig. 3.
The observed peaks at 1.51, 1.56, and 1.70 eV always exist, and their intensities decreased with aging without shifting, whereas the peak at 1.81 eV shifts to the blue region with increasing intensity in time. In the literature, this blue shift has been attributed to the decrease in dimension of PS pillars resulting from the oxidization of the surface with time, which supports the quantum confinement effect. The peak at 1.70 eV has also been reported to be related to presence of a Si/SiO2 interface . We assume that the origin of the peaks at 1.51 and 1.56 eV is not due to the QC effect but rather to chemical impurity of the PS. The reason for this is that only the intensities of the peaks decrease while the peak positions do not shift and become stable at a certain position. The existence of the three unstable peaks at 1.51, 1.56, and 1.70 eV increases instability significantly. The shift of the peak at 1.81 eV [Fig. 3(a)] to 1.90 eV [Fig. 3(e)] after 1.5 months is probably due to the QC effect as reported in the literature. The PL spectra of the PS samples prepared with the second type of solution are given in Fig. 4.
As shown in Fig. 4, while the peak intensity decreases, the peak position shifts less to the blue region when compared with the PS samples prepared with the first type solution. Although this case is more stable, we achieved for the first time a much more stable luminescence by keeping the samples in H2O2 for 45 minutes following rinsing in ethanol. The PL spectra obtained with this process are illustrated in Fig. 5.
Here not only are the three peaks observed in Fig. 3 removed, but also the Gaussian peak intensity at 2.01 eV remained stable even after 3 months. The possible reasons behind chemical impurity can be explained as follows: a surface area of 1000 m2/cm3 of PS is open to chemical impurities coming from the environment and the electrolyte . Whereas hydrogen and fluorine impurities arise from the electrolyte, carbon and oxygen impurities appear in response to the exposure of the sample to the environment after the anodization. The source of the carbon is not ethanol, because carbon impurities have been reported to occur on the PS samples formed in only a diluted HF solution . As environmental conditions vary from lab to lab, the effect of the amount of carbon on the optical and electrical properties of PS can also change. As a result, the chemical impurities can cause optical instabilities when surface passivation has not been performed immediately.
4. Summary and conclusions
Fabrication and aging effects of PS formed on n-Si have been reported by use of two types of solution. We studied the aging effect of PS samples fabricated in the first type of solution (HF:EtOH) and obtained Gaussian convolution peaks of the PL spectra at 1.51, 1.56, 1.70, and 1.81 eV. The positions of all the deconvoluted peaks were stable independently of time, except that the peak located at 1.81 eV shifted to 1.90 eV after 1.5 months. Also, the intensity of the three peaks decreased except for the peak at 1.81 eV. This possibly resulted because the positions of the three peaks did not change and the peak intensities decreased owing to chemical impurities. The change in the position and intensity of the peak at 1.81 eV could be related to the QC effect.
To ensure the stability of the PL spectra, the surface of the PS samples was intentionally passivated. For this purpose, to the best of our knowledge, the PS samples were kept in H2O2 for 45 min. for the first time, by repeating of the same process as above in the second type solution (6HF:4EtOH:H2O2). Finally, the three peaks disappeared, and a very stable luminescence at 2.01 eV was achieved.
This work has been funded in part by the Ataturk University Research Fund (PN 2001/115) project.
References and links
1. A. Uhlir, “Electrolytic shaping of germanium and silicon,” Bell Syst. Tech. J. 35, 333 (1956).
2. K. Yamagata and T. Yonehara, “Bonding, splitting and thinning by porous Si in ELTRAN ® SOI-Epi Wafer TM,” http://www.canon.co.jp/eltran
3. W.-K. Chen, ed., The VLSI Handbook, ISBN 0-8493-8593-8 (2000).
4. L. T. Canham, “Silicon quantum wire array fabrication by electrochemical and chemical dissolution of wafers,” Appl. Phys. Lett. 57, 1046 (1990). [CrossRef]
5. V. Lehman and U. Gösele, “Porous silicon formation: a quantum wire effect,” Appl. Phys. Lett. 58, 856 (1991). [CrossRef]
6. A. G. Cullis, L. T. Canham, and P. D. J. Calcott, “The structural and luminescence properties of porous silicon,” J. Appl. Phys. 82, 909 (1997). [CrossRef]
7. V. Mulloni and L. Pavesi, “Electrochemically oxidized porous silicon microcavities,” Mater. Sci. Eng. B 69, 59 (2000). [CrossRef]
8. H. Elhouichet and M. Oueslati, “The role of ambient ageing on porous silicon photoluminescence: evidence of phonon contribution,” Appl. Surf. Sci. 191, 1 (2002). [CrossRef]
9. M. V. Wolkinet al., “Electronic states and luminescence in porous silicon quantum dots: the role of oxygen,” Phys. Rev. Lett. 82, 197 (1999). [CrossRef]
10. V. Lehman, Electrochemistry of Silicon ISBN:3-527-60027-2 (2002).
11. Z. Yamaniet al., “Red to green rainbow photoluminescence from unoxidized silicon nanocrystallites,” J. Appl. Phys. 83, 3929 (1998). [CrossRef]
12. T. Yoshidaet al., “Near-IR LEDs fabricated with monodispersed nanocrystallite Si,” Solid State Technol. 45, 41 (2002).
13. O. Bisi, S. Ossicini, and L. Pavesi, “Porous silicon: a quantum sponge structure for silicon based optoelectronics,” Surf. Sci. Rep. 38, 1 (2000). [CrossRef]