Doping efficiency of phosphorus doped silicon nanocrystals

0 downloads 0 Views 557KB Size Report
Jun 6, 2012 - photoluminescence spectra from passivated and unpassivated samples. Finally .... passivated and non-passivated samples IH/I. Clearly, the.
Doping efficiency of phosphorus doped silicon nanocrystals embedded in a SiO2 matrix S. Gutsch, A. M. Hartel, D. Hiller, N. Zakharov, P. Werner et al. Citation: Appl. Phys. Lett. 100, 233115 (2012); doi: 10.1063/1.4727891 View online: http://dx.doi.org/10.1063/1.4727891 View Table of Contents: http://apl.aip.org/resource/1/APPLAB/v100/i23 Published by the American Institute of Physics.

Related Articles Electronegativity and doping in semiconductors J. Appl. Phys. 112, 046101 (2012) High phosphorous doped germanium: Dopant diffusion and modeling J. Appl. Phys. 112, 034509 (2012) A novel method for effective sodium ion implantation into silicon Rev. Sci. Instrum. 83, 075116 (2012) Superconductor-insulator transition controlled by annealing in Ga implanted Si Appl. Phys. Lett. 100, 262602 (2012) Accurate calculation of field and carrier distributions in doped semiconductors AIP Advances 2, 022135 (2012)

Additional information on Appl. Phys. Lett. Journal Homepage: http://apl.aip.org/ Journal Information: http://apl.aip.org/about/about_the_journal Top downloads: http://apl.aip.org/features/most_downloaded Information for Authors: http://apl.aip.org/authors

Downloaded 11 Sep 2012 to 192.108.69.177. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

APPLIED PHYSICS LETTERS 100, 233115 (2012)

Doping efficiency of phosphorus doped silicon nanocrystals embedded in a SiO2 matrix S. Gutsch,1 A. M. Hartel,1 D. Hiller,1 N. Zakharov,2 P. Werner,2 and M. Zacharias1 1

IMTEK, Faculty of Engineering, Albert-Ludwigs-University Freiburg, Georges-Ko¨hler-Allee 103, 79110 Freiburg, Germany 2 Max-Planck-Institute of Microstructure Physics, Weinberg 2, 06120 Halle

(Received 7 November 2011; accepted 21 May 2012; published online 6 June 2012) Strongly size controlled silicon nanocrystals in silicon oxynitride matrix were prepared using plasma enhanced chemical vapor deposition following the superlattice approach. Doping was achieved by adding diluted phosphine as a precursor gas. Phosphorus quantification was done by secondary ion mass spectrometry. A model based on Poissonian distributions of interface defects and dopants is proposed to calculate the defects and the dopants per silicon nanocrystal as a function of phosphorus concentration. The model requires the comparison between the photoluminescence spectra from passivated and unpassivated samples. Finally, the doping C 2012 efficiency of silicon nanocrystals embedded in silicon oxynitride is estimated to be >20%. V American Institute of Physics. [http://dx.doi.org/10.1063/1.4727891] The observation of visible light extracted from porous silicon1,2 has triggered hopes and efforts for optoelectronic applications of silicon. Especially, the tunability of the Si bandgap by precisely controlling the size of Si nanocrystals (SiNCs) within the superlattice approach3 is expected to increase the efficiency of third generation all silicon tandem solar cells.4 Indeed, an all Si quantum dot based photovoltaic device has been recently demonstrated.5 However, a number of issues especially regarding doping and transport are still in controversial debate. Physical doping of silicon nanostructures with phosphorus is thought to be unlikely because of the large impurity formation energy.6,7 Hence, the doping efficiency is expected to be very low as was already demonstrated for gas phase synthesized SiNC powders.8,9 Therefore, usually very high dopant concentrations are employed.10,11 Up to now, the main processes used for phosphorus (P) doping of SiNCs in a SiO2 matrix are the co-sputtering technique10–12 which normally results in a continuous doping of the whole layer, or ion implantation13–15 resulting in a non-uniform dopant depth profile. However, a selected doping of ultra thin layers (either the barrier or the SiNCs containing active layer) during the deposition process is not possible with these methods. In this work, we will concentrate on phosphorus (ntype) doping of size controlled SiNCs fabricated by PECVD using the superlattice approach.3,16 Phosphorus doping is realized by selectively adding diluted phosphine gas (PH3) during the deposition process. With this technique, we can add dopants either to the nanocrystal precipation layer or to the matrix separation layer or both which allows for careful adjustment of the total dopant concentration. The superlattices used in our experiments consisting of 30 bilayers (3.5 nm/4 nm) of silicon rich silicon oxynitride (SRON) with a composition of SiO1N0.22 and stoichiometric SiO2 were deposited on wet chemically cleaned n-type (100)-Si substrates. The films were grown using an Oxford Instruments “Plasmalab 100” PECVD system with a 13.56 MHz driven parallel plate reactor. SiH4 diluted by high 0003-6951/2012/100(23)/233115/4/$30.00

purity Ar (5% SiH4 in Ar) and N2O were used as precursor gases. Further process details can be found in a recent paper.16 For the phosphorus doping, part of the Ar flow was replaced by a mixture of 1% PH3 in Ar. The flow rate of the diluted phosphine was chosen such that the PH3 to SiH4 flow ratios were either 0.1% or 0.5% for the SRON layers and 2.5% for the SiO2 layers, respectively. Note that the addition of PH3 was checked to have no significant influence on the growth rate. The higher ratio for the SiO2 layer doping is due to a smaller needed SiH4 flow compared to the SRON layers. Further deposition details are given in Table I. Subsequently, the samples were annealed for 1 h in N2 ambient at 1100  C in a quartz tube furnace to form and crystallize the SiNCs within the SRON layers. Some samples were further subjected to an additional forming gas (5% H2 in N2) annealing (450  C, 1 h) to reduce the defect density. Photoluminescence (PL) was measured at room temperature with a LN2-cooled CCD camera attached to a single grating monochromator using a HeCd laser (325 nm line) as excitation source. Dynamic secondary ion mass spectrometry (SIMS) using 5 keV Csþ ions was used to measure the P concentration using a calibrated quantification standard (P in thermal SiO2). Time-of-flight SIMS (ToF-SIMS, 1 keV Csþ) depth profiles were measured to obtain the P distribution within the SRON/SiO2 superlattice structure. In Fig. 1 the ToF-SIMS profiles showing the Si2 and the SiP signal of samples S2 (doped SRON layers) and S3 (doped SiO2 layers) are presented. The superlattice structure is clearly visible in the depth profile by the oscillating waveform of the secondary ion signals. The SiNC layer (formed in the SRON) are indicated by the peaks of the periodical Si2 signal. It turns out that the SiP signal is well in phase with the Si2 signal for both samples. This is particularly interesting, because in sample S2 the SRON was doped whereas in sample S3 the SiO2 was doped instead. The results indicate a preferred P localization in the SRON layers during annealing. A similar result was previously reported for e-beam doped SiNCs.17 The estimated atomic P

100, 233115-1

C 2012 American Institute of Physics V

Downloaded 11 Sep 2012 to 192.108.69.177. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

233115-2

Gutsch et al.

Appl. Phys. Lett. 100, 233115 (2012)

TABLE I. PECVD process parameters and dopant concentrations as measured by quantitative SIMS and PL intensity ratio between passivated and unpassivated samples IH/I. All samples consist of 30 bilayers with 3.5 nm SRON and 4 nm SiO2. H defines the PH3/SiH4 flow ratio in SRON and SiO2 layer, Cp is the total P concentration measured by dynamic SIMS, and RPL is the PL intensity ratio of passivated and unpassivated samples. Sample name S0 S1 S2 S3 S4

HSRON (%)

HSiO2 (%)

CP (at. %)

RPL of IH/I

0 0.1 0.5 0 0.5

0 0 0 2.5 2.5

0 0.15 0.65 1.1 1.75

2.33 1.55 1.22 1.09 1.02

percentages from dynamic SIMS are included in Table I. Our method allows to tune the P concentration from around 0.15 at. % to 1.75 at. % by making use of doping both the SRON layer and the SiO2 barrier layer. Fig. 2 shows the PL spectra for all samples after crystallization and forming gas treatment. The PL intensity decreases as a function of doping and the peak maximum exhibits a clear blueshift of up to 70 meV. This trend can be explained by the interplay of defect passivation and P doping of SiNCs. It has been shown before that P can increase the SiNCs PL intensity at lower concentration,10,18 whereas at higher P concentrations the PL is usually quenched.14,19 The PL increase was attributed to either matrix relaxation effects induced by P incorporation in the surrounding SiO2 matrix, passivation of dangling bonds directly at the SiNC/SiO2 interface, or the charge compensation of P donor electrons by dangling bonds.20 On the other hand, the PL quenching was ascribed to non-radiative Auger recombination of doped or charged SiNCs.21–23 The expected Auger recombination lifetimes are in the range of ns, whereas the exciton lifetime is in the ls range. Measured PL lifetimes for all samples are similar within the time resolution of our setup (1ls). In addition, temperature dependent PL data show no remarkable difference between undoped, passivated, or doped samples. The size distribution of SiNCs within our samples is about 3.5 6 0.5 nm.16 The larger a SiNC, the more efficiently

FIG. 2. PL spectra of samples passivated with hydrogen. A clear decrease of PL intensity accompanied by a blueshift is observed for increased dopant concentrations. The inset shows an underfocus bright field TEM image of sample 4, where the superlattice structure is conserved.

it will be doped and thus will not contribute to the PL intensity as discussed above. Since the PL emission energy is increasing as the size of the SiNC is decreased, the net PL peak will shift to higher energies and the integrated PL intensity is expected to be reduced as observed in our samples, too. In addition, we observe an increase of high energy side of the PL spectrum for lower doping concentration that may be attributed to a predominant reduction of defects for smaller SiNCs. Note that a possible P induced size distribution change cannot be excluded completely at present. However, high resolution TEM imaging shows no SiNC size dependence on increasing P dopant concentration within the measurement accuracy as was observed elsewhere already.11,12 In the inset of Fig. 2, a bright field TEM image of sample S4 (with the highest dopant concentration) is shown, demonstrating the conservation of the superlattice structure. In Table I, we also show the forming gas passivation efficiency that is calculated by taking the PL intensity ratio of passivated and non-passivated samples IH/I. Clearly, the hydrogen passivation effect decreases as a function of doping, highlighting the fact that P does indeed affect the passivation. Along with an overall decreasing PL intensity, we can conclude that in our case, P is beneficial to Si dangling bond reduction and induces a non-H2-passivatable PL quenching effect that is preferably attributed to Auger recombination of P doped SiNCs. In the low excitation regime, the PL intensity ratio IH/I is equal to the ratio of luminescent defect-free SiNCs in the sample before and after H2 passivation. The probability of finding k defects at a SiNC is given by a Poissonian distribution PD ðkÞ ¼

FIG. 1. ToF-SIMS depth profile of samples S2 and S3. For both samples, the Si2 (attributed to the SiNC) is well in phase with the SiP indicating P localization within the SRON layers.

enD  nD k ; k!

(1)

where nD is the average number of defects per SiNC. The ratio IH/I can then be expressed as the ratio of “no defect” probabilities (k ¼ 0). This leads to the following equation: DnD ¼ lnðIH =IÞ;

Downloaded 11 Sep 2012 to 192.108.69.177. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

(2)

233115-3

Gutsch et al.

Appl. Phys. Lett. 100, 233115 (2012)

FIG. 3. Analysis of the PL intensity data correlated to the P dopant concentration from quantitative SIMS measurements: (a) defect reduction per SiNC DnD , (b) P dopants per SiNC nP, and (c) doping efficiency (the lines connecting the data points are guide to the eye).

where the quantity DnD corresponds to the defect reduction per SiNC after H2 passivation. In Fig. 3(a), we plot DnD as a function of P doping, and we can clearly see the reduction of H2-passivatable dangling bond defects as reported before.24 If the amount of dangling bond defect is not affected by adding P, the PL ratio IH/I will remain constant. Similar to Eq. (1), the probability of finding k P atoms within a SiNC is described by a Poissonian distribution as well14 PP ðkÞ ¼

enP  nP k ; k!

(3)

where nP ¼ VNC NSi is the number of dopants per SiNC. VNC is the volume of a single SiNC and NSi is a uniform dopant density. Assuming now that each dopant within a SiNC quenches its luminescence, we find the average number of dopants per SiNC in analogy to Eq. (2) nP ¼ lnðIref ;H =IP;H Þ:

(4)

Here, Iref,H and IP,H are the PL intensities of the passivated, undoped reference sample and passivated, doped samples, respectively. The result is shown in Fig. 3(b). As expected, the number of dopants increases by increasing the doping concentration. An average of one dopant per SiNC is reached at approximately 1 at. % P. From the value of nP one can calculate the doping concentration NSi within the SiNC by dividing by the SiNC volume. Normalizing NSi to the dopant concentration measured by SIMS gives then an estimate for the doping efficiency that is shown in Fig. 3(c). Initially, the doping efficiency is

about 60% but drops significantly as the P concentration is increased. For even higher concentration, the doping efficiency recovers and reaches about 50%. We note that the doping efficiency is surprisingly high for all data points. In particular, the increase at high P concentrations is not intuitive. Furthermore, in Fig. 3(b), we observe a superlinear increase of dopants per SiNC. This observation may indicate a second PL quenching effect that is caused by the saturation of P atoms which might influence the present data. On the one hand, theoretical work by a number of groups6,7 claim that it is energetically unfavorable for P atoms to reside within the SiNC due to the large impurity formation energy in SiNCs, but on the other hand, the diffusion and solubility of P in the SiO2 matrix is significantly reduced with respect to silicon. Experimental studies have shown strong evidence previously that SiNCs can be doped with P if the SiNCs are prepared by a high temperature solid-phase crystallization as is the case in our SiNCs.17,19 We believe therefore that interface segregation kinetics will become more and more significant as the P content is increased. Moreover, we can speculate that at high P concentration, P atoms may either cluster at the interface or induce matrix defects that act as additional PL quenching centers. Noting that the P defect compensation is beneficial for the reduction of interface defects and n-type doping is required for PV applications, the presented work represents an important step towards the realization of Si quantum dot based solar cells. In conclusion, the P doping of size controlled SiNCs using PECVD was demonstrated by either doping the barrier layer or the SRON layer or both during the deposition. The dopant dependent blueshift and the clear decrease of the PL after hydrogen passivation is attributed to the preferential doping of larger sized SiNCs, effectively removing their PL contribution due to Auger recombination induced PL quenching. The optical sensitivity of SiNCs to single P atoms allows us to introduce comparative PL spectroscopy as a simple method to estimate the dopant concentration within the SiNCs which is a most difficult task by other techniques. Within this framework, the doping efficiency was estimated by combining quantitative SIMS with PL intensity data. Furthermore, the interface defect reduction is illustrated by means of a reduced hydrogen passivation efficiency for P doped samples. This work was financially supported by German Research Foundation (Nos. ZA191/24-1 and ZA191/27-1) and by the European Commission (project NASCEnT, FP7-245977). 1

L. T. Canham, Appl. Phys. Lett. 57, 1046 (1990). V. Lehmann and U. Go¨sele, Appl. Phys. Lett. 58, 856 (1991). 3 M. Zacharias, J. Heitmann, R. Scholz, U. Kahler, M. Schmidt, and J. Bla¨sing, Appl. Phys. Lett. 80, 661 (2002). 4 G. Conibeer, M. Green, R. Corkish, Y. Cho, E.-C. Cho, C.-W. Jiang, T. Fangsuwannarak, E. Pink, Y. Huang, T. Puzzer, T. Trupke, B. Richards, A. Shalav, and K.-L. Lin, Thin Solid Films 511–512, 654 (2006). 5 I. Perez-Wurfl, X. Hao, A. Gentle, D.-H. Kim, G. Conibeer, and M. A. Green, Appl. Phys. Lett. 95, 153506 (2009). 6 G. M. Dalpian and J. R. Chelikowsky, Phys. Rev. Lett. 96, 226802 (2006). 7 S. Ossicini, F. Iori, E. Degoli, E. Luppi, R. Magri, R. Poli, G. Cantele, F. Trani, and D. Ninno, IEEE J. Sel. Top. Quantum Electron. 12, 1585 (2006). 8 A. R. Stegner, R. N. Pereira, R. Lechner, K. Klein, H. Wiggers, M. Stutzmann, and M. S. Brandt, Phys. Rev. B 80, 165326 (2009). 2

Downloaded 11 Sep 2012 to 192.108.69.177. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

233115-4 9

Gutsch et al.

X. D. Pi, R. Gresback, R. W. Liptak, S. A. Campbell, and U. Kortshagen, Appl. Phys. Lett. 92, 123102 (2008). 10 M. Fujii, A. Mimura, S. Hayashi, and K. Yamamoto, Appl. Phys. Lett. 75, 184 (1999). 11 X. J. Hao, E.-C. Cho, G. Scardera, E. Bellet-Amalric, D. Bellet, Y. S. Shen, S. Huang, Y. D. Huang, G. Conibeer, and M. A. Green, Thin Solid Films 517, 5646 (2009). 12 X. J. Hao, E.-C. Cho, G. Scardera, Y. S. Shen, E. Bellet-Amalric, D. Bellet, G. Conibeer, and M. A. Green, Sol. Energy Mater. Sol. Cells 93, 1524 (2009). 13 K. Murakami, R. Shirakawa, M. Tsujimura, N. Uchida, N. Fukata, and S.-I. Hishita, J. Appl. Phys. 105, 054307 (2009). 14 A. L. Tchebotareva, M. J. A. de Dood, J. S. Biteen, H. A. Atwater, and A. Polman, J. Lumin. 114, 137 (2005). 15 G. A. Kachurin, S. G. Yanovskaya, D. I. Tetelbaum, and A. N. Mikhailov, Semiconductors 37, 713 (2003).

Appl. Phys. Lett. 100, 233115 (2012) 16

A. M. Hartel, D. Hiller, S. Gutsch, P. Lo¨per, S. Estrade´, F. Peiro´, B. Garrido, and M. Zacharias, Thin Solid Films 520, 121 (2011). M. Perego, C. Bonafos, and M. Fanciulli, Nanotechnology 21, 025602 (2010). 18 M. Fujii, A. Mimura, S. Hayashi, K. Yamamoto, C. Urakawa, and H. Ohta, J. Appl. Phys. 87, 1855 (2000). 19 A. Mimura, M. Fujii, S. Hayashi, D. Kovalev, and F. Koch, Phys. Rev. B 62, 12625 (2000). 20 A. R. Stegner, R. N. Pereira, K. Klein, H. Wiggers, M. S. Brandt, and M. Stutzmann, Physica B 401–402, 541 (2007). 21 C. Delerue, M. Lannoo, G. Allan, E. Martin, I. Mihalcescu, J. C. Vial, R. Romestain, F. Muller, and A. Bsiesy, Phys. Rev. Lett. 75, 2228 (1995). 22 I. Mihalcescu, J. C. Vial, A. Bsiesy, F. Muller, R. Romestain, E. Martin, C. Delerue, M. Lannoo, and G. Allan, Phys. Rev. B 51, 17605 (1995). 23 M. Lannoo, C. Delerue, and G. Allan, J. Lumin. 70, 170 (1996). 24 M. Fujii, A. Mimura, S. Hayashi, Y. Yamamoto, and K. Murakami, Phys. Rev. Lett. 89, 206805 (2002). 17

Downloaded 11 Sep 2012 to 192.108.69.177. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions