Home
Search
Collections
Journals
About
Contact us
My IOPscience
The effect of tensile and bending strain on the electrical properties of p-type 〈110〉 silicon nanowires
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2015 Nanotechnology 26 265703 (http://iopscience.iop.org/0957-4484/26/26/265703) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 165.123.34.86 This content was downloaded on 08/09/2015 at 10:25
Please note that terms and conditions apply.
Nanotechnology Nanotechnology 26 (2015) 265703 (6pp)
doi:10.1088/0957-4484/26/26/265703
The effect of tensile and bending strain on the electrical properties of p-type 〈110〉 silicon nanowires Ruiwen Shao1, Pan Gao1 and Kun Zheng1,2 1
Institute of Microstructure and Properties of Advanced Materials, Beijing University of Technology, Beijing 100124, People’s Republic of China 2 Beijing Key Lab of Microstructure and Property of Advanced Material, Beijing University of Technology, Beijing 100124, People’s Republic of China E-mail: [email protected] Received 17 March 2015, revised 1 May 2015 Accepted for publication 15 May 2015 Published 10 June 2015 Abstract
In this study, electromechanical responses induced by uniaxial tensile and bending deformation were obtained for p-type 〈110〉-oriented Si whiskers by in situ transmission electron microscopy (TEM). Ohmic contacts between the nanowires (NWs) and electrodes were achieved using electron-beam-induced carbon deposition. Results show that enhancements in the carrier transport properties were achieved under both uniaxial tensile and bending strains. With the strain increased to 1.5% before fracture, the improvement in the conductance reached a maximum, which was as large as 24.2%, without any sign of saturation. On the other hand, under 5.8% bending strain, a 67% conductivity enhancement could be achieved. This study should provide important insight into the performance of nanoscale-strained Si. Keywords: piezoresistance effect, electrical properties, silicon nanowires, tensile strain, bending strain (Some figures may appear in colour only in the online journal) complementary metal oxide semiconductor (CMOS) performance improvements [12, 13]. Strain engineering has been considered to be one of the most promising strategies for developing high-performance silicon nanodevices [14]. In addition, there have been attempts to tailor the optical properties of SiNWs by strain engineering. SiNWs have shown a shift in photoluminescence with strain levels of up to 2%, suggesting strain-induced band structure changes [15, 16]. As one of the primary semiconductors, the relationship between the electrical properties of SiNWs and applied strain has attracted extensive attention. In the past decade, tremendous efforts have been dedicated to altering the straininduced carrier transport properties to develop highly sensitive strain gauges or a more effective mode for enhancing the carrier mobility based on theoretical calculations [17–22] or experiments [23–30]. It has been demonstrated that the orientation, size, doping type and level, as well as the strain state, are the primary factors [17, 19, 21, 22] affecting the band structure, effective carrier mass, and electrical properties
One-dimensional nanostructures such as semiconductor nanowires (NWs) are attractive building blocks for the assembly of nanoelectronic and nanophotonic systems due to their tremendous technological potential [1, 2]. Among these materials, silicon nanowires (SiNWs) have attracted widespread attention as a promising structure for future integrated circuits because of their excellent scaling capability and compatibility with Si-based electronic technology [3–5]. SiNWs are also considered attractive building blocks for developing flexible electronics [6–8]. The piezoresistance effect of silicon [9], whereby the electrical resistance changes due to an applied mechanical stress, has been widely exploited in mechanical sensors [10, 11]. Strain plays an important role in determining the band structure and, thereby, elementary physical properties like the effective mass, carrier mobility, or electrical conductivity. More importantly, strained Si on silicon–germanium substrates has been explored in an effort to boost carrier mobility since the early 1990s and is currently accepted as the key solution for 0957-4484/15/265703+06$33.00
1
© 2015 IOP Publishing Ltd Printed in the UK
R Shao et al
Nanotechnology 26 (2015) 265703
corresponding selected area electron diffraction (SAED) pattern, from which the axial direction of the Si whisker can be confirmed to lie along a 〈110〉 direction. In situ electrical measurements were performed using a commercial scanning tunneling microscope–transmission electron microscope (STM–TEM) probing system (Nanofactory Instruments) inserted into a JEOL-2010 TEM (operating at 200 kV). Before the in situ experiment, a fresh tungsten tip was placed on a movable terminal of the STM holder, and Si whiskers were placed on the fixed Ag electrode. The Si whiskers were prepared by FIB and then transferred to the silver wire. The tungsten tip could be moved precisely in three dimensions. Figure 1(a) shows a schematic drawing of the experimental setup, in which a NW is bonded to a W tip and a Ag electrode by electron-beam-induced deposition (EBID). To form good electrical contacts between the SiNWs and the W electrode, a W tip was freshly prepared. Before the experiment, by controlling the piezo-motor, the tungsten tip was moved close and contacted to the Ag electrode. Then, a large current was introduced to the tungsten tip. Due to the Joule heating, the oxide layers on the surface of the tungsten tip would be melted. After this fast melt-quench process, the sharp points of the W tip became hemispherical in shape, as shown in the inset of figure 1(b). EBID was then used to weld the NWs to two electrodes. The details of this process can be found in [32–34]. The contacts between the NW and two electrodes were both ohmic and mechanically robust. By accurately controlling the piezo-motor, the individual NWs could be either stretched or compressed. Simultaneously, the corresponding I–V curves of the deformed whiskers were recorded. Figure 1(b) is a low-magnification TEM image showing two electrodes and a bonded SiNW. Because the piezo-motor could be accurately controlled to move forward and backward, and its repeatability was carefully examined and confirmed, we designed the in situ experiment as follows. After recording the TEM image shown in figure 2(a), the electron beam was turned off, and deformation (compressed or stretched) was induced by advancing the piezomotor four fine steps for each deformation. For each deformation, the corresponding I–V characteristics were measured by sweeping the voltage from −4–4 V. This procedure eliminated the effect of the electron beam on the measured I–V characteristics. The entire experiment was then repeated (with identical deformation control achieved by repeating the stepping of the piezo-motor) with the electron beam switched on (the intensity of the beam was weak but sufficient for capturing TEM images), during which the I–V characteristics were measured again, and the corresponding TEM images of the strained SiNWs were recorded. Figure 2 shows an individual SiNW with a diameter of 181 nm that was fixed between a Ag electrode and a W tip. By placing the W tip far from the Ag electrode, a uniaxial tension test was performed on the SiNW, and the electrical properties were simultaneously measured. Figures 2(a)–(f) show six sequential TEM images of a typical tensile process, and figure 2(g) shows the corresponding I–V curves. By measuring the gauge length of the NW, the strain could be calculated by the DigitalMicrograph software program (version 1.70.16). Although the accuracy of the measured
Figure 1. (a) Experimental setup showing a NW connected to two
electrodes placed inside an STM–TEM probing system. (b) W tip, Ag electrode, and a bonded SiNW. The inset is the corresponding selected area electron diffraction (SAED) pattern, from which the axial direction of the Si whisker can be confirmed to lie along a 〈110〉 direction.
of strained crystalline Si. In terms of directional considerations, 〈100〉, 〈110〉, and 〈111〉 are the most important directions in practical applications. Detailed research about the effect of different orientations on the strain-induced carrier transport properties of SiNWs is desirable. Although the electrical transport properties of strained 〈110〉-oriented, onedimensional Si nanostructures have been studied experimentally [23–29], the results show wide divergences. We note that a slightly enhanced piezoresistance effect has been previously reported for Si beams defined by electron-beam lithography [24]. While He and Yang observed that the longitudinal piezoresistance of p-type SiNWs could be extremely high, exceeding 100-fold the bulk value [27], in addition, A Lugstein et al demonstrated that under ultrahigh strain conditions, p-type single-crystal silicon NWs exhibit an anomalous negative piezoresistance effect [30]. The divergence may result from different conditions of SiNWs in [23–29], such as doping types, doping concentrations, defects, etc To exclude the effects above, a better way is to use the same type of SiNWs with different growth directions. Based on this consideration, in our work, SiNWs with different orientations were prepared from the same Si wafer by a using focused ion beam (FIB) technique. In our previous work, p-type 〈100〉oriented Si whiskers were prepared [31], and in this study, p-type 〈110〉-oriented Si whiskers were obtained. Using in situ transmission electron microscopy (TEM), the effects of strain on the electrical transport properties along the 〈110〉 direction of single-crystalline Si whiskers were studied. A p-type doped Si 〈001〉 film with an electrical resistance of 14 ∼ 22 Ω · cm, lateral dimensions of 1 mm × 100 μm (with side edges of {100} facets), and a thickness of 300 nm was used to prepare Si 〈110〉-oriented whiskers using FIB technology. The fabricated Si 〈110〉-oriented whiskers showed lateral dimensions on the order of several hundred nanometers. The prepared Si whiskers could be transferred to a silver wire, which was used as an electrode. Figure 1(b) shows a TEM image of a typical Si whisker; the inset is the 2
R Shao et al
Nanotechnology 26 (2015) 265703
Figure 2. Images of SiNWs under tensile strain and the corresponding I–V curves. (a)–(e) Images of the SiNWs under tensile strain and (f) image of the fractured NW. (g) The corresponding I–V curves.
length obtained for any nanostructure using this software may vary with the imaging conditions, the relative change in length and, in turn, the corresponding tensile strain can be accurately determined by comparing the change in the measured lengths during deformation. The initial length was 958 nm, as shown in figure 2(a). The measured lengths were 962, 966, 970, and 973 nm after uniaxial tension was applied, as shown in figures 2(b)–(e), and the corresponding calculated strains were 0.42, 0.74, 1.16, and 1.36%, respectively. Compared with that obtained in our previous tensile experiments for SiC nanowires [33], the fractured strain is far lower. This discrepancy may be due to the defects induced by FIB. The electrical properties of the NW were measured under tensile strain, and the corresponding I–V curves are shown in figure 2(g). The voltage was swept from −4–4 V. In the strain-free case, the black I–V curve (curve ‘a’) indicates the intrinsic electrical properties. The linear I–V characteristic of the NW indicates that good ohmic contacts were achieved. It is clear that the conductance was enhanced with an increase in the tensile strain. When we pulled the NW again, it broke at the junction between the NW and the Ag electrode, which is shown in figure 2(f). In this tensile test, the final measured strain approaches 1.5% before fracture. This failure may have been induced by torsional force. At the same time, it can be observed that there was no current signal in the NW’s I–V curve (curve ‘f’ in figure 2(g)). The above-described analysis qualitatively demonstrated the change in current with an increase in the tensile strain. Next, we discuss the quantitative relationship between the rate
Figure 3. The current improvement rate versus the strain.
at which the current was improved and the strain. In this case, the current values were measured at an applied voltage of 4 V. Figure 3 shows the relationship between the rate of current improvement and the strain. When the strain approached 1.5% before fracture, the change in the conductance reached a maximum, reaching a value as large as 24.2%, without any sign of saturation. In fact, the slope of the red curve determined by fitting the data varies with the strain state and exhibits a gentle rise with the increase in the tensile strain. The rate of current enhancement reached a maximum when 3
R Shao et al
Nanotechnology 26 (2015) 265703
Figure 4. Images of the bent SiNWs and the corresponding I–V curves. (a)–(e) Images of the bent SiNW and (f) image of the released NW. (g) The corresponding I–V curves.
bent NW. A series of strains could then be obtained, namely, 2.4% (state 2), 2.8% (state 3), 3.4% (state 4), and 3.9% (state 5), respectively. Figure 4(g) shows the corresponding I–V curves that were simultaneously measured. Similarly to the case in which the NWs were placed in uniaxial tension, an enhancement in the electrical conductivity with increasing bending strain was observed. When the NW end was returned to its original position, as shown in figure 4(g), although the NW still showed some deformation, the conductance was nearly the same as the original value (curve ‘f’ in figure 4(g)). We then investigated the reproducibility and stability of the strain effect on the current in the SiNWs, and the results are shown in figure 4. In this test, we obtained the current response of the NWs over an entire stretch–release cycle under a fixed bias of 3 V. Initially, no stress was applied, and the current was maintained at 42.9 ± 0.2 nA. By controlling the movement of the W tip and applying a bending stress to the NW, the current increased gradually. The current values were 43.5, 44.7, 46.0, and 47.6 nA, which correspond to strains of 0.7%, 1.4%, 2.1%, and 2.8%. Then, when we released the applied stress, the current decreased and finally returned to the initial state. Whole cycles were performed, and the results could be repeated. Our results indicate that the SiNWs are highly reproducible and demonstrate good stability in the case of a stress-induced change in the electrical properties. These linear I–V curves and the reversible modulation further confirm that the contacts tested in our electromechanical experiments were stable and ohmic in nature. Next, we further investigated the current improvement rate of individual SiNWs as a function of the strain applied. In
the strain approached 1.5% just before fracture. No sign of reduction in the current enhancement rate was observed. In our previous work, enhancements of the carrier transport properties of p-type 〈100〉-oriented Si whiskers are observed under uniaxial tensile and compressive strains. It has been found that over 400% enhancement of electrical conductivity is achieved under a 2% tensile strain, while a 2% compressive strain can only cause 80% conductivity enhancement [31]. Compared with our previous results [31], it can be observed that the variation in the electrical properties of 〈110〉-oriented SiNWs under strain is much lower than that of 〈100〉-oriented NWs. For example, at a strain level of 1.5%, a ∼360% change in the electrical current could be observed in 〈100〉-oriented SiNWs under tensile strain, whereas only a 24% change in the electrical current can be observed under the same strain in 〈100〉-oriented SiNWs. Similarly, we also measured the electrical properties of p-type Si 〈110〉 whiskers under bending strain. Figures 4(a)– (f) show six sequential TEM images of the SiNWs under bending strain, and figure 4(g) shows the corresponding I–V curves. In this experiment, we used tungsten tips to bend the 〈110〉 Si whiskers. Similarly to the measurements performed in the tensile experiment, the TEM images were recorded under different bending strains. The initial length of this Si whisker was measured to be 912 nm. By controlling the tip’s movement, bending strain could be applied to this whisker. In this manner, we correlated the bending strain (D/2R) with the change in conductance, where D is the diameter of the NWs (181 nm in this sample) and R is the radius of curvature of the 4
R Shao et al
Nanotechnology 26 (2015) 265703
dimensional changes and piezoresistivity. Thus, the relation between the relative change in resistance in a piezoresistive element, R/R0, subjected to uniaxial stress is given by R ρ = (1 + 2v) ε L + R0 ρ0
(1)
where υ = 0.15 [20] is the Poisson’s ratio and εL = 1.48% is the strain along the piezoresistor with resistance R. The resistance and the resistivity in the unstressed material are denoted R0 and ρ0, respectively. Because the observed changes in resistance (R/R0 = 1.24) were significantly greater than the changes in dimension (εL = 1.48%), we converted the relative changes in resistance to relative changes in resistivity Δρ0, subtracting the dimensional parameters but neglecting the dimensional changes in the NWs. Traditionally, when a semiconductor crystal is subjected to strain, the change in resistance is typically referred to as the piezoresistance effect. The change in the resistance is given by ΔR Δl = GF , R l
(2)
where R is the resistance, GF is the gauge factor, and l is the original length. In our experiment, the gauge factor of the NWs was calculated to be −11. The gauge factors were used to calculate the piezoresistance coefficients (and vice versa) using the following relation: GF = π × Y,
(3)
where π is the piezoresistance coefficient and Y is the Young’s modulus of a 〈110〉-oriented SiNW (i.e., 170 GPa, which is similar to that of bulk Si) [20, 35]. The calculated piezoresistance coefficient of a [110] SiNW is −14.4 × 10−11 Pa−1. We then examined the mechanisms that are responsible for the change in the resistivity Δρ0. It is generally expected that the piezoresistance can be explained by strain-induced changes in band structure and shifts in mobility in the bulk material. For Si, the change in transport properties is mainly attributed to the variation in carrier mobility [20, 24, 25, 36]. Fischetti et al [22] calculated the band structure of strained bulk Si using nonlocal empirical pseudo-potentials. Considering a transport effective mass and intervalley scattering, the results indicated a change in electron/hole mobility with applied strain. The authors’ calculations, based on measured bulk silicon piezoelectric coefficients, predicted an enhancement in both the electron and hole mobilities under uniaxial tensile strain. The main causes of the enhancement of the hole mobility are the breaking of the degeneracy of the v2 (heavy hole) and v1 (light hole) valence bands at Γ and the reduction in the transport effective mass. Daryoush Shiri et al [37] computationally investigated the electromechanical properties of [100] and [110] silicon nanowires under uniaxial strain. It was observed that the bandgap decreases with both increasing tensile strain and large compressive strain, which is well consistent with our experimental results. On the other hand, bending strain in SiNWs is much more complex. Bending can also change the band structure of
Figure 5. (a) Current responses under applied stress over time. (b)
The current improvement rate versus the bending strain.
this test, we increased the bending strain continuously until the SiNW broke. The current values were measured at an applied voltage of 3 V. Figure 5(b) shows the relationship between the current improvement rate and the strain. When the strain approached 6% before fracture, the change in the conductance reached a maximum, which was as large as 67%. It is worth noting that the slope of the red curve obtained by fitting the data varies with the strain state. The rate at which the current increased reached a maximum around a bending strain of 4.2%. Before the bending strain reached 4.2%, the slope of the red curve determined by fitting the data showed a fast rise, whereas at higher strain the slope decreased gradually and showed signs of saturation. This interesting relationship between the resistance of the NW and the strain applied may be induced by the complexity of the bending strain (outer tensile and inner compressive). In semiconductors, particularly indirect bandgap semiconductors, mechanical stress affects the electronic band structure, thus altering the effective electron mass, the mobility, and the resistivity ρ. Because NWs become elongated and slightly thinner when pulled, the change in resistance caused by an applied stress is the result of both 5
R Shao et al
Nanotechnology 26 (2015) 265703
References
SiNWs, which will consequently affect the electronic transport properties of individual SiNWs. In particular, because the bending strain is inhomogeneous, tensile and compressive strain and shear strain (even in the middle part) coexist, and the bottom of the conduction band will split into broad subbands, each representing a different type of strain effect. Daryoush Shiri et al [37] observed that for 〈110〉 silicon nanowires both the tensile strain and compressive strain tend to narrow the bandgap. Paul W Leu et al [20] investigated the mechanical and electronic properties of strained SiNWs using ab initio density functional theory calculations. In 〈110〉 NWs, compressive strain tended to first increase the bandgap due to rising of the Δ2 state. However, the bandgap increase is followed by a decrease, due to the v1 state rising, while the bottom of conduction band Δ4 remains fairly constant. Tensile strain causes a monotonous closing of the bandgap due to lowering of the Δ2 state, while the top of the valence band v2 remains constant. Based on the simulations, in our experiment, bending strain was believed to narrow the bandgap, increase the concentration of lighter carriers in the axial direction, and, finally, enhance the carrier mobility and result in the bending-induced conductance enhancement measured in our electromechanical experiments. Because the bending strain in SiNWs itself is quite complex, the way it affects the band structure and enhances carrier mobility requires further investigation. In summary, we performed in situ manipulation and electrical transport measurements on individual p-type Si 〈110〉 whiskers using a piezo-driven STM–TEM probing system within a TEM. Ohmic contacts were established via the EBID technique. An enhancement in the conductance of the NWs with an increase in the tensile and bending strain was observed. When the strain approached 1.5% before fracture, the change in the conductance reached a maximum, which was as large as 24.2%, without any sign of saturation. On the other hand, under 5.8% strain, a 67% enhancement in conductivity could be achieved under the applied bending strain. The conductance variations induced by strain in p-type 〈110〉-oriented SiNWs were far lower than those observed for p-type 〈100〉-oriented SiNWs. The underlying mechanism for the change in conductance was attributed to strain-induced changes in band structure. This study provides important insight into how variations in carrier mobility caused by strain affect the charge transport properties of silicon nanowires.
[1] Hu J, Odom T W and Lieber C M 1999 Acc. Chem. Res. 32 435 [2] Lieber C M 2003 MRS Bull. 28 4 [3] Cui Y, Zhong Z, Wang D, Wang W U and Lieber C M 2003 Nano Lett. 3 149 [4] Majima H, Saito Y and Hiramoto T 2001 Tech. Dig.—Int. Electron Devices Meet. 733 33.3.1–4 [5] Wang J, Polizzi E and Lundstrom M 2004 J. Appl. Phys. 96 2192 [6] McAlpine M, Friedman R, Jin S, Lin K, Wang W and Lieber C M 2003 Nano Lett. 3 1531 [7] Koo J, Jeon Y, Lee M and Kim S 2011 Japan. J. Appl. Phys. 50 065001 [8] Liu X, Long Y Z, Liao L, Duan X and Fan Z 2012 ACS Nano 6 1888 [9] Smith C S 1954 Phys. Rev. 94 42 [10] Tortonese M, Barrett R C and Quate C F 1992 Appl. Phys. Lett. 62 834 [11] Wee K W, Kang G Y, Park J, Kang J Y, Yoon D S, Park J H and Kim T S 2005 Biosens. Bioelectron. 20 1932 [12] Lee M L, Fitzgerald E A, Bulsara M T, Currie M T and Lochtefeld A 2005 J. Appl. Phys. 97 011101 [13] Haugerud B M, Bosworth L A and Belford R E 2003 J. Appl. Phys. 94 4102 [14] Ieong M, Doris B, Kedzierski J, Rim K and Yang M 2004 Science 306 2057 [15] Lyons D M, Ryan R M, Morris M A and Holmes J D 2002 Nano Lett. 2 811 [16] Audoit G, Mhuircheartaigh T N, Lipson S M, Morris M A, Blau W J and Holmes J D 2005 J. Mater. Chem. 15 4809 [17] Hong K, Kim J, Lee S and Shin J K 2008 Nano Lett. 8 1335 [18] Sajjad R N and Alam K 2009 J. Appl. Phys. 105 044307 [19] Zhang J, Huang Q, Yu H and Lei S 2009 Sensor 9 2746 [20] Leu P W, Svizhenko A and Cho K 2008 Phys. Rev. B 77 235305 [21] Niquet Y M, Delerue C and Krzeminski C 2012 Nano Lett. 12 3545 [22] Fishchetti M V and Laux S E 1996 J. Appl. Phys. 80 2234 [23] Feste S F, Knoch J, Habicht S, Buca D, Zhao Q T and Mantl S 2009 Solid-State Electron. 53 1257 [24] Toriyama T, Funai D and Sugiyama S 2003 J. Appl. Phys. 93 561 [25] Yang Y L and Li X X 2011 Nanotechnology 22 015501 [26] Neuzil P, Wong C C and Rebound J 2010 Nano Lett. 10 1248 [27] He R H and Yang P D 2006 Nat. Nanotechnology 1 42 [28] Milne J S, Rowe A C H, Arscott S and Renner C 2010 Phys. Rev. Lett. 105 226802 [29] Toriyama T and Sugiyama S 2003 Sensors Actuators A 108 244 [30] Lugstein A, Steinmair M, Steiger A, Kosina H and Bertagnolli E 2010 Nano Lett. 10 3204 [31] Zheng K, Shao R W, Deng Q S, Zhang Y F, Li Y J, Han X D, Zhang Z and Zou J 2014 Appl. Phys. Lett. 104 013111 [32] Shao R W, Zheng K, Zhang Y F, Li Y J, Zhang Z and Han X D 2014 Nanoscale 6 4936 [33] Shao R W, Zheng K, Zhang Y F, Li Y J, Zhang Z and Han X D 2012 Appl. Phys. Lett. 101 233109 [34] Shao R W and Zheng K 2015 RSC Advances 5 34447 [35] Li X, Ono T, Wang Y and Esashi M 2003 Appl. Phys. Lett. 83 3081 [36] Cao J X, Gong X G and Wu R Q 2007 Phys. Rev. B 75 233302 [37] Shiri D, Kong Y, Buin A and Anantram M P 2008 Appl. Phys. Lett. 93 073114
Acknowledgments This work was supported by the National Natural Science Foundation of China (11374029), the Foundation for the Author of National Excellent Doctoral Dissertation of PR China (201214), the Beijing Nova Program (Z121103002512017), the Key Project of the National Natural Science Foundation of China (11234011), Henry Fok of the Ministry of Education Fund (141008), and the Beijing City Board of Education Project (KM201310005009).
6
Search
Collections
Journals
About
Contact us
My IOPscience
The effect of tensile and bending strain on the electrical properties of p-type 〈110〉 silicon nanowires
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2015 Nanotechnology 26 265703 (http://iopscience.iop.org/0957-4484/26/26/265703) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 165.123.34.86 This content was downloaded on 08/09/2015 at 10:25
Please note that terms and conditions apply.
Nanotechnology Nanotechnology 26 (2015) 265703 (6pp)
doi:10.1088/0957-4484/26/26/265703
The effect of tensile and bending strain on the electrical properties of p-type 〈110〉 silicon nanowires Ruiwen Shao1, Pan Gao1 and Kun Zheng1,2 1
Institute of Microstructure and Properties of Advanced Materials, Beijing University of Technology, Beijing 100124, People’s Republic of China 2 Beijing Key Lab of Microstructure and Property of Advanced Material, Beijing University of Technology, Beijing 100124, People’s Republic of China E-mail: [email protected] Received 17 March 2015, revised 1 May 2015 Accepted for publication 15 May 2015 Published 10 June 2015 Abstract
In this study, electromechanical responses induced by uniaxial tensile and bending deformation were obtained for p-type 〈110〉-oriented Si whiskers by in situ transmission electron microscopy (TEM). Ohmic contacts between the nanowires (NWs) and electrodes were achieved using electron-beam-induced carbon deposition. Results show that enhancements in the carrier transport properties were achieved under both uniaxial tensile and bending strains. With the strain increased to 1.5% before fracture, the improvement in the conductance reached a maximum, which was as large as 24.2%, without any sign of saturation. On the other hand, under 5.8% bending strain, a 67% conductivity enhancement could be achieved. This study should provide important insight into the performance of nanoscale-strained Si. Keywords: piezoresistance effect, electrical properties, silicon nanowires, tensile strain, bending strain (Some figures may appear in colour only in the online journal) complementary metal oxide semiconductor (CMOS) performance improvements [12, 13]. Strain engineering has been considered to be one of the most promising strategies for developing high-performance silicon nanodevices [14]. In addition, there have been attempts to tailor the optical properties of SiNWs by strain engineering. SiNWs have shown a shift in photoluminescence with strain levels of up to 2%, suggesting strain-induced band structure changes [15, 16]. As one of the primary semiconductors, the relationship between the electrical properties of SiNWs and applied strain has attracted extensive attention. In the past decade, tremendous efforts have been dedicated to altering the straininduced carrier transport properties to develop highly sensitive strain gauges or a more effective mode for enhancing the carrier mobility based on theoretical calculations [17–22] or experiments [23–30]. It has been demonstrated that the orientation, size, doping type and level, as well as the strain state, are the primary factors [17, 19, 21, 22] affecting the band structure, effective carrier mass, and electrical properties
One-dimensional nanostructures such as semiconductor nanowires (NWs) are attractive building blocks for the assembly of nanoelectronic and nanophotonic systems due to their tremendous technological potential [1, 2]. Among these materials, silicon nanowires (SiNWs) have attracted widespread attention as a promising structure for future integrated circuits because of their excellent scaling capability and compatibility with Si-based electronic technology [3–5]. SiNWs are also considered attractive building blocks for developing flexible electronics [6–8]. The piezoresistance effect of silicon [9], whereby the electrical resistance changes due to an applied mechanical stress, has been widely exploited in mechanical sensors [10, 11]. Strain plays an important role in determining the band structure and, thereby, elementary physical properties like the effective mass, carrier mobility, or electrical conductivity. More importantly, strained Si on silicon–germanium substrates has been explored in an effort to boost carrier mobility since the early 1990s and is currently accepted as the key solution for 0957-4484/15/265703+06$33.00
1
© 2015 IOP Publishing Ltd Printed in the UK
R Shao et al
Nanotechnology 26 (2015) 265703
corresponding selected area electron diffraction (SAED) pattern, from which the axial direction of the Si whisker can be confirmed to lie along a 〈110〉 direction. In situ electrical measurements were performed using a commercial scanning tunneling microscope–transmission electron microscope (STM–TEM) probing system (Nanofactory Instruments) inserted into a JEOL-2010 TEM (operating at 200 kV). Before the in situ experiment, a fresh tungsten tip was placed on a movable terminal of the STM holder, and Si whiskers were placed on the fixed Ag electrode. The Si whiskers were prepared by FIB and then transferred to the silver wire. The tungsten tip could be moved precisely in three dimensions. Figure 1(a) shows a schematic drawing of the experimental setup, in which a NW is bonded to a W tip and a Ag electrode by electron-beam-induced deposition (EBID). To form good electrical contacts between the SiNWs and the W electrode, a W tip was freshly prepared. Before the experiment, by controlling the piezo-motor, the tungsten tip was moved close and contacted to the Ag electrode. Then, a large current was introduced to the tungsten tip. Due to the Joule heating, the oxide layers on the surface of the tungsten tip would be melted. After this fast melt-quench process, the sharp points of the W tip became hemispherical in shape, as shown in the inset of figure 1(b). EBID was then used to weld the NWs to two electrodes. The details of this process can be found in [32–34]. The contacts between the NW and two electrodes were both ohmic and mechanically robust. By accurately controlling the piezo-motor, the individual NWs could be either stretched or compressed. Simultaneously, the corresponding I–V curves of the deformed whiskers were recorded. Figure 1(b) is a low-magnification TEM image showing two electrodes and a bonded SiNW. Because the piezo-motor could be accurately controlled to move forward and backward, and its repeatability was carefully examined and confirmed, we designed the in situ experiment as follows. After recording the TEM image shown in figure 2(a), the electron beam was turned off, and deformation (compressed or stretched) was induced by advancing the piezomotor four fine steps for each deformation. For each deformation, the corresponding I–V characteristics were measured by sweeping the voltage from −4–4 V. This procedure eliminated the effect of the electron beam on the measured I–V characteristics. The entire experiment was then repeated (with identical deformation control achieved by repeating the stepping of the piezo-motor) with the electron beam switched on (the intensity of the beam was weak but sufficient for capturing TEM images), during which the I–V characteristics were measured again, and the corresponding TEM images of the strained SiNWs were recorded. Figure 2 shows an individual SiNW with a diameter of 181 nm that was fixed between a Ag electrode and a W tip. By placing the W tip far from the Ag electrode, a uniaxial tension test was performed on the SiNW, and the electrical properties were simultaneously measured. Figures 2(a)–(f) show six sequential TEM images of a typical tensile process, and figure 2(g) shows the corresponding I–V curves. By measuring the gauge length of the NW, the strain could be calculated by the DigitalMicrograph software program (version 1.70.16). Although the accuracy of the measured
Figure 1. (a) Experimental setup showing a NW connected to two
electrodes placed inside an STM–TEM probing system. (b) W tip, Ag electrode, and a bonded SiNW. The inset is the corresponding selected area electron diffraction (SAED) pattern, from which the axial direction of the Si whisker can be confirmed to lie along a 〈110〉 direction.
of strained crystalline Si. In terms of directional considerations, 〈100〉, 〈110〉, and 〈111〉 are the most important directions in practical applications. Detailed research about the effect of different orientations on the strain-induced carrier transport properties of SiNWs is desirable. Although the electrical transport properties of strained 〈110〉-oriented, onedimensional Si nanostructures have been studied experimentally [23–29], the results show wide divergences. We note that a slightly enhanced piezoresistance effect has been previously reported for Si beams defined by electron-beam lithography [24]. While He and Yang observed that the longitudinal piezoresistance of p-type SiNWs could be extremely high, exceeding 100-fold the bulk value [27], in addition, A Lugstein et al demonstrated that under ultrahigh strain conditions, p-type single-crystal silicon NWs exhibit an anomalous negative piezoresistance effect [30]. The divergence may result from different conditions of SiNWs in [23–29], such as doping types, doping concentrations, defects, etc To exclude the effects above, a better way is to use the same type of SiNWs with different growth directions. Based on this consideration, in our work, SiNWs with different orientations were prepared from the same Si wafer by a using focused ion beam (FIB) technique. In our previous work, p-type 〈100〉oriented Si whiskers were prepared [31], and in this study, p-type 〈110〉-oriented Si whiskers were obtained. Using in situ transmission electron microscopy (TEM), the effects of strain on the electrical transport properties along the 〈110〉 direction of single-crystalline Si whiskers were studied. A p-type doped Si 〈001〉 film with an electrical resistance of 14 ∼ 22 Ω · cm, lateral dimensions of 1 mm × 100 μm (with side edges of {100} facets), and a thickness of 300 nm was used to prepare Si 〈110〉-oriented whiskers using FIB technology. The fabricated Si 〈110〉-oriented whiskers showed lateral dimensions on the order of several hundred nanometers. The prepared Si whiskers could be transferred to a silver wire, which was used as an electrode. Figure 1(b) shows a TEM image of a typical Si whisker; the inset is the 2
R Shao et al
Nanotechnology 26 (2015) 265703
Figure 2. Images of SiNWs under tensile strain and the corresponding I–V curves. (a)–(e) Images of the SiNWs under tensile strain and (f) image of the fractured NW. (g) The corresponding I–V curves.
length obtained for any nanostructure using this software may vary with the imaging conditions, the relative change in length and, in turn, the corresponding tensile strain can be accurately determined by comparing the change in the measured lengths during deformation. The initial length was 958 nm, as shown in figure 2(a). The measured lengths were 962, 966, 970, and 973 nm after uniaxial tension was applied, as shown in figures 2(b)–(e), and the corresponding calculated strains were 0.42, 0.74, 1.16, and 1.36%, respectively. Compared with that obtained in our previous tensile experiments for SiC nanowires [33], the fractured strain is far lower. This discrepancy may be due to the defects induced by FIB. The electrical properties of the NW were measured under tensile strain, and the corresponding I–V curves are shown in figure 2(g). The voltage was swept from −4–4 V. In the strain-free case, the black I–V curve (curve ‘a’) indicates the intrinsic electrical properties. The linear I–V characteristic of the NW indicates that good ohmic contacts were achieved. It is clear that the conductance was enhanced with an increase in the tensile strain. When we pulled the NW again, it broke at the junction between the NW and the Ag electrode, which is shown in figure 2(f). In this tensile test, the final measured strain approaches 1.5% before fracture. This failure may have been induced by torsional force. At the same time, it can be observed that there was no current signal in the NW’s I–V curve (curve ‘f’ in figure 2(g)). The above-described analysis qualitatively demonstrated the change in current with an increase in the tensile strain. Next, we discuss the quantitative relationship between the rate
Figure 3. The current improvement rate versus the strain.
at which the current was improved and the strain. In this case, the current values were measured at an applied voltage of 4 V. Figure 3 shows the relationship between the rate of current improvement and the strain. When the strain approached 1.5% before fracture, the change in the conductance reached a maximum, reaching a value as large as 24.2%, without any sign of saturation. In fact, the slope of the red curve determined by fitting the data varies with the strain state and exhibits a gentle rise with the increase in the tensile strain. The rate of current enhancement reached a maximum when 3
R Shao et al
Nanotechnology 26 (2015) 265703
Figure 4. Images of the bent SiNWs and the corresponding I–V curves. (a)–(e) Images of the bent SiNW and (f) image of the released NW. (g) The corresponding I–V curves.
bent NW. A series of strains could then be obtained, namely, 2.4% (state 2), 2.8% (state 3), 3.4% (state 4), and 3.9% (state 5), respectively. Figure 4(g) shows the corresponding I–V curves that were simultaneously measured. Similarly to the case in which the NWs were placed in uniaxial tension, an enhancement in the electrical conductivity with increasing bending strain was observed. When the NW end was returned to its original position, as shown in figure 4(g), although the NW still showed some deformation, the conductance was nearly the same as the original value (curve ‘f’ in figure 4(g)). We then investigated the reproducibility and stability of the strain effect on the current in the SiNWs, and the results are shown in figure 4. In this test, we obtained the current response of the NWs over an entire stretch–release cycle under a fixed bias of 3 V. Initially, no stress was applied, and the current was maintained at 42.9 ± 0.2 nA. By controlling the movement of the W tip and applying a bending stress to the NW, the current increased gradually. The current values were 43.5, 44.7, 46.0, and 47.6 nA, which correspond to strains of 0.7%, 1.4%, 2.1%, and 2.8%. Then, when we released the applied stress, the current decreased and finally returned to the initial state. Whole cycles were performed, and the results could be repeated. Our results indicate that the SiNWs are highly reproducible and demonstrate good stability in the case of a stress-induced change in the electrical properties. These linear I–V curves and the reversible modulation further confirm that the contacts tested in our electromechanical experiments were stable and ohmic in nature. Next, we further investigated the current improvement rate of individual SiNWs as a function of the strain applied. In
the strain approached 1.5% just before fracture. No sign of reduction in the current enhancement rate was observed. In our previous work, enhancements of the carrier transport properties of p-type 〈100〉-oriented Si whiskers are observed under uniaxial tensile and compressive strains. It has been found that over 400% enhancement of electrical conductivity is achieved under a 2% tensile strain, while a 2% compressive strain can only cause 80% conductivity enhancement [31]. Compared with our previous results [31], it can be observed that the variation in the electrical properties of 〈110〉-oriented SiNWs under strain is much lower than that of 〈100〉-oriented NWs. For example, at a strain level of 1.5%, a ∼360% change in the electrical current could be observed in 〈100〉-oriented SiNWs under tensile strain, whereas only a 24% change in the electrical current can be observed under the same strain in 〈100〉-oriented SiNWs. Similarly, we also measured the electrical properties of p-type Si 〈110〉 whiskers under bending strain. Figures 4(a)– (f) show six sequential TEM images of the SiNWs under bending strain, and figure 4(g) shows the corresponding I–V curves. In this experiment, we used tungsten tips to bend the 〈110〉 Si whiskers. Similarly to the measurements performed in the tensile experiment, the TEM images were recorded under different bending strains. The initial length of this Si whisker was measured to be 912 nm. By controlling the tip’s movement, bending strain could be applied to this whisker. In this manner, we correlated the bending strain (D/2R) with the change in conductance, where D is the diameter of the NWs (181 nm in this sample) and R is the radius of curvature of the 4
R Shao et al
Nanotechnology 26 (2015) 265703
dimensional changes and piezoresistivity. Thus, the relation between the relative change in resistance in a piezoresistive element, R/R0, subjected to uniaxial stress is given by R ρ = (1 + 2v) ε L + R0 ρ0
(1)
where υ = 0.15 [20] is the Poisson’s ratio and εL = 1.48% is the strain along the piezoresistor with resistance R. The resistance and the resistivity in the unstressed material are denoted R0 and ρ0, respectively. Because the observed changes in resistance (R/R0 = 1.24) were significantly greater than the changes in dimension (εL = 1.48%), we converted the relative changes in resistance to relative changes in resistivity Δρ0, subtracting the dimensional parameters but neglecting the dimensional changes in the NWs. Traditionally, when a semiconductor crystal is subjected to strain, the change in resistance is typically referred to as the piezoresistance effect. The change in the resistance is given by ΔR Δl = GF , R l
(2)
where R is the resistance, GF is the gauge factor, and l is the original length. In our experiment, the gauge factor of the NWs was calculated to be −11. The gauge factors were used to calculate the piezoresistance coefficients (and vice versa) using the following relation: GF = π × Y,
(3)
where π is the piezoresistance coefficient and Y is the Young’s modulus of a 〈110〉-oriented SiNW (i.e., 170 GPa, which is similar to that of bulk Si) [20, 35]. The calculated piezoresistance coefficient of a [110] SiNW is −14.4 × 10−11 Pa−1. We then examined the mechanisms that are responsible for the change in the resistivity Δρ0. It is generally expected that the piezoresistance can be explained by strain-induced changes in band structure and shifts in mobility in the bulk material. For Si, the change in transport properties is mainly attributed to the variation in carrier mobility [20, 24, 25, 36]. Fischetti et al [22] calculated the band structure of strained bulk Si using nonlocal empirical pseudo-potentials. Considering a transport effective mass and intervalley scattering, the results indicated a change in electron/hole mobility with applied strain. The authors’ calculations, based on measured bulk silicon piezoelectric coefficients, predicted an enhancement in both the electron and hole mobilities under uniaxial tensile strain. The main causes of the enhancement of the hole mobility are the breaking of the degeneracy of the v2 (heavy hole) and v1 (light hole) valence bands at Γ and the reduction in the transport effective mass. Daryoush Shiri et al [37] computationally investigated the electromechanical properties of [100] and [110] silicon nanowires under uniaxial strain. It was observed that the bandgap decreases with both increasing tensile strain and large compressive strain, which is well consistent with our experimental results. On the other hand, bending strain in SiNWs is much more complex. Bending can also change the band structure of
Figure 5. (a) Current responses under applied stress over time. (b)
The current improvement rate versus the bending strain.
this test, we increased the bending strain continuously until the SiNW broke. The current values were measured at an applied voltage of 3 V. Figure 5(b) shows the relationship between the current improvement rate and the strain. When the strain approached 6% before fracture, the change in the conductance reached a maximum, which was as large as 67%. It is worth noting that the slope of the red curve obtained by fitting the data varies with the strain state. The rate at which the current increased reached a maximum around a bending strain of 4.2%. Before the bending strain reached 4.2%, the slope of the red curve determined by fitting the data showed a fast rise, whereas at higher strain the slope decreased gradually and showed signs of saturation. This interesting relationship between the resistance of the NW and the strain applied may be induced by the complexity of the bending strain (outer tensile and inner compressive). In semiconductors, particularly indirect bandgap semiconductors, mechanical stress affects the electronic band structure, thus altering the effective electron mass, the mobility, and the resistivity ρ. Because NWs become elongated and slightly thinner when pulled, the change in resistance caused by an applied stress is the result of both 5
R Shao et al
Nanotechnology 26 (2015) 265703
References
SiNWs, which will consequently affect the electronic transport properties of individual SiNWs. In particular, because the bending strain is inhomogeneous, tensile and compressive strain and shear strain (even in the middle part) coexist, and the bottom of the conduction band will split into broad subbands, each representing a different type of strain effect. Daryoush Shiri et al [37] observed that for 〈110〉 silicon nanowires both the tensile strain and compressive strain tend to narrow the bandgap. Paul W Leu et al [20] investigated the mechanical and electronic properties of strained SiNWs using ab initio density functional theory calculations. In 〈110〉 NWs, compressive strain tended to first increase the bandgap due to rising of the Δ2 state. However, the bandgap increase is followed by a decrease, due to the v1 state rising, while the bottom of conduction band Δ4 remains fairly constant. Tensile strain causes a monotonous closing of the bandgap due to lowering of the Δ2 state, while the top of the valence band v2 remains constant. Based on the simulations, in our experiment, bending strain was believed to narrow the bandgap, increase the concentration of lighter carriers in the axial direction, and, finally, enhance the carrier mobility and result in the bending-induced conductance enhancement measured in our electromechanical experiments. Because the bending strain in SiNWs itself is quite complex, the way it affects the band structure and enhances carrier mobility requires further investigation. In summary, we performed in situ manipulation and electrical transport measurements on individual p-type Si 〈110〉 whiskers using a piezo-driven STM–TEM probing system within a TEM. Ohmic contacts were established via the EBID technique. An enhancement in the conductance of the NWs with an increase in the tensile and bending strain was observed. When the strain approached 1.5% before fracture, the change in the conductance reached a maximum, which was as large as 24.2%, without any sign of saturation. On the other hand, under 5.8% strain, a 67% enhancement in conductivity could be achieved under the applied bending strain. The conductance variations induced by strain in p-type 〈110〉-oriented SiNWs were far lower than those observed for p-type 〈100〉-oriented SiNWs. The underlying mechanism for the change in conductance was attributed to strain-induced changes in band structure. This study provides important insight into how variations in carrier mobility caused by strain affect the charge transport properties of silicon nanowires.
[1] Hu J, Odom T W and Lieber C M 1999 Acc. Chem. Res. 32 435 [2] Lieber C M 2003 MRS Bull. 28 4 [3] Cui Y, Zhong Z, Wang D, Wang W U and Lieber C M 2003 Nano Lett. 3 149 [4] Majima H, Saito Y and Hiramoto T 2001 Tech. Dig.—Int. Electron Devices Meet. 733 33.3.1–4 [5] Wang J, Polizzi E and Lundstrom M 2004 J. Appl. Phys. 96 2192 [6] McAlpine M, Friedman R, Jin S, Lin K, Wang W and Lieber C M 2003 Nano Lett. 3 1531 [7] Koo J, Jeon Y, Lee M and Kim S 2011 Japan. J. Appl. Phys. 50 065001 [8] Liu X, Long Y Z, Liao L, Duan X and Fan Z 2012 ACS Nano 6 1888 [9] Smith C S 1954 Phys. Rev. 94 42 [10] Tortonese M, Barrett R C and Quate C F 1992 Appl. Phys. Lett. 62 834 [11] Wee K W, Kang G Y, Park J, Kang J Y, Yoon D S, Park J H and Kim T S 2005 Biosens. Bioelectron. 20 1932 [12] Lee M L, Fitzgerald E A, Bulsara M T, Currie M T and Lochtefeld A 2005 J. Appl. Phys. 97 011101 [13] Haugerud B M, Bosworth L A and Belford R E 2003 J. Appl. Phys. 94 4102 [14] Ieong M, Doris B, Kedzierski J, Rim K and Yang M 2004 Science 306 2057 [15] Lyons D M, Ryan R M, Morris M A and Holmes J D 2002 Nano Lett. 2 811 [16] Audoit G, Mhuircheartaigh T N, Lipson S M, Morris M A, Blau W J and Holmes J D 2005 J. Mater. Chem. 15 4809 [17] Hong K, Kim J, Lee S and Shin J K 2008 Nano Lett. 8 1335 [18] Sajjad R N and Alam K 2009 J. Appl. Phys. 105 044307 [19] Zhang J, Huang Q, Yu H and Lei S 2009 Sensor 9 2746 [20] Leu P W, Svizhenko A and Cho K 2008 Phys. Rev. B 77 235305 [21] Niquet Y M, Delerue C and Krzeminski C 2012 Nano Lett. 12 3545 [22] Fishchetti M V and Laux S E 1996 J. Appl. Phys. 80 2234 [23] Feste S F, Knoch J, Habicht S, Buca D, Zhao Q T and Mantl S 2009 Solid-State Electron. 53 1257 [24] Toriyama T, Funai D and Sugiyama S 2003 J. Appl. Phys. 93 561 [25] Yang Y L and Li X X 2011 Nanotechnology 22 015501 [26] Neuzil P, Wong C C and Rebound J 2010 Nano Lett. 10 1248 [27] He R H and Yang P D 2006 Nat. Nanotechnology 1 42 [28] Milne J S, Rowe A C H, Arscott S and Renner C 2010 Phys. Rev. Lett. 105 226802 [29] Toriyama T and Sugiyama S 2003 Sensors Actuators A 108 244 [30] Lugstein A, Steinmair M, Steiger A, Kosina H and Bertagnolli E 2010 Nano Lett. 10 3204 [31] Zheng K, Shao R W, Deng Q S, Zhang Y F, Li Y J, Han X D, Zhang Z and Zou J 2014 Appl. Phys. Lett. 104 013111 [32] Shao R W, Zheng K, Zhang Y F, Li Y J, Zhang Z and Han X D 2014 Nanoscale 6 4936 [33] Shao R W, Zheng K, Zhang Y F, Li Y J, Zhang Z and Han X D 2012 Appl. Phys. Lett. 101 233109 [34] Shao R W and Zheng K 2015 RSC Advances 5 34447 [35] Li X, Ono T, Wang Y and Esashi M 2003 Appl. Phys. Lett. 83 3081 [36] Cao J X, Gong X G and Wu R Q 2007 Phys. Rev. B 75 233302 [37] Shiri D, Kong Y, Buin A and Anantram M P 2008 Appl. Phys. Lett. 93 073114
Acknowledgments This work was supported by the National Natural Science Foundation of China (11374029), the Foundation for the Author of National Excellent Doctoral Dissertation of PR China (201214), the Beijing Nova Program (Z121103002512017), the Key Project of the National Natural Science Foundation of China (11234011), Henry Fok of the Ministry of Education Fund (141008), and the Beijing City Board of Education Project (KM201310005009).
6
