Influence of fabrication parameter on the nanostructure and photoluminescence of highly doped p-porous silicon

Journal of Luminescence 146 (2014) 76–82

Contents lists available at ScienceDirect

Journal of Luminescence journal homepage: www.elsevier.com/locate/jlumin

Influence of fabrication parameter on the nanostructure and photoluminescence of highly doped p-porous silicon Shaoyuan Li a,b, Wenhui Ma a,b,n, Yang Zhou a,b,n, Xiuhua Chen c, Mingyu Ma a,b, Yongyin Xiao c, Yaohui Xu a,b a

National Engineering Laboratory for Vacuum Metallurgy, Kunming University of Science and Technology, Kunming 650093, China Faculty of Metallurgical and energy engineering, Kunming University of Science and Technology, Kunming 650093, China c Faculty of Physical Science and Technology, Yunnan University, Kunming 650091, China b

art ic l e i nf o

a b s t r a c t

Article history: Received 8 July 2013 Received in revised form 9 September 2013 Accepted 11 September 2013 Available online 25 September 2013

Porous silicon (PS) was prepared by anodizing highly doped p-type silicon in the solution of H2O/ethanol/ HF. The effects of key fabrication parameters (HF concentration, etching time and current density) on the nanostructure of PS were carefully investigated by AFM, SEM and TEM characterization. According to the experimental results, a more full-fledged model was developed to explain the crack behaviors on PS surface. The photoluminescence (PL) of resulting PS was studied by a fluorescence spectrophotometer and the results show that PL peak positions shift to shorter wavelength with the increasing current density, anodisation time and dilution of electrolyte. The PL spectra blue shift of the sample with higher porosity is confirmed by HRTEM results that the higher porosity results in smaller Si nanocrystals. A linear model (λPL/nm ¼ 620.3–0.595P, R¼ 0.905) was established to describe the correlation between PL peak positions and porosity of PS. & 2013 Elsevier B.V. All rights reserved.

Keywords: Porous silicon Structure characterization Crack behaviors Photoluminescence Porosity-PL peak shift

1. Introduction Since the intense visible photoluminescence and electroluminescence of porous silicon (PS) were discovered [1–3], an extremely large amount of efforts have been focused on this attractive material to integrate optical and electronic functions on-chip [4,5]. Meanwhile, due to its unique properties such as huge specific area, good compatibility with silicon-based IC technology and adjustable physical–chemical structure, the applications of PS in many other areas have also been largely reported. It mainly included chemical sensors and biosensors [6–8], solar cell [9,10], thermal isolation material [11] and biomedical field [12,13], etc. PS has been primarily produced by the electrochemical etching method, which yields a reproducible pore structure if the key parameters can be accurately controlled. The controllable preparation is significant to the application of PS in optical or other fields. It is well known that the photoluminescence (PL) performance of PS is affected by its nanostructure, such as porosity, pores and Si branches sizes and distributions. Although numerous literatures

n Corresponding authors at: National Engineering Laboratory for Vacuum Metallurgy, Kunming University of Science and Technology, Kunming 650093, China. Tel.: þ86 871 6516 1583. E-mail addresses: [email protected] (W. Ma), [email protected] (Y. Zhou).

0022-2313/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jlumin.2013.09.024

[14–16] have already been reported concerning fabrication and optical properties of PS, the relationship between fabrica tion parameters and optical properties of PS is not yet well established [17]. The recent reports on this topic show that the researchers are still attempting to optimize the PS for special optical applications [18,19].

Scheme 1. Schematic of the electrochemical cell with two Pt electrodes.

S. Li et al. / Journal of Luminescence 146 (2014) 76–82

The p þ single crystalline Si wafers are mostly reported as the substrates for fabricating Bragg gratings sensor [20], waveguides [21] and optical biosensor [22]. For this reason, the highly boron doped silicon wafer was selected as the substrate material in

Table 1 The etching conditions and the analysis results of PS sample.

77

present work. The effects of key fabrication parameters (HF concentration, etching time and current density) on the porous structures and PL behaviors were carefully investigated. Finally, a linear model for describing the relationship between porosity of PS and the PL peak position was established.

2. Experimental

Sample no.

Current density (J, mA/cm2)

Etching time (t, min)

HF concentration (C, %)

Porosity (P, %)

Roughness (R, nm)

1-a 1-b 1-c 1-d

20 20 20 20

25 25 25 25

10 8 6.67 5.71

31.2 45.6 61.2 69.6

5.99 10.3 11.8 15.6

2.1. PS preparation A serial of PS samples were fabricated through electrochemical etching in alcoholic HF electrolyte. The double-polished single crystalline p þ -Si (100) wafers with thickness of 480 720 μm and resistivity of 0.01–0.09 Ω cm were used as the original material.

Fig. 1. 2D and 3D AFM photographs of the samples prepared under various HF concentrations.

78

S. Li et al. / Journal of Luminescence 146 (2014) 76–82

Prior to the etching, the wafer was cut into a constant area of 1 cm  1 cm, and cleaned in ethanol and distilled water for 10 min under sonication, respectively. After that, the native SiO2 was removed by immersing the specimens in 10% HF aqueous for 5 min. The treated wafers were installed in the self-made Teflon apparatus (Scheme 1), and two platinum plates were used as the counter electrode. A mixture of 40 wt% hydrofluoric acid (HF), 99.8 wt% ethanol and deionized water (18.25 Ω cm) was used as an electrolyte, and the concentration of HF changed from 10 to 5.71%. The etching was performed at the room temperature under the magnetic stirring. The etching current density (from 10 to 60 mA/cm2) was imposed on two Pt electrodes by dc power, and anodization was performed for 15–60 min. After etching, the PS was rinsed with absolute ethanol and dried under a stream of dry nitrogen. The physical morphology of PS layer was characterized by a QUANTA 200 scanning electron microscope (SEM, FEI) and a 5500 model atomic force microscope (AFM, Agilent). The details of the representative sample were studied by the JEM-2100 high-resolution transmission electron microscope (HRTEM, JEOL). The samples were prepared by spreading PS into ethanol and then salvaging with copper grids. The HRTEM characterization was operated at 200 kV, which was analyzed to obtain interplanar spacing (d) and the selected area electron diffraction pattern. The photoluminescence spectra of the samples were measured by F320 fluorescence spectrophotometer at the room temperature, and the excitation wavelength was 365 nm.

which is attributed to the capillary stress during drying [26]. After mild ultrasound treatment (The power of ultrasonic: 100 W, ultrasonic frequency: 53 KHz, duration time: 30 s), the crack layers are removed in different extents and the pores gradually become visible (as shown in Fig. 1(b)–(d)). For the sample prepared in electrolyte containing 5.71% HF, numerous pores with an average diameter of  100 nm are found after the ultrasound treatment, which are bigger than that of sample 1-b and 1-c. The increased dissolution of silicon in the diluting HF system could be attributed to the increased OH  species with dilution [27].

3. Results and discussion 3.1. Effect of HF concentration on PS surface morphology The effect of HF concentration on the surface morphology was investigated by fixing etching current density and anodization time. The porosity of PS was obtained by the gravimetric method [23] and the results were shown in Table 1. The porosity of PS increases with the decreasing concentration, which is in agreement with previous reports [24,25]. After the etching, the samples showed regular changes of color and contour, the colors gradually became shallow, which changed from black brown to brown and finally to yellow. The sample prepared in 10% HF solution showed nearly uniform smooth surface, however, the samples in lower HF concentrations cracked in various degrees, and the cracks became more serious with the decreasing HF concentration. From the 2D morphology images of the samples, as shown in Fig. 1, no visible pore is observed on the PS surface treated in higher HF system (Fig. 1(a)), which is consistent with the compact and bright macro-morphology. It is attributed to that the less cracks form on the PS surface. For the samples 1-b, 1-c and 1-d, the cracks are observed with naked eye on the top of sample surface,

Table 2 The etching conditions and the analysis results of PS sample. Sample no.

2-a 2-b 2-c 2-d

Current density (J, mA/ cm2)

Etching time (t, min)

15 15 15 15

15 25 45 60

Porosity HF concentration (P, %) (C, %)

Porous layer thickness (h,

5.71 5.71 5.71 5.71

28.1 44.7 66.2 72.4

μm)

6.8 11.5 23 6.5

Main pores diameter (d, nm) 30 710 40 710 65 715 90 720

Fig. 2. Cross sectional images of the samples prepared under various etching time, the contraction cracks are marked by circle and the arrows indicate the direction of stress, the diameters of representative pore channels were marked in (a1)–(d1).

S. Li et al. / Journal of Luminescence 146 (2014) 76–82

79

Table 3 The etching conditions and the analysis results of PS sample. Sample no.

3-a 3-b 3-c 3-d

Current density (J, mA/cm2)

Etching time (t, min)

HF concentration (C, %)

Porosity (P, %)

PS layer

10 25 40 60

50 50 50 50

5.71 5.71 5.71 5.71

61.8 72.9 80.7 -

15.9 15.3 6.2 -

thickness (h,

μm)

Main pores diameter (d, nm) 40 710 80 720 110 720 -

Fig. 3. Cross sectional images of the samples prepared under various etching current density, the shelling layer was marked by circle in (a), (b) and (c), and the direction of stress is marked by the arrows.

80

S. Li et al. / Journal of Luminescence 146 (2014) 76–82

In order to better study the change of PS surface structure, the scale of height of all 3D-view was uniformly set as 110 nm. From Fig. 1(a-1), The PS sample treated in the higher HF concentration is smoother. However, when the electrolyte is diluted massive “hills” are found on the sample surface and the height increases with the decreasing HF concentration. The roughnesses of the samples are counted as 5.99 nm, 10.3 nm, 11.8 nm and 15.6 nm, respectively. It is attributed to that the crack layers obtained in lower HF concentration are more prone to be removed by the sonication. 3.2. Effect of anodization time on the pore structures In order to investigate the effect of anodization time on the thickness and nanostructure of PS layer, we selected the HF concentration of 5.71% as electrolyte. And the preparation parameters and analysis results have been summarized in Table 2. After etching the samples were directly characterized by SEM without any ultrasonic treatment and all of images were shown in Fig. 2. The pore channels of all samples are perpendicular to the silicon substrate. The obvious changes of surface structure can be found by comparing the main views of PS (Fig. 2(a)–(d)). The surface has no crack when etching time is 15 min. But, as anodization proceeds, the cracks appear and become more serious, and fall off after anodization of 60 min. Meanwhile, the cracks have characteristic collapsed “mesa” structures, with a compressed top, which is caused by lateral stress during the evaporation process [28]. The increase of cracks can be attributed to increasing capillary stress with porosity. From the magnification of PS sample (Figs. 2(a-1)–(d-1)), the aligned column-like pores are clearly seen within all of the PS samples and the pore diameters increase with etching time. The mesoporous structure (2–50 nm) can be obtained when the etching time is less than 25 min, for samples anodized for 60 min, the pore diameters increase to about 100 nm. Finally, the relationship between PS thickness and anodization time was investigated in Fig. 2(e). For the short anodization (tr45 min), the thickness of PS increases at a rate of  0.85 μm/min. But the thickness is dramatically reduced when the anodization time is 60 min, which is caused by the drop of crack layers. 3.3. Effect of current density on the pores structure In order to explore how etching current densities affect the pore structures of PS layers, four different current densities (10, 25, 40 and 60 mA/cm2) were selected and the etching time and HF concentration were set as 50 min and 5.71%, respectively. The preparation parameters and analysis results have been summarized in Table 3. The cross-sectional views of PS samples treated in different current density are depicted in Fig. 3. Sample 3-a has a characteristic double-deck structure, the surface layer is slightly cracked and crooked, but it still attaches to the porous layer. For sample 3b and 3-c, the surface layers crack seriously and flake off. The

shelling of the top layer is attributed to the combined effects of the lateral stress and vertical stress during liquid evaporation [28]. Meanwhile, the thickness of PS layer is reduced with the increasing anodization current, which is caused by the formation of thicker shelling layers on PS surface. The electrochemical polish is observed when the current density increases to 60 mA/cm2, and no porous structure is found in the interface layer. The pore structures are observed in the corresponding magnified images, which have significant changes by comparing with the cylinder-like pores in Figs. 2(a)–(d). The consistent pore channels and silicon wires elongation can't be observed in the sample, instead of sponge-like or web-like pore channels, which is responsible for the formation of shelling layer [28]. According to the results presented in Sections 3.2 and 3.3, we summarized the effects of fabrication parameters on the pore structures and the crack behaviors in Table 4. We can find that different pores structures are introduced by various etching currents which lead to different crack behaviors. The relevant explanation model has been presented by Buratto et al. in literature of [28], the authors have explained how the lateral stress and vertical stress were induced by the columnar and weblike pores. In this paper, the crack types induced by the pore geometry are generally consistent with their model. But, compared with their work, the effects of porosity on the cracks were additionally investigated in present work. The results show that the cracks don't occur when the porosity is relatively small, especially for the sample with columnar pores, with the increase of porosity, the cracks become more serious and finally fall off. The change trend of pore diameters with the etching current is just the opposite to that of Burattos’ results [28], but the results similar to ours have been reported by many other researchers [29,30]. We would have preferred to accept the fact observed in this paper, and then a more full-fledged model for explaining the crack behaviors of PS will be expected. The samples prepared at lower current have little pore diameter and pore density, with no cracks after drying, it is attributed to the massive existence of silicon skeleton that can effectively withstand capillary pressures. Instead, the PS samples with bigger pore diameter and porosity are susceptible to pressures, so the surface layers are shrinking or peeling and finally falling off. The conclusion can be drawn that the crack behaviors of PS surface layer are combined effect of the pore geometry and porosity. The pore geometry determines the crack types and the porosity determines the cracks extent.

3.4. TEM and HRTEM characterization of PS nanostructure TEM characterization was employed to investigate more detailed morphology and structure of representative PS samples. According to SEM results above, mesoporous structure can be obtained with the lower etching current and shorter anodization time. The typical HRTEM images and corresponding (001) orientation electron diffraction patterns were shown in Fig. 4.

Table 4 Summary of the pores geometry and the crack behaviors of PS prepared under different etching current and time. Current density (J, mA/cm2)

Thickness (h,

15 15 15 15 10 25 40 60

6.8 11.5 23 6.5 15.9 15.3 6.2 -

μm)

Main pores diameter (d, nm)

Porosity (P, %)

Pore geometry

Crack behaviors

307 10 407 10 657 15 907 20 407 10 807 20 1107 20 -

28.1 44.7 66.2 72.4 61.8 72.9 80.7 -

Column-like Column-like Column-like Column-like Sponge-like Web-like Web-like No

No Shrink Shrink Fall off Little shelling Shelling Shelling Polishing

S. Li et al. / Journal of Luminescence 146 (2014) 76–82

81

Fig. 4. TEM and HRTEM images of the pore structure and Si nanoctrystal (marked by circle in (a-1) and (b-1)).

From Fig. 4(a), we can find an array of pores about 30 nm, often having square shapes. The PS skeleton has single crystal structure. The electron diffraction pattern is closely similar to that of characteristic bulk silicon material, and the diffraction spot of (400) and (220) are assigned in the inset, respectively. There are weak arc-shaped streaks around the main spots, which is attributed to the presence of disoriented structure in the sample. Fig. 4(b) shows larger pore diameters about 80 nm, and the electron diffraction pattern shows more curving spots with larger diffuse scattering. Lots of lattice distortions are found in the Si nanocrystal, which are likely to increase the diffuse scattering and diffraction angle. The capillary pressures within PS is the cause of distortion [31], therefore, it can be concluded that the PS samples with higher porosity have been subjected to higher stresses. Meanwhile, from Figs. 4(a-1) and (b-1), the interplanar distances (d) are respectively 3.318 Å and 3.326 Å for each sample. When they are compared with the report by Martin-Palma et al. [32], the results are very similar and slightly larger than that of (111) plane of bulk Si ( 3.145 Å), which is caused by lattice expansion. The intersecting angle of 70.21 obtained from the Si nanocrystals in Fig. 4(a-1) has a good agreement with that of (111) plane of bulk Si (70.51) [33].

Meanwhile, it is noticed that the intensity of the PL peak also increases with the increasing dilution of the solution, which can be explained by the increasing surface area. The PL peak positions have smaller blue shift from 600 to 570 nm with the increasing anodisation time. The sizes of the residual silicon skeleton decrease with the increasing etching time, which is consistent with the model of the Q.C. effect. From Fig. 5(c), The PL peak positions only move a little on shorter wavelength with the increase of current density from 10 to 40 mA/cm2, which may be attributed to a relatively small change of porosity. The polished sample has no PL peak during the test range. According to the results above, we can make the conclusion that the PL peak positions shift to shorter wavelength with the increasing porosity, which is confirmed by HRTEM results of that the higher porosity brings the smaller Si nanocrystals. Finally, the effect of porosity of PS on PL behaviors was studied and a simple linear equation (λPL =nm ¼ 620:3 0:595P; R ¼ 0:905) was established to describe the relationship between the porosity and the PL peak positions.

3.5. Effect of fabrication parameters on the PL behaviors

In summary, the effects of key fabrication parameters (HF concentration, etching time and current density) on the nanostructure and PL behaviors of PS were carefully investigated by AFM, SEM, TEM and photospectroscopy, respectively. The results show that the pore diameters and porosity are increased with the increasing current density, anodisation time and dilution of electrolyte, while the thickness of PS layer decreases with the increasing current. A more full-fledged model for explaining the crack behaviors of PS is proposed in this paper, which can be confirmed by the fact that the lattice mismatch and smaller Si nanocrystals can be seen more often in the sample with higher porosity, it is responsible for the PL spectra blue shift of PS with

In this part, we studied the effect of fabrication parameters on the PL spectra of PS, which were recorded in the range between 500 and 700 nm, as presented in Fig. 5. The PL spectra have a blue shift from 665 to 580 nm by decreasing HF concentration from 10% to 5.71%. The dimension of Si wires is smaller in the samples prepared with the lower HF concentration than that of sample with higher HF concentration [34]. According to the quantum confinement (Q.C.) effects [35], the mean size of silicon nanoparticles is expected to get smaller with the increasing porosity, and which make PL peaks blue shift.

4. Conclusions

82

S. Li et al. / Journal of Luminescence 146 (2014) 76–82

Fig. 5. Room temperature photoluminescence spectra of PS prepared at (a) various HF concentration, (b) various etching time, (c) various current density and (d) the relationship of porosity and the PL peak positions.

higher porosity. The PL results show that the PL peaks shift to the shorter wavelength with the increasing current density, anodisation time and dilution of the electrolyte. Besides qualitative description, a linear model (λPL =nm ¼ 620:3  0:595P; R ¼ 0:905) is finally established to describe the relationship between the PL behaviors and the porosity, which has great guiding significance to fabricate PS with special PL performance that is desired for different application fields. Acknowledgments The authors are grateful to ShiXing Wang for his help in PL spectrum test. Financial supports of this work from NSFC (50903041), Natural Science Foundation of Yunnan province (2009CD026) and Science Fund of Yunnan Province Education Department (2011J074) were gratefully acknowledged. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]

A.G. Cullis, L.T. Canham, Nature 353 (1991) 335. N. Koshida, H. Nobuyoshi, Appl. Phys. Lett. 60 (1992) 347. T. Karacali, K. Cicek, J. Lumin. 131 (2011) 2100. W. Lang, P. Steiner, F. Kozlowski, J. Lumin. 57 (1993) 341. R.T. Collins, P.M. Fauchet, M.A. Tischler, Phys. Today 50 (1997) 24. C. Baratto, G. Faglia, E. Comini, G. Sberveglieri, A. Taroni, V. La. Ferrara, L. Quercia, G Di. Francia, Sens. Actuators B 77 (2001) 62. S.E. Lewis, J.R. DeBoer, J.L. Gole, P.J. Hesketh, Sens. Actuators B 110 (2005) 54. V.S.-Y. Lin, K. Motesharei, K.-P.S. Dancil, M.J. Sailor, Science 278 (1997) 840. P. Vitanov, M. Kamenova, N. Tyutyundzhiev, M. Delibasheva, E. Goranova, M. Peneva, Thin Solid Films 297 (1997) 299. K.A. Salman, K. Omar, Z. Hassan, Sol. Energy 86 (2012) 541. V. Lysenko, S. Périchon, B. Remaki, D. Barbier, Sens. Actuators A 99 (2002) 13.

[12] E.J. Anglin, L.Y. Cheng, W.R. Freeman, M.J. Sailor, Adv. Drug Deliver. Rev 60 (2008) 1266. [13] L. Vaccari, D. Canton, N. Zaffaroni, R. Villa, M. Tormen, E Di. Fabrizio, Microelectron. Eng. 83 (2006) 1598. [14] H. Elhouichet, B. Bessais, O.B. younes, H. Ezzaouia, M. Oueslati, Thin Solid Films 304 (1997) 358. [15] D.R. Kwon, S. Ghosh, C. Lee, Mater. Sci. Eng. B 103 (2003) 1. [16] T.A. El-Brolossy, Indian J. Phys. 86 (2012) 39. [17] D. Verma, F. Khan, S.N. Singh, P.K. Singh, Sol. Energy Mat. Sol. C 95 (2011) 30. [18] M. Zare, A. Shokrollahi, F.E. Seraji, Appl. Surf. Sci 257 (2011) 9507. [19] N. Chiboub, N. Gabouze, J.N. Chazalviel, F. Ozanam, S. Moulay, A. Manseri, Appl. Surf. Sci. 256 (2010) 3826. [20] I. Rea, M. Iodice, G. Coppola, I. Rendina, A. Marino, L. De Stefano, Sens. Actuators B 139 (2009) 39. [21] A. Shokrollahi, M. Zare, Optik 124 (2013) 855. [22] L De Stefano, P. Arcari, A. Lamberti, C. Sanges, L. Rotiroti, I. Rea, I. Rendina, Sensors 7 (2007) 214. [23] C. Pickering, M.I.J. Beale, D.J. Robbins, P.J. Pearson, R. Greef, J. Phys. C 17 (1984) 6535. [24] R. Herino, G. Bomchil, K. Barla, C. Bertrand, J. Electrochem. Soc. 134 (1987) 1994. ́ [25] J. Dian, A. Macek, D. Nižňanský, I. Němec, V. Vrkoslav, T. Chvojka, I. Jelınek, Appl. Surf. Sci. 238 (2004) 169. [26] O. Belmont, C. Faivre, D. Bellet, Y. Brenchet, Thin Solid Films 276 (1996) 219. [27] X.G. Zhang, Electrochemistry of Silicon and Its Oxide, Kluwer Academic Publisher, New York, 2001. [28] M.D. Mason, D.J. Sirbuly, S.K. Buratto, Thin Solid Films 406 (2002) 151. [29] T. Nakagawa, H. Sugiyama, N. Koshida, Jpn. J. Appl. Phys. 37 (1998) 7186. [30] V. Lehmann, R. Stengl, A. Luigart, Mat. Sci. Eng. B 69–70 (2000) 11. [31] T.L Sudesh, L. Wijesinghe, S.Q. Li, M.B.H. Breese, D.J. Blackwood, Electrochim. Acta 54 (2009) 3671. [32] R.J. Martín-Palma, L. Pascual, P. Herrero, J.M. Martínez-Duart, Appl. Phys. Lett. 81 (2002) 25. [33] H. Kohno, N. Ozaki, H. Yoshida, K. Tanaka, S. Takeda, Cryst. Res. Technol. 38 (2003) 1082. [34] C.-H. Lin, S.C. Lee, Y.-F. Chen, J. Appl. Phys. 75 (1994) 7728. [35] L.T. Canham, Appl. Phys. Lett. 57 (1990) 1046.