Sup

UV-

attac

urac

over

nond

DNA

DNA

also

the e

abso

the

anth

pplementary

-Vis absorp

ched to the

cil moiety i

rlapping π–

denaturing p

A. Longer D

A ladder sam

referred to

extra Aq mo

orption peak

anthraquino

hraquinone w

y Figure 1

ption spect

e DNA back

is located i

–orbitals. b,

polyacrylam

DNA seque

mple (DNA

as DNA ru

oiety. c, Bo

k for DNA.

one n to π

was success

│Structure

tra of Aq-D

kbone. The

in the mino

, The form

mide gel elec

ences have

A mixture co

uler). Aq-DN

th spectra h

The Aq-DN

π* transition

sfully introd

e, Nondena

DNA and u

e Aq moiety

or groove1,

mation of d

ctrophoresis

slower mob

onsisting of

NA has a sli

have strong

NA has a we

n absorptio

duced into th

aturing PAG

u-DNA. a,

y stacks wi

allowing c

double-helic

s. Column 2

bility as ind

10 bp-200 b

ight slower m

absorption a

eak absorpti

n. The UV

he DNA.

GE gel elec

Anthraquin

ith adjacent

charge trans

al DNA w

2 is u-DNA.

dicated by

bp dsDNA

mobility tha

at 260 nm, w

ion at aroun

V-Vis spectr

ctrophoresi

none (Aq) m

t bases whi

sport throug

was confirm

. Column 3

column 1 f

in 10bp inte

an u-DNA,

which is a t

nd 340 nm,

ra confirme

1

is and

moiety

ile the

gh the

med by

is Aq-

for the

ervals,

due to

typical

due to

ed the

Sup

with

(CV

peak

poss

lowe

histo

Note

any

pplementary

h two thiola

V) as the con

ks in the con

sibility of so

er limit of th

ograms und

e that the CV

peaks eithe

y Figure 2│

ate linkers a

ntrol exper

nductance h

olvent-based

he current a

der different

V does not

r because th

│Conducta

at different

riments. a, C

histogram de

d molecular

amplified in

gate voltag

show redox

he conductan

nce histogr

t gate volta

Conductanc

etected with

r junction fo

the STM sc

ges with the

x peaks. Con

nce of ssDN

rams of cac

ges with th

ce histogram

hin the condu

ormation. Th

canner. b, S

CV for ssD

nductance hi

NA is below

codylate buf

he cyclic vol

m of cacodyl

uctance ran

he noise leve

tructure and

DNA with tw

istograms a

w the measur

ffer, and ss

ltammogra

late buffer. N

nge, ruling o

el is due to

d conductan

wo thiolate l

lso do not h

rable range.

2

sDNA

m

No

out the

the

nce

linkers.

have

Sup

subs

eyes

quas

stud

surfa

with

to th

shift

volta

(Sup

erro

vs. p

vers

pplementary

strate. a, Pl

s. In the elec

si reference

dy the variat

face for 3 ho

hin 3 hours)

he positive d

fted to the ne

ammogram

pplementary

rs in the gat

potential sw

sus potential

y Figure 3│

lot of the red

ctrochemica

electrode in

tions, we rec

ours (all the

in 10 minu

direction wi

egative direc

s before and

y Fig. 4-5). H

te voltages (

weeping rate

l sweeping r

│Cyclic vol

dox potentia

al gate contr

ntroduced v

corded the c

conductanc

tes intervals

ith time, from

ction. This p

d after the E

Hence, we u

(Vg). b, Sep

. Red line is

rate. The lin

ltammogram

al changes v

rolled condu

variations in

cyclic voltam

ce measurem

s. We found

m which we

phenomeno

EC-STM bre

used the dif

aration betw

s a guide to

near relation

m study of

vs. time, wh

uctance mea

the actual a

mmograms

ments on the

d the redox p

e concluded

on is robust a

eak junction

fferences in

ween the ox

eyes. c, The

n confirms th

the Aq-DN

here the red

asurement, t

applied gate

for Aq-DNA

e Au surface

potential of

d that the ref

as we can se

n conductanc

the redox p

idation and

e cathodic p

hat the redo

NA modified

line is a gui

the use of si

e voltages. T

A modified

e can be fini

f Aq-DNA sh

ference pote

ee from the

ce measurem

otential as t

reduction p

peak current

ox peaks we

3

d Au

ide to

ilver

To

Au

ished

hifted

ential

cyclic

ments

the

peaks

t Ipc

re due

4

surface bound Aq-DNA molecules. Fitting the data with a linear dependence (red line)

provides a surface coverage of 1.48±0.03 pmol·cm-2. Ag/AgCl reference electrode was

used in b and c. d, Cyclic voltammogram obtained with 0.1 V·s-1 sweeping rate, from

which a cathodic peak and anodic peak were determined after baseline correction. Note:

A silver wire was used as quasi-reference electrode to be consistent with the STM break

junction experiments. The redox potential was determined by averaging the cathodic peak

and anodic peak. e, Cathodic peak in the CV. f, Anodic peak in the CV. Note: The

potential in e and f are quoted with respect to the redox potential of anthraquinone. The

areas of the cathodic and anodic peaks provide the number of molecules participated in

the redox reaction.

Sup

elec

cond

each

cond

Ag w

take

the r

Whe

in ea

pplementary

trochemica

ductance h

h conductan

ductance me

wire quasi-r

en as error b

redox poten

en the appli

ach of the co

y Figure 4│

al gate cont

istograms o

nce histogram

easurement,

reference ele

bars in the ga

ntial of Aq, o

ed gate volt

onductance

│Cyclic vol

trolled cond

of Aq-DNA

m. The blac

, respectivel

ectrode is sh

ate voltages

only one pe

tage is close

histograms

ltammogram

ductance m

A. Note that

ck and red cu

ly. Redox po

hown in the

s. When the

ak was give

e to the redo

s.

ms (CVs) b

measuremen

the applied

urves are th

otential of A

e CV. The sh

applied gat

en in each of

ox potential

before and a

nts and the

gate voltag

he CVs befor

Aq/H2Aq wi

hifts in the r

te voltage is

f the conduc

of Aq, two

after

correspond

ge was show

re and after

ith respect t

redox peaks

s far away fr

ctance histo

peaks were

5

ding

wn in

the

to the

s were

rom

ograms.

given

Sup

elec

cond

The

curv

pote

the C

bias

cond

pola

are n

pplementary

trochemica

ductance h

applied gat

ves are the C

ential of Aq/

CV. The shi

voltages (n

ductance of

arity. c, The

no conducta

y Figure 5│

al gate cont

istograms a

te voltage w

CVs before a

/H2Aq with

ifts in the re

not 5 mV) ar

f low and hig

applied gat

ance peaks i

│Cyclic vol

trolled cond

and the pea

was shown in

and after the

respect to t

edox peaks w

re shown in

gh conducta

te voltages w

in the range

ltammogram

ductance m

ak position

n each cond

e conductan

the Ag wire

were taken a

n the conduc

ance peak is

were marked

from ~10-5.

ms (CV) of

measuremen

vs. gate vo

ductance hist

nce measure

quasi-refere

as error bars

ctance histog

s not depend

d in the con

.5 to ~10-4 G

f Aq-DNA b

nts, the corr

ltage plot f

togram. The

ement, respe

ence electro

s in the gate

grams. Thes

dent on bias

nductance hi

G0. Note that

before and

responding

for Aq-DNA

e black and

ectively. Red

ode is shown

e voltages. b

se show that

voltage or b

istograms. T

t the whole

6

after

g

A. a,

red

dox

n in

b, The

t the

bias

There

cond

not c

Sup

cycl

cont

Con

Ag w

gate

ductance ran

change whe

pplementary

lic voltamm

trol experim

nductance hi

wire quasi-r

e voltage. Th

nge is from

en gate volta

y Figure 6│

mograms (C

ments. a, St

istogram un

reference ele

he red line i

~10-5.5 to ~

age is from

│Conducta

CV) and con

tructure of u

nder differen

ectrode. d, C

s a guide to

10-2.8 G0. d,

0.2 to 0.55 V

nce histogr

nductance v

u-DNA. b, C

nt gate volta

Conductanc

o eyes.

, The low on

V.

ram at diffe

vs. gate volt

CV does not

ages where t

ce does not c

nductance o

erent applie

tage plot fo

t show redo

the Vg is wit

change with

of Aq-DNA

ed gate volt

or u-DNA a

ox peaks. c,

th respect to

h respect to t

7

does

tages,

as the

o the

the

Sup

is th

corr

leng

valu

whic

histo

pplementary

he distance o

responds to t

gth is repres

ue of the dis

ch does not

ograms for u

y Figure 7│

over which a

the length o

ented by the

tance on the

have any ph

u-DNA, hig

│Step lengt

a molecular

of the condu

e distance b

e x-scale me

hysical mea

gh conductan

th analysis

r junction ca

uctance plate

etween the

eans the mo

aning to the

nce state of

of Aq-DNA

an be stretch

eau in the cu

two black d

ovement of t

molecular j

f Aq-DNA a

A and u-DN

hed before b

urrent-distan

dash lines. T

the piezo sc

junctions. b

and low cond

NA. a, Step l

breakdown,

nce traces. T

The absolute

canner in the

-d, Step len

ductance sta

8

length

which

This

e

e STM,

ngth

ate of

9

Aq-DNA. The peaks were fitted with a lognormal distribution and the mean values were

taken as the step length. Errors were the fitting errors.

Sup

volt

mea

(bef

f, 2D

only

volta

pplementary

tages and cy

asurements

fore and afte

D G-t histog

y one band,

ages in Sup

y Figure 8│

yclic voltam

) for u-DNA

er the condu

grams at Vg

which is co

plementary

│2D conduc

mmograms

A as the co

uctance mea

of -0.65 V,

nsistent wit

Fig. 6.

ctance-time

(before and

ntrol exper

asurements)

-0.60 V, -0.

th the condu

e (G-t) histo

d after the

riments. a,

for u-DNA

.55 V, -0.50

uctance mea

ogram at di

conductan

Cyclic volta

, showing n

0 V and -0.4

asurement at

ifferent gat

ce

ammograms

no redox pea

45 V, showin

t difference

10

te

s

aks. b-

ng

gate

Sup

state

mea

dete

curr

trace

swit

the

appe

low

pplementary

e of Aq-D

asurements).

ected. Then

rent was rec

es were ob

tching event

low conduc

ear. e, 1D c

conductanc

y Figure 9

DNA. a, c

. b, Condu

the tip was

orded. c, Si

served (bla

ts occurring

ctance band

onductance

ce state to hi

│2D G-t hi

cyclic volta

uctance-dista

s fixed in p

imilar to Fig

ack, red and

g (magenta t

d fades aw

e histograms

igh conduct

stogram stu

ammograms

ance trace

position (gr

gure 5b, thre

d blue trace

trace). d, 2D

ay with tim

s at t = 0 an

tance state.

udy started

s (before

with platea

ey points) a

ee major typ

es). A few

D G-t histog

me, and the

nd t = 0.1 s,

d with the l

and after

au at low

and held fo

pes of condu

of the trac

gram showi

e high cond

showing th

low conduc

the condu

conductanc

or 0.1 s whi

uctance-tim

ces have mu

ing two ban

ductance sta

he switching

11

ctance

ctance

ce was

ile the

me (G-t)

ultiple

nds. As

arts to

g from

Sup

volt

mea

d, V

the C

posi

pplementary

tages and cy

asurements

Vg = -0±5 mV

CV. Notice

itive.

y Figure 10

yclic voltam

) for Aq-DN

V. e, Vg = 9±

that there w

0│2D condu

mmograms

NA. a, Vg =

±10 mV. Ga

will be more

uctance-tim

(before and

= -30±17 mV

ate voltage i

e switching e

me (G-t) his

d after the

V. b, Vg = -2

is vs. the re

events when

togram at d

conductan

21±7 mV. c,

dox potentia

n gate voltag

different ga

ce

, Vg = -13±8

al of Aq/H2

ge is more

12

ate

8 mV.

Aq in

Sup

Aq-

3’ or

redo

show

betw

geom

pplementary

DNA. a, Str

r Aq-DNA-

ox peak of A

wing no pea

ween the two

metry as sho

y Figure 11

ructure of A

-5’ respectiv

Aq moiety. c

aks, which in

o electrodes

own in Fig.

1│Structur

Aq-DNA wi

vely. b, CV

c, Conducta

ndicates the

s. Note: The

1a.

e, CV and c

th monothio

for Aq-DNA

ance histogra

e need of bo

e only possib

conductanc

ol at 3’-end

A-3’ and Aq

ams of Aq-D

oth thiol grou

ble binding

ce histogram

or 5’-end, n

q-DNA-5’,

DNA-3’ and

ups to bridg

geometry is

ms of mono

named Aq-D

showing the

d Aq-DNA-

ge Aq-DNA

s the uprigh

13

othiol

DNA-

e

-5’

A

ht

Sup

diffe

beha

curv

Sup

with

mea

diffe

pote

Whe

pplementary

erent voltag

avior. c, 2-d

ves, showing

pplementary

h 10 base pa

asurements,

erent electro

ential, only l

en gate volta

y Figure 12

ge polarities.

dimensional

g that the co

y Figure 13

airs. a, Stru

showing the

ochemical g

low conduct

age is close

2│Current-

. b, Represe

conductanc

onductance

3│Structur

ucture of Aq

e redox pea

gate voltages

tance peak o

to the redo

-voltage (I-

entative I-V

ce-voltage (

does not dep

e, CV and c

q-DNA-2. b

ak of Aq mo

s. When gat

or high cond

x potential,

V) characte

curves for A

G-V) constr

pend on bia

conductanc

, CV before

iety. c, Con

te voltage is

ductance pe

both peaks

eristics of A

Aq-DNA, sh

ructed by th

as voltage or

ce histogram

e and after th

nductance hi

s above or b

eak shows u

show up. T

Aq-DNA. a,

howing line

ousands of I

r its polarity

m of Aq-DN

he conducta

istograms un

elow the red

up, respectiv

The low and

14

, Two

ear

I-V

y.

NA-2

ance

nder

dox

vely.

high

cond

and

char

the t

Sup

orbi

sem

311+

Gua

the o

ductance va

high condu

rge transpor

two states b

pplementary

ital spatial

miempirical

+G(p,d) lev

anines comp

orbital main

alues are 2.5

uctance valu

rt involves th

by changing

y Figure 14

distributio

level. Simil

vel of theory

paring to the

n localized o

5×10-4 G0 an

es for Aq-D

he DNA seq

the DNA se

4│ Energy d

n in a) oxid

lar to Fig. 6

y, the HOMO

e HOMO of

on the Aq m

nd 1.4×10-3 G

DNA. These

quences and

equence.

diagram fo

dation state

obtained fr

O-1 of H2Aq

f Aq. Molecu

moiety for bo

G0, respecti

findings de

d one can tu

or the charg

e and b) red

rom DFT ca

q is closer t

ular orbital

oth cases.

ively, smalle

emonstrate t

une the cond

ge transpor

duction stat

alculations a

o the HOM

spatial distr

er than the l

that the mea

ductance val

rt and mole

te at ZINDO

at the M06-2

O levels of

ribution indi

15

low

asured

lues of

ecular

O/S

2X/6-

icates

Sup

orbi

a, T

for 0

Both

posi

pplementary

ital spatial

he Aq and H

0.2 Å. Resu

h energy lev

ition of Aq m

y Figure 15

distributio

H2Aq were

ults are from

vel alignmen

moiety.

5│Energy d

n in both st

moved dow

m DFT calcu

nts and spat

diagram for

tates when

wn for 0.2 Å

ulations at th

tial distribut

r the charg

modifying

. b, The Aq

he M06-2X/

tion are wea

ge transport

the positio

q and H2Aq w

/6-311+G(p,

akly depend

t and molec

on of Aq mo

were moved

,d) level of t

dent on the

16

cular

oiety.

d up

theory.

Sup

orbi

H2A

for 2

Both

of A

pplementary

ital spatial

Aq were mov

20º. Results

h energy lev

Aq moiety.

y Figure 16

distributio

ved clockwi

are from D

vel alignmen

6│Energy d

n in both st

ise for 20º. b

DFT calculat

nts and spat

diagram for

tates when

b, The Aq a

tions at the M

tial distribut

r the charg

rotating th

and H2Aq w

M06-2X/6-3

tion are wea

ge transport

he Aq moiet

were moved

311+G(p,d)

akly depend

t and molec

ty. a, The A

countercloc

) level of the

dent on the a

17

cular

Aq and

ckwise

eory.

angle

18

Supplementary Table 1│Conductance values with experimental errors under

different electrochemical gate voltages.

19

Relative Gate Voltage Vg/V

Error in Gate Voltage/V

Conductance Value/10-4 G0

Error in Conductance Value/10-4 G0

-0.0114 0.0021 30.2 0.7-0.0114 0.0021 4.0 0.10.014 0.020 30.2 0.7

0.0144 0.020 4.4 0.10.0013 0.020 33.1 0.80.0013 0.020 3.1 0.10.024 0.011 29.5 1.40.024 0.011 3.1 0.1-0.045 0.013 29.5 0.70.0414 0.0022 3.1 0.1-0.0020 0.0087 30.2 0.7-0.0020 0.0087 4.4 0.1-0.0251 -0.0026 37.2 0.9-0.0251 -0.0026 4.5 0.1-0.032 -0.013 37.2 0.9-0.032 -0.013 4.0 0.20.011 -0.021 25.7 0.60.011 -0.021 3.3 0.10.030 -0.017 27.5 2.50.030 -0.017 3.2 0.1-0.015 -0.016 33.1 0.8-0.015 -0.016 3.5 0.1-0.227 0.042 28.8 0.70.139 0.019 3.7 0.10.239 0.019 3.8 0.1-0.078 0.019 26.9 0.60.1320 0.0066 3.1 0.1-0.1010 0.0092 28.8 0.70.085 0.013 4.4 0.1-0.146 0.015 31.6 0.70.187 0.034 3.2 0.1

-0.0410 0.0092 30.9 0.70.503 0.042 3.8 0.10.239 0.019 3.8 0.10.297 0.034 3.3 0.10.339 0.025 4.0 0.10.390 0.016 3.9 0.10.440 0.016 4.6 0.1

20

Supplementary Table 2│kf and kb values, fitted gate voltages and the applied gate

voltages from 2D G-t histograms. By fitting the curves in Fig. 5e (The peak area of the

high conductance states versus time plot in the 2D G-t histograms) with equation (4), we

were able to obtain kf and kb. Combing these values with equation (3), the fitted gate

voltages can be determined to compare with the applied gate voltages.

kf kb Fitted Vg/mV Applied Vg/mV 8.49+0.30 4.0+0.9 9+3 9+10 9.78+0.34 10.0+1.0 0+2 0+5 5.65+0.38 15.3+2.2 -13+3 -13+8 4.17+0.37 12.0+2.7 -13+4 -21+7 2.09+0.58 16+9 -25+11 -30+17

Supplementary Table 3│Experimental error for Aq-DNA and u-DNA when EC gate

voltage is off

1st time 2nd time 3rd time Result Aq-DNA -3.41 -3.41 -3.39 -3.40±0.01 u-DNA -2.82 -2.83 -2.87 -2.85±0.02

Supplementary Note 1│Analysis of the relationship between the conductance peak

area and the surface coverage of the redox species

To confirm that there are only two conductance states, we checked the conductance

ranging from 3.2×10-6 to 2.5×10-2 G0 and did not detect any other conductance peaks (see

Supplementary Fig. 5l-5n). Also, it is unlikely that more than one DNA molecule could

bridge between the tip and substrate, due to repulsion between the negatively charged

DNA. Thus we attribute each of the individual current-distance traces in the conductance

21

histograms to one single DNA molecular junction. In this way, the peak area S in a

conductance histogram can be expressed by2,

(1)

where nj is the number of molecular junctions (traces that have plateau), L is the length of

the plateau regime, or step length, f is the sampling frequency, v is the pulling rate in

nm·s-1 and U is the bin number in the conductance histogram. The step lengths L for the

two conductance states are the same (Supplementary Fig. 7, consistent with Bruot et al.’s

results3), so the peak area S is proportional to the number of molecular junctions, nj.

In STM break junction experiment, nj can be expressed as:

jn N Y (2)

Where N is the total current-distance trace collected and Y is the chance of forming a

molecular bridge during one current-distance trace. The oxidized state and reduced state

were measured under the same circumstance (e.g. same solution, temperature, Au tip and

substrate), and their structures only differ by the Aq moiety. Therefore, we believe that

the Y is proportional to the surface coverage of the species Γ. Thus ideally the peak area S

is proportional to Γ.

Supplementary Note 2│Switching probabilities of G-t trace under different gate

voltages

j

LfS n

vU

22

The conductance in the G-t trace can only be either in the high conductance state (Aq is

in oxidized state) or in the low conductance state (Aq is in reduced state), which can be

described as:

1 ox redG p G p G , p = 0 or 1. (3)

However, the probability of finding a single molecule in the reduced state Peq(red) or in

the oxidized state Peq(ox) under an equilibrium depends on the gate voltage Vg according

to Nernst equation.

eq

1(red)

1g

nFV

RT

P

e

(4)

eq (ox)

1

g

g

nFV

RT

nFV

RT

eP

e

(5)

Notice that Peq(red) + Peq(ox) = 1. The Equation (4) in the main text describes how the

percentage of reduced state changes with respect to time when starting from Pt(red) = 1.

As time t goes to infinity, the Equation (4) will eventually become Supplementary

Equation (4). This is reasonable as one would expect an equilibrium state being reached

after infinite amount of time.

( )

redeq

red

lim (red) lim (red)b fk k t

f b btt t

b f b f ox

k e k kP P

k k k k

(6)

Supplementary Note 3│Experimental errors of conductance measurements

23

To examine the reproducibility of the conductance measurements in the conductance

measurements when electrochemical gate voltage was off, the experiments for u-DNA

and Aq-DNA were repeated three times, each consisting of ~4000 curves. The results

were listed in the table below, showing the reproducibility of the conductance

measurements. The experimental error was calculated by: 2

1

1 N

ii

xN

, where N

is the number of sets, xi is the peak position in each individual set of experiment (on a

logarithm scale) and μ is the peak position obtained by compiling all the 3 histograms

(Supplementary Table 3). For conductance measurements under different gate voltages,

we used the fitting errors (Gaussian fitting) in each of the histograms at different gate

voltages as the experimental errors4, as shown in Fig. 3g. The broad distribution (the

width in the Gaussian fit) in the conductance histogram is an inherent property of single

molecule measurement originated from the variation in the molecule-electrode contact

coupling, and dependence of the conductance on the couplings5, 6, rather than an

experimental error.

Supplementary References:

1.  Patra, A. & Richert, C. High Fidelity Base Pairing at the 3'‐Terminus. J. Am. Chem. Soc. 131, 12671‐12681 (2009). 

2.  Arielly, R., Vadai, M., Kardash, D., Noy, G. & Selzer, Y. Real‐Time Detection of Redox Events in Molecular Junctions. J. Am. Chem. Soc. 136, 2674‐2680 (2014). 

3.  Bruot, C., Xiang, L., Palma, J.L. & Tao, N. Effect of Mechanical Stretching on DNA Conductance. ACS Nano 9, 88‐94 (2015). 

4.  Richter, P. Estimating errors in least‐squares fitting. Telecommun. Dat. Acqu. Progres. Rep. 42‐122, 107‐137 (1995). 

5.  Quan, R., Pitler, C.S., Ratner, M.A. & Reuter, M.G. Quantitative Interpretations of Break Junction Conductance Histograms in Molecular Electron Transport. ACS Nano 9, 7704‐7713 (2015). 

24

6.  Malen, J.A. et al. The Nature of Transport Variations in Molecular Heterojunction Electronics. Nano Lett. 9, 3406‐3412 (2009).