Electrical contact

Abstract

The present invention relates to an electrical contact. In particular, it relates to an electrical contact capable of establishing an electrical contact with a soft material. More particular, the electrical contact comprises (a) a non-Newtonian liquid metal alloy, the non-Newtonian liquid metal alloy is formed in a polymer insulator, wherein the contact surface of the electrical contact that contacts the soft material is a smooth flat non-patterned surface, the surface comprising the non-Newtonian liquid metal alloy sandwiched between the polymer insulator. The microfluidic device comprising the electrical contact and a method for forming the electrical contact are also disclosed.

Claims

1. An electrical contact comprising: (a) a top electrode comprising a non-Newtonian liquid metal alloy; and (b) a bottom electrode comprising a self-assembled monolayer of molecules (SAM) formed on a metal substrate, wherein the surface of the SAM layer of the bottom electrode contacting the top electrode is a template-stripped non-patterned surface and the electrical contact has no edge effect; and the surface of the liquid metal alloy contacting the SAM layer is contained in a polymer insulator and the area of the electrical contact between the liquid metal alloy surface and the SAM layer is determined by modulating the diameter of the liquid metal alloy surface contacting the SAM layer, the diameter being between 15 m and 55 m.

2. The electrical contact according to claim 1, wherein the non-Newtonian liquid metal alloy is EGaIn.

3. The electrical contact according to claim 2, wherein the EGaIn contains 75.5 wt % Ga and 24.5 wt % In.

4. The electrical contact according to claim 1, wherein the polymer insulator is PDMS.

5. The electrical contact according to claim 4, wherein the PDMS is transparent.

6. The electrical contact according to claim 1, wherein the metal of the metal substrate is selected from the group consisting of silver, copper, nickel, platinum, palladium, and gold.

7. The electrical contact according to claim 1, wherein the SAM layer is formed from a material of 99.9% purity.

8. The electrical contact according to claim 1, wherein the SAM layer has a thickness of 1 nm to 2 nm.

9. The electrical contact according to claim 1, wherein the SAM layer is formed of molecules of formula S(CH.sub.2).sub.n1CH.sub.3, n being 10, 12, 14, 16, or 18.

10. A method for forming the electrical contact of claim 1, the method comprising: (a) providing the top electrode comprising the non-Newtonian liquid metal alloy; (b) preparing the bottom electrode by forming the self-assembled monolayer of molecules (SAM) on a metal substrate; and (c) contacting the liquid metal alloy of the top electrode with the surface of the SAM layer of the bottom electrode.

11. The method according to claim 10, wherein the non-Newtonian liquid metal alloy is EGaIn.

12. The method according to claim 11, wherein the EGaIn contains 75.5 wt % Ga and 24.5 wt % In.

13. The method according to claim 10, wherein the polymer insulator is PDMS.

14. The method according to claim 13, wherein the PDMS is transparent.

15. An electrical contact comprising: a top electrode comprising a non-Newtonian liquid metal alloy; and a bottom electrode comprising a self-assembled monolayer of molecules (SAM) formed on a metal substrate, wherein the surface of the SAM layer of the bottom electrode contacting the top electrode is a template-stripped non-patterned surface and the electrical contact has no edge effect; and the surface of the liquid metal alloy contacting the SAM layer is contained in a polymer insulator and the area of the electrical contact between the liquid metal alloy surface and the SAM layer is determined by modulating the diameter of the liquid metal alloy surface contacting the SAM layer, the diameter being between 15 m and 55 m; and the top electrode is prepared by stabilizing the non-Newtonian liquid metal alloy in a microfluidic device.

Description

BRIEF DESCRIPTION OF THE FIGURES

(1) In order that the present invention may be fully understood and readily put into practical effect, there shall now be described by way of non-limitative examples only preferred embodiments of the present invention, the description being with reference to the accompanying illustrative figures.

(2) In the Figures:

(3) FIG. 1 is a schematic diagram showing a comparison of the molecular junction according to an embodiment of the present invention with other molecular junctions;

(4) FIG. 2 is a schematic diagram showing the definition of the accuracy and precision of the electrical measurements for SAM-based junctions;

(5) FIG. 3 is a schematic diagram showing the fabrication of a top-electrode according to an embodiment of the present invention;

(6) FIG. 4 shows SEM images of the microfluidic device according to an embodiment of the present invention;

(7) FIG. 5 is a schematic diagram showing the electric contact according to an embodiment of the present invention;

(8) FIG. 6 shows graphs of data obtained from using the electric contact according to an embodiment of the present invention;

(9) FIG. 7 shows a graph of data obtained from using the electric contact according to an embodiment of the present invention;

(10) FIG. 8 shows a histogram of data obtained from using the electric contact according to an embodiment of the present invention;

(11) FIG. 9 shows histograms of data obtained from using the electric contact according to an embodiment of the present invention;

(12) FIG. 10 shows graphs showing the stability of the junctions using the electric contact according to an embodiment of the present invention;

(13) FIG. 11 shows the fabrication steps in the fabrication of the mold for microchannels in PDMS according to an embodiment of the present invention;

(14) FIG. 12 shows AFM images of a template-stripped Ag surface and an as-deposited Ag surface;

(15) FIGS. 13 to 17 show histograms of data obtained from using the electric contact according to an embodiment of the present invention;

(16) FIG. 18 show Plots of the average values of log|J| measured at 0.50 V using different EGaIn-based techniques for n-alkanethiolate self-assembled monolayer (SAMs)fthe solid line represent the fit to the Simmons equation;

(17) FIG. 19 show J(V) curves of junctions with SAMs of (a) SC.sub.13CH.sub.3, or (b) SC.sub.17CH.sub.3 right after the devices according to embodiments of the present invention were prepared and after the devices were left for several days; and

(18) FIG. 20 show the GC-MS and the corresponding MS spectra of HSC.sub.n1CH.sub.3 with n=10 (a and b), 12 (c and d), 14 (e and f), 16 (g and h), and 18 (i and j).

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

(19) The present invention relates to a method to fabricate SAM-based tunnel junctions that generate highly reproducible J(V) data in terms of precision and accuracy, in good yields of working junctions, with a value of the tunneling decay coefficient close to the consensus value, and good electrical stability. This method works well because it i) does not require patterned bottom-electrodes, ii) is compatible with ultra-flat template stripped bottom-electrodes that contain large grains, iii) does not expose the SAMs to harsh fabrication conditions, and iv) minimizes potential contamination of the bottom-electrode from the ambient. These improvements in the fabrication process resulted in SAM-based junctions of high quality and reproducibility that are (nearly) independent of the users or top-electrodes.

(20) By far most studies have focused to develop techniques to maximize yields of working junctions, or to produce them on large scales, but the reproducibility of the systems has been rarely defined and reported. It is well-known that some electronic properties of SAMs have been reproduced across several test-beds, but with a large spread in the current densities of eight to nine orders of magnitude. The width of the distributions indicates the precision of the data. The closeness of the average value of the distribution to a reference value, or standard value, indicates the accuracy (see

(21) FIG. 2). Standards for electrical characteristics across test-beds have not been defined because often standards have not been established within a single laboratory or, for a given technique, across different laboratories.

(22) It has been reported that clean template-stripped (TS) metal surfaces are ultra-flat (three to four times lower rootmean-square (rms) surface roughness than the surfaces fabricated by direct metal deposition) and readily available in ordinary laboratory conditions: the metal surface can be stripped off the template and immediately (within a few seconds) immersed into a solution with the SAM precursor to minimize contamination from the ambient environment. It has been shown that these TS metal surfaces resulted in SAM-based junctions in higher yields and a smaller spread in the J(V) data than those junctions formed with bottom-electrodes obtained by direct metal deposition. Hence, a fabrication technique to construct SAM-based devices that is compatible with TS surfaces, that is, a technique that does not require patterning of the bottom-electrode, is highly desirable.

(23) Cone-shaped tips of a liquid-metal alloy (eutectic mixture of 75.5% Ga and 24.5% In by weight with a thin 0.7 nm surface layer of conductive GaOx, abbreviated as GaOx/EGaIn) have been used to form electrical contacts to SAMs in various physical-organic studies of charge transport across SAMs. This method produces highly reproducible data in good yields and is very easy to set-up in a laboratory. This method has also disadvantages and it suffers from user dependent variations in the details of the formation of tips and the SAM//GaOx/EGaIn contacts, and the stability of the junctions is limited by the details of the micromanipulator on which the top-electrode is mounted.

(24) Here we describe a new type of top-electrode that allows us to form molecular junctions without the need for patterning of the bottom-electrode that is compatible with metal surfaces obtained by TS (as shown in FIG. 1c). As shown in FIG. 1(c), The GaOx/EGaIn (10) was stabilized in a microfluidic device (15) made of a transparent rubber of polydimethylsiloxane (PDMS) (20), which acts as an insulator, which we placed on the self-assembled monolayer of molecules SAMs (25). After electrical examination of the junctions, we removed the top-electrode (30) from the SAM (25) and placed it in contact with a different area of the SAM, or with a SAM on a different substrate (35), to form a new junction. Our method produces J(V) data that are very similar to data obtained by other EGaIn-based techniques, and are independent of temperature, from which we conclude that coherent tunneling dominates the mechanism of charge transport (see Table 1).

(25) TABLE-US-00001 TABLE 1 The average values of log|J|, , and J.sub.0, measured using different EGaIn-based techniques for n-alkanethiolate SAMs. <log|J|> [A cm.sup.2] J.sub.0 Techniques SC.sub.9CH.sub.3 SC.sub.11CH.sub.3 SC.sub.13CH.sub.3 SC.sub.15CH.sub.3 SC.sub.17CH.sub.3 [n.sub.C.sup.1] [A cm.sup.2] this work 1.95 2.92 3.63 4.63 5.44 1.00 0.03 2.4 10.sup.2 tips.sup.a) 1.77 2.47 3.70 4.32 5.31 1.02 0.09 3.4 10.sup.2 modified tips.sup.b) 1.250 1.60 2.30 3.270 4.10 0.91 0.02 25 10.sup.2 cross-bar.sup.c) NA 2.70 3.20 4.50 5.20 0.92 0.24 3.4 10.sup.2 Reference values.sup.d) 1.7 0.4 2.4 0.6 3.2 0.6 4.2 0.6 5.0 0.6 1.00 0.02 0.2-2 10.sup.3

(26) As such, as can be seen in FIG. 1(c), an electrical contact such as an electrical contact (30) can be fabricated and prepared for establishing an electrical contact with a soft material such as SAMs (25). The SAMs may be one cell thick (about 1 nm to 2 nm in thickness), and may be purified by any techniques known to the skilled person. Such an electrical contact may be known as a top-electrode (30) comprising a non-Newtonian liquid metal alloy (10) that is formed in a polymer insulator (20). In the embodiment of the present invention, the alloy used is GaOx/EGaIn. As described above, any liquid metal alloy that has conductive properties may be used. From a side view of the device (15), the liquid metal alloy (10) is formed such that it is sandwiched between two portions of the polymer insulator (20) to form a smooth flat non-patterned surface that contacts the soft material SAM (25). The SAM (25) is formed on an Ag substrate (35) to form a bottom-electrode (40). The advantage of the smooth flat surface is that, unlike other electrical contacts shown in FIGS. 1(a) and 1(b), there is no edge effecta disadvantage described above. FIG. 5 shows the electrical contact in usecontact between top electrode and bottom electrode.

(27) The sandwiched portion of the liquid metal alloy (10) may be any suitable diameter. In the present invention, the diameter may be between 15 m to 55 m. The amount of liquid metal alloy (10) in relation to the amount of polymer insulator (20) sandwiching it may be any amount a skilled person may find suitable to achieve the electrical characteristics of the electrical contact (30) which will be described in detail below.

(28) In an embodiment of the present invention, the polymer (or rubber) insulator (20) is a transparent PDMS. The liquid metal alloy (10) may be any non-Newtonian liquid metal alloy. In the present case, the alloy is a eutectic mixture of Ga and In (EGaIn) comprising 75.5 wt % Ga and 24.5 wt % In. The substrate (35) may be silver, copper, nickel, platinum, palladium or gold.

(29) The advantages of the present method is that encapsulation of the metal top-electrodes in PDMS eliminates instabilities associated with micromanipulators, e.g., drift or vibrations, and minimizes user-to-user variations in the details of the formation of the top-electrode and the SAM//GaOx/EGaIn contacts resulting in data with high precision and replicability. These features made it possible to study the electrical characteristics of the junctions over a period of time of ten days, bias stressing up to 1.010.sup.5 s, and over the range of temperatures of 160-297 K. Cone-shaped tips of GaOx/EGaIn can only be prepared one at a time per EGaIn-set-up, while the fabrication process reported here can be performed in parallel to fabricate large numbers of junctions.

EXAMPLE

(30) Junctions with GaOx/EGaIn Top-Electrodes

(31) The EGaIn spontaneously forms a self-limiting layer of GaOx in air with a thickness of 0.7 nm and because of its non-Newtonian properties this material can be shaped. Therefore, unlike Hg, GaOx/EGaIn forms stable structures in PDMS micro-channels. The oxide layer also prevents the bulk EGaIn from alloying with the gold or silver bottom electrode which adds stability to junctions. The oxide layer is defective and contains oxygen vacancies, and it is highly conductive.

(32) The precision of the data, that is, the width of the distributions of the values of J (see FIG. 2 for definitions), generated using junctions formed with cone-shaped tips of GaOx/EGaIn relies on the operator because the formation of the tips and bringing the tip in contact with the SAMs are usually performed with a manually operated manipulator. (Alternatively, piezo-controlled manipulators may be used.) For instance, the contact size, tip roughness, and the speed of the tip used to approach the SAMs, differ in details from user-to-user. It has been shown that these factors broaden the distributions of the current densities significantly. Recently, it was reported that flattening the cone-shaped tips by molding the tips against flat and clean Si/SiO.sub.2 surfaces followed by voltage cycling (three cycles of 2 V) resulted in smoother tips and higher reproducibility between users than using unmodified cone-shaped tips of GaOx/EGaIn. Stabilization of the GaOx/EGaIn in a micro-channel in a cross-bar configuration resulted in well-defined geometrical contact areas, but the improvement in the width of the distributions of the values of J was only marginal because the bottom-electrodes contained edges at which SAMs cannot pack well.

(33) Despite the (small) differences between the details of the formation of the GaOx/EGaIn top-contacts, Table 1 shows that the J(V) characteristics of SAM-based junctions with GaOx/EGaIn top-electrodes across laboratories differ slightly (less than one order of magnitude) compared to the eight to nine orders of magnitude difference in J(V) characteristics across test-beds. Thus EGaInbased techniques produce data that are replicable (in spite of the different levels of precision) across laboratories and platforms.

(34) Precision and Accuracy

(35) According to Equation (1) (see below), the values of log|J| are normally distributed when the error in d follows a normal distribution because J depends exponentially on d. The error in d certainly depends on many factors including defects in the electrode materials, for example, step edges, vacancy islands, or grain boundaries, defects in the SAMs, for example, phase domains, physisorbed or chemisorbed materials, or errors during the fabrication process, for example, (partial) penetration of the SAMs by the top-electrode, or damage to the SAMs inflicted by solvents or high temperatures during fabrication. These potential defects that result in uncertainties in the effective values of d and all may result in batch-to-batch or user-to-user variations and consequently introduce error that cause the data to deviate from Gaussian distributions and increase the standard deviation. Thus, one way to compare the precision of different techniques for junction measurements is to compare the standard deviations () of the values of J for normal distributions, or the analogues log-standard deviations ( log) for log-standard distributions (see FIG. 2). Data that follow narrow distributions make it possible to separate informative data from non-informative data more accurately than those data that follow broad distributions.

(36) As shown in FIG. 2, the accuracy of the data is defined as the difference between the data obtained from the measurement and the true, or defined, value. This Figure shows that data may be very precise but not accurate, but all other combinations are also possible and, for instance, data may be accurate but not precise. Although the width of the histograms of the values of J may be very narrow for a given test-bed, they do so with values that differ by eight to nine orders of magnitude across different test-beds. In the present invention, we do not aim to define the standards for the absolute values of J.sub.0 for junctions with SAMs of n-alkanethiolates because the factors that contribute to J.sub.0 for a given test-bed have, in general, not been identified. Here we wish to establish the replicability of our method relative to EGaIn-based techniques using reference values of the current densities (Table 1). This comparison helps to identify sources of error that are important to consider in general to maximize both precision and replicability (see below).

(37) Error Analysis

(38) As mentioned above, normally the values of log|J| (for a given voltage) are plotted versus n C followed by fitting this data to the Simmons equation. Previous work included the comparison of compared different statistical methods to determine the values of and J.sub.0 and the differences and limitations of these methods thoroughly discussed. These methods either used average values of log|J| (Gaussian mean, median, or arithmetic mean) to which a line was fitted using a least-squares fitting algorithm, or by plotting all data to which a line is fitted using either a least-squares algorithm or by minimizing the sum of the absolute error. Here we chose two methods to determine the values of and J.sub.0: i) plotting the Gaussian means of the value of log|J| vs n.sub.C followed by least squares fitting of Equation 1 (method 1) and ii) plotting all data (all values of log|J| except data that was obtained for junctions that shorted) followed by fitting to Equation (1) by minimizing the sum of the absolute values of the error (method 2). The first method assumes the data follow random distributions, or, in other words, the data are normally distributed, while the second method does not make any assumptions regarding the type of distribution.

(39) Results and Discussion

(40) 1. Fabrication of the Top-Electrode

(41) In essence, we first fabricated the 3D microfluidic channel and then injected the liquid metal into the channel to complete the formation of the top-electrode. The procedures of fabricating the microfluidic device as the top electrode is shown in FIG. 3. The mold which consists of a line and a pillar made of photoresists on a Si/SiO.sub.2 wafer (85), as shown in FIG. 4.

(42) The fabrication process is described in detailed below. Briefly, PDMS was spin-coated on the mold to fully cover the photoresist line, but not over the pillar. A channel in PDMS was aligned over the pillar perpendicularly with respect to the thin channel with the assistance of a thin layer of uncured PDMS as the glue. More uncured PDMS was added to stabilize the thin layer of PDMS. The microfluidic device was peeled off from the mold and a hole was punched at the end of the small channel (the connection). The width of this small second microfluidic channel was controlled to be smaller than 10 m to prevent the injection or diffusion of GaO.sub.x/EGaIn into this channel. We then place the microfluidic device on an ITO substrate (90) and injected GaO.sub.x/EGaIn into the PDMS channel. The outlet of the second microfluidic channel or through-hole (>20 m) was filled with GaO.sub.x/EGaIn by applying vacuum to the small channel. Separation of the microfluidic device from the ITO yielded a complete top-electrode. The electrical contact between the top-electrode and the soft material or matter can be formed by simply placing the top-electrode on the soft material or matter.

(43) The final product of the microfluidic device having an electrical contact according to an embodiment of the present invention is shown in FIG. 3(h). The device comprising a first microfluidic channel (channel 1-50) having an inlet (55) and an outlet (60); and a second microfluidic channel (channel 2-65) in a plane that is lower than the plane of channel 1. Both first and second microfluidic channels are in fluid communication with each other. The second microfluidic channel (channel 2-65) further comprising a first end in communication with the first microfluidic channel (channel 1-50) to form a connection (through-hole 70); and a second end opposite the first end to form an outlet (75). The connection (through-hole 70) further comprises an outlet (80) for exposing the liquid metal alloy (10) when the alloy (10) is formed in the microfluidic channel 1. The diameter of the outlet (80) may be between 15 m to 55 m.

(44) From the figures in FIG. 3, it can be seen that microfluidic channels 1 and 2 (50 and 65) on different plane but are in fluidic communication. The connection (70) is intermediate the inlet (55) and outlet (60) of the first microfluidic channel 1 (50). In an embodiment of the present invention, the microfluidic channels are perpendicular to each other.

(45) The process as set out in FIG. 3 is described in detail below.

(46) FIG. 3 shows the fabrication process of the top-electrode of GaOx/EGaIn stabilized in a microfluidic chip made of PDMS. We fabricated the mold to shape the PDMS which consisted of a pillar (45) (with a height of 60 m and the diameter of 45 m) connected to a line (42) (1.0 cm10 m10 m; FIG. 3a) via a two-step photolithography process.

(47) In essence, we followed a procedure reported in Kartalov et al. (P. Natl. Acad. Sci. USA 2006, 103, 12280) to fabricate the mold for PDMS microchannel (channel 2) connected with a through-hole (pillar).

(48) The fabrication steps of the fabrication of the mold for micro-channels in PDMS are shown in FIG. 11. In summary, FIG. 11(a) shows a SU8-2015 photoresist with thickness of 10 m was spin-coated on a Si wafer. FIG. 11(b) shows that the substrate was exposed to UV light through a photomask. FIG. 11(c) shows the unexposed photoresist was removed by developing. FIG. 11(d) shows a AZ-50XT photoresist with thickness of 60 m was spin-coated on the substrate. FIG. 11(e) shows that the substrate was covered with another photomask using a mask aligner and exposed to UV light. FIG. 11(f) shows the exposed photoresist when removed by developing.

(49) The Si wafer was exposed to hexamethyldisilazane (HMDS) vapor in a bake oven (YES 310TA) at 150 C. for 5 minutes. We deposited 10 m thick of SU8-2015 photoresist (Microchem) on the wafer by spin-coating the photoresist at 4500 rpm for 1 minute (FIG. 11a). The substrate was baked at 65 C. for 1 minute and 95 C. for 3 minutes on a hotplate. The photoresist was exposed to UV light (5 mW) through a mask for 22 seconds using a mask aligner (Suss Microtech; FIG. 11b), and followed by a post-exposure bake at 65 C. for 1 minute and 95 C. for 3 minutes. After cooling down to room temperature, the structures were developed in SU8 developer (Microchem) to yield a line (FIG. 11c). The substrate was then rinsed with copious amount of isopropyl alcohol, blown to dryness in a stream of N2, and baked at 150 C. for 5 minutes. To form the structure of the pillar at one end of this line, we first deposited 60 m thick AZ-50XT photoresist (AZ Electronic Materials) by spin-coating the photoresist at 700 rpm for 20 seconds on the wafer with the line-feature, and then baked the substrate for 2, 5, 2, and 9 minutes at 65, 115, 65, and 115 C., respectively (FIG. 11d). The mask was aligned with respect to the wafer followed by exposure to UV light (5 mW) for 2 minutes (FIG. 11e). The substrate was developed in the mixture of 1:1 ratio of AZ 400K developer:water for 5 minutes, and was rinsed with copious amount of water, and blown to dryness with N.sub.2 (FIG. 11f).

(50) To fabricate the PDMS channel 1 (that was used to over the pillar), we first fabricated a mold of photoresist with the dimensions of 1.0 cm300 82 m120 m on a Si wafer by spin-coating SU8-3050 at 1000 rpm/s for 1 minute, followed by baking the substrate for 1 and 5 minutes at 65 and 95 C., respectively. The photoresist was then exposed to UV light through a shadow mask with the mask size of 1.0 cm300 m. After developing as described above, the resulting mold was treated with FOTS vapor in a vacuum desiccator for 30 minutes. A mixture of 10:1 of PDMS and curing agent was poured on the mold and cured at 80 C. for 1 hour. After curing, we peeled off the PDMS layer from the mold and punched two holes (2 mm) at the ends of the channel 1 prior to alignment over the pillar.

(51) FIG. 4 a shows a scanning electron micrograph (SEM) of the mold. The image shows that the pillar had a larger base-diameter (55 m) than top-diameter (35 m) and had a depression at top-center due to over-developing of the thick layer of photoresist during the lithography process.

(52) The fabrication process of the top-electrode of GaO.sub.x/EGaIn starts with it being stabilized or formed or cured in a microfluidic device made of PDMS. We fabricated the mold for the PDMS microfluidic chip which consisted of a pillar (with a height of 60 m and the diameter of 35 m) connected to a line (1.0 cm10 m10 m; FIG. 3a). The pillar and the line were made of photoresist by a two-step photolithography process. The mold was treated with 1H,1H,2H,2H-perfluorooctyltrichlorosilane (Cl.sub.3Si(CH.sub.2).sub.2(CF.sub.2).sub.5CF.sub.3, FOTS) to minimize the interaction of the PDMS with the wafer to ensure the defect free separation of the PDMS from the mold. A layer of 20 m of uncured PDMS (Sylgard 184) with curing agent (10:1) was formed by spin-coating to fully cover the line but not the pillar (FIG. 3b). After curing of the PDMS at 80 C. for 30 minutes, we formed an addition layer of 5 m uncured PDMS (with curing agent) on which a channel in PDMS (1.0 cm300 m120 m), which we fabricated in a separate step, with an inlet and an outlet was aligned over the pillar perpendicularly with respect to the thin line (FIG. 3c). A seal was formed between the layers by curing the 5 m thick PDMS layer at 80 C. for 30 minutes. This thin PDMS layer prevented air from being trapped between the layers of PDMS and improved the mechanical stability of the devices (which are required to perform measurements as a function of temperature in vacuum; see supporting information). We added more uncured PDMS (with curing agent) with curing agent to stabilize the 20 m PDMS film (FIG. 3d). After curing the PDMS, we separated the microfluidic device from the mold and punched a hole at the end of the small channel (FIG. 3e). To inject GaO.sub.x/EGaIn into the microchannels, we placed the microfluidic device on indium tin oxide (ITO). The transparent and conductive properties of ITO allowed us to follow all subsequent stages of the fabrication process by optical microscopy and conductivity measurements. The large micro-channel was filled with EGaIn by applying vacuum (620 Torr) to the outlet of the channel with a drop of GaOx/EGaIn present at the inlet (FIG. 3f). Once the large channel was filled with the liquid metal, we applied vacuum to the small channel to fill the through-hole with GaO.sub.x/EGaIn (FIG. 3g). The diameter of the small channel was chosen such that the high surface tension of GaO.sub.x/EGaIn prevented it to fill this small channel in the range of the applied pressures. By simply measuring the resistivity between the ITO and the GaO.sub.x/EGaIn present at the inlet using a multi-meter we determined if the GaO.sub.x/EGaIn filled the through-hole and formed good electrical contacts with the ITO. Examination of the footprint of the contact of the GaO.sub.x/EGaIn with ITO by optical microscope allowed us to determine the geometrical area of the contact accurately. Finally, we separated the top-electrode from the ITO (FIG. 3h). During this step, no liquid-metal was left behind on the ITO surface.

(53) From the optical micrographs of FIG. 4, one can see the molds that were used to mold the rubber. Once the mold is prepared it can be used endlessly. FIG. 4(b) shows an optical micrograph of the rubber mold and FIG. 4(c) shows a close up of the small micrometer sized hole without liquid metal and FIG. 4(d) with liquid metal. FIG. 4(a) shows a scanning electron micrograph of the mold recorded from an angle with respect to the surface normal of 60. The image shows the pillar had a larger base-diameter (70 m) than top-diameter (40 m) due to the over-development during the fabrication. FIG. 4(b) shows the optical micrograph of the cross section of PDMS device (without liquid metal in the microfluidic channels) and that the 3D structures of FIG. 4(a) were successfully replicated by the PDMS. We placed the PDMS microfluidic channel on an ITO substrate to monitor the filling of the microfluidic channels with liquid metal. Thus the optical properties of this rubberthe fact that this rubber is transparentmake it possible to follow the fabrication closely and improve yields and reproducibility. FIG. 4 show the optical micrographs before, and after, filling the microfluidic channels with the liquid-metal, respectively. The small channel remained empty because of the high surface tension of the liquid-metalthis small channel is needed to fill the hole. We derived the geometrical contact area of the GaOx/EGaIn with the ITO from these imagesagain to be able to do so the topical transparent properties of the rubber mold are important.

(54) The mold was treated with 1H,1H,2H,2H-perfluorooctyltrichlorosilane (Cl.sub.3Si(CH.sub.2).sub.2(CF.sub.2).sub.5CF.sub.3, FOTS) to minimize the interaction of the PDMS with the wafer to ensure the defect free separation of the PDMS from the mold (see below). A layer of 20 m of uncured PDMS (Sylgard 184) with curing agent (in a ratio of 10:1) was formed by spin-coating which covered the line but not the pillar (FIG. 3b). After curing of the PDMS at 80 C. for 30 min, we spin coated an additional layer of 5 m uncured PDMS (with curing agent) and aligned a channel (channel 1) in PDMS (1.0 cm300 m120 m), which we fabricated in a separate step, with an inlet and an outlet over the pillar perpendicularly with respect to the line of the mold (FIG. 3c). A covalent seal was formed between the layers by curing the 5 m thick PDMS layer at 80 C. for 30 min. This thin PDMS layer improved the mechanical stability of the devices. We added more uncured PDMS with curing agent to stabilize the 20 m PDMS film (FIG. 3d). After curing the PDMS, we separated the microfluidic chip which contained two perpendicular channels with a microscale through-hole at the intersection from the mold and punched a hole at the end of the channel 2 (FIG. 3e). FIG. 4b shows the optical micrograph of the cross section of the PDMS device (without liquid metal in the micro-channels) and that the 3D structures were successfully replicated by the PDMS.

(55) To inject GaOx/EGaIn into the micro-channel 1 and the through-hole, we placed the microfluidic chip on indium tin oxide (ITO). The transparent and conductive properties of ITO allowed us to follow all subsequent stages of the fabrication process by optical microscopy and conductivity measurements. Channel 1 was filled with EGaIn by applying vacuum (500 Torr) to the outlet of the channel with a drop of GaOx/EGaIn present at the inlet (FIG. 3f), after which we applied vacuum to channel 2 to fill the through-hole with GaOx/EGaIn (FIG. 3g). The diameter of channel 2 was chosen so that the high surface tension of GaOx/EGaIn (624 mN/m) prevented it to fill this channel in the range of the applied pressures. FIG. 4c shows that the liquid-metal did fill the through-hole but not channel 2. We derived the geometrical contact area of the GaOx/EGaIn with the ITO from these images. We found that the diameter of the GaOx/EGaIn-ITO contact (35 m) was smaller than the diameter of the through-hole (55 m) because of a gap of 10 m between the GaOx/EGaIn and the walls of the hole. The formation of the gap was likely caused by the high surface-tension of the GaOx/EGaIn. By simply measuring the resistance between the ITO and the GaOx/EGaIn present at the inlet using a multi-meter, we determined if the GaOx/EGaIn filled the through-hole and formed good electrical contact with the ITO. Finally, we separated the top-electrode from the ITO (FIG. 3h). During this step, no liquid-metal was left behind on the ITO surface.

(56) If required, the dimensions of the top-electrodes can be easily reduced by reducing the lengths of channels 1 and 2. As an alternative to the above, a top-electrode with both channels 1 and 2 having a length of 0.5 cm may be fabricated. As a further, alternative, the length of the channels 1 and 2 may be any length between 1.0 to 0.5 cm.

(57) 2. Fabrication of the Junctions

(58) FIG. 5 shows schematically the reversible placement of the top-electrode on a SAM on Ag.sup.TS electrodes (FIG. 5a).

(59) The figure shows schematically the rubber stamp that stabilizes the non-Newtonian liquid metal. Non-Newtonian liquids behave as solid or liquid depending on the pressure. This liquid metal behave as liquid and can be injected in to microfluidic channels when a pressure difference is applied, but returns to its solid state in ambient conditions. We chose a rubber stamp because this material forms conformal contacts with all target surfaces we tested. The top-panel shows the top-electrode in contact with a monolayer of molecules that are exactly only ONE molecule thick (1-2 nm). The metal is indicated as GaOx/EGaIn, the rubber is indicated as PDMS, Ag indicate the silver surface that supports the self-assembled monolayer (SAM). The inset show schematically the metal-molecule-metal structure. The panel in the middle show the separation of the top-electrode from the SAM and the bottom panel shows how the electrode is placed back in contact with the target. This step can be repeated up to 20-30 times.

(60) We found that the electrodes formed good electrical contacts with the SAMs in most cases; in cases a good contact did not form (which resulted in either an open circuit or J(V) characteristics with values of J that were more than two orders of magnitude lower than the log-mean value), we simply applied vacuum to channel 2 to restore good electrical contact of the GaOx/EGaIn with the SAM. FIG. 4d shows a photograph of a complete device. The PDMS is flexible and forms a reversible conformal contact with the substrate through van der Waals interactions. We observed that the seal between the top-electrode and the SAM-Ag.sup.TS substrate allowed the GaOx/EGaIn to form good electrical contacts with the monolayers. Because the EGaIn surface is exposed to air in our experiments, we believe that the GaOx film that forms spontaneously in air on the bulk metal is continuous and very similar in composition to that in other types of EGaIn-based junctions. After recording the J(V) curves, we separated the top-electrode from the substrate (FIG. 5b), and placed it in contact with the SAM in an area that had not been in contact with either the PDMS (to avoid potential contamination of the SAMs by, for instance, low molecular weight or uncured PDMS) or the GaOx/EGaIn previously, or in contact with another Ag.sup.TS-SAM substrate (FIG. 5c).

(61) The bottom-electrode (eg. Shown in FIG. 5with the SAM-AG) may be ultra-flat and/or smooth. It may have a root-mean-square (r.m.s.) roughness around 0.9 nm over an 11 m.sup.2 area and large grains (0.2-1 m.sup.2).

(62) The procedure to remove the top-electrode from the surface and to form a new junction typically takes 5-10 s. The rate at which junctions can be formed for a single electrode is similar to that for cone-shaped based techniques (once a cone-shaped tip of GaO.sub.x/EGaIn has been formed). We used 4 wafers which allowed us to prepare six top-electrodes at once per wafer, but top-electrodes with shorter channels can also be prepared to yield for instance 18 top-electrodes per wafer. The top-electrodes lasted for about 15-25 junctions before they failed and did not form good electrical contacts with the SAMs. Optical micrographs of these failing top-electrodes revealed that the small channels were blocked by dust particles and therefore good contacts could not be restored by applying vacuum to channel 2. Occasionally (in about 1 out of 20 top-electrodes), the thin PDMS membrane surrounding the GaOx/EGaIn ruptured resulting in ill-defined junction areas. Hence, the number of junctions that can be formed in parallel is only limited by the number of available molds and the rate of data acquisition is only limited by the electronic equipment.

(63) 3. Proposed Reference Values of J for EGaIn-Based Techniques

(64) Table 1 shows the values of <log|J|> for Ag.sup.TSSC.sub.n//GaO.sub.x/EGaIn junctions with n=10, 12, 14, 16, and 18. The reference values of <log|J|> were obtained by averaging the values of <log|J|> obtained from earlier and the present work. We determined reference values of and J.sub.0 by least squares fitting the average values of <log|J|> to the Simmons equation (see FIG. 6 and discussion below). The values of J.sub.0 depend on many factors, including the effective contact area or contact resistance. Here, we do not aim to compare the absolute values of J.sub.0 across test-beds but only across EGaIn-based techniques. As we show here, these proposed reference values are useful to compare EGaIn-based techniques to each other, or to identify sources of error.

(65) 4. The Electrical Characteristics of the Devices

(66) In molecular electronics, it is a common practice to determine the tunneling decay coefficient, (n.sub.C.sup.1), by measuring the value of J at a given voltage, V(V), as a function of the thickness of the SAMs, d ( or n.sub.C which is the number of carbon atoms in the back bone of the molecules), when through bond tunneling is the dominant mechanism of charge transport. By fitting the data to the Simmons equation (Equation (1)), one can derive the values of and of the hypothetical current density, J.sub.0 (A cm.sup.2), for a junction with d=0. This procedure has been used across several test-beds using SAMs of n-alkanethiolates of the form S(CH.sub.2).sub.n1CH.sub.3 and it is now commonly believed that the correct value for is 1.0 n.sub.C.sup.1.
J=J.sub.0e.sup.d(1)

(67) We formed junctions with SAMs of S(CH).sub.n1CH.sub.3 (n=10, 12, 14, 16, or 18) on Ag.sup.TS. Prior to the self-assembly of the monolayers, we purified the n-alkanethiolates. Although we did not test the performance of the junctions as a function of thiol purity, this procedure minimizes potential variations in the batch-to-batch quality of the thiols as received from the suppliers which could influence replicability and/or precision of the electrical characteristics of the junctions. Using a single top-electrode, we measured a complete graph of |J| against n.sub.C determined at 0.50 V as follows. We recorded the values of J over the range of biases of 0.50 and 0.50 V (one trace=0 V.fwdarw.0.50 V.fwdarw.0.50 V (2) 0 V) at intervals of 25 mV. We recorded a total of 20 J(V) traces for each junction and measured three junctions for each type of SAM using a single top-electrode. Thus, we formed 15 junctions with five different SAMs and recorded in total 300 traces and 600 values of |J| at each applied bias using a single top-electrode. This procedure was repeated with five different top-electrodes to yield a total of five plots of versus n.sub.C operated by one out of three different users. This procedure allows us to determine the replicability and precision of data across individual users and top-electrodes.

(68) TABLE-US-00002 TABLE 2 The total number of non-shorting junctions (N.sub.J), the yields of the working devices and .sub.log of J(V) measurements for the n-alkanethiolate-based junctions. Non-shorting Non-shorting Molecules Junctions N.sub.J junctions yield [%] .sub.log SC.sub.9CH.sub.3 21 600 15 71 0.12 SC.sub.11CH.sub.3 19 600 15 79 0.25 SC.sub.13CH.sub.3 20 600 15 75 0.22 SC.sub.15CH.sub.3 19 600 15 79 0.16 SC.sub.17CH.sub.3 17 600 15 88 0.15 total 96 3000 75 .sup.78.sup.a 0.18.sup.a .sup.aThese numbers are average values.

(69) Table 2 summarizes the characteristics of the junctions. We excluded shorts and open circuits and kept the number of working junctions constant so the data for all junctions carry the same weight in our analysis. FIG. 6 shows the averaged J(V) curves over all users for each type of junction, the histograms of log|J| at 0.50 V, and plots of <|J|> versus n.sub.C. Fitting the data to the Simmons equation gave values of of 1.000.03 n.sub.C.sup.1 and J.sub.0 of 2.410.sup.2 A cm.sup.2 (using method 1: fitting the Simmons equation to the Gaussian means of log J by least squares fitting). We note that the value of J.sub.0 is not precise because of the long extrapolation. The value of is very close to the consensus value, and the value of J.sub.0 is very close to previously reported values obtained for other types of junctions with GaO.sub.x/EGaIn top-electrodes (See Table 3 and below). These results indicate that the dominant mechanism of charge trans-port across our junctions is through-bond tunneling.

(70) One may argue that if the distribution of log|J| deviates from normality, it is more accurate to estimate trend statistics by fitting all values of log|J| by a least-absolute-errors algorithm (using method 2: Fitting the Simmons equation to all values of log J by minimizing the absolute error) and) rather than by fitting Gaussian means of log|J| with a least-squares algorithm (method 1) since the former method does not assume any type of distribution while the latter method does. FIG. 6d shows the plot of all values of log|J| at 0.50 V versus n.sub.C with a fit to the Simmons equation using method 2. The values of and J.sub.0 were found to be 1.000.02 n.sub.C.sup.1 and 2571.6 A cm.sup.2, respectively. Considering the small differences in the values of and J.sub.0 obtained by methods 1 and 2, we believe that the assumption that our data follow a normal distribution introduces a negligibly small error in the analysis of our data.

(71) 5. Precision of the Data

(72) The striking difference of the current fabrication method with respect to other methods is that the values of log-standard deviation ( log) are very small and fall in the range of 0.12 to 0.25 with an average of 0.180.05; these values are one of the lowest in general (see Table 3). Thus our method generates J(V) data with very high precision.

(73) TABLE-US-00003 TABLE 3 Comparison of .sub.log, , and yield of different tunneling junctions with SAMs. Type of junction Technique N .sub.log [n.sub.C.sup.1] yield [%] N.sub.max.sup.k) Refs. Ag-SAM//SAM-Hg Hg-drop 1-5 =1.0-1.5.sup.c) 0.80 NA 13 [32] Hg-SAM//Hg Hg-drop 5-10 =0.11-0.6.sup.c) 1.06 0.04 NA 10 [38] Hg-SAM//SAM-Hg Hg-drop 5-10 =0.11-0.43.sup.c) 1.02 0.07 NA 10 [38] Si-SAM//Hg Hg-drop >7 =0.37-0.70.sup.c) 0.76 0.09 NA NA [88, 97] Al/Al.sub.2O.sub.3-SAM//Hg Hg-drop 12-18 0.25-0.75 .sup.1.34 0.004.sup.h) 25-75 18 [41] M-SAM//M.sup.a) CP AFM 5-10 =0.28-1.0.sup.c,d) 1.1 NA 10 [42] Au-SAM//Au STM break junction 3000 0.02-10.sup.d) 0.94-0.96 10-40 NA [94] Ag-SAM//Ag STM break junction 7000 =0.16-0.32.sup.f) .sup.0.98 0.05.sup.i) NA NA [95] Au-SAM//Au nanoskiving >10 =0.05-0.28.sup.c) 0.94 36-67 32 [85] Si-SAM//Au PALO.sup.b) >10 =0.05-0.26.sup.c) NA NA NA [92] Si-SAM//Au flip chip lamination >30 =0.06-0.1.sup.c,g) NA 90 NA [89] Au-SAM//Au crossed-wires NA =0.22.sup.c,e) NA NA NA [90] Au-SAM//polymer/Au PEDOT:PSS/micropore >17 =0.11-0.15.sup.c) 0.71 0.06 >95 100 [93] Au-SAM//polymer/Au PEDOT:PSS/micropore 74 =0.23.sup.c) 1.33 0.05 58 74 [39] Au-SAM//Au SiO.sub.2 micropole 33-63 0.23-0.527 1.04-1.08 1.2-1.75 NA [26] Au-SAM//Au Si.sub.3N.sub.4 nanopore =160 =0.27-0.32.sup.c) 1.07 0.02 7.1 NA [14] Au-SAM//Au wedging transfer 200-340 0.57-1.13 0.73 0.06 38-50 NA [86] Au-SAM//graphene/Au graphene/micropore 258 =0.27-0.67.sup.c) 1.06 0.14 90 2000 [87] graphene-SAM//graphene graphene//micropore >50 =0.17-0.35.sup.c) 0.54 0.01 >80 NA [91] Ag-SAM//GaO.sub.x/EGaIn cone-shaped tip 376-3892 0.23-1.1 .sup.1.04 0.06.sup.j) 82-100 NA [40] Ag-SAM//GaO.sub.x/EGaIn Flattened cone-shaped tip 360-480 0.3-0.7 0.92 0.02 =90% NA [46] Ag-SAM//GaO.sub.x/EGaIn cross-bars 400-756 0.21-0.85 0.98 0.2 70-85 NA [16] Ag-SAM//GaO.sub.x/EGaIn through-hole 600 0.12-0.25 1.00 0.03 78 2500 this work .sup.a)A metal-coated (Au, Ag, or Pt) AFM tip was contacted with a SAM on a Au-, Ag-, or Pt-coated Si substrate; .sup.b)Polymer-assisted lift-off method; .sup.c)Roughly estimated from the J(V) curves in the corresponding references; .sup.d)The log standard deviations of the resistance instead of current density were measured; .sup.e)The standard deviations of the conductance instead of current density were reported; .sup.f)The log standard deviations of the conductance instead of current density were reported; .sup.g)The standard deviations of the current at 1 V instead of current density were reported; .sup.h)This value was reported for C8-C12 junctions. The value was reported to be 0.77 0.005 for the junctions of C12-C16; .sup.i)The value was measured to be 0.93 0.05 n.sub.C.sup.1 when the Au tip was used to measure the junctions of SAMS on Au substrates; .sup.j)This value was obtained by measuring the SAMs of n-alkanethiolate with even number of carbons in the molecules. The value of was found to be 1.19 0.08 n.sub.C.sup.1 for SAMs of n-alkanethiolate with odd number of carbons in the molecules; .sup.k)The maximum number of continuous scans without shorting or becoming open circuit for a single junction. These numbers are either shown or indicated in the papers.

(74) To determine reproducibility between different top-electrodes and different investigators we performed two experiments: three investigators measured the electrical characteristics of the junctions using i) five different top-electrodes each operated by one of the three investigators, or ii) one top-electrode operated by all three investigators. FIG. 7 shows the corresponding five plots of log|J| versus n.sub.C.sup.1 determined by three different users using five different top-electrodes. Each plot was measured by one user and top-electrode combination as indicated in the Figure. FIGS. 13 to 17 show the corresponding histograms of log(|J|) determined at 0.50 V for all Ag.sup.TSSC.sub.n//GaO.sub.x/EGaIn junctions for each user and top-electrode combination. The values of obtained from these plots were all in the range of 0.95-1.05 n.sub.C.sup.1 with values of J.sub.0 ranging from 109-560 A cm.sup.2 (see inset of FIG. 7). Note that we give a range of values rather than standard deviations because the number of J(V)-curves is per user lower than the total number of data; these numbers are not precise but reasonably replicable because they all are in the range of previously reported values of 0.2-210.sup.3 A cm.sup.2 (Table 1). FIGS. 13 to 17 show the histograms of log|J| measured at 0.50 V for junctions with SAMs of SC.sub.9CH.sub.3, SC.sub.11CH.sub.3, SC.sub.13CH.sub.3, SC.sub.15CH.sub.3, and SC.sub.17CH.sub.3, respectively. In our measurements, three junctions with 40 J(V) traces for each junction were collected for each type of SAMs by using the same top-electrode conducted by three different users. This procedure was repeated for five times with five different top-electrodes. By fitting the histograms of log|J| at 0.50 V to Gaussians, we determined the average log|J| values and log-standard deviations following previously reported procedures.

(75) To investigate if the data depend on the users, we also examined the histograms of the values of log|J| at 0.50 V for junctions with SAMs of SC.sub.17CH.sub.3 obtained by three different users using the same top-electrode. FIG. 7 shows that the data are independent of the user who conducted the measurement. These results indicate that the values of log|J| and are narrowly distributed and independent of the user or top-electrode: the data produced by our technique is precise with respect to different operators and top-electrodes.

(76) 6. Replicability of the Data

(77) As mentioned earlier that junctions with TS bottom-electrodes result in junctions with higher yields in non-shorting junctions with smaller log-standard deviations than those junctions with direct metal deposited bottom-electrodes, but the authors did not discuss whether the topography of the bottom-electrode is important in the replicability of the data. To determine if the topography of the bottom-electrode is an important source for lowering the replicability, we formed SAMs of SC.sub.17CH.sub.3 (using the same batch of the thiol precursor) on as-deposited Ag substrates, which had a rms roughness of 3.3 nm and small grains of <310.sup.2 m.sup.2, and on TS surfaces, which had a rms roughness of 0.9 nm and large grains of 0.05-0.9 m.sup.2 in agreement with previously reported data (see FIG. 12 for AFM images). Subsequently, we formed junctions using one top-electrode operated by one investigator. FIG. 8 shows that the values of J increased nearly two orders of magnitude and the log increased from 0.26 to 0.58 as we changed the bottom-electrode from Ag.sup.TS to as-deposited Ag. This change in the topography of the bottom-electrode resulted in large decrease in the precision and replicability of the J(V) data. The exact rms values and grain sizes of the bottom-electrodes depends on many factors including deposition rate, base-pressure of the vacuum chamber, pre-treatment of the target surface, and so forth. Thus, small variations in the topography of the bottom-electrode can be a source of data broadening and may cause shoulders or even new peaks in histograms of J. To maximize the precision and replicability of the J(V) data generated by our method, we record AFM images for every new batch of electrodes and only use surfaces similar to that shown in FIG. 12a before we start experiments. FIG. 12(a) shows an AFM image of a template-stripped Ag surface, while FIG. 12(b) shows an AFM image of a as-deposited (thermal) Ag surface. The rms roughness of the template-stripped and the as-deposited Ag surfaces were determined to be 0.9 and 3.3 nm (over an area of 11 m) respectively.

(78) To show that stabilization of the GaOx/EGaIn in the through-hole in PDMS contributes to the precision of the data, we measured the J(V) characteristics of junctions with SAMs of SC.sub.7CH.sub.3 with top-electrodes of GaOx/EGaIn stabilized in PDMS (FIG. 9a) or with cone-shaped tips of GaO/EGaIn operated by three investigators (FIG. 9b). FIG. 9 shows that both data sets have their log-mean value of J close to the reference value. The widths of both distributions are comparable, but shoulders on both sides of the main peak are visible which mainly originated from one of the three operators for junctions with cone-shaped tips of GaO.sub.x/EGaIn. Thus, in this experiment user-to-user correlation was significant. These results are similar to those reported earlier who collected large values of N.sub.J of up to a few thousand by multiple investigators and reported broad distributions that contained multiple peaks. Junctions prepared with cone-shaped tips of GaO.sub.x/EGaIn vary in details of the formation of the tip and the formation of the junctions that differ from user-to-user. The stabilization of the top-electrode in microfluidic device minimizes the user-to-user variations in the formation of the top-electrodes, the geometric area of the junctions, and the potential error associated with vibrations and drift of the cone-shaped tip of GaOx/EGaIn mounted on a micro-manipulator as top-electrodes resulting in precise data.

(79) 7. Stability of the Devices

(80) To ensure our test-bed can be used as a reliable platform for studying charge transport across SAMs, it is crucial to know the electrical stability and the lifetime of these devices. We tested the electrical stabilities of the devices incorporating SAMs of SC.sub.9CH.sub.3, SC.sub.13CH.sub.3 and SC.sub.17CH.sub.3 against continuous cycling of voltage (2500 cycles of 0 V.fwdarw.0.50 V.fwdarw.0.50 V.fwdarw.0 V), bias stress (by applying a constant bias of 0.50 V for 10.sup.5 seconds), and aging (ambient conditions at room temperature) over a period of time of ten days. FIG. 10a shows that these devices are electrically stable and did not short during voltage cycling. FIG. 19 shows the values of J (at 0.50 V) as a function of cycle number. The value of J for the junction with SC.sub.9CH.sub.3 was more stable than those junctions with SAMs of SC.sub.13CH.sub.3 (noisy around cycle number 975-985) and SC.sub.17CH.sub.3 (noisy for cycle number>1500). FIG. 10b shows the retention characteristics of the devices. The junctions with SAMs of SC.sub.9CH.sub.3 and SC.sub.13CH.sub.3 were more stable than the junction with SC.sub.17CH.sub.3 which became noisy after 3.010.sup.3 s. One possible reason for the difference in the electrical stability between the devices is that SAMs with long alkyl chains are more crystalline and therefore contain more defects from, for example, phase domains boundaries than the short liquid-like SAMs. FIG. 10c shows the J(V) curves of the device with SAMs of SC.sub.9CH.sub.3 determined at t=0, 1, 2, 5, and 10 days. Over this period of time, the values of J decreased by approximately a factor of seven. A similar behavior was observed for devices with SC.sub.13CH.sub.3, and SC.sub.17CH.sub.3 SAMs (see FIG. 20). The reason for the change in current densities is unclear, but it may involve oxidation of the metal-thiolate bonds or the formation of silver sulfides.

(81) Measurements of J(V) as a function of temperature T(K) are important to establish the mechanism of charge transport across tunnel junctions. To test the stability of the devices against changes in temperature, we studied the electrical characteristics of the devices over a range of values of T of 160-297 K. These measurements were performed in a probe station at a pressure of 110.sup.5 bar. In agreement with previous observations, both the change of pressure from ambient to vacuum and solidification of the bulk EGaIn at T=220-240 K did not result in shorts, open circuits, or changed the electrical char-acteristics of the devices notably in any other way. FIG. 10d shows that the J(V) curves of devices with SAMs SC.sub.9CH.sub.3 are (almost) independent of temperature as expected when tun-neling is the dominant mechanism of charge transport. The devices shorted at the temperature lower than 160 K likely due to the differences in the thermal expansion coefficients of the different components in the devices (310.sup.4 K.sup.1 for PDMS, 0.0810.sup.4 K.sup.1 for glass, 0.04210.sup.4 K.sup.1 for Ga.sub.2O.sub.3, 0.1810.sup.4 K.sup.1 for EGaIn).

(82) The following is a description of the electrical measurements: (a) Electrical Measurements Using GaO.sub.x/EGaIn Cone-shaped Tips

(83) The experiments involving junctions with cone-shaped tips of GaO.sub.x/EGaIn were conducted using a home-built set-up. The set-up contained a micromanipulator (Leica) equipped with a 10-l glass syringe (Hamilton, 1701 RNR) with a metallic needle (Hamilton, conical shape 26s) and a tungsten probe (Signatone, SE-T) connected to a Keithley 6430 source meter. The formation of tips and the junctions have been described previously. Briefly, we filled the syringe with GaO.sub.x/EGaIn and pushed a drop of GaO.sub.x/EGaIn out of the syringe and brought the GaO.sub.x/EGaIn droplet hanging from the needle in contact with a sacrificial substrate. Once the GaO.sub.x/EGaIn droplet stuck on the substrate, we slowly pulled the syringe needle from the droplet using the micromanipulator. A conical shaped tip of GaO.sub.x/EGaIn formed suspended from the syringe needle once it disconnected from the GaO.sub.x/EGaIn at the surface. (b) Proposed Reference Values for EGaIn-based Techniques

(84) We determined the reference values of and J.sub.0 by least squares fitting the average <log|J|> (the values are shown in Table S1 below) using the Simmons equation (see FIG. 18).

(85) TABLE-US-00004 TABLE S1 Summary of GC-MS area percent report of SC.sub.nCH.sub.3 (n = 10, 12, 14, 16, 18). Retention time Total Compound Peak (minutes) Area (%) C.sub.10SH 1 11.95 4.3 10.sup.8 98.2 2 21.92 7.8 10.sup.6 1.8 C.sub.12SH 1 14.75 1.6 10.sup.8 100.0 C.sub.14SH 1 16.77 3.2 10.sup.8 100.0 C.sub.16SH 1 18.24 1.5 10.sup.7 100.0 C.sub.18SH 1 19.45 9.3 10.sup.7 100.0

(86) It is well-known that thiols can decompose to form disulfides and sulfonates in ambient conditions. All the as-received n-alkanethiols (Sigma-Aldrich) were recrystallized from ethanol (AR grade) under atmosphere of N.sub.2 at 20 C. followed by quick filtration prior to use. In case of insoluble impurities, the ethanolic thiol solutions were filtered before recrystallization. FIG. 20 and Table S1 show the GC-MS spectra and the corresponding analytical results, respectively of the n-alkanethiolates after the purification. These results reveal that after the purification, the concentration of HSC.sub.9CH.sub.3 was more than 98.7% with trace amount of disulfides, while no impurities were found in HSC.sub.11CH.sub.3, HSC.sub.13CH.sub.3, HSC.sub.15CH.sub.3 and HSC.sub.17CH.sub.3.

(87) 8. Comparison to Other Test-Beds

(88) To judge the performance of our method against previously reported test-beds, we compared i) yields in working junctions, ii) log-standard deviations as an indicator of reproducibility or precision, iii) values of as an indicator of the replicability or quality of the junctions (lower or higher values than the con sensus value of 0.9-1.1 n.sub.C.sup.1 are likely caused by artifacts), and iv) the stability against voltage cycling (crudely judged from the number of scans), and v) the ability to generate statistically large numbers of data. Our fabrication technique give devices with i) yield larger than 75%, ii) values of log smaller than 0.3, iii) of 1.000.03 n.sub.C.sup.1, iv) good electrical stability (2500 times of voltage cycling), and v) produces statistically large numbers of data (N>600). In Table 3 we highlighted techniques that have comparable or better characteristics than ours in bold. Although these criteria are arbitrary chosen and this comparison gives a crude impression at best how different test-beds perform relative to ours, this effort hopefully serves as a starting point to judge methods not only by yields in non-shorting junctions, or the value of , but also by more criteria including stability and more importantly reproducibility and replicability.

(89) Table 3 is not comprehensive, but we included data obtained by large-area SAM-based junctions that contain large numbers of molecules, and techniques based on scanning probes that contain small numbers of molecules or even single molecules. In typical scanning tunneling microscope (STM) measurements, the air or vacuum gap between the tip and the molecules complicates evaluating the true conductance of the molecules. The so-called STM break junction technique forms junctions by capturing the molecules between the STM tip and the bottom-electrode in situ from solution. Although this technique produces large numbers of data, little information is available regarding the supra-molecular structure of the junctions. Direct deposition of the top-electrodes on SAMs resulted in low yields of non-shorting junctions and is prone to metal filament formation, and other types of defects. Using a conductive layer (polymer or graphene based materials) between the SAM and the top-electrode protects the SAM during metal deposition and increased the yields. Other techniques have avoided metal deposition by using liquid-metal top-electrodes (Hg or GaO.sub.x/EGaIn) which deform and conform to, rather than penetrate, the SAM once brought into contact with the SAM. Others have deposited solid electrodes from solution or used bending wires to form junctions.

(90) Among the methods that generate values of that are close to 1.0 n.sub.C.sup.1, our method has amongst the smallest log values (0.12-0.25). Nanoskiving also generates comparably small log values (0.05-0.28) with 0.94 n.sub.C.sup.1 but with low numbers of data. Junctions with PDOT:PSS protection layers are generated in very high yields with small errors and good stabilities, but with very low values of . Junctions with graphene as protection layer perform also well and produce large numbers of data in high yields with close to 1.0 n.sub.C.sup.1, but with a larger error than our method. For most fabrication methods, the stability of the junctions against voltage cycling has not been reported, but our method compares well in stability to that of the rigorously tested junctions with graphene protection layers. Thus, we conclude that our fabrication method generates junctions with high reproducibility, replicability, good electrical stability, and generates statistically large numbers of data in good yields.

CONCLUSION

(91) Here we report a new technique to from electrical top-contacts to SAMs that relies on a top-electrode of a non-Newtonian liquid metal alloy stabilized in a micro-scale through-hole in PDMS. This top-electrode can be directly placed onto the SAMs, removed from the SAMs once the measurements are completed, and used again to form a new junction. Typically 15-25 junctions can be formed with a single top-electrode.

(92) Thus, this method provides the opportunity to investigate the reproducibility of the electrical characteristics of SAM-based junctions as a function of the top-electrodes and users. We found that the electrical characteristics are highly reproducible between different users and top-electrodes: the values of obtained by three investigators using five different top-electrodes ranged only from 0.95 to 1.05 n.sub.C.sup.1 with an average value of 1.000.03 n.sub.C.sup.1.

(93) Unlike other methods to fabricate SAM-based devices (of the sort shown in FIG. 1), our technique is compatible with template-stripped bottom-electrodes and does not require patterning of the bottom-electrode. This ensures that the electrodes supporting the SAMs are clean and never had been exposed to photoresist (which is often difficult to remove completely) and only briefly exposed to the ambient (few seconds), and do not contain edges at which SAMs cannot pack well. The stabilization of the top-electrode minimizes the user-to-user variation in contacting the SAMs, defines the geometrical area of the junctions. To improve replicability, prior to the SAM formation, we purified the thiols and characterized the template-stripped electrodes by AFM to confirm the quality of the bottom-electrodes. All these factors resulted in very narrow log-normally distributed values of J ( log=0.12-0.25), i.e., the data generated by our junctions is highly reproducible in terms of precision with high replicability relative to other EGaIn-techniques.

(94) This method minimizes the potential error associated with cone-shaped tips of GaO.sub.x/EGaIn suspended from a syringe such as vibrations, pressure at which the tip is brought in contact with the SAM, or drift of the tip with respect to the SAM. Therefore it is possible to measure J(V) curves over a range of temperatures of T=160-297 K which confirmed that the dominant mechanism of charge transport is coherent tunneling. We conclude that all EGaIn-based techniques produce J(V) data that agree with one another (because the values of J.sub.0 vary only a factor of ten which is small relative to the eight to nine orders of magnitude difference across text-beds), but our data are more precise (the distributions of J have small log-standard deviations; Table 1). To date we cannot explain the absolute values of the values of J (or J.sub.0) in every detail. It is reported that the GaOx layer is highly conductive and does not affect J significantly, but the effective electrical contact area is smaller than the geometrical contact area and therefore results in an underestimation of the value of J.sub.0. We will discuss the role of molecule-electrode contact resistances and defects on the values of J.sub.0 elsewhere.

(95) The term reproducibility is ill-defined and so is the quality of SAM-based junctions. Therefore it is difficult to compare one test-bed to another, but by far most approaches have only focused on the yields in non-shorting junctions and standard deviations. We found that comparing yields and (log) standard deviations only provide marginal information regarding the quality of the junctions. For instance, techniques that produce significantly lower or higher values of than 1.0 n.sub.C.sup.1 with very small standard deviations in high yields are perhaps precise, but are not accurate and likely probe defective junctions. We used the following parameters to evaluate test-beds against each other: i) yields in non-shoring junctions (78%), ii) log-standard deviation (0.12-0.25), iii) value of (1.000.03 n.sub.C.sup.1), iv) electrical stability (voltage cycling for 2500 cycles), and v) ability to generate large numbers of data (600). We found that our devices exhibit very good overall performance relative to other test-beds and that our junctions are of good quality and pro-duce data that are both precise (Table 3) and replicable (relative to other EGaIn-based techniques; Table 1). Although this list is not exhaustive, we believe it serves as a good starting point to evaluate test-beds against each other.

(96) Many fabrication methods use protective layers (to protect the SAMs during the metal deposition process to form the top-contacts) that are deposited by solution based processes, or on the deposition of the electrode from solution. We believe that our method to form electrical contacts to SAMs of n-alkanethiolates can be readily extended to other types of SAMs, monolayers of biomolecules, or other types of materials that may not be compatible with direct deposition methods of metals, or exposure to solvents, to form high quality junctions in good yields with high reproducibility.

(97) The area of the electrical contact may be between 100 to 300 m.sup.2, limited by the surface tension of liquid metal. This could be overcome by using different type of liquid metal that has smaller surface tension.

(98) The mechanical stability can be improved by down-sizing the whole devices. The smaller the channels, the more solid the material behaves. Alternatively, any suitable polymers other than PDMS known to the skilled person may be used.

(99) Whilst there has been described in the foregoing description preferred embodiments of the present invention, it will be understood by those skilled in the technology concerned that many variations or modifications in details of design or construction may be made without departing from the present invention.