Ionogel‐Electrode for the Study of Protein Tunnel Junctions under Physiologically Relevant Conditions

The study of charge transport through proteins is essential for understanding complicated electrochemical processes in biological activities while the reasons for the coexistence of tunneling and hopping phenomena in protein junctions still remain unclear. In this work, a flexible and conductive ionogel electrode is synthesized and is used as a top contact to form highly reproducible protein junctions. The junctions of proteins, including human serum albumin, cytochrome C and hemoglobin, show temperature‐independent electron tunneling characteristics when the junctions are in solid states while with a different mechanism of temperature‐dependent electron hopping when junctions are hydrated under physiologically relevant conditions. It is demonstrated that the solvent reorganization energy plays an important role in the electron‐hopping process and experimentally shown that it requires ≈100 meV for electron hopping through one heme group inside a hydrated protein molecule connected between two electrodes.


Introduction
Numerous fundamental studies and electronic applications including the pathways of electrons in biosystem, [1] electronic biomaterials, [2] and bio(semi-)conductors [3] are associated with the phenomenon of tunneling across biomolecules. In particular, the investigation of tunneling through proteins is important for us to understand the protein-associated electrochemical processes, for instance, respiration, [4] photosynthesis, [5] and biocatalytic reactions. [6] There are two types of tunneling found in protein junctions: i) the temperature-inactive coherent tunneling process, meaning that even with complex structures, the protein only works as a certain tunneling barrier between electrodes; and ii) the temperature-active incoherent tunneling that Apart from the single molecule junctions, large-area molecular junctions can provide platforms with unmodified proteins adsorbed on SAMs, but most of the time, the mechanism of tunneling dominates in the junction formed and measured in solid states. For example, Cahen and co-workers constructed protein junctions in solid states and suggested the long-range tunneling mechanism observed by measurements of azurin [16] and cytochrome C. [17] Recently, they improved their platform by using Au nanowire-electrode-protein monolayer-microelectrode junctions, and found out a very weak negative temperature-dependent current at 160-300 K. [18] Nijhuis et al. confirmed this mechanism with monolayers of ferritin using a liquid metal top electrode of EGaIn (gallium-indium alloy). [19] This long-range tunneling process happens with proteins inside solid-state devices, but whether it would happen with those proteins in their biological environment remains opaque. There are also other techniques to study the tunneling or hopping through proteins, such as photoinduced electron transfer, [20] molecular dynamic simulations of electron transfer, [21] but those approaches are either indirect electrode/protein contacts or only theoretical calculations. How electrons flow through proteins, what mechanisms the tunneling is governed by, and why proteins can conduct electrons with two different mechanisms in junctions at diverse conditions, those questions are needed to be answered urgently. Therefore, it requires a testbed that can measure electrical responses of proteins in both solid states and hydrated forms, in order to explain the coexistence of two distinct mechanisms of tunneling and hopping in protein molecules.
In this work, we polymerize a mixture of monomers in an ionic liquid to form an ionogel (abbreviated as IG) material, which has been proven to show superiorities of high thermal [22] and elec-trochemical stability, [23] relatively good ionic conductivity, [24] and remarkable mechanical properties. [25] We further add the powders of carbon black (BP2000; Cabot Corporation; abbreviated as CB) into IG polymer (abbreviated as IG-CB) to yield a rubberlike IG-CB electrode ( Figure 1A) with high conductivity, which makes it useful to form electrical contacts with monolayer of molecules. The proteins we studied are human serum albumin (HSA; electrochemically inactive), cytochrome C (Cyt C; electrochemically active with one redox center), and hemoglobin (HGB; electrochemically active with four redox centers), and we grow them on the negatively charged surfaces of SAMs ( Figure 1B). The structural analysis of proteins adsorbed on SAMs is conducted by surface spectroscopy and microscopy indicating that the proteins have been adsorbed on SAMs and packed into a homogeneous single layer. The temperature-dependent charge-transport measurements of protein junctions in air and in water, showcase the switch of transport mechanism through proteins by changes of the environment. Our IG-CB electrode for large-area protein junctions provides a neat and universal method to gain fundamental understanding of tunneling through proteins.

The Fabrication and Characterization of IG-CB Top Electrode
The ionogel was synthesized by light-initiated radical polymerization, the procedure which we followed was enlightened by others [26] (see reagents and synthesis procedure in the section of Experimental Section). The resistivity of the polymerized IG is about 5000 Ω cm measured by four-probes resistivity measurements. By grinding the conductive carbon black ( Figure 1A)  Figure S3 in the Supporting Information.) F) Plots of <log|J|> at −0.5 V versus n C for junctions incorporating SAMs on Ag TS substrates (the dashed line is replotted from ref. [29].).
into the IG polymer, we could reduce the resistivity of the bulky IG-CB material down to ≈10 Ω cm (Table S1, Supporting Information). We chose CB as the conductive dopants because: i) they are very conductive with resistivity << 2.5 Ω cm, [27] and ii) more importantly they are mechanically and chemically stable against redox or acid/base environments. [28] Next, we fabricated a thin poly(dimethylsiloxane) (PDMS) film and punched nine through-holes with each of the size of 1.76 mm in diameter ( Figure S1 and Table S2, Supporting Information) by a PDMS hole-puncher. We shaped the IG-CB materials to be cylinderlike pillars and filled them into the through-holes as shown in Figure 2A,B. Finally, we flattened the surface of IG-CB by pressing the IG-CB//PDMS film on top of an ultraflat SiO 2 /Si wafer (rms roughness < 0.2 nm), which was heated at 80°C, for 30 s. The scanning electron microscopy (SEM) images of the IG-CB surfaces ( Figure 2C) showed that the IG-CB surfaces were smooth enough for the monolayer/IG-CB contacts. For the balance between a high conductivity and low surface roughness of the IG-CB electrode, we chose 15% (in weight percentage) dopants of CB mixed with the IG polymer to fabricate our top electrodes for the charge-transport measurements (see all SEM images and resistivity of all percentage of dopants in Figure S2 and Table S1 (Supporting Information). Note that we found the IG-CB surfaces became rougher by adding more than 20% CB dopants). Finally, a layer of flexible polymer-based top electrode was ready to be stamped on the top of SAMs to form a molecular junction either in solid states or immersed in water. This simple method avoids sophisticated microfabrication processes, and allows us to measure tunneling at the molecular level in a highly stable and reproducible manner at the same time.

Measurements of Ag TS -S(CH 2 ) n−1 CH 3 //IG-CB Junctions
To verify that our IG-CB electrode can be used to conduct SAMbased charge-transport measurements, we tested the junctions in the form of Ag TS -S(CH 2 ) n−1 CH 3 //IG-CB with n = 10, 12, …, 22 (the Ag TS stands for template-stripped silver electrode) and fitted our data with simplified Simmons equation where d is the thickness of the SAM, represents for tunneling decay coefficient and J 0 (A cm −2 ) is a constant that depends on the contacts and system. We chose to measure SAMs of Ag TS -S(CH 2 ) n−1 CH 3 (abbreviated as Ag TS -SC n ) because they were the most well-studied molecular junctions worldwide for comparisons across different testbeds. We used the previously reported method to form high-quality SAMs of Ag TS -SC n on templatestripped silver surfaces. [29] Figure 2D shows all <log|J(V)|> traces from nine junctions through nine Ag TS -SC 12 /IG-CB contacts in one PDMS sheet were overlapped within one standard deviation. We conclude that the reproducibility of the IG-CB junctions is of the highest standard. Next, we recorded a statistically large number of J(V) curves (20 junctions and 480 J(V) traces for each sample, the statistics are summarized in Table S3 in the Supporting Information) to plot the <log|J(V)|> curves ( Figure 2E), the mean values <log|J|>|, and associated Gaussian log-standard deviations ( log ). We plotted the histograms of log|J| at different voltages fitted with Gaussians ( Figure S3, Supporting Information, histograms of log|J| at −0.5 V with Gaussian fits). All histograms are equally distributed with only small overlaps at the tails of Gaussians, which again indicates the high reproducibility of our method. Figure 2F shows the <log|J|> at −0.5 V against the number of carbons in chains, which is fitted by the Equation 1. The value of we obtained with the IG-CB junctions is 0.93±0.04 per carbon that is comparable to the values (0.9-1.0 per carbon) reported by other techniques, for instance, the Hg junction, [30] the conductive probe AFM junction, [31] the graphene top-electrode junction, [32] and the EGaIn junction. [29] We noticed that the values of <log|J|> measured by the IG-CB junctions (black square in Figure 2F) were similar with that measured by the EGaIn junctions (red dashed line in Figure 2F from our previous work [29] ), using the same SAMs. Although the contact area of EGaIn junctions ranged from 1000 to 4000 μm 2 that was at least three orders smaller than the contact area of IG-CB junctions, the similar values of current densities inferred that the ratio of electrical contacted molecules in SAMs by IG-CB top electrode and EGaIn top electrode was in the same magnitude. Furthermore, we measured charge transport through SAMs of Ag TS -SC 12 in top contact with IG-CB electrodes containing four different sizes of throughholes, and the current densities remained the same ( Figure S4, Supporting Information). We also performed the measurements of current densities as a function of scan rate ( Figure S5, Supporting Information) to confirm the values of measured current were dominated by charge, not ions, transport, which is important for our further discussions on the mechanisms of protein junctions.
In conclusion, this IG-CB electrode can provide charge-transport information at the molecular level as other testbeds can, and later we show this electrode can allow us to perform SAM-based junctions in water with high stability and reproducibility for the first time while other testbeds cannot. Before we come to discuss the results of protein junctions, the Figure 3 (all histograms in Figure S6 in the Supporting Information) and Figure S7 (Supporting Information) shows the comparison of the charge-transport data in air and in water obtained from the junctions of SAMs of HSC n , HSC n COOH and HSC 11 SO 3 − Na + . The <log|J(V)|> traces remain very similar between the dataset in air and in water, which indicates that there is not an influential Faradaic leakage-current with our junctions in water.

The Formation of Protein Layers on SAMs
Direct adsorptions of proteins on the bare surface of the electrodes, such as Au, Ag, Hg, Pt, etc., will change the conformation of proteins or even lead to denaturation due to the high surface energy of the metals. [33] On the other hand, physical adsorptions of proteins on the charged surfaces of SAMs can form homogenous layers of proteins and largely prevent the possibility of denaturation. [34] Therefore, we formed the layer of proteins www.advancedsciencenews.com www.advmat.de supported by SAMs for the investigation of charge-transport measurements. The three proteins we chose are positively charged at part of their surfaces, which can be adsorbed via electrostatic forces on the top of the SAMs that are composed of negatively charged terminal groups, and the chemical form of the SAMs are Ag TS -SC 10 COOH and Ag TS -SC 11 SO 3 − Na + (see the synthesis of molecules in Figures S8−S14 in the Supporting Information). We chose these three proteins because: i) they are important proteins in the metabolism of living bodies involving charge transfer; [35] ii) they all are considered as "small" proteins in size that have high charge-to-area ratio benefiting for physical adsorptions to form uniformly packed monolayer; and iii) HSA, Cyt C, and HGB contain the heme groups with the number of zero, one and four, respectively, which will benefit us in deeply investigating the role of redox center in the protein electrical junctions. The preparation of the samples of SAMs//proteins followed by previously reported methods. [36] Briefly, we immersed the chip with a SAM of Ag TS -SC 10 COOH (or -SC 11 SO 3 − Na + ) in a buffer solution containing proteins for 30 min (see the details in the Experimental Section). We have considered the different isoelectric points of the three proteins, and the concentration of proteins and the pH values of buffer solutions were the optimized conditions for the growth of densely packed protein layer based on the previous studies. [37] Before the electrical measurements, the chip was picked up from the solution with tweezers and then rinsed with buffer to remove loosely adsorbed proteins. For solid-state experiments, we gently blew dry the chip after rinsing, and for non-solid-state experiments, we immediately immersed the chip in water or corresponding buffer solution in preparing protein SAMs (see in the Experimental Section) after rinsing.
Although adsorptions of proteins on the surface of SAMs by electrostatic forces have been proven to be a successful method to grow protein monolayer, [38] we performed the X-ray photoelectron spectroscopy (XPS), surface plasmon resonance (SPR), and atomic force microscopy (AFM) to ensure we have formed the protein layer with high quality. The high-resolution XPS spectra of N 1s are shown in Figure 4A,B, and we do not observe the signal of N 1s before the adsorption of proteins, while after adsorption, the signal of N 1s appeared significantly, which indicates the successful adsorption of proteins on SAMs. The peak of N 1s is assigned to the amido bond and other residues containing nitrogen such as tryptophan and histidine. The HGB shows the highest intensity of N 1s because this protein contains the most porphyrin rings among HSA, Cyt C, and HGB. The C 1s signals, shown in Figure 4C,D, show two extra peaks at binding energy of 286.9 and 288.6 eV from samples adsorbed with proteins compared to pure SAMs which are assigned to the chemical environments of C-N and C-O (see Figure 4E for the examples of peak assignments of C 1s signal) originated from proteins. Based on the analysis of N 1s and C 1s spectra, we confirmed the three proteins, HSA, Cyt C, and HGB, can all be adsorbed on SAMs. In addition, we calculated the elemental ratios of N 1s /C 1s shown in the Figure 4F from the XPS data, in which the dashed line is a literature value of the universal N/C ratio of a protein. [39] Figure 4F shows that the ratios of N/C from each protein sample are slightly less than the universal N/C ratio, which is due to the contribution of increased amounts of C from SAMs. The adsorption strength of each protein is mainly due to the difference in electric charge, hydrophobicity and surface energetics among proteins. [40] The survey spectra and high-resolution spectra of O 1s and S 2p are given in Figures S15 and S16 (Supporting Information), and they are in line with our interpretations from N 1s and C 1s spectra. Figure 4G-I shows AFM images of HSA, Cyt C, and HGB adsorbed on SAMs. The presence of proteins on top of SAMs was confirmed by visual observation, and the height profiles showed the average height of each protein was in between the molecular long-axis and short-axis of proteins from the protein data bank (PDB) database ( Figure S17 and Table S4, Supporting Information) and in good agreement with the previous studies, [36,41] which implied a monolayer of proteins formed on SAMs. Although we performed the AFM measurements in air, there are abundant studies [42] showing that the tightly bonded water molecules can maintain the structures of HSA, Cyt C, and HGB in solid states. The results from SPR are shown in Figure S18 (Supporting Information) which qualitatively confirmed the protein layer adsorbed on SAMs. All characterization results confirmed the high-quality monolayers of proteins have been formed on SAMs.

The Comparison between Junctions of Proteins in Solid States and in Water
After successful adsorption of the protein layer on SAMs, we formed and measured the junctions in the forms of Ag TS -SAMs/proteins//IG-CB in either air or water, and we analyzed the J(V) data in the same way we described for n-alkanthiolates SAMs. Figure 5 shows the <log|J(V)|> traces of three proteins in solid states (data in black) or in water (data in red), and the histograms of log|J| are shown in Figure S19 (Supporting Information) and the statistics are summarized in Table S5 in the Supporting Information. The revisability of our testing data is also verified shown in Figure S20 (Supporting Information), which is a plot of 20 continuous J(V) traces of one protein junction to show there is no changes inside junctions induced by the sweeps of bias. From the transport data, we make four major observations: i) The values of <log|J|> from proteins are lower than that of SC 10 COOH and SC 11 SO 3 − Na + SAM, which indicates the protein serves as an extra charge-transport barrier in addition to the SAMs. ii) The values of <log|J|> through protein junctions in solid states are consistently larger than those in water and the conductance calculated from low bias can confirm this conclusion ( Figure S21, Supporting Information), which may be induced by the different mechanisms of charge transport: tunneling in solid states and hopping in water. Another possible reason is that the capacitive coupling strength between the ionogel electrode and proteins at solid states is stronger than the protein junctions in water. When the protein junctions are in solid states, the ionogel electrode is directly in contact with the surface of the protein; on the other hand, when the protein junctions are immersed in water, some water molecules may sneak in between the surfaces of the ionogel and proteins that may weaken the coupling. This also can cause favorable direct tunneling in solid states become impossible in water environment. Therefore, we performed temperature-dependent measurements and discussed the results in the next paragraph. iii) The curvature of <log|J(V)|> traces and values of <log|J|> for all junctions in solid states are very similar which indicates that the monolayer of three different proteins served as similar tunneling barriers for electrons tunneling through. iv) The curvature of <log|J(V)|> traces for junctions in water differs from them in junctions of solid states. Since water is a polar liquid, which should be a better conductor that possibly results in increases in the measured J but we observe a reduction of J. The results in Figure 3 show almost identical <log|J(V)|> measured with junctions of alkanethiolates in solid states and in water, and these results suggest that water may not be directly involved in tunneling measurements. Therefore, the observed differences in the shape and values of <log|J(V)|> between junctions in air and in water, are likely caused by the difference in tunneling mechanisms through dehydrated proteins and hydrated proteins in a physiologically relevant microenvironment and we continue to discuss it in the following paragraph.

Temperature-Dependent Measurements
In order to investigate the mechanisms of tunneling through proteins, we measured the protein junctions in situ with heating the bottom electrodes from room temperature (≈25°C) to ≈50°C at maximum (the temperature higher than 50°C will rapidly denature the proteins in this study [43] ) monitored by a thermocouple immersed in water. We tested the thermal stability of the IG-CB electrodes and Ag TS electrodes under different temperatures with junctions of Ag TS -SC 12 SAMs (see <ln|J|>-T −1 curves in Figure S22 in the Supporting Information) in water and concluded that the contact of the junctions was stable under different temperatures. The temperature-independent tunneling behaviors of the junctions of Ag TS -SC 12 SAMs were observed as expected. In sharp contrast to the junctions of Ag TS -SC 12 SAMs, Figure 6 shows the protein junctions switched from temperature independent tunneling process in the air to temperaturedependent hopping process in water. To fit the data obtained from the temperature-dependent measurements, we used the Arrhenius equation in integration form as follows where k B is the Boltzmann constant (k B = 8.62 × 10 −5 eV K −1 ), J 0 is the pre-exponential factor, and E a is the activation energy. As shown in Figure 6, the values of <ln|J|> at ±1.0 V from junctions in solid states are temperature independent, while those from junctions in water are temperature dependent with different values of E a for different proteins. Obviously, the mech-anisms of tunneling through proteins can be switched by environmental conditions. The values of E a at −1.0 and +1.0 V for the same protein are similar, and the values of E a of Cyt C are in similar magnitude compared with other reported values by measurements from electrochemistry or theoretical calculations. [44] Figure 6A,D shows that, although HSA does not contain a heme group, the HSA junctions in water still generate an E a of 219−244 meV, which commonly is not observed by electrochemical measurement. In molecular tunnel junctions, weak temperaturedependent hopping process can be observed, where charge carriers couple to vibrational modes of the molecules. [45] The theory also predicts the surrounding solvent molecules will reorganize when charges hop to a protein molecule. [42e,46] Therefore, we believe, when electrons pass through HSA, they can couple to the vibrational modes of peptide chains and alter the distribution of charges at HSA, which in turn reorganizes the surrounding dipoles, such as water molecules and ions, and thus we observed an apparent E a . [47] The values of E a for three proteins show the trend of HSA < Cyt C < HGB, and this trend can be correlated to the number of heme groups they containing, which indicates that the hopping of charge to heme group may play a vital role in the hopping through hydrated proteins. Therefore, we plotted Figure 7 to explain the process of hopping through proteins. The values of E a we obtained could be contributed by many factors, such as the nature of proteins, the charge around the microenvironment of proteins, the geometry of junctions etc. The Figure 7A shows the plot of E a at ±1.0 V against N H (N H : number of heme groups in one protein). Clearly, E a has a linear relationship with N H , in which we obtained the slope of 105.6 meV/N H at −1.0 V and 130.8 meV/N H at +1.0 V, which represents how much activation energy per heme group needs for the hopping process happening when the charge hops at −1.0 V and at +1.0 V. The Figure 7B (see <ln|J|>-T -1 plots at different applied voltages in Figure S23 in the Supporting Information) shows a weak voltage dependence of E a with junctions of HSA, but for the junctions of Cyt C and HGB, the values of E a increase with applied bias above a threshold value of −0.6 V and below −0.6 V the values of E a are weakly dependent on applied voltages. According to Marcus theory, the extent of the dependence of E a on applied voltages is proportional to the charge in the molecule. [48] In our case, the HSA does not contain a redox center which makes it cannot be strongly charged with increased bias, so we observe a near steady E a at around 200 meV as a function of V, which would be mainly from the reorganization of charges and dipoles surrounding the proteins. In the cases of the Cyt C and HGB proteins, we believe the charge, are likely to hop to heme groups, as the redox centers inside proteins with applied voltage above −0.6 V. The highest occupied molecular orbital (HOMO) of the heme group was reported to be −4.7 to −4.9 eV relative to the energy level of vacuum, [49] which is about 0.7-0.9 eV difference from the work function (WF) of our samples (see Figure S24 of the WF data in the Supporting Information). Considering the potential drop of 0.1-0.2 eV at the van der Waals's contact of the protein/electrode interface, this energy difference between HOMO of heme and WF of electrode is in excellent agreement with the data in Figure 7B and confirms the transition of the hopping process from the charge hopping to heme groups involved at V > −0.6 V, to the hopping process only involving vibrational mode of peptides at V < −0.6 V. Shipps et al. observed a switch of contact resistance at high bias in amyloid proteins due to a barrier bending effect similarly. [50] This analysis can provide us with a powerful tool to study the hopping process through proteins, and we schematically draw our understandings of the mechanism of hopping in Figure 7C,D to illustrate the hopping process through a protein in our junction setup.
To demonstrate our junction measurements can be conducted in PBS buffer solution, we measured the junctions of SAMs and proteins in 1× PBS buffer (Figure 8). We observed very similar <log|J(V)|> traces in values and shapes compared with junctions measured in water. This demonstration showcases we can reproduce physiologically relevant conditions for proteins to study the charge-transport phenomenon.

Conclusion
We have shown a new type of top electrode using ionogel that enables us to conduct stable and reproducible SAM-based  molecular tunneling junctions in solid states and in water environment. The junctions with proteins of HSA, Cyt C, and HGB demonstrated transitions of mechanisms between tunneling and hopping by changes of the environment. We experimentally explained the apparent E a , obtained by Arrhenius fits to temperature-dependent data, is partially contributed by the charge hopping of redox centers, which is strongly dependent on the applied external voltage and the number of redox centers inside one protein. Another part of the activation energy is contributed by the reorganization energy of the surrounding solvent molecules. For ionogel electrode, the ability to carry out measurements with the junction formed and immersed in water would make it possible to study a wide range of interesting biologically relevant systems. Moreover, its outstanding advantages like high www.advancedsciencenews.com www.advmat.de conductivity, low cost and easy fabrication shed light on the realizations of (bio)molecular electronic devices.

Experimental Section
Fabrication of IG-CB Materials: Ionogel was synthesized by lightinduced radical polymerization. First, we dissolved the monomer of acrylic acid (AA, 90%, 0.1 mL, 1.46 mmol) without further purification in an ionic liquid ([C 2 mim][EtSO 4 ], 1.46 mL) at a concentration of 1.00 mol L −1 . Then, poly(ethylene glycol) diacrylate (PEGDA, M.W.400, 0.0035 g, 8.75 μmol) was added at 0.60 mol% of AA as the cross-linker, andketoglutaric acid (0.0021 g, 14.6 μmol) at 1.00 mol% of AA as the initiator. The mixed solution was stirred at 800 rpm for 30 min under room temperature. After stirring, the solution was transferred to a 20.0 mL transparent glass vial, and the vial was placed in an ultraviolet exposure chamber (SCIENTZ 03-II, Ningbo Scientz Biotechnology Co., LTD) for UVlight-induced cross-linking. After 1 h of radiation by UV lamp with power of 50 W and wavelength of 365 nm, the solution turned into a stretchable and transparent film of poly(acrylic acid) (PAA) ionogel. In order to increase the conductivity, 15 wt% conductive carbon powders (BP2000, 0.2 g) were added to this ionogel and ground continuously. After grinding for 20 min, the final IG-CB electrodes were obtained and the surface was hydrophilic. The hydrophobic surface of the IG-CB can be achieved by using the monomer of ethyl acrylate (0.1 mL, 0.92 mmol), ionic liquid of [C 2 mim][NTf 2 ] (0.92 mL), cross-linking agent (0.0022 g, 5.52 μmol), and initiator of 1-hydroxycyclohexyl phenyl ketone (7.00 mol%, 0.0132 g, 64.4 μmol) with the same synthesis procedure. Both hydrophilic and hydrophobic IG-CB electrodes can be used for SAM-based molecular junctions in solid states with generating the same J(V) results ( Figure S25, Supporting Information). The hydrophobic IG-CB electrodes were used for junction measurements in water because water could swell the hydrophilic IG-CB electrodes slowly.
Fabrication of Top Electrodes: First, the mold of a polymer sheet made of the poly(dimethylsiloxane) (PDMS) with filling of IG-CB electrode. The curing agent was added into the monomer at 10 wt%. Then the mixture was stirred evenly and turned to vacuum to remove bubbles. Next, the mixture was transferred into an oven at 80°C and cured for 1 h. After cure, The PDMS was cut into a 0.5 cm × 0.5 cm square sheet, and then this PDMS sheet was punched by a PDMS puncher (Model: WH-CF-13, Wenhao Co., Ltd., Suzhou, China) to create nine holes with a diameter of 1.76 mm size. The holes were filled with IG-CB. After removing the excess parts of IG-CB by a scalpel, the whole PDMS sheet was placed on a smooth silicon wafer which was sitting on the heating table and heated to 80°C. The PDMS sheet was pressed for 30 s to flatten the bottom of IG-CB. Finally, a top electrode with a flat bottom surface of IG-CB was obtained.
Adsorptions of Protein Layer: The template-stripped Ag surfaces were made following the reported process. [51] After cleaving the Ag TS substrates from the silicon wafer, they were immersed into 1.0 × 10 −3 m ethanolic solutions of SC 10 COOH or SC 11 SO 3 − Na + for 3 h at room temperature. Then by a gentle flow of ethanol they were rinsed and dried in a steam of N 2 gas. Next, they were immersed into solution of protein for 30 min, in which for Cyt C and HSA the buffer is 0.1 × 10 −3 m PBS while for HGB it is 0.1 × 10 −3 m phosphate buffer with pH = 5.8. The concentration of proteins was 20.00 mg mL −1 for HSA, 15.00 mg mL −1 for Cyt C and 0.25 mg mL −1 for HGB. The different concentration of protein solutions were in consideration of the formation of densely packed monolayer. Cyt C and HSA layer were formed according to the concentration from the reference. [34a] It was, however, experimentally found that HGB cannot form a monolayer at the concentration more than 1.0 mg mL −1 ( Figure S26, Supporting Information). Therefore, the concentration of HGB solution was optimized to 0.25 mg mL −1 for a best packing of HGB monolayer. The protein samples were rinsed with buffer right before the characterizations or junction measurements.
X-ray Photoelectron Spectroscopy: X-ray photoelectron spectroscopy (XPS) was conducted with the NCESBJ (National Center of Electron Spectroscopy in Beijing). The energy of the incident X-ray beam (1486.6 eV) was used by the Thermo Scientific K-Alpha XPS system. All measurements were performed in an ultrahigh vacuum chamber with a base pressure of 1 × 10 −8 Pa. Survey spectra and high-resolution spectra of C 1s , S 2p , N 1s , O 1s ,Au 4f were recorded.
Atomic-Force Microscopy: The AFM images were recorded by a cypher S Oxford instrument from Asylum Research with tapping mode tips (AC200TS-R3, resonant frequency: 150 kHz, force constant: 200 N m −1 ). The AFM software Gwyddion Analysis (version 2.59) was used to obtain the height profiles of protein layers.
Surface Plasmon Resonance Spectroscopy: SPR was performed on a Biacore 8K plus instrument. The substrate containing the SAM to be analyzed was mounted in a SPR cartridge. The SPR protocol for measuring the adsorption of protein to SAMs consisted of the following: i) flowing 0.01× PBS for 120 s, then replacing the flow of buffer with a flow of a solution of Cyt C (1.00 mg mL −1 in 0.01× PBS) for 1200 s, and finally injecting 0.01× PBS for an additional 600 s; and ii) flowing a solution of sodium dodecyl sulfate (40 × 10 −3 m in 0.01× PBS) over the SAM surface for 30 s to reproduce followed by rinsing the surface with 0.01× PBS for 5 min. The flow rate used for all experiments was 10 μL min -1 .
Ultraviolet Photoelectron Spectroscopy: Ultraviolet photoelectron spectroscopy (UPS) was carried out at the NCESBJ (National Center of Electron Spectroscopy in Beijing). All measurements were performed in an ultrahigh vacuum chamber with a base pressure of 1 × 10 −8 Pa. All UPS spectra were referred to the Fermi edge of Au. To probe the valence band, the photon energy at 21.22 eV was used and −10 V bias was applied to the sample to overcome the work function of the analyzer.
Formation of Protein Junction: The chip was placed inside a glass cell, where deionized water was added slowly into the cell until submerging the surface of the chip. Next, the IG-CB top-electrode was brought in contact with the surface of the chip and ensured the water would not immerse the top of the electrode which was wired to the electrometer. Finally the junction was formed by placing a probe, as the grounding electrode, penetrating the protein/SAM layer and in contact with the Ag bottom electrode.
Temperature-Dependence Measurements: For solid-state junctions, a polyimide (PI) film embedded with multiple heating resistors was used to heat the Ag TS substrate to increase temperature (ΔT = 0-25 K) from room temperature. The thermocouple associated with a voltage controller to control the heating was contacted with the bottom electrode to control the temperature of the junctions. Meanwhile, for the junctions in water, two PI films were placed under the cell and the thermocouple was added into water to detect the temperature of water. The temperature difference could also reach to the same level as the junctions in solid states. Under each temperature, 10 traces were recorded of the junction and the current density shown in the Figure 6 was averaged from the 10 traces.
Statistical Analysis: Junctions containing each type of SAM were formed on three to four different substrates. Six to eight junctions were formed on each substrate. For each type of SAM the J(V) characteristics were recorded of 20 junctions (see Tables S3 and S5 in the Supporting  Information). Each J(V) trace was recorded from 0 V → +0.5 V → 0 V →−0.5 V → 0 V with a step size of 50 mV and a delay of 0.1 s for SAM junctions and from 0 V → +1.0 V → 0 V →−1.0 V → 0 V with a step size of 50 mV and a delay of 0.1 s for protein junctions. The molecular junctions were very stable over 7000 scans ( Figure S27, Supporting Information) within the applied bias with all junctions in 100% yield (Tables S3 and S5, Supporting Information).

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.