Sie sind auf Seite 1von 8

ARTICLES

Effects of hydration on molecular junction transport


DAVID P. LONG1 *, JASON L. LAZORCIK1 , BRENT A. MANTOOTH1 , MARTIN H. MOORE2 , MARK A. RATNER3 , ALESSANDRO TROISI4 , YUXING YAO5 , JACOB W. CISZEK5 , JAMES M. TOUR5 AND RANGANATHAN SHASHIDHAR1
1 2

Research and Development Center, Science Applications International Corporation (SAIC), 9460 Innovation Drive, Manassas, Virginia 20110, USA Center for Bio/Molecular Science and Engineering (Code 6900), Naval Research Laboratory, Washington, District of Columbia 20375, USA 3 Department of Chemistry, Center for Nanofabrication and Molecular Self-Assembly, Northwestern University, Evanston, Illinois 60208, USA 4 Department of Chemistry and Centre of Scientic Computing, University of Warwick, Gibbet Hill Rd, Coventry CV4 7AL, UK 5 Department of Chemistry and Smalley Institute for Nanoscale Science and Technology, Rice University, MS 222, 6100 Main Street, Houston, Texas 77005, USA * e-mail: longdp@saic.com

Published online: 15 October 2006; doi:10.1038/nmat1754

The study of charge transport through increasingly complex small molecules will benet from a detailed understanding of how contaminants from the environment affect molecular conduction. This should provide a clearer picture of the electronic characteristics of molecules by eliminating interference from adsorbed species. Here we use magnetically assembled microsphere junctions incorporating thiol monolayers to provide insight into changing electron transport characteristics resulting from exposure to air. Using this technique, currentvoltage analysis and inelastic electron tunnelling spectroscopy (IETS) demonstrate that the primary interaction affecting molecular conduction is rapid hydration at the goldsulphur contacts. We use IETS to present evidence for changing mechanisms of charge transport as a result of this interaction. The detrimental effects on molecular conduction discussed here are important for understanding electron transport through goldthiol molecular junctions once exposed to atmospheric conditions.

he study of charge transport through single and small assemblies of conductive molecules is now an area of intense research1,2 . Although signicant progress has been made in understanding electron transport through molecular junctions37 , the fabrication of reliable devices continues to be problematic, with many obstacles remaining to simply reproduce results from dierent laboratories or varying test-bed architectures8 . Uncertainty over the self-assembled monolayer (SAM) defect density, electron transport mechanism, nature of the electrode molecule interface and molecular environment have all been suggested as possible explanations for divergent results813 . Another important variable is the inuence of adsorbate penetration (for example, water, oxygen, solvent) into monolayers or test-bed architectures, a topic that has received little attention. Although studies examining chemical reactions on monolayer surfaces have appeared in the literature14,15 , few have focused on molecular adsorption into the monolayer structure or the resulting eects on charge transport1618 . Given the sensitivity of molecular conduction to changing environments, and the continuing use of many test structures that require exposure to either solvent or air, understanding the eects caused by varying chemical exposures could help explain some of the discrepancies found in the literature. We have reported a method for fabricating molecular junctions based on the magnetic-directed assembly of metallized silica microspheres onto SAM-functionalized sourcedrain electrodes19 . This method exploits localized magnetic elds generated in a ferromagnetic array to direct the assembly of susceptible species to the high-eld regions between opposing micrometre-scale contacts. Entrapment of magnetic colloid by SAM-functionalized electrodes results in the parallel fabrication of stable molecular junctions without the use of chemicals or processes that might erode or damage organic monolayers. Although contact asymmetry in our devices caused by variability in the self-assembly process (formation of one molecular junction versus two in series; see below) prevents the quantitative assessment of current
901

nature materials VOL 5 NOVEMBER 2006 www.nature.com/naturematerials

2006 Nature Publishing Group

ARTICLES
a
108

106 107 108 109 1010 1011


0 250 Voltage (mV) 500
Vacuum

30 25
Argon

t-1 argon and H2O

109

H 2O

H2O
Current (nA)

20 15 10 5 0 0 100
Hydrated t-2 argon OPE monothiol

Current (A)

1011

1012

Vacuum Ar/N2 1:1 Ar/O2 Ar/H2O

Current (A)

1010

Ar/N2 1:1 Ar/O2 Ar/H2O OPE dithiol

C-12 monothiol 1013 0 250 Voltage (mV) 500

200 Time (s)

300

400

106 S 107 Vac 108 Current (A) 109 1010 1011 C-9 monothiol 1012 0 250 Voltage (mV) 500 H 2O 1.00 Current (A)

106
S S

106
Vac

106
1.00 Vac

107 108

Vac

107 108
Current (A) 2.07

107 108
Current (A)

2.22

H 2O

109 1010 1011 H2O

109 1010 1011

109 1010 1011 S

H2O

S OPE monothiol

S OPE dithiol

C-9 dithiol
1012

250 Voltage (mV)

500

1012
0 250 Voltage (mV) 500

1012
0 250 Voltage (mV) 500

Figure 1 Effects of hydration on microsphere junction I (V ) characteristics. a, I(V ) analysis of C-12 monothiol and OPE dithiol SAMs under various atmospheres. Data representing vacuum (black), argon/nitrogen (green) and 1:1 argon/oxygen (red) overlay; the introduction of water vapour (blue) causes a decrease in conductance. b, Current as a function of time for OPE monothiol junction during exposure to water vapour (saturated in argon gas) at 23 C/760 torr. The device was exposed to water from time t-1 to t-2 (35 s). Data points were acquired every 100 ms while applying a 100 mV bias. c,d, I(V ) analysis of mono- and dithiol microsphere junctions under anhydrous and hydrated conditions. C-9 alkanethiol (c) and OPE (d) shown for comparison. We have scaled the data, so the curves representing vacuum conditions are equivalent. The magnitude of current change in c and d is normalized to the monothiol systems and is calculated from the average of data points taken every 100 mV along the curves. Molecules are shown without their accompanying thiol protons. Acetyl protecting groups were removed from OPE during monolayer assembly to generate the free thiol. Parts a, c and d are plotted using a logarithmic scale.

density, we have found this method to be ideal for examining molecular conductance switching or other nonlinear transport behaviours independent of junction size20,21 . Microsphere junctions also incorporate an open architecture, which allows the rapid diusion of the surrounding atmosphere into each device. This presented the opportunity to study how molecular conduction is inuenced by chemical interactions initiated when gaseous species diuse between molecules supporting charge transport. Here we demonstrate how microsphere molecular junctions can be used to monitor molecular conduction as atmospheric components interact with a thiol monolayer and provide in situ vibrational information by means of inelastic electron tunnelling spectroscopy (IETS) on the chemical interactions and mechanisms of charge transport inside a device. Taken together, these results provide a better understanding of the varying electronic behaviour in molecular junctions under dierent environmental conditions. Our preliminary evaluation of the microsphere test bed used a variety of prototypical molecular wire candidates under both inert and ambient conditions19 . We observed a marked decrease in molecular conductance when exposing inert microsphere junctions to air. This eect was observed in all alkanethiol and
902

oligo(phenylene-ethynylene) (OPE) SAMs tested and found to be reversible, with junctions returning to their original conducting state after applying vacuum, although repeated cycling caused deterioration of device performance. To identify the atmospheric constituent causing this eect, we monitored the conductance of alkanethiol and OPE junctions while changing the surrounding gaseous environment (Fig. 1a). All major atmospheric components (N2 , O2 , Ar, CO2 and H2 O) were tested by measuring the currentvoltage, I (V ), response of our devices under vacuum then backlling our probe station with the desired gas or gas mixture followed by further I (V ) measurements. No change in conductance was observed in any device until water vapour (in argon) was admitted to the chamber. This demonstrates that the observed eect is due to the interaction of water with our devices and not caused by changing pressure (1 106 to 760 torr) or the presence of oxygen. We also observed a loss in electronic stability in hydrated devices compared with the excellent stability shown under anhydrous conditions. Similar instability has been reported using molecular-resolution scanning tunnelling microscopy of hydrated hydrocarbon monolayers on gold and attributed to rearrangements in the SAM domain structure18,22 . Also, the rate of conductance
nature materials VOL 5 NOVEMBER 2006 www.nature.com/naturematerials

2006 Nature Publishing Group

ARTICLES
a
1,200 Wavenumbers (cm1) 1,400

b
1,600 60 55 50

60 K 2.5 (d2I /dV 2)/(dI /dV ) (V1) 50 K

FWHM (mV)

40 K 2.0

45 40 35

30 K

20 K

C9 s(CH2)
1.5 10 K 140 160 Voltage (mV) Wavenumbers (cm1) 1,400 1,600 180 200

C9 s(CH2)
30 0 10 20 30 Temperature (K) 40 50 60

c
1,200

d
1,800

10
OPE (8a)

50

8
(d2I /dV 2)/(dI /dV ) (V1)

r.m.s. 24 mV 40 20 mV FWHM (mV)

16 mV 12 mV

30

2 8 mV 0
140 160 180 200 Voltage (mV) 220 240

20 OPE (8a) 10
5 10 15 20 a.c. modulation (r.m.s. value) (mV) 25

Figure 2 Temperature and a.c. modulation dependence of microsphere junction IETS. a, Successive IETS scans of a C-9 alkanethiol s (CH2 ) vibrational mode at constant 6 mV modulation amplitude obtained under increasing temperature (1060 K). b, IETS linewidth (FWHM) as a function of temperature. Linear regression was used to determine the slope of this line (reported as the average of high and low slope values taking into consideration error bars) to be 0.40. The inset shows a scanning electron micrograph of a microsphere junction, scale bar = 1 m. c, Successive IETS scans at a xed temperature of 4.2 K showing the OPE (8a) vibrational mode acquired under increasing modulation amplitude (824 mV). d, OPE IETS linewidth as a function of increasing modulation amplitude. We used linear regression to determine the slope of this line to be 1.5. All IET spectra shown are plotted as the average of the forward and reverse traces. All peaks were tted using a gaussian distribution function to obtain FWHM values and estimated error (standard deviation values) for each data point.

change in our devices was determined by monitoring current as a function of time while exposing inert junctions to argon containing water vapour (Fig. 1b). Approximately one order of magnitude decrease in current is observed in 2 s on introduction of water, the magnitude of which matches results when similar devices are exposed to air. Junctions incorporating alkanethiols were found to react with water vapour on average four times faster than OPE and consistently showed a greater decrease in conductance (see Fig. 1c,d). The higher reactivity found in alkyl junctions is consistent with stronger interactions between polar water molecules and the negatively charged sulphur atom of the thiolate and is caused by increased electron-donor ability of the sulphur induced by the alkyl versus the aromatic moieties. Aryl systems such as OPE can delocalize charge across their ring systems, reducing charge on the sulphur atom.
nature materials VOL 5 NOVEMBER 2006 www.nature.com/naturematerials

To test whether water was indeed interacting at the sulphur atom in our devices, we studied a series of alkane and OPE monolayers and examined them under anhydrous and hydrated conditions. Figure 1c compares the I (V ) characteristics of C-9 mono- and dithiol monolayers before and after hydration. A distinct 2:1 ratio is observed when comparing the magnitude of current before and after hydration, with dithiol-based junctions showing twice the eect of the monothiol. Figure 1d shows a similar result, in which mono- and dithiol OPE junctions were studied for their corresponding response, again indicating that similar interactions with water are occurring in both alkyl and aryl SAMs. These experiments demonstrate that the eects of hydration are intimately associated with the number of goldthiol contacts in our devices, indicating that the interaction of water is occurring at the thiolate moiety. A mechanism involving only
903

2006 Nature Publishing Group

ARTICLES
a
2.5 S 2.0
2.0 0 (cm1) 500 1,000 1,500

Wavenumbers (cm1) 2,000 2,500 3,000 S

Wavenumbers (cm1) 2,000 2,500 3,000

c
2.5

Wavenumbers (cm1) 2,000 2,500 3,000

3.0

(CH)

2.0

(CH)

(d2I /dV 2)/(dI /dV ) (V 1)

(d2I /dV 2)/(dI /dV ) (V 1)

(d2I /dV 2)/(dI /dV ) (V 1)

IETS

1.5

1.0

2.0

(SH) 2,412 cm1

1.5 (SD) 1,936 cm1

1.0

100 (mV)

200

1.0

1.0

0.5 0.5 (CH) 0 200


0

0 400
200 250 300 350 Bias voltage (mV) 400 200 250 300 350 Bias voltage (mV) 400

250 300 350 Bias voltage (mV)

Figure 3 IETS analysis of anhydrous and hydrated alkanethiol microsphere junctions. All spectra show the SH to CH stretching region (200400 mV) after normalizing the d2 I/dV 2 signal by the differential conductance. Spectra are plotted from experimental data without the use of smoothing techniques and shown as the average of the forward and reverse traces. a, IETS showing the empty SH region in the anhydrous C-9 control junction; the inset shows anhydrous C-9 low-frequency modes. AuS shoulder mode (29 mV) on zero-bias feature not shown. b, IETS of a hydrated C-9 microsphere junction (30 min H2 O/argon, 760 torr/23 C) showing prominent red-shifted SH stretching vibration. c, SD stretching vibration after device exposure to D2 O (30 min D2 O/argon, 760 torr/23 C), showing expected isotopic shift.

simple adsorption of water into the SAM structure without specic binding sites would be dependent on the free volume inside the monolayer and not expected to produce strict 2:1 ratios in the molecules examined here16 . Also, the parallel shape of the I (V ) data presented in (Fig. 1a,c,d) strongly indicates that the added water changes the spectral density (mixing with the electrodes) term in the conductance. This is consistent with the analysis presented here and in agreement with modication of the interfacial structure. As thiols share many chemical properties with alcohols, strong interaction between water and the sulphur atoms is expected. Calculations have shown how the proton-acceptor capability of sulphur, despite having a lower electronegativity, is similar to that of oxygen23,24 . This eect is due to the extended nature of sulphurs lone pair of electrons allowing strong interaction with protons. The dierence in polarizability between the terminal sulphur groups and the hydrophobic molecules used in this study restricted dipole interactions to predominantly the electrodemolecule interface, permitting a study of hydration of the goldsulphur contacts. IETS has recently been used to study the vibrational spectra of molecules inside various test-bed architectures2528 . We used IETS to monitor changes in the vibrational spectra of alkanethiol and OPE microsphere junctions after exposure to dierent environmental conditions to verify independently the association of water at the thiolate moiety in our devices. Our method to acquire IETS used microsphere junctions with standard a.c. modulation techniques and dual lock-in ampliers, similar to the method reported in the literature26 . Anhydrous alkanethiol and OPE monolayers were found to yield spectra similar to those previously reported using crossed-wire tunnel junctions26 , and predicted using Greens function derivatives and density functional theory2932 . However, our spectra did not show bias oset or overlapping double spectra as expected for two molecular junctions in series originally predicted for the dual microsphereelectrode contacts19 . These eects would result from a perceived dierence in voltage drop in each molecular junction because they lie in series.
904

To probe this question further and to verify that the microsphere test bed produced valid spectra, we carried out control experiments to measure peak broadening as a function of increasing temperature and modulation amplitude10,33 . Thermal broadening, being a fundamental property of inelastic electron tunnelling, is a particularly sensitive test to verify the origin of the second-harmonic signal recorded during IETS34 . We acquired IET spectra from C-9 alkanethiol microsphere junctions from 10 to 60 K using a constant modulation voltage of 6 mV (Fig. 2a). We used a gaussian function to t the experimental data and determine the full-width at half-maximum (FWHM) of each peak (Fig. 2b). We then used linear regression to determine the slope of the resulting line, which we compared with the theoretical value calculated from nite-temperature (4.2 K) and voltage2 modulation (6 mV) eects according to W 2 theoretical = Wmodulation + 2 Wthermal (where W = FWHM; calculations and plots shown in the Supplementary Information). We found our variable-temperature control experiment to be in agreement with the thermal broadening predicted for inelastic electron tunnelling processes on the basis of the correlation of our experimental rate of FWHM broadening with that calculated (b = 0.40experimental versus 0.375theoretical ). Similarly, we also veried our OPE spectra using variable a.c. modulation at a xed temperature of 4.2 K. Figure 2c shows the modulation dependence of the OPE (8a) vibrational mode from 8 to 24 mV and Fig. 2d shows FWHM as a function of modulation voltage. The slope for this line was again in agreement with the theoretical value (b = 1.5experimental versus 1.695theoretical ). IETS linewidths from microsphere junctions were found to be approximately twice the width predicted by theory. We attribute this to the force applied to the SAM during deposition of the microsphere, which could cause loss in molecular orientation in the monolayer, appearance of further vibrational modes and peak broadening35 . Microsphere junction linewidths for C-9 (CH) and OPE (8a) vibrational modes, acquired at 4.2 K and 8.0 mV modulation amplitude, yielded IETS peaks with FWHM of 2122 mV, placing
nature materials VOL 5 NOVEMBER 2006 www.nature.com/naturematerials

2006 Nature Publishing Group

ARTICLES
this technique similar in resolution to the crossed-wire test bed but yielding peaks slightly broader than those generated using the nanopore system under similar conditions10,26,33 . Measurement of the predicted IETS broadening conrms that our spectra originate from inelastic electron tunnelling processes in microsphere junctions, thus validating the spectra as genuine. However, the correlation of our spectra with previous IETS data from techniques known to form exclusively single junctions and the complete lack of dual or oset spectra, which should result from two molecular junctions in series, indicates that our devices also incorporate single molecular junctions rather than two as previously indicated19 . We again attribute this to asymmetric deposition of the microsphere and formation of metalmetal shorts on one side of our devices, caused by the sphere always striking one electrode before the other during self-assembly. This process would disperse the energy of the depositing sphere (accelerated by magnetic attraction) into predominantly one of the monolayers, causing a short, whereas on the other electrode the sphere can settle into position more softly to form a viable molecular junction. Fortuitously, asymmetric shorting in our devices yields simplied IET spectra, which correlate well with recently published data and permit the analysis of hydration in our devices. IET spectra of C-9 dithiol SAMs after exposure to dierent environmental conditions are shown in Fig. 3. All vibrational modes discussed are summarized in Table 1. IETS analysis of hydrated devices was carried out by exposing junctions to H2 O vapour in argon for 30 min, then applying vacuum with simultaneous cooling to cryogenic temperatures with liquid helium to retain bound water molecules for analysis. Anhydrous devices were formed through prolonged exposure to high-vacuum conditions (106 torr for 2 days) before IETS. Anhydrous spectra show an intense peak centred at 362 mV (2,920 cm1 ), originating from CH stretching and observed at identical frequencies in all alkane IET spectra. The absence of vibrational modes indicative of free thiol (SH, 320 mV) indicates that the molecule has assembled on the surface as the thiolate and no longer incorporates a proton. Figure 3b shows IETS of a similar C-9 junction after hydration. A new peak appears, centred at 299 mV (2,412 cm1 ), which we attribute to an induced SH stretching vibration from associated water at the sulphur atom. Free alkanethiols commonly yield SH stretching vibrations between 2,550 and 2,600 cm1 in infrared spectroscopy, well above the location of our observed peak. However, thiols undergoing hydrogen-bonding interactions can shift their stretching vibration to signicantly lower frequencies due to weakening of the SH bond when the proton interacts with another electronegative species such as oxygen or nitrogen24,36 . Further evidence for this interaction was obtained by exposing our devices to D2 O vapour followed by IETS analysis (Fig. 3c). As expected, the spectrum does not contain the previously observed SH mode but includes a new peak, centred at 240 mV (1,936 cm1 ), which we attribute to the SD stretching vibration from interacting D2 O. This peak is found to redshift 476 cm1 relative to the SH vibration discussed in Fig. 3b, in accordance with the new isotopic ratio24 . These results establish that the observed peaks in the SH region of our hydrated IET spectra are due to interactions at the sulphur atom by water and not residual thiol remaining in the junctions after magnetic assembly. To verify that -conjugated OPE monolayers undergo similar interactions with water, given their analogous I (V ) response, we also carried out IETS analysis on OPE dithiol microsphere junctions under anhydrous and hydrated environmental conditions. Anhydrous OPE microsphere junctions were found to yield prominent IETS vibrational modes similar to those previously reported26,29,30 . Figure 4a shows a spectrum of an anhydrous OPE dithiol monolayer from 250 to 340 mV showing an intense peak
nature materials VOL 5 NOVEMBER 2006 www.nature.com/naturematerials

Table 1 Summary of experimentally obtained IETS vibrational modes for monolayers investigated. (Note: Mode assignments made on the basis of previous IET, infrared, Raman and high-resolution electron energy-loss spectroscopic methods 2931,37,4345 .) Molecule C9 and C12 mV 362 299 240 175 140 118 81 29 314 274 197 135 101 73 50 30 Peak position cm1 2,920 2,412 1,936 1,412 1,129 952 653 234 2,533 2,210 1,589 1,089 815 589 400 240 Vibrational mode (CH) (SH) (SD) w (CH2 ), s (CH2 ) (CC) r (CH2 ) (CS) (AuS) (SH) (CC) (8a), (19a) (18a), (18b), (9a) aryl CH oop bend CCC ip bend CCC ip bend (AuS)

OPE

(Anhydrous) (Hydrated) (Anhydrous) (Anhydrous)

Peak visible as a shoulder on zero-bias feature, location approximate. oop = out of plane. ip = in plane

at 274 mV (2,210 cm1 ) known to arise from CC stretching vibrations. Again, the absence of peaks associated with free thiol indicates that the molecule has assembled on the surface as the thiolate and does not incorporate a bound proton. IETS of hydrated OPE (Fig. 4b) shows the triple-bond mode with a further peak centred at 314 mV (2,533 cm1 ), which we attribute to stretching vibrations for an SH bond from associated water at the sulphur atom. This new mode is again observed at lower frequencies than predicted for the OPE free thiol (2,596 cm1 ), indicating a weakened SH vibration and sharing of the proton between electronegative atoms. The magnitude of redshift is found to be smaller than those observed in alkane junctions, possibly due to resonance stabilization with the aromatic ring system. To eliminate the possibility that the new mode observed at 314 mV was caused by the use of a deprotecting agent during SAM formation (30% ammonium hydroxide in H2 O), we carried out a control experiment where SAMs of C-12 alkanethiol were deposited in the presence of 2 l of aqueous ammonium hydroxide. IETS of these devices under anhydrous conditions shows no evidence of sulphur hydrogen vibrations. This result supports the conclusion that the observed OPE mode at 314 mV is due to the in situ hydration of our molecular junctions immediately before IETS analysis and not the result of the deprotecting agent. Figure 4c shows the complete vibrational spectrum of another hydrated OPE junction, showing all prominent modes for this molecule. The low-frequency peak marked with an asterisk is observed only in hydrated OPE junctions and is discussed in the next section. The low-energy regions in our OPE spectra contained vibrational information, which could be used to follow the change in the electron transport mechanism initiated by the diusion of water into our devices. Figure 5a compares experimentally obtained anhydrous OPE IET spectra with a spectrum calculated using the method described in ref. 29. The theoretical spectrum in Fig. 5a was computed assuming a chemical bond between the terminal sulphur and the gold surface. An excellent correlation is found between our experimental results and the calculated spectrum, which is composed primarily of OPE ring modes centred at 1,089 cm1
905

2006 Nature Publishing Group

ARTICLES
a
5.0 S 4.5 S Wavenumbers (cm1) 2,250 2,500 2,750

b
3.5

Wavenumbers (cm1) 2,250 2,500

c
2,750 2.5 0

Wavenumbers (cm1) 1,200 2,400

ZBF 3.0 (C C)
2.0

(C C) (8a) (18a)

(d2I /dV 2)/(dI /dV ) (V 1)

(d2I /dV 2)/(dI /dV ) (V 1)

4.0

2.5

(SH) 2,530 cm1

(d2I /dV 2)/(dI /dV ) (V 1)

(C C)

1.5

3.5

2.0

1.0

3.0

1.5
0.5

2.5 250

275 300 Bias voltage (mV)

325

1.0 250

275

300 325 Bias voltage (mV)

350

100 200 Bias voltage (mV)

300

Figure 4 IETS analysis of anhydrous and hydrated OPE microsphere junctions. All spectra are normalized by the differential conductance. Spectra are plotted from experimental data without the use of smoothing techniques and are shown as the average of the forward and reverse traces. a, IETS showing the empty SH region (2,600 cm1 ) of an OPE microsphere junction maintained under anhydrous environmental conditions. b, IETS showing prominent SH stretching vibration after device exposure to H2 O (30 min H2 O/argon, 760 torr/23 C). c, Full IET vibrational spectrum of a hydrated OPE dithiol microsphere junction. Zero-bias feature shown off scale. The peak marked with an asterisk is expanded in Fig. 5a.

(18a, 18b and 9a)37 and low-frequency electrodesulphur interactions, appearing as shoulders on the zero-bias feature38 . The spectrum also contains a peak at 101 mV (815 cm1 ), which is clearly composed of two distinct vibrations. These separate vibrations match the location of calculated modes (the only nontotally symmetric modes identiable in this region of the spectrum and marked with arrows in Fig. 5a) known to be active when the electrode is strongly coupled with the -orbital of the OPE sulphur atom39 . Appearance of these vibrations supports the expectation that anhydrous OPE microsphere junctions transport electrons through strong goldsulphur coupling at the electrode surface. IET spectra can be rationalized if we consider tunnelling through the molecular wire as a superposition of many possible tunnelling paths. A tunnelling path is a series of adjacent atomic orbitals that mediate ecient electronic coupling between two electrodes, and a given tunnelling path is dominant only when the couplings of its initial and nal orbitals with the electrodes are both important. We showed recently that dierent tunnelling paths are associated with dierent IET signals39 ; in particular, there are dierent IET peaks associated with direct coupling of the electrode with - and -orbitals of the molecule. In the case of a hydrated OPE interface, we expect the coupling with the -orbitals of the molecule to be much weaker (owing to absence of the SAu bond), so that the most important tunnelling path involves coupling of the molecular -orbitals with the electrode. The -tunnelling path is also favoured because the molecular axis can form a larger angle with the normal to the surface, exposing the molecular -orbitals to a more favourable interaction with the electrode. In Fig. 5b, we compare the hydrated OPE IET experimental data with the computed spectrum obtained by suppressing the electronic coupling between the molecular -orbital and the electrode on one side of the molecule. Once again, the hydrated OPE spectrum includes a large peak at 1,089 cm1 arising from several ring modes, but the prominent CH bending vibration at 815 cm1 , associated with the electronic coupling between the electrode and the sulphur -orbitals, is absent. There
906

is also a clear absence of the low-frequency shoulder modes on the zero-bias feature, attributed to goldsulphur interactions and consistent with weakening of this bond due to water associated with the thiolate moiety. Replacing these absent modes is a new peak, centred at 73 mV (589 cm1 ), which matches the position for a calculated vibration at 590 cm1 (see arrow, Fig. 5b) associated with direct coupling of the electrode and the OPE -orbitals. We obtained a better correlation between our experimentally obtained IET spectra of hydrated OPE and calculated spectra if -orbital coupling was suppressed on only one side of the junction. We attribute this to contact asymmetry in our devices brought about by weaker coupling of the molecule to the microsphere surface relative to the self-assembled goldsulphur bond on the electrode, indicating that the microspheremolecule AuS contact is more susceptible to hydration. Observation of these dierent vibrational modes signals a change in transport mechanism from electron injection into the sulphurs -orbital in anhydrous junctions to direct coupling of the electrode and the molecules -orbitals, bypassing the degraded thiolate bond, in hydrated junctions. This proposed mechanism of charge transport through hydrated OPE molecules is consistent with the measured increase in junction resistance. In addition, thiol-based OPE monolayers on gold are known to undergo AuS-aryl hybridization changes owing to resonance stabilization from the conjugated molecules40 . This can initiate changes in molecular tilt relative to the surface and one proposed mechanism for stochastic switching observed in scanning tunnelling microscopy. The results presented here indicate that hydration at the thiolate moiety can induce similar eects. Recent studies using organometallic complexes that incorporate a metalsulphur bond have shown how the presence of even a single hydrogen bond interacting with the coordinated thiolate can drastically alter the sulphurs chemistry to the metal centre by decreasing its reactivity towards electrophiles by two orders of magnitude41 . Similarly, our results using the microsphere test bed indicate that, on exposure to air, molecules of water rapidly associate with the electronegative thiolate moiety, disrupting the
nature materials VOL 5 NOVEMBER 2006 www.nature.com/naturematerials

2006 Nature Publishing Group

ARTICLES
a
3.0
OPE (anhydrous) Computed spectrum

(d2I /dV 2)/(dI /dV ) (V 1)

Symmetrical mode

These results have important implications for any research into molecular electronics or SAMs using goldthiol contacts that are unable to work entirely under anhydrous conditions, and could help explain some of the discrepancies found in the literature concerning molecular conduction.

*
2.0

METHODS
FORMATION OF SELF-ASSEMBLED MONOLAYERS ON MAGNETIC ARRAYS

*
1.0

*
200

*
400

* * * *
1,000 1,200 600 800 Wavenumbers (cm1)

b
OPE (hydrated)

2.0 (d2I /dV 2)/(dI /dV ) (V 1)

Computed spectrum

Symmetrical mode

*
1.0

We cleaned magnetic arrays composed of evaporated gold-coated nickel and incorporating 0.5 m spacing between the sourcedrain electrodes by immersing them for 30 min in 30% hydrogen peroxide solution, followed by sonication in anhydrous ethanol for 20 min and nal argon plasma cleaning for 5 min (Plasma Prep II SPI Supplies). We immediately transferred the substrates to a Vacuum Atmospheres glove box maintained below 0.2 p.p.m. contamination levels of oxygen and water. We deposited alkane monolayers from anhydrous tetrahydrofuran (inhibitor free, 5 ml) using 2 mM solutions for 18 h, followed by copious rinsing with clean tetrahydrofuran and drying in the glove-box atmosphere. We deposited OPE SAMs from 0.5 mM anhydrous tetrahydrofuran solutions from the acetyl-protected molecule and converted them to the free thiol using 2.0 l of 30% NH4 OH(aq) solution immediately before self-assembly. We carried out OPE deposition for 18 h, followed by standard rinsing and drying procedures.
MICROSPHERE JUNCTION FABRICATION AND MEASUREMENTS

* *
200

*
400

* * * *
1,000 1,200 600 800 Wavenumbers (cm1)

Figure 5 Comparison of low-energy region in anhydrous and hydrated OPE IET spectra. a, Comparison of experimentally obtained IET spectra from an anhydrous OPE microsphere junction with calculated vibrational modes on the basis of efcient electron injection into the -orbital of the OPE sulphur atom. b, Comparison of experimentally obtained IET spectra of hydrated OPE with predicted modes when only direct electron injection into the -orbitals of the molecule is allowed on one of the moleculemetal interfaces. Experimental IET spectra are shown in blue and plotted as the average of the forward and reverse traces. All calculated vibrational modes are shown in red29 . Symmetrical modes are shown as red lines with asterisks, whereas the unlabelled red lines correspond to non-totally symmetrical modes. Vibrational modes are located corresponding to the calculated frequencies with heights proportional to computed IETS intensity.

The fabrication of metallized silica colloid dispersed in anhydrous ethanol has been described previously19 . We further diluted this solution by dispersing 0.3 ml of stock solution into 100 ml of anhydrous ethanol (200 proof), followed by sonication for 30 min. We used approximately 10 ml of this diluted solution for each deposition. We carried out magnetic assembly in a test tube by immersing the 1 1 cm2 SAM-functionalized magnetic array in the colloid solution for 30 min while it was placed in an external 225 G magnetic eld oriented parallel to the long axis of the features in the array. After assembly, we dried devices under nitrogen and transferred them to a cryogenic vacuum probe station tted with an Agilent 4155B parametric analyser under computer control for electrical (I /V and IETS) analysis.
IETS

goldsulphur bond to the electrode. The weakening (or complete loss) of this crucial bond, as shown by the lack of AuS modes in IETS, initiates a cascade of detrimental eects on charge transport through our devices. We attribute the loss of ecient charge injection into the sulphur atom of hydrated OPE junctions to this association, which forces electrons to nd an alternative pathway through the device to bypass the degraded goldthiol contacts. The prominent peak in IETS appearing at 73 mV (589 cm1 ) indicates that the dominant route for electron transport through hydrated OPE junctions becomes direct injection of electrons into the molecules -orbitals. Detachment from the gold surface is consistent with the decrease in observed current and loss of electronic stability in hydrated devices. Although the association of water seems to be reversible, it may yet be found that this interaction is but the rst step in the oxidation process known to occur in SAMs on gold left exposed to ambient conditions42 .
nature materials VOL 5 NOVEMBER 2006 www.nature.com/naturematerials

We acquired IETS in microsphere molecular junctions using standard a.c. modulation techniques and dual lock-in ampliers to simultaneously record the currentvoltage characteristics and the rst and second harmonics at 4.2 K (proportional to dI /dV and d2 I /dV 2 , respectively) as previously described26 . We optimized IETS parameters for individual junctions to accommodate changing current levels and device quality under dierent environmental conditions. Typical instrument settings were as follows: temperature, 4.2 K; lock-in time constant, 1 s; a.c. modulation voltage, 410 mV; step size, 12 mV. Before each data point was acquired, a delay of 24 s was applied to allow the lock-in amplier to stabilize. Each data point is the average of 1,0002,000 samples with a delay of 1 ms between each. Analysis of IET spectra initially made use of a bias/2 scale for the energy axis because of the expected two molecular junctions in series, but failed to yield peaks that corresponded to known vibrational frequencies. Using a full bias scale, we found measured peaks to occur at the expected frequencies for both alkanethiol and OPE junctions, and these correlated well with previously published vibrational spectra and theoretical calculations26,29,30 .

Received 24 March 2006; accepted 1 September 2006; published 15 October 2006. References
1. Tour, J. M. Molecular Electronics: Commercial Insights, Chemistry, Devices, Architectures and Programming (World Scientic, River Edge, New Jersey, 2003). 2. Carroll, R. L. & Gorman, C. B. The genesis of molecular electronics. Angew. Chem. Int. Edn 41, 43784400 (2002). 3. Joachim, C. & Ratner, M. A. Molecular electronics: Some views on transport junctions and beyond. Proc. Natl Acad. Sci. 102, 88018808 (2005). 4. Rakshit, T., Liang, G. C., Ghosh, A. W. & Datta, S. Silicon-based molecular electronics. Nano Lett. 4, 18031807 (2004). 5. Seminario, J. M., Cordova, L. E. & Derosa, P. A. An ab initio approach to the calculation of current-voltage characteristics of programmable molecular devices. Proc. IEEE 91, 19581975 (2003). 6. Seminario, J. M. & Yan, L. Ab initio analysis of electron currents in thioalkanes. Int. J. Quantum Chem. 102, 711723 (2005). 7. Seminario, J. M., Cruz, C. D. L., Derosa, P. A. & Yan, L. Nanometer-size conducting and insulating molecular devices. J. Phys. Chem. B 108, 1787917885 (2004).

907

2006 Nature Publishing Group

ARTICLES
8. Salomon, A. et al. Comparison of electronic transport measurements on organic molecules. Adv. Mater. 15, 18811890 (2003). 9. Selzer, Y. et al. Eect of local environment on molecular conduction: Isolated molecule versus self-assembled monolayer. Nano Lett. 5, 6165 (2005). 10. Wang, W., Lee, T. & Reed, M. A. Electronic transport in molecular self-assembled monolayer devices. Proc. IEEE 93, 18151824 (2005). 11. Moore, A. M. et al. Molecular engineering and measurements to test hypothesized mechanisms in single molecule conductance switching. J. Am. Chem. Soc. 128, 19591967 (2006). 12. Selzer, Y., Salomom, A. & Cahen, D. The importance of chemical bonding to the contact for tunneling through alkyl chains. J. Phys. Chem. B 106, 1043210439 (2002). 13. Basch, H., Cohen, R. & Ratner, M. A. Interface geometry and molecular junction conductance: Geometric uctuation and stochastic switching. Nano Lett. 5, 16681675 (2005). 14. Chechik, V., Crooks, R. M. & Stirling, C. J. M. Reactions and reactivity in self-assembled monolayers. Adv. Mater. 12, 11611171 (2000). 15. Ulman, A. Formation and structure of self-assembled monolayers. Chem. Rev. 96, 15331554 (1996). 16. Karpovich, D. S. & Blanchard, G. J. Vapor adsorption onto metal and modied interfaces: Evidence for adsorbate penetration of an alkanethiol monolayer on gold. Langmuir 13, 40314037 (1997). 17. Anderson, M. R., Evaniak, M. N. & Zhang, M. Inuence of solvent on the interfacial structure of self-assembled alkanethiol monolayers. Langmuir 12, 23272331 (1996). 18. Pi, U. H. et al. Current ow through dierent phases of dodecanethiol self-assembled monolayer. Surf. Sci. 583, 8893 (2005). 19. Long, D. P. et al. Magnetic directed assembly of molecular junctions. Appl. Phys. Lett. 86, 153105153107 (2005). 20. Blum, A. S. et al. Molecularly inherent voltage-controlled conductance switching. Nature Mater. 4, 167172 (2004). 21. Kushmerick, J. G., Blum, A. S. & Long, D. P. Metrology for molecular electronics. Anal. Chim. Acta 568, 2027 (2006). 22. Poirier, G. E., Pylant, E. D. & White, J. M. Crystalline structures of pristine and hydrated mercaptohexanol self-assembled monolayers on Au(111). J. Chem. Phys. 104, 73257328 (1996). 23. Rablen, P. R., Lockman, J. W. & Jorgensen, W. L. Ab initio study of hydrogen-bonded complexes of small organic molecules with water. J. Phys. Chem. A 102, 37823797 (1998). 24. Steinwender, E. & Mikenda, W. Hydrogen bonding in 2-hydroxybenzoyl and 2-hydroxythiolbenzoyl compounds: Spectroscopic characterization and spectroscopic bond strength sequences. Monatsh. Chem. 125, 695705 (1994). 25. Cai, L. et al. Reversible bistable switching in nanoscale thiol-substituted oligoaniline molecular junctions. Nano Lett. 5, 23652372 (2005). 26. Kushmerick, J. G. et al. Vibronic contributions to charge transport across molecular junctions. Nano Lett. 4, 639642 (2004). 27. Wang, W., Lee, T., Kretzschmar, I. & Reed, M. A. Inelastic electron tunneling spectroscopy of an alkanedithiol self-assembled monolayer. Nano Lett. 4, 643646 (2004). 28. Ho, W. Single molecule chemistry. J. Chem. Phys. 117, 1103311061 (2002). 29. Troisi, A. & Ratner, M. A. Modeling the inelastic tunneling spectra of molecular junctions. Phys. Rev. B 72, 033408033411 (2005). 30. Paulsson, M., Frederiksen, T. & Brandbyge, M. Inelastic transport through molecules: Comparing rst-principles calculations to experiments. Nano Lett. 6, 258262 (2006). 31. Jiang, J., Kula, M., Lu, W. & Luo, Y. First-principles simulations of inelastic electron tunneling spectroscopy of molecular electronic devices. Nano Lett. 5, 15511555 (2005). 32. Seminario, J. M. & Cordova, L. E. Theoretical interpretation of intrinsic line widths observed in inelastic electron tunneling scattering experiments. J. Phys. Chem. A 108, 51425144 (2004). 33. Lee, T., Wang, W. & Reed, M. A. Intrinsic electronic transport through alkanedithiol self-assembled monolayer. Japan. J. Appl. Phys. 44, 523529 (2005). 34. Lambe, J. & Jaklevic, R. C. Molecular vibration spectra by inelastic electron tunneling. Phys. Rev. 165, 821832 (1968). 35. Solomon, G. C. et al. Understanding the inelastic electron-tunneling spectra of alkanedithiols on gold. J. Chem. Phys. 124, 094704 (2006). 36. Krepps, M. K., Parkin, S. & Atwood, D. A. Hydrogen bonding with sulfur. Cryst. Growth Design 1, 291297 (2001). 37. Varsanyi, G. Assignments for Vibrational Spectra of Seven Hundred Benzene Derivatives (Wiley, New York, 1974). 38. Agnolet, G., Savitski, S. R. & Zimmerman, D. T. Zero bias features in self-assembling tunnel junctions. Physica B 284288, 18401841 (2000). 39. Troisi, A. & Ratner, M. A. Molecular transport junctions: Propensity rules for inelastic electron tunneling spectra. Nano Lett. 6, 17841788 (2006). 40. Donhauser, Z. J. et al. Conductance switching in single molecules through conformational changes. Science 292, 23032307 (2001). 41. Smith, J. N., Homan, J. T., Shirin, Z. & Carrano, C. J. H-bonding interactions and control of thiolate nucleophilicity and specicity in model complexes of zinc metalloproteins. Inorg. Chem. 44, 20122017 (2005). 42. Willey, T. M. et al. Rapid degradation of alkanethiol-based self-assembled monolayers on gold in ambient laboratory conditions. Surf. Sci. 576, 188196 (2005). 43. Stapleton, J. J. et al. Self-assembled oligo(phenylene-ethynylene) molecular electronic switch monolayers on gold: Structures and chemical stability. Langmuir 19, 82458255 (2003). 44. Kudelski, A. Characterization of thiolate-based mono- and bilayers by vibrational spectroscopy: A review. Vib. Spectrosc. 39, 200213 (2005). 45. Richter, L. J. et al. Optical characterization of oligo(phenylene-ethynylene) self-assembled monolayers on gold. J. Phys. Chem. B 108, 1254712559 (2004).

Acknowledgements
The authors are grateful to J. Kushmerick of the National Institute of Standards and Technology for helpful comments and density functional theory calculations and G. Malliaras of Cornell University for the fabrication of magnetic arrays. This work was supported by the Defense Advanced Research Projects Agency (DARPA), the Oce of Naval Research (ONR), the Air Force Oce of Scientic Research (AFOSR) and the National Science Foundation (NSF). Correspondence and requests for materials should be addressed to D.P.L. Supplementary Information accompanies this paper on www.nature.com/naturematerials.

Author contributions
D.P.L. is the inventor of the microsphere test bed and performed all surface preparations, device fabrications, environmental I (V ) analysis, variable-temperature IETS and project planning. J.L.L. and B.A.M. performed environmental IETS. M.H.M., Y.Y., J.W.C. and J.M.T. performed organic synthesis of OPE molecules. M.A.R. and A.T. conducted theoretical modelling of IETS spectra. R.S. supplied funding support.

Competing nancial interests


The authors declare that they have no competing nancial interests.
Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

908

nature materials VOL 5 NOVEMBER 2006 www.nature.com/naturematerials

2006 Nature Publishing Group

Das könnte Ihnen auch gefallen