Home
Search
Collections
Journals
About
Contact us
My IOPscience
Influence of silicon dioxide capping layers on pore characteristics in nanocrystalline silicon membranes
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2015 Nanotechnology 26 055706 (http://iopscience.iop.org/0957-4484/26/5/055706) View the table of contents for this issue, or go to the journal homepage for more Download details: IP Address: 128.227.24.141 This content was downloaded on 14/06/2017 at 00:20 Please note that terms and conditions apply.
You may also be interested in: Methods for controlling the pore properties of ultra-thin nanocrystalline siliconmembranes D Z Fang, C C Striemer, T R Gaborski et al. Lift-off of large-scale ultrathin nanomembranes Joshua J Miller, Robert N Carter, Kelly B McNabb et al. Ballistic and non-ballistic gas flow through ultrathin nanopores M N Kavalenka, C C Striemer, D Z Fang et al. Tribological and thermal stability study of nanoporous amorphous boron carbide films prepared by pulsed plasma chemical vapor deposition Shahira Liza, Naoto Ohtake, Hiroki Akasaka et al. Fabrication of nanopore arrays and ultrathin silicon nitride membranes byblock-copolymer-assisted lithography Ana-Maria Popa, Philippe Niedermann, Harry Heinzelmann et al. Low-cost fabrication technologies for nanostructures: state-of-the-art and potential A Santos, M J Deen and L F Marsal Effects of anodic aluminum oxide membrane on performance of nanostructured solar cells Hongmei Dang and Vijay Singh Ultrathin transparent membranes for cellular barrier and co-culture models Robert N Carter, Stephanie M Casillo, Andrea R Mazzocchi et al. Tunable nanoporous silicon oxide templates by swift heavy ion tracks technology E Yu Kaniukov, J Ustarroz, D V Yakimchuk et al.
Nanotechnology Nanotechnology 26 (2015) 055706 (7pp)
doi:10.1088/0957-4484/26/5/055706
Influence of silicon dioxide capping layers on pore characteristics in nanocrystalline silicon membranes Chengzhu Qi1,2, Christopher C Striemer3, Thomas R Gaborski4, James L McGrath5 and Philippe M Fauchet2 1
Materials Science Program, University of Rochester, Rochester, NY 14627, USA Department of Electrical Engineering and Computer Science, Vanderbilt University, Nashville, TN 37235, USA 3 SiMPore West Henrietta, NY 14586, USA 4 Department of Chemical and Biomedical Engineering, Rochester Institute of Technology, Rochester, NY 14623, USA 5 Department of Biomedical Engineering, University of Rochester, Rochester, NY 14627, USA 2
E-mail: [email protected] Received 28 July 2014, revised 6 November 2014 Accepted for publication 16 December 2014 Published 15 January 2015 Abstract
Porous nanocrystalline silicon (pnc-Si) membranes are a new class of membrane material with promising applications in biological separations. Pores are formed in a silicon film sandwiched between nm thick silicon dioxide layers during rapid thermal annealing. Controlling pore size is critical in the size-dependent separation applications. In this work, we systematically studied the influence of the silicon dioxide capping layers on pnc-Si membranes. Even a single nm thick top oxide layer is enough to switch from agglomeration to pore formation after annealing. Both the pore size and porosity increase with the thickness of the top oxide, but quickly reach a plateau after 10 nm of oxide. The bottom oxide layer acts as a barrier layer to prevent the a-Si film from undergoing homo-epitaxial growth during annealing. Both the pore size and porosity decrease as the thickness of the bottom oxide layer increases to 100 nm. The decrease of the pore size and porosity is correlated with the increased roughness of the bottom oxide layer, which hinders nanocrystal nucleation and nanopore formation. Keywords: porous membrane, rapid thermal annealing, crystallization (Some figures may appear in colour only in the online journal) 1. Introduction
advantages over other nanoporous membranes. First, the ultrathin nature of this membrane, only a few tens of nanometers thick, is comparable to the size of many biomolecules. The thickness of other porous membranes, such as anodic aluminum oxide membrane [5–7] and trach-etched membrane [8–10], are typically in the range of several hundred nanometers to hundred microns [6–9]. As a result, mass transport through pnc-Si membrane is greatly enhanced [11]. Another advantage is the simple and inexpensive fabrication process, involving a bottom-up approach, rather than a top-down complicated and expensive lithographic approach that porous silicon nitride membranes rely on [12].
Spontaneous formation of pores during annealing of thin amorphous silicon films sandwiched between thin silicon dioxide layers was first reported by Striemer et al in 2007 [1]. Although the pore diameter and porosity are controllable by changing the annealing temperature, a-Si film thickness and capping layer material [1–3], or by coating the pores with graphene [4], the pore formation process remains poorly understood. If the pore formation were understood, it may become possible to extend the range of applications of these porous nanocrystalline silicon (pnc-Si) membranes.This new nanoporous membrane offers several 0957-4484/15/055706+07$33.00
1
© 2015 IOP Publishing Ltd Printed in the UK
Nanotechnology 26 (2015) 055706
C Qi et al
wafers with 100 nm thick thermal oxide is patterned using standard photolithography technique. Three-layer stacks of silicon dioxide/amorphous silicon/silicon dioxide (OSO) are then deposited on the front side of patterned wafers using rfmagnetron sputtering (ATC-2000, AJA International, North Scituate, MA). The deposition temperature is 150 °C for all films. The deposition pressure is 3 mTorr and the substrate bias is 5 W for amorphous Si deposition, and 5 mTorr and 50 W bias for oxide deposition. The deposited amorphous silicon films are kept at 15 nm in all stacks while the thickness of the silicon dioxide capping layers are the variable parameters. We systematically investigated the influence of both the top and bottom oxide layers on pore formation. When we investigated the effect of the top oxide, its thickness was 1 nm, 5 nm, 10 nm, 20 nm, 40 nm, or 100 nm, while the bottom oxide layer was kept at 20 nm. When we investigated the effect of the bottom oxide, its thickness was 5 nm, 20 nm, 40 nm or 100 nm, while the top oxide layer was kept at 20 nm. The bottom oxide layer also serves as an etch stop for the removal of the bulk silicon. We found that a 1 nm bottom oxide is too thin to act as an etching stop. As a result, we could not test the effect of a 1 nm bottom oxide on pore formation. The thickness of each deposited film is determined by the corresponding deposition rate. To obtain the deposition rate of each film, several depositions with different deposition time have been tested and a spectroscopic ellipsometer (VASE, J A Woollam, Lincoln, NE) was used to measure the thickness of deposited films. The deposition rate was determined as the slope of the data points in a thickness versus deposition time graph and deposition time was calculated according to desired thickness. After deposition, solid phase crystallization (SPC) of the amorphous silicon films was induced by rapid thermal annealing (RTA) (Solaris 150, Surface Science Integration, El Mirage, AZ). Samples were annealed in an Ar atmosphere at 1000 °C with a 100 °C s−1 ramp rate and a soak time of 1 min at 1000 °C. After annealing the wafers were etched from the patterned side using a customized single-side Si etch cell and a wet anisotropic etchant, ethylene diamine pyrocatechol with pyrazine (EDP-300 F, Transene Company, Danvers, MA) at 110 °C. Because EDP has a very high silicon-to-oxide etch ratio [21], the bottom oxide layer acted as an etch stop. Finally, the capping oxide layers were removed using buffered oxide etchant (6:1 BOE) to expose the porous silicon membranes.
Figure 1. Process flow for the fabrication of porous nanocrystalline silicon membranes [1].
Current applications of these membranes are listed below. Gaborski et al [13] investigated the separation of nanoparticles with pnc-Si membranes, demonstrating 5 nm resolution with minimal sample loss. Agrawal et al [14] showed that pnc-Si membranes can be used as highly permeable and molecularly thin substrates for cell culture. Snyder et al [15] studied diffusive molecular separations through pnc-Si membranes and their recent result showed that eletroosmotic pump based on pnc-Si membranes could operate at an unprecedented low voltage of 250 mV [16]. Kavalenka et al [17, 18] investigated gas flow behavior through pnc-Si membranes and chemical capacitive vapor sensor using pnc-Si membranes as flexible conductive electrodes. Johnson et al [19] demonstrated a microfluidic dialysis device developed with pnc-Si membranes achieving no resistance to urea passage. DesOrmeaux et al [20] very recently demonstrated nanoporous silicon nitride membranes fabricated from pnc-Si templates with higher burst pressure and greater permeability. In this paper, we investigate the influence on pore formation of a parameter yet unexplored, that is, the thickness of both the top and bottom capping oxide layers in the 1–100 nm range. We find that the porosity and average pore diameter (APD) remain constant until the top oxide layer thickness decreases below 5 nm, when these two characteristics start to rapidly decrease. In contrast, both porosity and APD increase gradually with a decrease in the thickness of the bottom oxide.
2.2. Characterization of pnc-Si membranes
A transmission electron microscope (Hitachi 7650 TEM, Schaumburg, IL) was used to characterize the morphology of the pnc-Si membranes. A customized MATLAB image process script [3] was employed to quantify the pore characteristics from TEM images, including the APD, porosity, and pore distribution (available at http://nanomembranes.org/ resources/software/). The mean values of pore characteristics in each membrane were calculated from at least three individual TEM images taken at different locations. A scanning electron microscope (Zeiss Supra 40VP SEM, Thornwood, NY) and an atomic force microscope (Nanoscan III
2. Methods 2.1. Fabrication of pnc-Si membranes
As previously reported by Striemer et al [1], fabrication of pnc-Si membranes involves a straightforward bottom-up approach (figure 1). First, the backside of (100) n-type silicon 2
Nanotechnology 26 (2015) 055706
C Qi et al
Figure 2. TEM images of annealed 15 nm thick pnc-Si membranes with 20 nm bottom oxide and different thickness top oxide layers ranging from (a) 1 nm, (b) 5 nm, (c) 10 nm, (d) 20 nm, (e) 40 nm, and (f) 100 nm. Pore characteristic plots of annealed samples, (g) porosity, (h) APD (error bar represents the standard deviation of pore size distribution), dash line is for visual guidance and (i) pore size distribution.
3
Nanotechnology 26 (2015) 055706
C Qi et al
Figure 3. SEM images of no-top oxide sample after RTA, (a) top-down view, the white areas represent three-dimensional crystalline Si
islands, and (b) cross-sectional view of Si islands.
capping oxide layer was annealed using the same condition. Figure 3 shows SEM images of this sample after annealing. Instead of a nanoporous silicon film, we observe the formation of large three-dimensional crystalline Si islands. These Si islands are spherical or rodlike and their shapes are irregular. The size of these islands ranges from several tens to a few hundred nanometers in diameter and the height is around 100 nm. This phenomenon is the well-known process of agglomeration [22–27]. The agglomeration of the silicon films is driven by the minimization of surface energy [26, 27]. Our result demonstrates a transition from isolated silicon island formation to continuous silicon film containing pores even when the top oxide layer increase is as little as 1 nm. The top oxide film is essential for the pore formation and acts as a confining layer to prevent the underlying silicon film from undergoing surface-energy-driven agglomeration [26, 27].
AFM, Duebendorf, Switzerland) were used to characterize the surface morphology and roughness.
3. Results and discussion 3.1. Influence from the top oxide layer
Figure 2 shows TEM images and pore characteristics of 15 nm thick pnc-Si membranes with top oxide layers of different thicknesses. The TEM images clearly show that only a few small pores are formed in the sample when the top oxide layer is 1 nm. Both the pore size and pore density increase rapidly as the top oxide layer increases from 1 nm to 5 nm and then seem to saturate when the top oxide layer is between 10 and 100 nm. The porosity and APD plots quantify the trend of increasing porosity and pore size with increasing top oxide layer thickness observed from TEM images. The porosity increases rapidly from 2.9% to 10.5% as the top oxide grows from 1 nm to 5 nm, then reaches an apparent plateau at approximately 12.5% when the top oxide exceeds 10 nm. The highest porosity is 13.1% from samples with a 100 nm top oxide. The APD plot is similar to the porosity plot. The APD starts at 16 nm for a 1 nm top oxide and quickly increases to 22.5 nm with a top oxide of 5 nm. Samples with 100 nm top oxide produce the highest APD of 24.3 nm. The pore distribution plots clearly reveal the difference between the sample with a 1 nm top oxide and the other samples and support the porosity and APD plots. The sharp decrease of the porosity and pore size as the top oxide thickness decreases from 5 nm to 1 nm is noteworthy and indicates that pore formation is very sensitive to the top oxide film in that range. We did not expect that an oxide layer having an effective thickness of 1 nm would still produce pores. To further investigate this effect, a two-layer structure consisting of an amorphous silicon film of 15 nm deposited on a 20 nm bottom silicon dioxide layer with no
3.2. Bottom oxide layer influence
The bottom oxide layer is also important for both pore formation and membrane fabrication. The bottom oxide acts as both the etch stop in the fabrication process and as a barrier layer between the amorphous silicon film and silicon substrate to prevent homo-epitaxial growth [28–31] of the amorphous silicon film during SPC. Bottom oxide layers with four different thicknesses (5 nm, 20 nm, 40 nm and 100 nm) were tested in the OSO stacks. Figure 4 shows the TEM images and pore characteristic plots of pnc-Si membranes with bottom oxide layers of different thicknesses. The influence of the bottom oxide on pore formation is the opposite of that of the top oxide. Pore size and density do not change significantly when the thickness of the bottom oxide layer increased from 5 nm to 40 nm. However, both pore size and density decrease measurably as the bottom oxide layer increased to 100 nm. The pore characteristic plots confirm this inverse trend. The porosity decreases from an average of 13.5% for a 5–40 nm thick bottom oxide, to 2.5% when the bottom oxide layer is 4
Nanotechnology 26 (2015) 055706
C Qi et al
Figure 4. TEM images of annealed 15 nm thick pnc-Si membranes with 20 nm top oxide and different thick bottom oxide layers ranging from (a) 5 nm, (b) 20 nm, (c) 40 nm and (d) 100 nm. Pore characteristic plots of annealed samples, (e) porosity, (f) APD (error bar represents the standard deviation of pore size distribution), dash line is for visual guidance and (g) pore size distribution.
5
Nanotechnology 26 (2015) 055706
C Qi et al
whereas both the pore size and porosity increase rapidly as the thickness of the top oxide layer increases starting at 1 nm and reach an apparent plateau when the top oxide exceeds 10 nm. The bottom oxide acts as a barrier layer to prevent homoepitaxial growth of silicon films during RTA. The pore size and porosity decrease with an increase in the bottom oxide layer thickness has been related to an increase in SiO2/a-Si interface roughness.
Acknowledgments The authors would like to thank D Fang for fabrication assistance, DesOrmeaux and C Chan of SiMPore for ellipsometry assistance; B McIntyre and K Bentley for electron microscopy assistance; M Bindschadler’s work on the pore processing program. Silicon membrane fabrication processing was conducted at the University of Rochester Integrated Nanosystem Center (URnano) and the Semiconductor and Microsystems Fabrication Laboratory (SMFL) at the Rochester Institute of Technology. This work was supported by the National Science Foundation (NSF CBET 1159579). As founders of SiMPore, CCS, TRG, JLM, and PMF declare a competing financial interest in this work.
Figure 5. RMS roughness of silicon dioxide films with different
thicknesses (dash line is for visual guidance).
100 nm. The APD plot shows a similar although less steep trend as the bottom oxide layer increased from 40 nm to 100 nm. The pore size distribution plot shows a clear shift to smaller sizes and lower pore densities for samples with a 100 nm bottom oxide as compared to the other samples. Our results show that silicon crystallization which is associated with pore formation is affected as the bottom oxide becomes thicker. It has been reported that interface roughness strongly affects nucleation during SPC of amorphous silicon films [32, 33]. A rough interface tends to decrease the diffusivity of silicon atoms and hinder nanocrystal nucleation [32], which would also suppress pore formation and growth. AFM was used to measure the roughness of silicon dioxide films deposited on silicon substrates with four different thicknesses (5 nm, 20 nm, 40 nm and 100 nm). Figure 5 plots the root mean square (RMS) roughness of the four different thick silicon dioxide films. The result shows that oxide layers from 5 nm to 40 nm in thickness have a constant RMS roughness of 1.35 Å. However, the RMS roughness of the 100 nm oxide film increases to 1.62 Å. TEM images also show that samples with a 100 nm bottom oxide layer yield smaller Si nanocrystals than the other samples. This supports the claim that a rougher SiO2/a-Si interface tends to suppress Si nanocrystal nucleation and inhibit pore formation and growth. It also explains why the pore characteristics change little even though the thickness of the top oxide layer increases to 100 nm, as the roughness of the top SiO2/a-Si interface remains unchanged as the silicon and bottom oxide thicknesses remain constant. We also did not observe obvious crystal size difference in the silicon films with top oxide layers of various thicknesses beyond 5 nm. However, further study is needed to understand the agglomeration-to-pore formation transition when a very thin top oxide is present.
References [1] Striemer C C, Gaborski T R, McGrath J L and Fauchet P M 2007 Nature 445 749 [2] Fang D Z, Striemer C C, Gaborski T R, McGrath J L and Fauchet P M 2010 J. Phys.: Condens. Matter. 22 454134 [3] Qi C, Striemer C C, Gaborski T R, McGrath J L and Fauchet P M 2014 Small 10 2946 [4] Fang D Z, Striemer C C, Gaborski T R, McGrath J L and Fauchet P M 2010 Nano Lett. 10 3904 [5] Ono S, Saito M and Asoh H 2005 Electrochim. Acta 51 827 [6] Li A P, Muller F, Birner A, Nielsch K and Gosele U 1999 Adv. Mater. 11 483 [7] Li A P, Muller F, Birner A, Nielsch K and Gosele U 1999 J. Vac. Sci. Technol. A 17 1428 [8] Apel P Y and Dmitriev S N 2011 Adv. Nat. Sci.: Nanosci. Nanotechnol. 2 013002 [9] Mukaibo H, Horne L P, Park D and Martin C R 2009 Small 5 2474 [10] Dobrev D, Neumann R, Angert N and Vetter J 2003 Appl. Phys. A 76 787 [11] van den Berg A and Wessling M 2007 Nature 445 726 [12] Tong H D, Jansen H V, Gadgil V J, Bostan C G, Berenschot E, van Rijn C J M and Elwenspoek M 2004 Nano Lett. 4 283 [13] Gaborski T R, Snyder J L, Striemer C C, Fang D Z, Hoffman M, Fauchet P M and McGrath J L 2010 ACS Nano 4 6973 [14] Agrawal A A, Nehilla B J, Reisig K V, Gaborski T R, Fang D Z, Striemer C C, Fauchet P M and McGrath J L 2010 Biomaterials 31 5408 [15] Snyder J L, Clark A Jr, Fang D Z, Gaborski T R, Striemer C C, Fauchet P M and McGrath J L 2010 J. Membr. Sci. 369 119 [16] Snyder J L, Getpreecharsawas J, Fang D Z, Gaborski T R, Striemer C C, Fauchet P M, Borkholder D A and McGrath J L 2013 Proc. Natl Acad. Sci. 110 18425
4. Conclusions In summary, we have demonstrated that both the top and bottom silicon dioxide films are essential for pore formation in pnc-Si membranes. Without the top oxide, silicon crystallization is accomplished by the process of agglomeration 6
Nanotechnology 26 (2015) 055706
C Qi et al
[24] Danielson D T, Sparacin D K, Michel J and Kimerling L C 2006 J. Appl. Phys. 100 083507 [25] Kondo H, Ueyama T, Ikenaga E, Kobayashi K, Sakai A, Ogawa M and Zaima S 2008 Thin Solid Films 517 297 [26] Capellini G, Ciasca G, De Seta M, Notargiacomo A, Evangelisti F and Nardone M 2009 J. Appl. Phys. 105 093525 [27] Nuryadi R, Ishikawa Y, Ono Y and Tabe M 2002 J. Vac. Sci. Technol. B 20 167 [28] Meyerson B S 1986 Appl. Phys. Lett. 48 797 [29] Hsieh T Y, Jung K H, Kwong D L and Lee S K 1991 J. Electrochem. Soc. 138 1188 [30] Pchelyak O P, Lovyagin R N, Krivorot E A, Toropov A I, Aleksand L N and Stenin S I 1973 Phys. Status Solidi A 17 339 [31] Miyazaki T and Adachi S 1992 J. Appl. Phys. 72 5471 [32] Song Y H, Kang S Y, Cho K I and Yoo H J 1997 Etri. J. 19 25 [33] Hadjadj A, Boufendi L, Huet S, Schelz S, Cabarrocas P R I, Estrade-Szwarckopf H and Rousseau B 2000 J. Vac. Sci. Technol. A 18 529
[17] Kavalenka M N, Striemer C C, Fang D Z, Gaborski T R, McGrath J L and Fauchet P M 2012 Nanotechnology 23 145706 [18] Kavalenka M N, Striemer C C, DesOrmeaux J P S, McGrath J L and Fauchet P M 2012 Sensors Actuators B 162 22 [19] Johnson D C, Khire T S, Lyubarskaya Y L, Smith K J P, DesOrmeaux J S, Taylor J G, Gaborski T R, Shestopalov A A, Striemer C C and McGrath J L 2013 Adv. Chronic Kidney Dis. 20 508 [20] DesOrmeaux J P, Winans J D, Wayson S E, Gaborski T R, Khire T S, Striemer C C and McGrath J L 2014 Nanoscale 6 10798 [21] Elwenspoek M C and Jansen H V 1998 Silicon Micromachining (Cambridge: Cambridge University Press) [22] Wakayama Y, Tagami T and Tanaka S 1999 Thin Solid Films 350 300 [23] Wakayama Y, Tagami T and Tanaka S 1999 J. Appl. Phys. 85 8492
7
Search
Collections
Journals
About
Contact us
My IOPscience
Influence of silicon dioxide capping layers on pore characteristics in nanocrystalline silicon membranes
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2015 Nanotechnology 26 055706 (http://iopscience.iop.org/0957-4484/26/5/055706) View the table of contents for this issue, or go to the journal homepage for more Download details: IP Address: 128.227.24.141 This content was downloaded on 14/06/2017 at 00:20 Please note that terms and conditions apply.
You may also be interested in: Methods for controlling the pore properties of ultra-thin nanocrystalline siliconmembranes D Z Fang, C C Striemer, T R Gaborski et al. Lift-off of large-scale ultrathin nanomembranes Joshua J Miller, Robert N Carter, Kelly B McNabb et al. Ballistic and non-ballistic gas flow through ultrathin nanopores M N Kavalenka, C C Striemer, D Z Fang et al. Tribological and thermal stability study of nanoporous amorphous boron carbide films prepared by pulsed plasma chemical vapor deposition Shahira Liza, Naoto Ohtake, Hiroki Akasaka et al. Fabrication of nanopore arrays and ultrathin silicon nitride membranes byblock-copolymer-assisted lithography Ana-Maria Popa, Philippe Niedermann, Harry Heinzelmann et al. Low-cost fabrication technologies for nanostructures: state-of-the-art and potential A Santos, M J Deen and L F Marsal Effects of anodic aluminum oxide membrane on performance of nanostructured solar cells Hongmei Dang and Vijay Singh Ultrathin transparent membranes for cellular barrier and co-culture models Robert N Carter, Stephanie M Casillo, Andrea R Mazzocchi et al. Tunable nanoporous silicon oxide templates by swift heavy ion tracks technology E Yu Kaniukov, J Ustarroz, D V Yakimchuk et al.
Nanotechnology Nanotechnology 26 (2015) 055706 (7pp)
doi:10.1088/0957-4484/26/5/055706
Influence of silicon dioxide capping layers on pore characteristics in nanocrystalline silicon membranes Chengzhu Qi1,2, Christopher C Striemer3, Thomas R Gaborski4, James L McGrath5 and Philippe M Fauchet2 1
Materials Science Program, University of Rochester, Rochester, NY 14627, USA Department of Electrical Engineering and Computer Science, Vanderbilt University, Nashville, TN 37235, USA 3 SiMPore West Henrietta, NY 14586, USA 4 Department of Chemical and Biomedical Engineering, Rochester Institute of Technology, Rochester, NY 14623, USA 5 Department of Biomedical Engineering, University of Rochester, Rochester, NY 14627, USA 2
E-mail: [email protected] Received 28 July 2014, revised 6 November 2014 Accepted for publication 16 December 2014 Published 15 January 2015 Abstract
Porous nanocrystalline silicon (pnc-Si) membranes are a new class of membrane material with promising applications in biological separations. Pores are formed in a silicon film sandwiched between nm thick silicon dioxide layers during rapid thermal annealing. Controlling pore size is critical in the size-dependent separation applications. In this work, we systematically studied the influence of the silicon dioxide capping layers on pnc-Si membranes. Even a single nm thick top oxide layer is enough to switch from agglomeration to pore formation after annealing. Both the pore size and porosity increase with the thickness of the top oxide, but quickly reach a plateau after 10 nm of oxide. The bottom oxide layer acts as a barrier layer to prevent the a-Si film from undergoing homo-epitaxial growth during annealing. Both the pore size and porosity decrease as the thickness of the bottom oxide layer increases to 100 nm. The decrease of the pore size and porosity is correlated with the increased roughness of the bottom oxide layer, which hinders nanocrystal nucleation and nanopore formation. Keywords: porous membrane, rapid thermal annealing, crystallization (Some figures may appear in colour only in the online journal) 1. Introduction
advantages over other nanoporous membranes. First, the ultrathin nature of this membrane, only a few tens of nanometers thick, is comparable to the size of many biomolecules. The thickness of other porous membranes, such as anodic aluminum oxide membrane [5–7] and trach-etched membrane [8–10], are typically in the range of several hundred nanometers to hundred microns [6–9]. As a result, mass transport through pnc-Si membrane is greatly enhanced [11]. Another advantage is the simple and inexpensive fabrication process, involving a bottom-up approach, rather than a top-down complicated and expensive lithographic approach that porous silicon nitride membranes rely on [12].
Spontaneous formation of pores during annealing of thin amorphous silicon films sandwiched between thin silicon dioxide layers was first reported by Striemer et al in 2007 [1]. Although the pore diameter and porosity are controllable by changing the annealing temperature, a-Si film thickness and capping layer material [1–3], or by coating the pores with graphene [4], the pore formation process remains poorly understood. If the pore formation were understood, it may become possible to extend the range of applications of these porous nanocrystalline silicon (pnc-Si) membranes.This new nanoporous membrane offers several 0957-4484/15/055706+07$33.00
1
© 2015 IOP Publishing Ltd Printed in the UK
Nanotechnology 26 (2015) 055706
C Qi et al
wafers with 100 nm thick thermal oxide is patterned using standard photolithography technique. Three-layer stacks of silicon dioxide/amorphous silicon/silicon dioxide (OSO) are then deposited on the front side of patterned wafers using rfmagnetron sputtering (ATC-2000, AJA International, North Scituate, MA). The deposition temperature is 150 °C for all films. The deposition pressure is 3 mTorr and the substrate bias is 5 W for amorphous Si deposition, and 5 mTorr and 50 W bias for oxide deposition. The deposited amorphous silicon films are kept at 15 nm in all stacks while the thickness of the silicon dioxide capping layers are the variable parameters. We systematically investigated the influence of both the top and bottom oxide layers on pore formation. When we investigated the effect of the top oxide, its thickness was 1 nm, 5 nm, 10 nm, 20 nm, 40 nm, or 100 nm, while the bottom oxide layer was kept at 20 nm. When we investigated the effect of the bottom oxide, its thickness was 5 nm, 20 nm, 40 nm or 100 nm, while the top oxide layer was kept at 20 nm. The bottom oxide layer also serves as an etch stop for the removal of the bulk silicon. We found that a 1 nm bottom oxide is too thin to act as an etching stop. As a result, we could not test the effect of a 1 nm bottom oxide on pore formation. The thickness of each deposited film is determined by the corresponding deposition rate. To obtain the deposition rate of each film, several depositions with different deposition time have been tested and a spectroscopic ellipsometer (VASE, J A Woollam, Lincoln, NE) was used to measure the thickness of deposited films. The deposition rate was determined as the slope of the data points in a thickness versus deposition time graph and deposition time was calculated according to desired thickness. After deposition, solid phase crystallization (SPC) of the amorphous silicon films was induced by rapid thermal annealing (RTA) (Solaris 150, Surface Science Integration, El Mirage, AZ). Samples were annealed in an Ar atmosphere at 1000 °C with a 100 °C s−1 ramp rate and a soak time of 1 min at 1000 °C. After annealing the wafers were etched from the patterned side using a customized single-side Si etch cell and a wet anisotropic etchant, ethylene diamine pyrocatechol with pyrazine (EDP-300 F, Transene Company, Danvers, MA) at 110 °C. Because EDP has a very high silicon-to-oxide etch ratio [21], the bottom oxide layer acted as an etch stop. Finally, the capping oxide layers were removed using buffered oxide etchant (6:1 BOE) to expose the porous silicon membranes.
Figure 1. Process flow for the fabrication of porous nanocrystalline silicon membranes [1].
Current applications of these membranes are listed below. Gaborski et al [13] investigated the separation of nanoparticles with pnc-Si membranes, demonstrating 5 nm resolution with minimal sample loss. Agrawal et al [14] showed that pnc-Si membranes can be used as highly permeable and molecularly thin substrates for cell culture. Snyder et al [15] studied diffusive molecular separations through pnc-Si membranes and their recent result showed that eletroosmotic pump based on pnc-Si membranes could operate at an unprecedented low voltage of 250 mV [16]. Kavalenka et al [17, 18] investigated gas flow behavior through pnc-Si membranes and chemical capacitive vapor sensor using pnc-Si membranes as flexible conductive electrodes. Johnson et al [19] demonstrated a microfluidic dialysis device developed with pnc-Si membranes achieving no resistance to urea passage. DesOrmeaux et al [20] very recently demonstrated nanoporous silicon nitride membranes fabricated from pnc-Si templates with higher burst pressure and greater permeability. In this paper, we investigate the influence on pore formation of a parameter yet unexplored, that is, the thickness of both the top and bottom capping oxide layers in the 1–100 nm range. We find that the porosity and average pore diameter (APD) remain constant until the top oxide layer thickness decreases below 5 nm, when these two characteristics start to rapidly decrease. In contrast, both porosity and APD increase gradually with a decrease in the thickness of the bottom oxide.
2.2. Characterization of pnc-Si membranes
A transmission electron microscope (Hitachi 7650 TEM, Schaumburg, IL) was used to characterize the morphology of the pnc-Si membranes. A customized MATLAB image process script [3] was employed to quantify the pore characteristics from TEM images, including the APD, porosity, and pore distribution (available at http://nanomembranes.org/ resources/software/). The mean values of pore characteristics in each membrane were calculated from at least three individual TEM images taken at different locations. A scanning electron microscope (Zeiss Supra 40VP SEM, Thornwood, NY) and an atomic force microscope (Nanoscan III
2. Methods 2.1. Fabrication of pnc-Si membranes
As previously reported by Striemer et al [1], fabrication of pnc-Si membranes involves a straightforward bottom-up approach (figure 1). First, the backside of (100) n-type silicon 2
Nanotechnology 26 (2015) 055706
C Qi et al
Figure 2. TEM images of annealed 15 nm thick pnc-Si membranes with 20 nm bottom oxide and different thickness top oxide layers ranging from (a) 1 nm, (b) 5 nm, (c) 10 nm, (d) 20 nm, (e) 40 nm, and (f) 100 nm. Pore characteristic plots of annealed samples, (g) porosity, (h) APD (error bar represents the standard deviation of pore size distribution), dash line is for visual guidance and (i) pore size distribution.
3
Nanotechnology 26 (2015) 055706
C Qi et al
Figure 3. SEM images of no-top oxide sample after RTA, (a) top-down view, the white areas represent three-dimensional crystalline Si
islands, and (b) cross-sectional view of Si islands.
capping oxide layer was annealed using the same condition. Figure 3 shows SEM images of this sample after annealing. Instead of a nanoporous silicon film, we observe the formation of large three-dimensional crystalline Si islands. These Si islands are spherical or rodlike and their shapes are irregular. The size of these islands ranges from several tens to a few hundred nanometers in diameter and the height is around 100 nm. This phenomenon is the well-known process of agglomeration [22–27]. The agglomeration of the silicon films is driven by the minimization of surface energy [26, 27]. Our result demonstrates a transition from isolated silicon island formation to continuous silicon film containing pores even when the top oxide layer increase is as little as 1 nm. The top oxide film is essential for the pore formation and acts as a confining layer to prevent the underlying silicon film from undergoing surface-energy-driven agglomeration [26, 27].
AFM, Duebendorf, Switzerland) were used to characterize the surface morphology and roughness.
3. Results and discussion 3.1. Influence from the top oxide layer
Figure 2 shows TEM images and pore characteristics of 15 nm thick pnc-Si membranes with top oxide layers of different thicknesses. The TEM images clearly show that only a few small pores are formed in the sample when the top oxide layer is 1 nm. Both the pore size and pore density increase rapidly as the top oxide layer increases from 1 nm to 5 nm and then seem to saturate when the top oxide layer is between 10 and 100 nm. The porosity and APD plots quantify the trend of increasing porosity and pore size with increasing top oxide layer thickness observed from TEM images. The porosity increases rapidly from 2.9% to 10.5% as the top oxide grows from 1 nm to 5 nm, then reaches an apparent plateau at approximately 12.5% when the top oxide exceeds 10 nm. The highest porosity is 13.1% from samples with a 100 nm top oxide. The APD plot is similar to the porosity plot. The APD starts at 16 nm for a 1 nm top oxide and quickly increases to 22.5 nm with a top oxide of 5 nm. Samples with 100 nm top oxide produce the highest APD of 24.3 nm. The pore distribution plots clearly reveal the difference between the sample with a 1 nm top oxide and the other samples and support the porosity and APD plots. The sharp decrease of the porosity and pore size as the top oxide thickness decreases from 5 nm to 1 nm is noteworthy and indicates that pore formation is very sensitive to the top oxide film in that range. We did not expect that an oxide layer having an effective thickness of 1 nm would still produce pores. To further investigate this effect, a two-layer structure consisting of an amorphous silicon film of 15 nm deposited on a 20 nm bottom silicon dioxide layer with no
3.2. Bottom oxide layer influence
The bottom oxide layer is also important for both pore formation and membrane fabrication. The bottom oxide acts as both the etch stop in the fabrication process and as a barrier layer between the amorphous silicon film and silicon substrate to prevent homo-epitaxial growth [28–31] of the amorphous silicon film during SPC. Bottom oxide layers with four different thicknesses (5 nm, 20 nm, 40 nm and 100 nm) were tested in the OSO stacks. Figure 4 shows the TEM images and pore characteristic plots of pnc-Si membranes with bottom oxide layers of different thicknesses. The influence of the bottom oxide on pore formation is the opposite of that of the top oxide. Pore size and density do not change significantly when the thickness of the bottom oxide layer increased from 5 nm to 40 nm. However, both pore size and density decrease measurably as the bottom oxide layer increased to 100 nm. The pore characteristic plots confirm this inverse trend. The porosity decreases from an average of 13.5% for a 5–40 nm thick bottom oxide, to 2.5% when the bottom oxide layer is 4
Nanotechnology 26 (2015) 055706
C Qi et al
Figure 4. TEM images of annealed 15 nm thick pnc-Si membranes with 20 nm top oxide and different thick bottom oxide layers ranging from (a) 5 nm, (b) 20 nm, (c) 40 nm and (d) 100 nm. Pore characteristic plots of annealed samples, (e) porosity, (f) APD (error bar represents the standard deviation of pore size distribution), dash line is for visual guidance and (g) pore size distribution.
5
Nanotechnology 26 (2015) 055706
C Qi et al
whereas both the pore size and porosity increase rapidly as the thickness of the top oxide layer increases starting at 1 nm and reach an apparent plateau when the top oxide exceeds 10 nm. The bottom oxide acts as a barrier layer to prevent homoepitaxial growth of silicon films during RTA. The pore size and porosity decrease with an increase in the bottom oxide layer thickness has been related to an increase in SiO2/a-Si interface roughness.
Acknowledgments The authors would like to thank D Fang for fabrication assistance, DesOrmeaux and C Chan of SiMPore for ellipsometry assistance; B McIntyre and K Bentley for electron microscopy assistance; M Bindschadler’s work on the pore processing program. Silicon membrane fabrication processing was conducted at the University of Rochester Integrated Nanosystem Center (URnano) and the Semiconductor and Microsystems Fabrication Laboratory (SMFL) at the Rochester Institute of Technology. This work was supported by the National Science Foundation (NSF CBET 1159579). As founders of SiMPore, CCS, TRG, JLM, and PMF declare a competing financial interest in this work.
Figure 5. RMS roughness of silicon dioxide films with different
thicknesses (dash line is for visual guidance).
100 nm. The APD plot shows a similar although less steep trend as the bottom oxide layer increased from 40 nm to 100 nm. The pore size distribution plot shows a clear shift to smaller sizes and lower pore densities for samples with a 100 nm bottom oxide as compared to the other samples. Our results show that silicon crystallization which is associated with pore formation is affected as the bottom oxide becomes thicker. It has been reported that interface roughness strongly affects nucleation during SPC of amorphous silicon films [32, 33]. A rough interface tends to decrease the diffusivity of silicon atoms and hinder nanocrystal nucleation [32], which would also suppress pore formation and growth. AFM was used to measure the roughness of silicon dioxide films deposited on silicon substrates with four different thicknesses (5 nm, 20 nm, 40 nm and 100 nm). Figure 5 plots the root mean square (RMS) roughness of the four different thick silicon dioxide films. The result shows that oxide layers from 5 nm to 40 nm in thickness have a constant RMS roughness of 1.35 Å. However, the RMS roughness of the 100 nm oxide film increases to 1.62 Å. TEM images also show that samples with a 100 nm bottom oxide layer yield smaller Si nanocrystals than the other samples. This supports the claim that a rougher SiO2/a-Si interface tends to suppress Si nanocrystal nucleation and inhibit pore formation and growth. It also explains why the pore characteristics change little even though the thickness of the top oxide layer increases to 100 nm, as the roughness of the top SiO2/a-Si interface remains unchanged as the silicon and bottom oxide thicknesses remain constant. We also did not observe obvious crystal size difference in the silicon films with top oxide layers of various thicknesses beyond 5 nm. However, further study is needed to understand the agglomeration-to-pore formation transition when a very thin top oxide is present.
References [1] Striemer C C, Gaborski T R, McGrath J L and Fauchet P M 2007 Nature 445 749 [2] Fang D Z, Striemer C C, Gaborski T R, McGrath J L and Fauchet P M 2010 J. Phys.: Condens. Matter. 22 454134 [3] Qi C, Striemer C C, Gaborski T R, McGrath J L and Fauchet P M 2014 Small 10 2946 [4] Fang D Z, Striemer C C, Gaborski T R, McGrath J L and Fauchet P M 2010 Nano Lett. 10 3904 [5] Ono S, Saito M and Asoh H 2005 Electrochim. Acta 51 827 [6] Li A P, Muller F, Birner A, Nielsch K and Gosele U 1999 Adv. Mater. 11 483 [7] Li A P, Muller F, Birner A, Nielsch K and Gosele U 1999 J. Vac. Sci. Technol. A 17 1428 [8] Apel P Y and Dmitriev S N 2011 Adv. Nat. Sci.: Nanosci. Nanotechnol. 2 013002 [9] Mukaibo H, Horne L P, Park D and Martin C R 2009 Small 5 2474 [10] Dobrev D, Neumann R, Angert N and Vetter J 2003 Appl. Phys. A 76 787 [11] van den Berg A and Wessling M 2007 Nature 445 726 [12] Tong H D, Jansen H V, Gadgil V J, Bostan C G, Berenschot E, van Rijn C J M and Elwenspoek M 2004 Nano Lett. 4 283 [13] Gaborski T R, Snyder J L, Striemer C C, Fang D Z, Hoffman M, Fauchet P M and McGrath J L 2010 ACS Nano 4 6973 [14] Agrawal A A, Nehilla B J, Reisig K V, Gaborski T R, Fang D Z, Striemer C C, Fauchet P M and McGrath J L 2010 Biomaterials 31 5408 [15] Snyder J L, Clark A Jr, Fang D Z, Gaborski T R, Striemer C C, Fauchet P M and McGrath J L 2010 J. Membr. Sci. 369 119 [16] Snyder J L, Getpreecharsawas J, Fang D Z, Gaborski T R, Striemer C C, Fauchet P M, Borkholder D A and McGrath J L 2013 Proc. Natl Acad. Sci. 110 18425
4. Conclusions In summary, we have demonstrated that both the top and bottom silicon dioxide films are essential for pore formation in pnc-Si membranes. Without the top oxide, silicon crystallization is accomplished by the process of agglomeration 6
Nanotechnology 26 (2015) 055706
C Qi et al
[24] Danielson D T, Sparacin D K, Michel J and Kimerling L C 2006 J. Appl. Phys. 100 083507 [25] Kondo H, Ueyama T, Ikenaga E, Kobayashi K, Sakai A, Ogawa M and Zaima S 2008 Thin Solid Films 517 297 [26] Capellini G, Ciasca G, De Seta M, Notargiacomo A, Evangelisti F and Nardone M 2009 J. Appl. Phys. 105 093525 [27] Nuryadi R, Ishikawa Y, Ono Y and Tabe M 2002 J. Vac. Sci. Technol. B 20 167 [28] Meyerson B S 1986 Appl. Phys. Lett. 48 797 [29] Hsieh T Y, Jung K H, Kwong D L and Lee S K 1991 J. Electrochem. Soc. 138 1188 [30] Pchelyak O P, Lovyagin R N, Krivorot E A, Toropov A I, Aleksand L N and Stenin S I 1973 Phys. Status Solidi A 17 339 [31] Miyazaki T and Adachi S 1992 J. Appl. Phys. 72 5471 [32] Song Y H, Kang S Y, Cho K I and Yoo H J 1997 Etri. J. 19 25 [33] Hadjadj A, Boufendi L, Huet S, Schelz S, Cabarrocas P R I, Estrade-Szwarckopf H and Rousseau B 2000 J. Vac. Sci. Technol. A 18 529
[17] Kavalenka M N, Striemer C C, Fang D Z, Gaborski T R, McGrath J L and Fauchet P M 2012 Nanotechnology 23 145706 [18] Kavalenka M N, Striemer C C, DesOrmeaux J P S, McGrath J L and Fauchet P M 2012 Sensors Actuators B 162 22 [19] Johnson D C, Khire T S, Lyubarskaya Y L, Smith K J P, DesOrmeaux J S, Taylor J G, Gaborski T R, Shestopalov A A, Striemer C C and McGrath J L 2013 Adv. Chronic Kidney Dis. 20 508 [20] DesOrmeaux J P, Winans J D, Wayson S E, Gaborski T R, Khire T S, Striemer C C and McGrath J L 2014 Nanoscale 6 10798 [21] Elwenspoek M C and Jansen H V 1998 Silicon Micromachining (Cambridge: Cambridge University Press) [22] Wakayama Y, Tagami T and Tanaka S 1999 Thin Solid Films 350 300 [23] Wakayama Y, Tagami T and Tanaka S 1999 J. Appl. Phys. 85 8492
7
