nems device for determination of shear stress between ... d… · 3) contact width: 50% of the tube...
TRANSCRIPT
Determination of Shear Stress Between Single-walled Carbon
Nanotubes and Substrates Using NEMS Devices
Student:Huiyan Pan
Advisor: Prof. McGruer Co-advisor: Prof. Adams
Electrical and Computer Engineering
1
Outline • Research background Introduction of carbon nanotubes (CNTs) Motivation of this work Shear stress measured by AFM in the literature
• Technique development to study the shear stress between CNTs and substrates Principle of the technique NEMS cantilever based technique Structure design Microfabrication Nanomanipulation tests Testing results Discussion
•Conclusions
2
Graphene
Background
Technique development
Conclusions
What are carbon nanotubes?
Single-walled carbon nanotube
(SWNT)
Multi-walled carbon
nanotube (MWNT) [1]
3 [1] A. R. Hall et al. Reports on Progress in Physics,2012.
Background
Technique development
Conclusions
Why are the interactions/shear stress between CNTs and substrates so
important?
4
1) Electrophoretic assembly [2] 2) Dielectrophoretic assembly [3] 3) Fluidic assembly [4]
The effect of CNT-substrate interactions on CNT assembly processes
A SWNT film got peeled off from the substrate by nitrogen drying after fluidic assembly [5]
In the fabrication of any CNT-based devices, the mechanical interactions
between tubes and substrates have to be strong enough for the devices to survive the subsequent processes
5 Background
Technique development
Conclusions
[2] A.R. Boccaccini et al, Carbon ,2006. [3] ] M. R. Islam, et al, Chemical Society Reviews, 2012. [4] S. Somu et al, ACS Nano, 2010. [5] A.R. Boccaccini et al, Carbon ,2006.
The effect of CNT-substrate interactions on CNT-based device performance
CNTs are attached to gold electrodes as probes for a nanotweezer [6]
CNTs contact with electrode during actuation for nanoswitches [7,8]
Background
Technique development
Conclusions
6
[6] P. Kim et al, Science, 2010. [7] S. W. Lee et al, Nano Lett, 2004. [8] C. Ke et al, Small, 2006.
A finite length of SWNT bundle slip into the trench region under electrostatic force during actuation.
AFM of device before testing
AFM of device after testing
The effect of shear stress at CNT-substrate interface on the actuation voltage of nanoswitches
Background
Technique development
Conclusions
SEM of the SWNT based nanoswitch [9] .
Originally taut SWNT bundle
Axial Slack (the increased length
relative to the originally taut
length)
7
[9] P. Ryan et al,Journal of Micromechanics and Microengineering, 2011.
AFM is in the use as lateral force microscopy [10]
Sliding of a MWNT on graphite by AFM (2MPa)[11] A MWNT was pushed on SiO2 surface by AFM (60MPa) [12]
Shear stress measured by AFM in the literature
Background
Technique development
Conclusions
.
Nanoscale dimension of CNTs
NanoNewton level interaction force
8
[10] P. Reimann et al, New J. Phys. ,2005. [11] M. R. Falvo et al, Nature, 1999. [12] J.-H. et al, Wear, 2009.
AFM manipulation of a SWNT bundle on a mica substrate. A small portion was extracted from the
bundle [14] . A shear stress of 10 MPa is estimated for a bent MWNT on a silicon substrate after AFM
manipulation[13] .
Disadvantages and limitations of the AFM-based technique:
1) Operation difficulty 2) Complicated force calibration 3) Substrate has to be very smooth 4) No shear stress data reported for SWNTs yet based on this technique
Background
Technique development
Conclusions
9
Shear stress measured by AFM in the literature
[13] T. Hertel, et al, J. Phys. Chem. B, 1998. [14] M. R. Falvo, et al. Microsc. Microanal., 1999
Objective
To develop a new efficient and easily controllable technique to study the shear
stress between SWNTs and substrates
Background
Technique development
Conclusions
10
Principle of the technique
A SWNT bundle is under tension T
A SWNT bundle gets axial slack S after the gap returns to its original size
Gap elongates by uG
SWNTs stretching SWNTs slipping
Elastic properties Shear stress
Background
Technique development
Conclusions
Axial slack S
Gap returns to its original size. SWNTs stretching recovers.
The axial slack results from the residual length slipped from the shoulders.
S < uG
Shear stress can be determined by measuring the axial slack
11
The testing device has a SWNT bundle contact with two shoulders
A microcantilever is pushed laterally by a nanomanipulator to cause the tubes to slip
Top view of the designed testing device
SWNTs
NEMS cantilever based testing device
Background
Technique development
Conclusions
Flexible cantilever
Stationary electrode
12
SEM of the fabricated cantilever structure (side view)
SEM of the device after SWNT bundle assembly (top view)
Problem: The contact length can not be measured based on the SEM images due to poor contrast.
The original cantilever structure
Background
Technique development
Conclusions
13
Contact length = Cantilever width
Improved structure
Background
Technique development
Conclusions
14
SEM of a fabricated improved structure SEM of the structure with an assembled SWNT bundle
Microfabrication flow
Si
SiO2
W
Au
a) Si/SiO2/ W sputtering
d) Cr/Au evaporation and lift-off
e) SiO2 wet etch for release
b) E-beam lithography and W etch
c) SiO2 wet etch (half of the SiO2 thickness)
Background
Technique development
Conclusions
15
Low yield (~5%) to assemble only one SWNT bundle between the electrodes. Reasons: 1) Average tube length of 0.6 um 2) Assembly can end up with too many
tube bundles or sometimes no tubes assembled
Side electrodes
Side electrodes
NEMS device for SWNT assembly
Dielectrophoretic assembly
Array structure to get more good devices for one assembly process
Background
Technique development
Conclusions
16
Assembly parameters: 1) Concentration of SWNT solution
(310 ng/ml ) ; 2) AC voltage (3.4 Vpp) and frequency
(500 KHz) 3) Electrode gap (~1.8 um) 4) Assembly time (30s)
Expose a spot of the SWNT bundle under electron beam at 500KX magnification for ~2 minutes (Supra 25 SEM).
Before cutting
SWNT cutting by Electron-beam
Background
Technique development
Conclusions
After cutting
17
Nanomanipulation by a Zyvex probe
18
After nanomanipulation Before nanomanipulation
Comparison of the same structure before and after nanomanipulation
Background
Technique development
Conclusions
19
Image analysis to obtain the axial slack
-20 degree view
70 degree view
After nanomanipulation Curve fitting in Matlab
Background
Technique development
Conclusions
20
Background
Technique development
Conclusions
21
1) SWNT: hollow cylinder with a radius of 0.65 nm and a wall thickness of 0.34 nm
2) Young’s Modulus: 1TPa Both experimental and theoretical data [15,16] reported in the literature for
SWNTs with the same geometrical details) 3) Contact width: 50% of the tube diameter It is the shear force per unit length instead of shear stress which is
directly determined by the experiments in concert with the theoretical model in this work.
Difficulties in obtaining the contact width: (a)The contact width between tubes and substrates is not well defined or
studied in the literature. (b)For different substrates, the contact width can be different due to
different interaction forces. In this work, we assume 50% of the tube diameter as the contact width
to convert the shear force per unit length to shear stress for all tested substrates.
Modeling assumptions
[15] M. F. Yu et al, Physical Review Letters, 2000. [16] J. P. Lu, Physical Review Letters, 1997.
Modeling assumptions
Background
Technique development
Conclusions
22
4) Nanotubes are assumed to spread out on the contact surface, so the shear stress is independent of the tube number in a bundle.
Fewer tubes contacting would correspond to a larger shear stress. Strong interactions in the bundle, weak interactions at the tube-substrate interface lead to fewer tubes contacting the substrate.
Current assumption: n1=n2 Circular cross section
Anything in between, shear stress τ* = (n1/n2 ) τ, where τ is based on the current assumption, n1: the number of tubes in the bundle, n2 : the number of tubes in
contact with substrate.
For a semicircular cross section with n1 =34 (based on a 5.6 nm bundle height measured by Peter) , n2 =5.6/1.3= 4 , τ* = 8.5 τ
1) Original taut SWNT bundle
Model (by Yu-Chiao Wu) Variables: L: original gap; Lc: contact length;
T: induced tension; f=T/Lc: shear force per unit length; uG: gap elongation; S: axial
slack; : displacement of the free end; EA: extensional rigidity of the tube
(1)
Partial-complete slip Displacement of the completely slipped tube bundle:
2) Partial-partial slip
3) Partial-complete slip
Background
Technique development
Conclusions
EATL /
)2( EATLc
)2()2( EATLEATLEATL ccuG ++=
1)2( uEATLc +
1)2()2( uu EATLEATLEATL ccG +++=
1u
23
Partial-partial slip Suspended tube bundle stretching:
Displacement of the tube bundle from each side:
(2)
4) The SWNT bundle gets some axial slack after the gap recovers to its original size
For partial-complete slip (2) +(3):
With gap elongation D, contact length Lc, original gap L, extensional rigidity EA of the SWNT, and axial slack after testing S all known, the shear stress τ can be
determined.
Background
Technique development
Conclusions
transition point
)4( EATLc
1)4( uEATLc +
1)4()4( uEATLcEATLcS ++=
)22(2)2( SLSEAf uG −=
)22()( cc LLLSEAf uG +⋅−=
( )( )cG LLLtS uu +== 2,01
24
Model (by Yu-Chiao Wu) Gap recovers : Residual displacement of the partially slipped tube bundle: Residual displacement of the completely slipped tube bundle:
(3)
For partial-partial slip (1) +(3):
The transition point:
Shear stress for gold substrates
Background
Technique development
Conclusions
25
Test Original gap L (nm)
Contact length LC
(nm)
Gap elongation
uG (nm)
Axial slack uS
(nm)
Shear stress τ (MPa)
1 338 380 45 32 100 2 338 380 89 80 75 3 270 373 30 20 91 4 275 370 22 13 79 5 301 335 60 53 71 6 361 364 42 28 108
An average shear stress of 87 MPa between SWNTs and gold cantilever surface was obtained.
Gold cantilever surface modification
NH2 SH
2-aminoethanethiol NH2(CH2)2SH
Gold
S
NH 2
S
NH 2
S
NH 2
S
NH 2
S
NH 2
S
NH 2
S
NH 2
S
NH 2
Functionalization (self-assembly) process: 1) Pre-cleaning in piranha for 10 mins 2) Functionalization/Self-assembly in 1
mM ethanolic alkanethiol for 24 hours
3) Rinse with ethanol
SH
2-phenylethanethiol C6H5(CH2)2SH
Background
Technique development
Conclusions
26
Motivation: people modify gold with alkanethiols to change the surface properties in the literature. It is important to understand how the mechanical interactions between tubes and gold surface are changed through the standard chemical functionalization.
Alkanethiols:
Shear stress for 2-aminoethanethiol surfaces
The average shear stress is 142 MPa with more scatter in the results compared to the untreated gold surfaces.
Background
Technique development
Conclusions
27
Test Original gap L (nm)
Contact length LC
(nm)
Gap elongation uG
(nm)
Axial slack uS
(nm)
Shear stress τ (MPa)
1 435 281 49 20 275 2 454 271 60 42 172 3 401 301 41 34 73 4 437 247 49 32 196 5 477 249 50 43 74 6 444 273 60 51 89 7 437 257 54 44 113
Results for 2-phenylethanethiol surfaces
Processes of a SWNT bundle slipping on the cantilever surface with a sufficiently low shear stress and the adhesion effect. (a) state before
testing, (b) during gap elongation, and (c) after gap recovery.
Background
Technique development
Conclusions
The axial slack S is greater than the gap elongation uG
A sufficiently low shear stress τ
(low elastic strain energy)
The work of adhesion Δγ
between tubes and the surface
The SWNT bundle is considered to be an inextensible string by neglecting the strain energy
+
μ = Δγ / τ be determined with an observed axial slack uS, a measured original contact
length LN, two measured curvatures Rw and RN and a given gap elongation uG.
Model (by Yu-Chiao Wu)
28
Shear stress for 2-phenylethanethiol surfaces
Background
Technique development
Conclusions
An average shear stress of 7.2 MPa is determined with an estimated work of adhesion of 0.5 J/m2 (calculated by Yu-Chiao based on a binding energy and a
packing density between a benzene molecule and a CNT). Even with a doubled work of adhesion, the shear stress is 14 MPa, which is still much smaller compared to
untreated gold surfaces.
Test Original gap L (nm)
LR (nm)
LL (nm)
uG (nm)
Axial slack uS
(nm)
RL (nm)
RR (nm)
μ = Δγ/τ (nm)
Shear stress τ (MPa)
1 387 315 467 38 52 93 60 102 4.9 2 358 346 456 31 54 74 44 181 2.8 3 378 352 450 40 44 62 45 64 7.8 4 442 276 334 59 85 128 69 113 4.4 5 463 281 334 63 71 92 79 59 8.5 6 437 316 385 32 49 67 75 122 4.1 7 415 301 409 49 63 45 80 111 4.5 8 459 232 340 44 47 61 45 35 14.3 9 448 275 353 54 57 63 65 37 13.6
29
Three post-assembly treatments of the devices
Process of the device treatments: (1) Nitrogen annealing at 350 ℃ for 5 minutes (2) E-beam irradiation for the tube-gold contact area with an exposure
dose of 0.7 C/cm2 (3) E-beam induced carbon deposition for different period of time (30
minutes and 60 minutes)
Background
Technique development
Conclusions
30
Motivation: it has been found that electrical contact resistance between metallic pads and SWNTs can be reduced by 3 to 4 orders of magnitude through either nitrogen annealing [17, 18] or e-beam irradiation [18, 19].
Intentional e-beam induced carbon deposition has been used for clamping CNTs onto AFM cantilevers [20, 21].
It is important for us to understand how the above processes affect the shear stress between SWNTs and gold surfaces.
[17] J. O. Lee et al, Journal of Physics D: Applied Physics, 2000. [18]P. J. Ryan, Doctor of Philosophy thesis, 2011. [19]A. Bachtold et al, Applied Physics Letters, 1998. [20]M.-F. Yu et al, Science, 2000. [21]W. Ding et al, Journal of Applied Physics, 2005.
Three post-assembly treatments of the devices
Background
Technique development
Conclusions
31
For the nitrogen annealed devices (2 out of 4 tests) and the e-beam irradiated devices (2 out of 5
tests), some tests resulted in a very nonuniform bundle in the gap, with the axial slack almost
equal to the gap elongation.
Ideas of what is going on: 1) Tubes break at different positions in the
bundle and the outer deposited carbon shell completely breaks
2) The tubes contacting the substrate break, and other tubes not contacting the substrate stay intact. They slip relative to each other in the
bundle 3) Some tubes contact with partial cantilever
width. Tubes contacting the whole cantilever width break and slip in the bundle
All tubes contact with the substrate
Some tubes in the bundle contact with substrate
SEM images for nitrogen annealed devices
Background
Technique development
Conclusions
32
Test 1
Test 4
Test 2
Test 3
Shear stress for nitrogen annealed devices
Background
Technique development
Conclusions
On average, the shear stress is estimated to be greater than 309 MPa based on a breaking strain of 5.3 % from the literature [22] . (b is the contact width in the
above equation)
Test Original gap L (nm)
Narrow cantilever
width (nm)
Gap elongation uG
(nm)
Axial slack uS
(nm)
Shear stress τ (MPa)
1 487 294 42 -- 280*
2 497 271 62 61 301*
3 526 260 68 68 314*
4 571 240 90 76 340*
EALb c
breakτε =
33
When a SWNT bundle starts to break, the exact slip length is unknown. However, it is for sure that the slip length Lc is smaller than the narrow cantilever width.
[22] M. F. Yu et al, Physical Review Letters, 2000.
SEM images for e-beam irradiated devices
Background
Technique development
Conclusions
Test 4 (2 tests show nonuniform bundle after nanomanipulation)
Test 2 (3 tests show uniform bundle after nanomanipulation)
34
Shear stress for e-beam irradiated devices
Background
Technique development
Conclusions
An average shear stress of 202 MPa based on 3 independent tests and an average estimated shear stress larger than 311 MPa based on 2 independent tests with partially broken tubes are obtained.
* Estimated based on breaking strain of 5.3%[22] .
Test Original gap L (nm)
Narrow cantilever
width (nm)
Gap elongation uG
(nm)
Axial slack uS
(nm)
Shear stress τ (MPa)
1 500 273 48 30 163
2 532 253 54 30 230
3 526 260 46 23 214
4 519 260 40 38 314*
5 494 266 53 53 307*
35 [22] M. F. Yu et al, Physical Review Letters, 2000.
SEM images for carbon deposited devices
Background
Technique development
Conclusions
Test 1 (Test 1,2,4 has devices with 30 minute carbon
deposition)
Test 3 (device with 60 minute carbon deposition)
36
Shear stress for carbon deposited devices
Background
Technique development
Conclusions
For the 30 min deposited devices, a shear stress of 124 MPa and 94 MPa are found for test 1,2 respectively. For the 60 min deposited
device in test 3, a shear stress of 200 MPa is obtained. The device in test 4 has tube bundle not covering the whole wide cantilever width,
but the result remains a mystery to us.
* Estimated based on breaking strain of 5.3%[22] .
Test Original gap L (nm)
Contact length LC
(nm)
Gap elongation
uG (nm)
Axial slack uS
(nm)
Shear stress τ (MPa)
1 487 279 46 32 124
2 474 279 52 42 94
3 494 274 49 27 200
4 503 284 52 65 --
37
Surfaces Average shear stress (MPa)
Number of Tests
Untreated gold surface 87 6
2-aminoethanethiol surface 142 7
2-phenylethanethiol surface 7 9
Discussion of the untreated and functionalized gold results
For untreated gold surfaces, the relatively high shear stress compared to silicon (10MPa) [13] , graphite (2MPa) [11] , and silica (60MPa) [12] surfaces, likely due to charge transfer interactions between the metal and the SWNTs.
Background
Technique development
Conclusions
38
The exposed phenyl groups of the 2-phenylethanethiol surface bind to the SWNT surfaces via relatively weak p-p stacking interactions, which causes a lower shear stress at the interface.
The SWNT-2-aminoethanethiol interface has a shear stress 2 times higher than that for untreated gold, which may result from enhanced bonding between the amino end-groups and the SWNTs.
[11] M. R. Falvo et al, Nature, 1999. [12] J.-H. et al, Wear, 2009. [13] T. Hertel, et al, J. Phys. Chem. B, 1998.
Surfaces Average shear stress (MPa)
Number of Tests
Untreated gold surface 87 6
N2 annealed devices 309* 4
E-beam irradiated devices
202 3 311* 2
Carbon deposited devices
107 2 (30mins deposition) 200 1 (60mins deposition)
Discussion of the post-assembly treatment results
For both N2 annealed and E-beam irradiated devices, the shear stress can be increased to over 300 MPa. It is possible that the greatly increased shear stress is due to the improved contact at the interface during those process. For the carbon deposited devices, the relatively high shear stress compared to the untreated gold surface may result from anchoring of SWNTs to the gold surfaces by the carbon layer coating . Also carbon deposition for longer period of time appears to increase the shear stress more, likely due to a thicker carbon layer coating.
Background
Technique development
Conclusions
39
Discussion of the experimental technique
Background
Technique development
Conclusions
40
1) The effect of incidental e-beam exposure on the shear stress
E-beam irradiation (mainly during SEM imaging before nanomanipulation): the overall exposure dose is estimated to be over 100 times lower than the intentional e-beam exposure dose in the post-assembly treatment tests. Carbon deposition (mainly during SWNT bundle cutting before nanomanipulation): the cutting usually takes ~ 2 mins VS intentional carbon deposition time of 30 mins or 60 mins in the post-assembly treatment tests. Also the SEM chamber is always cleaned by plasma before cutting to minimize the deposition effect.
2) Sensitivity of shear stress relative to experimental parameters
Overall the uncertainty in all experimental parameters contributes to a ~25% scatter in the shear stress for this technique.
Conclusions
1) A new technique based on a NEMS cantilever structure, a nanomanipulator, an SEM along with a theoretical model was developed to study the shear stress at SWNT-substrate interface.
2) For the first time, the interfacial shear stress between dielectrophoretically assembled SWNTs and substrates is determined.
3) It is demonstrated that we can vary the shear stress by a factor of 20 by functionalizing a gold surface with different alkanethiols before SWNT assembly.
4) Some preliminary results also indicate that through post-assembly treatments including N2 annealing, e-beam irradiation and carbon deposition the shear stress can be increased from 87 MPa to over 300 MPa for the case of annealing.
Background
Conclusions
Technique development
41
Recommendations for the future
Background
Technique development
Conclusions
42
1) It will be possible to further tune the shear stresses by varying the nature of the alkanethiol end group. Even finer control may be achieved by systematically varying the length of the carbon spacer between the thiol function and the associated end group.
2) It will be good to get more data points to confirm the consistency of the testing results for post-assembly physical treatments.
3) This technique can also be used for other materials by modifying the cantilever surface through different methods, such as metal layer deposition on the cantilevers or simply making the entire structure out of other materials. Further, instead of using SWNTs, the one-dimensional element can be MWNTs or other types of nanowires.
Publications from this work 1) Pan H Y, Wu Y C, Ryan P J, Adams G G and McGruer N E Measurement of
shear stress between single-wall carbon nanotubes and substrates using NEMS devices. In: Solid-State Sensors, Actuators and Microsystems Conference (TRANSDUCERS), 2011 16th International, pp 629-32
2) Pan H Y, Wu Y C, Adams G G, Miller G P and McGruer N E Interfacial Shear Stress between Single-Walled Carbon Nanotubes and Gold Surfaces with and without an Alkanethiol Monolayer, Journal of colloid and interface science, 2013, 407C, pp. 133-139
3) In preparation of another paper on the results of the post-assembly treatments
43
Acknowledgements • Advisors: Prof. Nicol McGruer and Prof. George Adams • Committee: Prof. Carol Livermore and Prof. Ahmed Busnaina • Siva Somu, Peter Ryan, Juan Aceros, Yu-Chiao Wu, and all other
colleagues in CHN; the staff Richard DeVito, Scott McNamara, David McKee, Matthew Botti, and Jessica Viator;
• Others: National Science Foundation and Brewer Science • Work was performed in the George J. Kostas Nanoscale
Technology and Manufacturing Research Center at Northeastern University.
44
Thanks for your attention!
Questions?
45