Carbon Nanotube and Graphene Chemical Sensors - The Nanotube ...

3 downloads 213 Views 2MB Size Report
Carbon Nanotube and. Graphene Chemical Sensors. Peter C. Eklund. Dept's of Physics and. Materials Science and Engineering. Pennsylvania State University.
Carbon Nanotube and Graphene Chemical Sensors Peter C. Eklund Dept’s of Physics and Materials Science and Engineering Pennsylvania State University University Park, PA 16802 USA

[email protected]

NT ‘06 , Nagano

Outline • SWNT and Nanowire Chemical Sensors – Where are we? – Thermoelectric Sensing of Gases with SWNT Mats • Physisorption (e.g., C6H2n ring molecules) • Gas Collisions (inert gases)

• On to Graphene – n-graphene layer films (nGLs) – Phonon properties of nGLs – Sensor properties—not yet!

• Summary and Conclusions NT ‘06 , Nagano

Red (before) Blue (after) PSA In2O3

SWNT

NT ‘06 , Nagano

In2O3

SWNT

NT ‘06 , Nagano

Static Gas-SWNT Interactions van der Waals

covalent

ionic

+

-

weak

Physi-sorption

strong

current

intermediate

Chemi-sorption

One might presume that only the chemisorbed atoms/molecules will significantly effect the transport of electrons in the tube wall…. NT ‘06 , Nagano

Structural Considerations surface channel pore groove

EB= 0.119 eV σ = 45 m2/g EB= 0.062 eV σ = 783 m2/g

EB= 0.089 eV σ = 22 m2/g EB= 0.049 eV σ = 483 m2/g

•Molecular adsorption at interstitial, groove and pore sites. •No “bulk atoms”…only surface atoms…electrical transport should be sensitive to gas:SWNT interactions •It takes time for atoms and molecules to saturate the surface via diffusion and to gain access to internal pores and channels

NT ‘06 , Nagano

Thermoelectric Power (TEP) T + ΔT

HEAT

T 1 mm x 1mm x 0.2 mm

SWNT spaghetti

Cu

{

V1

CN

Cu

CN (Constantan)

{

{

V3

V2

S = Seebeck Coefficient = V2/ΔT ; V2 = Open Circuit Voltage Hot

+

e

e e

-

Cold

c.f., H. Romero, G. U. Sumansekera, G.D. Mahan and P. C. Eklund, Phys. Rev. B., 65 (2002)

NT ‘06 , Nagano

TEP(S) and Conductivity(σ) of a SWNT Bundle σ 1 , S1 σ 2 , S2

metallic semiconducting

Two tubes in parallel (e.g., metallic and semiconducting) S = (1/σ)[S1σ1 +S2σ2] ~ S1

and σ =[σ1 +σ2]~ σ1

Metallic tube (σ1) dominates: i.e., σ1>> σ2

=> Metallic tubes dominate the bundle transport 2 ⎛ dln σ (E) ⎞ -π 2 k B S= T⎜ ⎟ ⎝ dE ⎠ E F 3e

NT ‘06 , Nagano

Mott relation for a Metal

Effect of additional impurity scattering channel 2 ⎛ dln σ (E) ⎞ -π 2 k B S= T⎜ ⎟ ⎝ dE ⎠ E F 3e

σ(E)= e2v(E) 2D(E)τ (E)

Mott relation for a Metal σ = conductivity

v- free carrier velocity, D-density of states, τ-carrier lifetime Matheissen’s rule ⇒ ⇒

ρ = ρ0 + ρI

ρI due to gas

2 π 2 k B T ⎛ ρ I ⎞⎛ 1 dτ I 1 dτ O ⎞ ⎜⎜ ⎟⎟⎜⎜ ⎟⎟ ρI Physisorption CH30H

C2H50H

C3H70H C4H90H

H20

NT ‘06 , Nagano

Resistive Response to C6H2n Adsorption

R0

Four probe resistance (Ω)

3.0

vapors in equilibrium with the liquid at 40 °C

n=3

2.8

n=4 n=5

2.6

n=6

2.4 pyridine N

2.2

0

1

2

3 4 Time (Hrs)

5

6

*Increase in R related to scattering via π-electron coupling. *Pyridine decreases the resistivity via creation of additional carriers. NT ‘06 , Nagano

Effects of Gas:SWNT Collisions

Gas molecule collisions with the tube wall should affect the electron mean free path: • ?Direct scattering of electrons by atom collisions? • ?Generate a transient “dent”; non-thermal phonons then scatter electrons?

NT ‘06 , Nagano

TEP Response to Inert Gases (Same sample: P=1atm and T=500 K) He

-40

S (µV/K)

-35

ΔS

-40 -45

-50

-50 4

8

12 16 Time (hrs)

20

24

-35 -40 -45 -50

0

4

-25

8

12 16 Time (hrs)

20

0

24

4

8

12 16 Time (hrs)

20

24

-25

Kr

-30 -35 -40

SWNTs: PLV,buckypaper, purified at Rice Univ., annealed at Penn State in vacuum at 1000°C, 8hr

Xe

-30

S (µV/K)

S (µV/K)

Ar

-30

-35

-45

0

Ne

-30

S (µV/K)

-30

S (µV/K)

-25

-25

-25

-35 -40 -45

-45

-50

-50 0

4

8

12 16 Time (hrs)

20

24

0

4

8

12 16 Time (hrs)

20

24

c.f., Romero, K. Bolton, A. Rosen and P. C. Eklund, Science (307) 2005 NTH.‘06 , Nagano

Inert Gas Collisions: TEP vs Resistivity S vs Δρ is linear, in accord with the Mott Equation and a new scattering channel (gas collisons)

P = 1atm T = 500 K Same film sample; successive exposure to series of inert gases

Slope depends on the mass of the colliding atom c.f., Romero, K. Bolton, A. Rosen and P. C. Eklund, Science (307) 2005 NTH.‘06 , Nagano

Computed Transient Power Spectrum (Local Vibrational Modes) Θi = 45°, KE =13kcal/mol

m=He,Ne,Xe

After 5 ps

(10,0)

•The amplitude of vibration, not the frequencies, are sensitive to the collider atom mass m •Radial component of kinetic energy is key parameter

40 cm-1

•The tube wall is dented, and then it rings in the low frequency “squash” mode

After 10 ps

0 cm-1

•Local energy spectrum; Derived from motion of C-atom nearest the collision (Ttube= 0 K).

500 cm-1

•Significant vibrational energy remains after 10 ps; 2-acoustic phonon decay

Simulations by K. Bolton and A. Rosen, Goteborg University/Chalmers Tech. Univ. c.f., Romero, K. Bolton, A. Rosen and P. C. Eklund, Science (307) 2005 NTH.‘06 , Nagano

TEP (S) vs Pressure (T = 500 K)

Saturation => Dent-Dent overlap at ~ 0.2 - 0.5 atm ?? c.f., Romero, K. Bolton, A. Rosen and P. C. Eklund, Science (307) 2005 NTH.‘06 , Nagano

Why do (S,ρ) saturate with pressure? •Use Kinetic theory of Gases; Ask: What collision rate Q causes two dents with lifetime τ and diameter d to overlap? • Q = P/(2πmkT)1/2

collisions/area.time,

• Overlapping Dents He dent lifetime (τ) ~100 psec dent diameter (d) 1.5 nm Psat(calc) 0.97 atm Psat(expt) 0.79 atm

P = pressure

QAτ = 1 collision, Area A~d2 Ar ~100 psec 4.3 nm 0.66 atm 0.56 atm

Xe ~100 psec 6.4 nm 0.54 atm 0.53 atm

c.f., Romero, K. Bolton, A. Rosen and P. C. Eklund, Science (307) 2005 NTH.‘06 , Nagano

diameter d

adjacent dents

Mn Power Laws : Expt and Theory T = 500 K P = 1 atm

Mass M D

“NanoDent”

All slopes in the plot represent M1/3 behavior c.f., Romero, K. Bolton, A. Rosen and P. C. Eklund, Science (307) 2005 NTH.‘06 , Nagano

Phonon Properties of Ultrathin Graphitic Films SWNT

3-layer graphene film

nGL Graphene Si : SiO2

How are the phonon properties related???

nGL=n-layer graphene film

Let’s study the phonon properties vs n! NT ‘06 , Nagano

nGL

Motivation

Si : SiO2

• Exciting transport results in ultrathin graphitic films containing a few atomic layers have been reported – `High carrier mobility; gate-controlled transport; Quantum Hall effect (Kim et al; de Heer et al)

• We call these carbon systems n-graphene layer films or nGL’s • Using phonons to probe the system vs number of layers (n)…..how will the system evolve? NT ‘06 , Nagano

nGL Film Preparation

HOPG n=5

n=2 n=8

n=1

n=19

n

nGL substrate

SiO2: Si

(side view)

(top view)

Film thickness measured by AFM z-scans NT ‘06 , Nagano

TEM image (a) and SAD (b) of nGL a)

b) θ'

θ

a) Plane view and b) electron diffraction pattern showing six-fold symmetry. In (a), the nGL has small contrast relative to the carbon TEM grid. Angles θ, θ’ in the figure are 120°, the angle between basal plane vectors in graphite NT ‘06 , Nagano

AFM z-scans of n=3 nGL substrate

n=3 nGL

Film thicknes via averaging the step heights of ~100 line scans

NT ‘06 , Nagano

nGL thickness h(n) vs assigned n

Height (nm)

8

h = C*n + D

•AFM z-scan used to measure the height relative to substrate

C = 0.35 ± 0.01 nm D = 0.33 ± 0.05 nm

6

•Least squares fit h vs n:

4

Slope= 0.35 nm ; slightly larger than c/2=0.335 nm

2

•h(1)= 0.35 + 0.33 = 0.68 nm extra thickness may reflect inherent difference in attractive AFM tip force

h(1)= 0.7 nm

0 0 NT ‘06 , Nagano

5 10 15 Estimated layers (n)

20

High Freq. Raman Spectrum of nGLs 2nd order

G-band

Intensity (A. U.)

8 6

HOPG n=19 n=8

4

n=5 n=4

2

n=3 n=2

0 1500 NT ‘06 , Nagano

n=1

2000

2500 -1 Wave Number ( cm )

3000

G-band vs n n

nGL

substrate

HOPG

n=19 n=8 n=5 n=4 n=3 n=2 n=1

“G”

~6 cm-1 1 6 0 0

Raman NT ‘06 , Nagano

Frequency Shift ( cm-1)

G-bands are all well fit with a single Lorentzian (Voight analysis)

G-Band Frequency vs 1/n ∞

10 5 4 3 I I I I I

2 I

n=1 I

nGL

n

substrate

ωG ~ 1/n 5.5 cm-1

“G” HOPG

NT ‘06 , Nagano

Phonon and Electron Dispersion in Graphene phonons Phonon DOS ~1500 cm-1 ~1375 cm-1 electrons

NT ‘06 , Nagano

~710 cm-1

c.f., C. Mapeli et al., Politecnico di Milano (1998) in A. Ferrari and J. Robertson, Phys Rev B61 (2000)

Hg line

Raman Intensity (Counts/mW-Sec)

0.35 0.30 0.25

G-band

D-band

HOPG n=19

n=8

“D”=Disorder-induced Scattering

n=5

•Dispersive mode

0.20

•Zone edge mode

n=4

0.15

n=3

0.10

n=2

0.05

Other weak 1st order Scattering

n=1

1200

•Intensity decreases with increasing n

1

2 “D”

1400

1600

NT ‘06 , Nagano Raman Frequency Shift ( cm-1)

D/G Raman Band Intensity vs 1/n n

nGL substrate

2 nm rms

n=1 substrate n=2

n=3

NT ‘06 , Nagano

High Freq. Raman Spectrum of nGLs 2nd order

G-band

Intensity (A. U.)

8 6

HOPG n=19 n=8

4

n=5 n=4

2

n=3 n=2

0 1500 NT ‘06 , Nagano

n=1

2000

2500 -1 Wave Number ( cm )

3000

2nd Order Raman Spectra of nGLs HOPG

HOPG

Intensity (A.U.)

n=19

n=19

n=8

n=8

n=5

n=5

n=4

n=4

n=3

n=3

n=2

n=2

n=1

2400 NT ‘06 , Nagano

2450

2500

2600

2700

n=1 2800

Raman Frequency Shift ( cm-1)

3200

3300

Summary: Chemical Sensors • Carbon Nanotube and Semiconducting Nanowire FETs show great promise as real-time analytical tools – There is much to be learned about the sensing mechanism (charge transfer vs electronic scattering)

• HOWEVER, the SWNT cylindrical shell geometry should be the most sensitive. We find it is even sensitive to: – physisorption (with little or no charge transfer) – even collisions of inert gas atoms with the tube wall

NT ‘06 , Nagano

Summary: Raman Scattering from nGLs • Raman Scattering useful to characterize the number of layers in an nGL film • 1st order G-band •

– Contains one Lorentzian – G-band upshifts with decreasing n: ωG(n) ~ ωGraphite + 7 cm-1(1/n) 2nd Order bands – Stronger than G-band for n