The Sensitivity Limits of Nanowire Biosensors Xuan Gao Dept of Chemistry and Chemical Biology, Harvard University Jan. 15 th, 2007 Texas A&M University
Why Nano for Bio-detection? Protein/DNA Virus Cell 0.1 1nm 10 100 1μm 10 Nanoarrays/assembles Nanodots, wires Nano high sensitivity! Sensitivity matters: early diagnostics of disease (low concentration protein detection); genomics (e.g. DNA sequencing); biophysics research (e.g. single molecule techniques)
utline 1. Motivation and introduction 2. Nanowire FET biosensors: sensitivity limits? A fundamental analysis: importance of length scales Pushing sensitivity limits: ph, protein detections Charge detection limits 3. Noise spectroscopy for bio-reaction 4. Perspective/Future directions
Nano for bio-detection: what have been demonstrated Bio-bar-coded nanoparticle Micromechanical device Surf plasmon reson. of nanoparticle Mirkin group, Science 2003 Fritz et al., Science 2000 Chilkoti, Anal Chem 2004 Carbon Nanotube FET Nanowire FET Main idea: Low concentration molecule binding on surface Large change in physical properties of volume (high surface-volume ratio) Dai group, PNAS 2003 Lieber group, Science 2003; Nat. Biotech 2005
Nanowire synthesis and FET fabrication CVD synthesis of Si-nanowire Nanowire Field-Effect-Transistor SiH 4 Au Si 3 N 4 Passivation layer (~60 nm) Doped p-type Si Nanowire 60 nm Ni for source and drain S D Substrate or back gate Channel length ~ 2 μm 5 nm I sd (na) 600 400 200 0-200 -400-600 -1.0-0.5 0.0 0.5 1.0 V sd (V) V sd (V) G (ns) 600 500 100 400 300 10 200 100 1 0-10 -5 0 5 10 V g (V) Back gate V g (V) e.g. L.J. Lauhon et al., Nature 420, 57 (2002). e.g. S.Jin et al., Nano Lett. 4, 915 (2004).
NW-FET sensor array chip Si-NW S D 100um G buffer sample time
Nanowire FET as biosensor: basic principle analyte - - - - s nanowire substrate d - s nanowire substrate d I sd or G I sd or G The field effect of molecule binding acts as a gate voltage time time
Nanowire FET as biosensor: sensitivity limits? analyte - - - - s nanowire substrate d - s nanowire substrate d I sd or G I sd or G The field effect of molecule binding acts as a gate voltage time time NanowireFET sensor: ultrasensitive, label-free, real-time, and multiplexible What is the fundamental limit of sensitivity/response? (That s why nano matters!)
NW-FET sensitivity: the role of carrier screening R E c ~λ E i E F E v High carrier density, λ<r NW carrier density&conductance changes in depth ~λ near surface. ΔG ~ Δp 2πR λ s d screening length λ (nm) 40 20 10 8 6 4 2 0.8 1 0.6 Volume conductance G p πr 2 Debye screening length: Debye ΔG/G λ/r (surface-volume ratio effect!) λ= εk B T pe 2 Screening length λ for p-silicon Typical R of NW Thomas-Fermi 10 16 10 17 10 18 10 19 10 20 p (cm -3 ) Typical doping level
NW-FET sensitivity: the role of carrier screening Δϕ R E F Low carrier density, λ>r Analyte molecules at surface gate the whole volume of nanowire s d screening length λ (nm) 40 20 10 8 6 4 2 0.8 1 0.6 Debye screening length: Debye λ= εk B T pe 2 Screening length λ for p-silicon Typical R of NW Thomas-Fermi 10 16 10 17 10 18 10 19 10 20 p (cm -3 )
Nanowire-FET sensors: the limiting response R s d Δϕ E F 1. Non-degenerate semiconductors: p = p i {exp(e i -E F )/k B T } 2. NW conductance G = p πr 2 eμ /L Low carrier density, λ>r Analyte molecules at surface gate the whole volume of nanowire NW sensor has the limiting (most sensitive) response in the long screening length regime: ΔG/G = exp(δϕ/kt)-1
NW-FET sensitivity: the role of carrier screening R R E c E i Reduce R, increase λ Δϕ ~λ E F E v Δϕ E F λ<r Molecules on surface gate ~λ depth into nanowire λ>r Molecules on surface gate the whole volume of nanowire ΔG/G exp(δϕ/kt)-1
λ>r is reached in the subthreshold regime Gate voltage V g G (ns) 1200 1000 800 600 400 200 V g s Linear regime g H 2 d Threshold voltage 0-0.6-0.4-0.2 0.0 0.2 0.4 0.6 water gate V g (V) solution gate V g (V) 1000 100 10 Subthreshold 1 screening length λ (nm) 60 40 20 10 8 6 4 2 0.8 1 Subthreshold regime λ>r R λ<r Linear regime 10 16 10 17 10 18 10 19 10 20 p (cm -3 ) NW sensor has the limiting (most sensitive) response in the long screening length (λ>r) regime: ΔG/G = exp(δϕ/kt)-1
Nanowire sensor: ph Sensing ~ V sd Δϕ - - - solution ph = -log 10 [H ] Gate/reference electrode - Double layer Si-H Si- - H NH 2 H NH 3
Pushing Sensitivity of Nanowire Sensors: ph sensing 700 10nm diameter p-type silicon NW 600 1000 1000 500 800 100 ΔG/G (%) 400 300 200 100 4 5 6 7 0 0 500 1000 1500 time (s) ph 9 40% G (ns) 600 400 200 0-0.4-0.2 0.0 0.2 0.4 Vg (V) 10 1 0.1
Pushing Sensitivity of Nanowire Sensors: ph sensing 700 10nm diameter p-type silicon NW 600 1000 1000 500 800 100 ΔG/G (%) 400 300 200 7 100 6 5 4 0 0 500 1000 1500 time (s) ph 9 150% 40% G (ns) 600 400 200 0-0.4-0.2 0.0 0.2 0.4 Vg (V) 10 1 0.1
Pushing Sensitivity of Nanowire Sensors: ph sensing 10nm diameter p-type silicon NW ΔG/G (%) 700 600 500 400 300 subthreshold near threshold linear regime 200 6 100 5 4 0 0 500 1000 1500 time (s) 7 ph 9 600% 150% 40% G (ns) 1000 800 600 400 200 0-0.4-0.2 0.0 0.2 0.4 Vg (V) 1000 100 10 1 0.1
Pushing Sensitivity of Nanowire Sensors: ph sensing ΔG/G = exp(δϕ /kt)-1 for Δϕ =30mV/pH 1000 Δϕ 1000 Linear 800 100 subthreshold G (ns) 600 400 Δϕ 10 200 1 0-0.4-0.2 0.0 0.2 0.4 Vg (V) 0.1 ΔG= Δϕ g m for Δϕ=30mV/pH Every ph change is equivalent to 30mV Vg shift in both regimes (also true for different NW diameter, doping level, etc.)
Pushing Sensitivity of Nanowire Sensors: ph sensing ΔG/G = exp(δϕ /kt)-1 for Δϕ =30mV/pH Electrochemical potential: μ ec = qϕ k B T ln ([H ] s /[H ] b ) Linear subthreshold Δϕ =k B T/q ln(10) ΔpH bulk ~59mV/pH If ph surface can be kept constant ph surface ΔG/G= Δϕ g m /G for Δϕ=30mV/pH ph bulk Si-H Si- - H NH 2 H NH 3 Δϕ
Pushing Sensitivity of Nanowire Sensors: ph sensing summary ΔG/G (%) 700 600 500 400 300 subthreshold near threshold linear regime 200 6 100 5 4 0 0 500 1000 1500 time (s) 7 ph 9 For chemical reaction induced Δϕ: The strongest response is indeed observed in subthreshold regime and ΔG/G = exp (Δϕ /kt)-1 chemistry doesn t change (Δϕ=const.)! E c E F E F ~λ E v Linear regime Subthreshold regime
Cancer marker protein sensing -surface modification HH H H H H Et Et Si C Et H (1% in 95% EtH) C Si H C Si H (1) Silane modification (2) Antibody modification NH 2 CNBH 3 in binding buffer (10mM, ph 8.4) F. Patolsky et al, Nat. Protocols 1, 1711 (2006). (3) Silane blocking H NH 2 CNBH 3 in binding buffer (10mM, ph 8.4) H H H NH C H H NH C H H NH C H H NH C H Si Si (4) Sensing Si Si in assay buffer (10uM, ph 7.1)
Cancer marker protein sensing - antibody - antigen G (μs) 1.50 1.45 1.40 1.35 ΔG PSA 150 nm 15 nm 1.5 nm 150 pm 15 pm 1.5 pm buffer 1.30 k a [Ab] [PSA] [Ab/PSA] k d 1.25 0 1000 Time (s) 2000 3000 [Ab/PSA] eq = [Ab]*[PSA] K d [PSA] Disassociation constant K d =k d /k a K d ~ nms (e.g. Saerens et al, J. Bio. Chem. 2004) ΔG (ns) 140 120 100 80 60 40 20 0 0 50n 100n 150n K d 1p 10p 100p 1n 10n 100n PSA concentration (M)
Pushing Sensitivity of Nanowire Sensors -cancer marker protein sensing - antibody - antigen 15 pm PSA buffer buffer G (ns) 1200 800 400 1700nS/V in linear regime 1000 100 110mV/dec 10 Linear regime 0-0.6-0.4-0.2 0.0 0.2 0.4 1 0.6 Vg (V) For a given protein concentration, detection is greatly improved in the subthreshold regime! Subthreshold regime (Δϕ~15mV for 15pM PSA in both regimes)
Pushing Sensitivity of Nanowire Sensors -cancer marker protein sensing ΔG/G (%) 30 25 20 15 10 5 12 - antibody - antigen 15pM PSA threshold voltage e Δϕ /kt -1 G (ns) 1200 800 400 1700nS/V in linear regime 0 1-0.6-0.4-0.2 0.0 0.2 0.4 0.6 Vg (V) 110mV/dec 1000 100 10 signal noise ratio 10 8 6 For a given protein concentration, detection is greatly improved in the subthreshold regime! 4 0.0 0.1 0.2 0.3 0.4 0.5 V g (V) (Δϕ~15mV for 15pM PSA in both regimes)
Nanowire Sensor detection limits -linear vs. subthreshold regime PSA detection in linear regime PSA detection in subthreshold regime PSA Time (s) Time (s) Detection limit reduced from pms to fms by operating nanowire sensor in the subthreshold regime where screening length λ>r!
Charge Sensitivity of Nanowire Sensors ΔQ Δϕ - - - - - - - - Surface charge detected: ΔQ= [C DL 1/(C -1 Si2 C -1 NW )] Δϕ L C DL Δϕ C NW C Si2
Charge Sensitivity of Nanowire Sensors ΔQ Δϕ 100 μm buffer R=λ Si R=5λ Si Surface charge detected: ΔQ= [C DL 1/(C -1 Si2 C -1 NW )] Δϕ L Solving Poisson-Boltzmann equation Δp(r), Δϕ(r). -60-40 -20 0 20 40 60 80 nm C DL Δϕ C NW C Si2
Charge Sensitivity of Nanowire Sensors ΔQ Δϕ 100 μm buffer R=λ Si R=5λ Si Surface charge detected: ΔQ= [C DL 1/(C -1 Si2 C -1 NW )] Δϕ L Solving Poisson-Boltzmann equation Δp(r), Δϕ(r). -60-40 -20 0 20 40 60 80 nm Q C NW NW R = Δ p e 2πrdr 0 = dq NW / dδϕ C DL Δϕ C Si2 λ DL C NW For R~10nm, L=1μm C NW : ff afs C DL : ffs C Si2 : ~ffs C NW e 2 (pπr 2 )/k B T 2λ Si /R, high p C NW e 2 (pπr 2 )/k B T, low p hole density per unit length of NW
Charge Sensitivity of Nanowire Sensors ΔQ Δϕ 100 μm buffer R=λ Si R=5λ Si Surface charge detected: ΔQ= [C DL 1/(C -1 Si2 C -1 NW )] Δϕ L C DL Δϕ L (in subthreshold) 2πε/ln(1λ DL /R) Δϕ L ε=80, L: device length, R: nanowire radius -60-40 -20 0 20 40 60 80 nm Δϕ min Linear Subthreshold 5mV 0.66mV C DL Δϕ C Si2 λ DL C NW For R~10nm, L=1μm C NW : ff afs C DL : ffs C Si2 : ~ffs ΔQ min (I.S.=10mM) ΔQ min (I.S.=10μM) ~300e ~40e ~50e ~6e * Calculated for per μm NW with R=10nm
utline 1. Motivation and introduction 2. Nanowire FET biosensors: sensitivity limits? A fundamental analysis: importance of length scales Pushing sensitivity limits: ph, protein detections Charge detection limits 3. Noise spectroscopy for bio-reaction 4. Perspective/Future directions
Nanowire FET noise as a probe for biochemical reactions--motivation Fluctuation dissipation theorem: the response of a system to an external perturbation is related to the fluctuations of the system in thermal equilibrium. cantilever stage R Noise power (A 2 ) ω reson ~Q -1 Johnson noise R or T AFM tip noise ω reson & Q Hutter, Bechhoefer RSI (1993). spin noise hyperfine splittings Crooker et al, Nature (2004). Noise can give useful information without disturbing the system!
Nanowire FET noise as a probe for biochemical reactions--motivation Fluctuation dissipation theorem: the response of a system to an external perturbation is related to the fluctuations of the system in thermal equilibrium. Can we obtain reaction kinetics from the noise of bio-sensors at equilibrium? binding unbinding Buffer Sample Buffer? Reaction Kinetics k a [Ab] [Ag] [Ab/Ag] k d
Protein binding induced Lorentzian noise G ( μs) 1.2 1.0 0.8 150 pm 150 pm real-time 5 pm 0.15 pm S 10-5 10-6 10-7 10-8 1/f buffer 1/f S 10-4 10-5 10-6 10-7 10-8 5 G 5nM = 4600 0 2 4 6 8 Time (s) 10 12x10 3 10 0 10 1 10 2 10 3 10 4 10 5 f (Hz) 10 1 10 2 10 3 10 4 10 5 f (Hz) S 10-5 10-6 150pM G = 3700 S 10-5 10-6 5 G 5pM = 4000 S 10-4 10-5 10-6 0.15 pm G = 3600 0.15pM PSA 10-7 10-7 10-7 10-8 10-8 10-8 10 1 10 2 10 3 10 4 10 5 1 1 f (Hz) 10 1 10 2 10 3 10 4 10 5 f (Hz) τ e t Time domain 2 2 τ f ; τ ~1 ms 10 1 10 2 10 3 10 4 10 5 f (Hz) PSA binding/unbinding at equilibrium adds a Lorentzian noise to the 1/f noise of NW. Lorentzian noise still presents in low concentrations which are not detectable in time domain
Summary 1. Sensitivity limits of nanowire biosensors Where to achieve the best sensitivity?-subthreshold. reduced carrier screening efficient molecule gating; ΔG/G exp(δϕ /kt)-1. Pushing sensitivity limits: ph, protein detections pm PSA (linear) fm PSA detection (subthreshold) Charge detection limits (a few e s) 2. Probing bio-reaction by device noise protein binding induced Lorentzian noise S 10-5 10-6 10-7 10-8 10-4 10-5 ~λ E c E F E v vs. 15 pm PSA subthreshold Linear regime 1/f E F Linear 10 0 10 1 10 2 10 3 10 4 10 5 f (Hz) buffer 1/f 0.15 pm G = 3600 PSA S 10-6 10-7 Subthreshold 10-8 10 1 10 2 10 3 10 4 10 5 f (Hz)
Nanowire Research: Future Directions (1) Towards single molecule detection and reaction dynamics study G Bio-Apps single molecule study reaction dynamics capacitance sensing biomolecule manipulation and electrokinetics protein/dna/cell Initial approaches: higher mobility / shorter channel devices; suthreshold regime sensing G τ on time τ off Label free & direct electrical study of single biomolecules and their reaction dynamics
Nanowire Research: Future Directions (2) capacitance sensing (ε effect); Bio-Apps single molecule study reaction dynamics capacitance sensing biomolecule manipulation and electrokinetics protein/dna/cell capacitance bridge Conductance charge of biomolecules Capacitance dielectric properties of biomolecules
Nanowire Research: Future Directions Physics G(μS) Quantized at low T! diffusive vs. ballistic transport in 1D single particle vs. many-body thermal/thermoelectric properties charge vs. spin transport 1D hole gas, Lu et al, PNAS 2005, Xiang et al, Nature 2006 V g (V) Hetero-structures will enable novel physics study in clean ( ballistic ) 1D nanostructures How to directly probe ballistic holes in NW? How about heat transport?...
Nanowire Research: Future Directions Development new materials, structures addressable arrays micro-fluidics, lab on a chip novel devices, energy solutions Bio-Apps proteomics, genomics drug discovery, cell biology single molecule study reaction dynamics capacitance sensing biomolecule manipulation Physics and electrokinetics protein/dna/cell G τ on τ off time diffusive vs. ballistic single particle vs. many-body charge vs. spin transport electric/thermal properties C
Acknowledgement Dr. GengFeng Zheng Prof. Charles Lieber The Lieber group Thank You!