A Label-Free Aptamer-Based Nanogap Capacitive Biosensor with Greatly Diminished Electrode Polarization Effects

A significant impediment to the use of impedance spectroscopy in bio-sensing is the electrode polarization effect that arises from the movement of free ions to the electrode-solution interface, forming an electrical double layer (EDL). The EDL screens the dielectric response of the bulk and its large capacitance dominates the signal response at low frequency, masking information particularly relevant for biological samples, such as molecular conformation changes and DNA hybridization. The fabrication of nanogap capacitors with electrode separation less than the EDL thickness can significantly reduce electrode polarization effects and provide enormous improvement in sensitivity due to better matching of the sensing volume with the size of the target entities. We report on the fabrication of a horizontal thin-film nanogap capacitive sensor with electrode separation of 40 nm that shows almost no electrode polarization effects when measured with water and ionic buffer solutions, thereby allowing direct quantification of their relative permittivity at low frequencies. Surface modification of the electrodes with thiol-functionalized single strand DNA aptamers transforms the device into a label-free biosensor with high sensitivity and selectivity towards the detection of a specific protein. Using this approach, we have developed a biosensor for the detection of human alpha thrombin. In addition, we also examine frequency dependent permittivity measurements on high ionic strength solutions contained within the nanogap and discuss how these support recent experimental observations of large Debye lengths. A large shift in the Debye relaxation frequency to lower frequency is also found, which is consistent with water molecules being in a rigid-like state, possibly indicating the formation of an ordered “ice-like” phase. Altogether, this work highlights the need for better understanding of fluids in confined, nanoscale geometries, from which important new applications in sensing may arise. Page 1 of 24 Physical Chemistry Chemical Physics P hy si ca lC he m is tr y C he m ic al P hy si cs A cc ep te d M an us cr ip t Pu bl is he d on 0 6 D ec em be r 20 18 . D ow nl oa de d by U ni ve rs ity o f H ul l o n 12 /6 /2 01 8 2: 12 :0 5 PM . View Article Online OI: 10.1039/C8CP05510F


Introduction
Label-free electrochemical 1 and field effect transistor 2 bio-sensors have gained enormous attention in the last decade because of their low cost, portability and high potential for impact in point-of-care applications. In combination with recent advances in nanoscale engineering and surface functionalization for immobilization and detection, 3 devices have attained extraordinary levels of sensitivity with new memristor-based Si nanowire sensors recently achieving detection levels as low as 23 attomole. 4 Nowadays, a wide range of nanogap electrodes for bio-sensing applications have been investigated (see reviews 5,6 ). However, the vast majority of these are resistive based sensors. 7,8,9 Capacitive based biosensors, such as those made in this study, have been investigated much less and although they have not yet achieved the levels of sensitivity of resistive based sensors, they can provide important additional knowledge (via the frequency dependence and phase information) as well as have the same advantages of small footprint, simple instrumentation, lowcost fabrication and real-time detection. Although macroscopic sized capacitive biosensors have been used for quite some time 10 their detection capabilities have been limited because of the large electrode polarization effect, which arises from the movement of free ions to the electrode-solution interface, forming an electrical double layer. The capacitance of this layer dominates the signal at low frequency, masking relaxation effects in the bulk 11 and leading to incorrect measurement of the relative permittivity as shown in Figure 1a). True quantification of the dielectric properties of the relevant part of the material or solution has therefore relied on methods of either correcting the measurement via analytical means, 12,13 which involves the fitting of various empirical functions or modelling of the electrode polarization via long-range many-body theory, or by physical compensation of the electrode double layer itself. The latter involving the use of reference electrodes or the insertion of thin electrically insulating blocking electrodes. 14,15 However, because the electrode polarization effect is heavily dependent on the surface chemistry, electrode topography, and the involvement of both short and long-range interactions of ions at the electrode/solution interface, no simple correction techniques exists and all of these approaches pose significant flaws in the correct determination of the relative permittivity.
3 Figure 1. a) Effect of electrode polarization on measurements of the electrical relative permittivity and b) overlapping electrical double layers leads to a constant electrical potential across a nanogap device, eliminating electrode polarization effects.
Nanogap electrodes having electrode separation less than the electrical double layer (EDL) thickness, typically <100 nm, can significantly reduce electrode polarization effects since the confined region causes the electrical double layers from opposite electrodes to overlap, leading to a constant electrical potential across the nanogap region as shown in Figure 1b). This has been shown theoretically by treating the distribution of ions in the EDL in terms of the Gouy-Chapman model 16,17 and solving the Poisson-Boltzmann equation for the electrical potential. In this case, when the electrode separation, L, is less than the electrical double layer thickness, such that L~0.1 κ -1 (where κ is the Debye-Hückel parameter and κ -1 is the Debye length) the electrical potential is almost constant across the nanogap region, 18 rather than dropping to zero in the middle of the gap. The small potential difference between the electrode interface and the middle region of the gap indicates an almost negligible contribution to the measured impedance from the electrical double layer and subsequently almost complete elimination of the electrode polarization effect. This has recently been demonstrated 19 in a vertical electrode nanogap structure whereby the relative permittivity of water at 10 Hz was measured to be ε r ~ 75 in contrast to macroscopic spaced electrodes where significant electrode polarization effects instead give a value many orders of magnitude larger, ε r ~ 1.5×10 6 .
Reduction of the gap size to nanoscale lengths in impedance spectroscopy also provides enormous advantages for the detection of surface bound species, since in comparison to macroscopic sized devices the size of the immobilized surface species makes up a significant proportion of the dielectric volume, thereby minimizing unwanted signal contribution from other material in the bulk. The use of short-chain single stranded nucleic acid aptamers for surface functionalization is particularly useful in this respect, since these molecules possess very high selectivity towards only the desired target species. In recent years 20,21,22 the use of aptamers in biological sensing and the early detection of diseases has rapidly increased due to increased availability of aptamer types, better understanding of surface-immobilization techniques and advances in nanotechnology that have greatly improved device sensitivity. Aptamers are single-stranded nucleic acids (RNA, DNA) that possess unique binding capabilities in contrast to more traditional recognition elements, such as antibodies and enzymes. Aptamers have much greater stability and can be used for a wide variety of target molecules, including proteins, peptides, small organic molecules, metal ions, and even large supramolecular complexes such as cells and viruses. Aptamers are also readily produced in high concentration with specific sequence structures and are easily modified with linkers and functional groups. Thus, the combination of a nanogap impedance spectroscopy with aptamer functionalized 4 electrodes provides a powerful but simple label-free approach for the detection of target analytes with very high specificity and selectivity. The possibility to use low frequency electrical measurements, as shown in this work, may also prove to be particularly important since it could be used to detect large-scale changes in molecular structures, such as binding events and conformation changes in DNA. We should also note that the use of electrical detection methods in contrast to optical methods, which tend to require a larger footprint due to the need of both optical sources and detectors, mitigates the potential development of small, low cost, electronic handheld sensors.
In this work we demonstrate a new, simpler and low cost approach for making a nanogap device with a planar geometry and with electrodes that are metallic. This work contrasts with two previously reported nanogap biosensor devices that have significant differences in their design, use and experimental results. The first (earlier) device had a horizontal oriented nanogap 18 but with electrodes made from silicon. This work also did not use aptamer functionalized electrodes and therefore was non-specific against the target species. The second nanogap device had a vertical oriented nanogap 19 with gold electrodes and used aptamer functionalized electrodes but had a more complicated fabrication process and a more difficult geometry for connecting the nanogap capacitors in parallel. Our results importantly discriminates between the two sets of different experimental data obtained in these earlier reports, showing the importance of using metallic electrodes rather than silicon, whilst also confirming that electrode polarization effects can indeed be greatly diminished with the use of nanogap electrodes.
To demonstrate a working prototype biosensor we have chosen human alpha thrombin as a target protein as this is a commonly used standard in the field. Thrombin is a Na + activated protease enzyme that is central to the formation of blood coagulants during injury. It is not present in blood under normally circumstances but is instead generated from an inactive form, prothrombin (coagulation factor II), following an injury. During the clotting process prothrombin is cleaved to form thrombin, which clots blood through the conversion of fibrinogen into fibrin. As well as many blood related illnesses involving the improper coagulation of blood the detection and monitoring of thrombin and prothrombin is important in understanding many illnesses such as Alzheimer's disease and cancer. The detection range of thrombin in blood varies widely from pM to μM depending on the injury or disease under investigation. 23

Sensor Design
A focus of the sensor design was to improve upon the two earlier reported nanogap sensors, utilizing the advantages of both but also making the device fabrication more suited for commercialization. In contrast to the first nanogap study, 18 which utilized silicon electrodes, it was important to use a fabrication methodology that permitted the use of metallic electrodes, since this allows the use of thiol functionalized aptamers, which would give the device both very high specificity and immobilization of the target protein directly within the active sensing region. Immobilization of the aptamer and target protein allows the device to be rinsed so as to remove unbound proteins and aptamers that affect both the specificity and sensitivity of the device. We particularly note here that aptamers were not used in the first nanogap biosensor study. Instead, the bio-sensor detected only whether DNA material was present in the sensor, meaning there was no specificity. The DNA was either measured in the solution or after evaporation of the buffer solution, the latter potentially problematic since this would leave behind ionic salts from the buffer in the process. Another potential problem of this study is that silicon and polysilicon electrodes were used that is likely to have influenced the dielectric response of both the water and ionic solutions measured as their response was not flat in the low frequency region and the relative permittivity value found for water was very different to the known theoretical value. In addition to a contribution to the overall permittivity from the native silicon oxide layer, the different value measured for water may also have resulted from strong van der Waals interactions with the underlying crystalline Si atoms, which influences ordering of molecules within the first 4 nm of the liquid layer. 24 Our sensor design also focussed on simplification of the fabrication process so that it was more suitable to commercialization. A horizontal nanogap architecture utilizing metallic crossbar electrodes was chosen for simplicity in fabrication, since standard photolithography techniques can be used. This contrasts with the second reported nanogap study 19 in the literature, which instead used a vertical nanogap architecture. However, this involved a complicated device fabrication process that included additionally depositing a silicon nitride layer and a convoluted final lift-off procedure involving planarization of the Au electrodes by chemical-mechanical polishing. Instead, our methodology uses only a simple two-step photolithography method and a single HF etching step.
Furthermore, low cost glass substrates can be used and only small quantities of gold are evaporated during the process, i.e. two 50 nm thick gold evaporation in contrast to two 1 micron thick gold evaporations.

Nanofabrication
The schematic of the nanogap bio-sensor device is shown in Figure 2 and consists of top and bottom gold electrodes separated by a SiO 2 support layer. Device fabrication involved standard photolithography processes for the metal electrodes and an etching step to form the nanogap cavity.
The simple techniques involved in the device fabrication are key advantages of this method and should enable low cost fabrication of devices in relatively large quantities. Figure 3 demonstrates the main sequence of steps used in the fabrication. A highly polished flat quartz substrate was first 6 cleaned using acetone, isopropanol and de-ionized water in an ultrasonic bath. The substrate was then coated with the adhesion promoter TI prime followed by the image reversal photoresist AZ5214E (MicroChemicals) for the first photolithography step. An image reversal photoresist was used to give inclined side-walls for improved removal of the metal layer during lift-off. Metallization of the first layer electrodes consisted of 5 nm (chromium) and 50 nm (gold), deposited using an HHV Auto 500 e-beam evaporator. Following this, a 40 nm SiO 2 dielectric layer, acting as a sacrificial/support layer, was deposited by RF magnetron sputtering deposition (at 100 W). The process gas consisted of an argon:oxygen ratio of 3:1 at 1.2x10 -2 mbar with the substrate held at 300°C, to improve stoichiometry and minimize leakage current. The second photolithography step was used to form the top metal electrode. A similar method to the first photolithography step was used for this procedure. To make the nanogap, wet etching of the sacrificial SiO 2 layer was carried out using a weak solution of DI water:HF in a ratio of 10:1. Figure 4a) shows an SEM image of a typical etched nanogap with 40 nm separation between the top and bottom electrodes. The final device is shown in Figure 4b) with each device consisting of 100 capacitors in parallel. This is used to increase the overall device capacitance so that is in an acceptable range (~ pF) for ease of measurement. A protective resist, observable in the optical photo as the outer region around the square window, was also used so that only the active sensing region came into contact with the solution.

Surface Functionalization
Surface functionalization by the DNA oligonucleotide was carried out using the following process.
The thiol modified aptamer for thrombin with sequence 25 oligo 5'-GGT TGG TTT GGT TGG TTT/3- 8 (CH 2 ) 3 SH-3' (Integrated DNA Technologies) was purchased in its protected S=S form. Reduction was carried out by treating the oligo (100 μM) with DTT (10 mM). The DTT was subsequently removed by passing the mixture through a Sephadex column. The aptamers were denatured at 90°C for 5 min and left to cool to room temperature to ensure correct folding. Functionalization involved a co-immobilization process with mercaptohexanol (MCH) acting as a surface blocking-agent, as well as a spacer between aptamers, providing sufficient room to enable hybridization with the target molecules. 26 The co-immobilization process involved incubating the device with 1 μM of aptamer + forming the G quadruplex structure that is necessary for specific binding, and therefore cannot bind to the human alpha thrombin. Experiments were carried out in exactly the same manner as above except that the control oligonucleotide was used instead to functionalize the devices.

AC Impedance Measurements
To characterise the devices impedance spectroscopy measurements were taken over the range of to the accepted value of the relative permittivity of water in macroscopic systems, which almost matches the ideal Debye relaxation model and has instead the step to lower permittivity at much high frequency, typically in the vicinity of 2-20 GHz. 27,28 The possible reasons behind these differences may be related to the confined environment of the water molecules within the nanogap and the potential formation of an ice-like phase. 18,19 The discussion section examines this possibility in more detail. The measured value of the static relative permittivity of water in the nanogap, ε r ≈93, is larger than the accepted value of water, 29 which ranges from ε r =78.30 to ε r =78.58 (at 25°C). The relative permittivity value is however very close to the measured static relative permittivity value of ice (Ih) 30,31 with a hexagonal crystalline structure,  s 91.5, which lends further support to the possibility that water in the nanogap may be in a highly ordered state. This is debated further in the discussion section. We should also stress that the value measured is likely to be affected by uncertainty in the length of the undercut etch on either side of the SiO 2 support, as well as changes in surface roughness and uniformity in the nanogap thickness across the device. The presence of air bubbles in the nanogap can also not be ruled out.  Figure 5 also shows the results of the relative permittivity of different concentration SSC buffer solutions. These show many similarities to deionized water in the nanogap. The dielectric response is flat below a frequency of approximately 1000 Hz for low ionic concentration, indicating the absence of electrode polarization effects, and the transition to a lower relative permittivity occurs at a similarly low frequency and with a similar shape. The static relative permittivity of the SSC buffer is however higher than water and increases with increasing SSC concentration, rising from ε r ≈167 in the case of 0.1SSC to ε r ≈262 for 1.0SSC. This behaviour is unexpected since the relative permittivity should instead decrease with increasing ion concentration since the presence of strong internal electric fields between ions inhibits polarization of the water molecules. However, interestingly these results are in perfect agreement with a previous report of the relative permittivity of SSC in a nanogap device. 19 The possible reasons for this occurring is detailed in the discussion section.  into consideration. The black curve in Figure 6 is the relative permittivity as a function of frequency for a bare (non-functionalized) nanogap device in the presence of the 0.5xSSC buffer solution. The same device was measured (again in a 0.5xSSC buffer) but after the aptamer functionalization process (red curve). The curve indicates an overall shift to lower relative permittivity, which is likely due to a reduction in the content of the higher permittivity buffer solution within the nanogap volume, due to the presence of the aptamers on both surface electrodes. The same device was then measured (blue curve) after a 2 hour exposure of the device to a 0.5xSSC buffer solution containing the human α-thrombin (concentration of 1 μM) and following a wash cycle with just pure 0.5xSSC buffer to remove any excess unbound proteins. The results shown in Figure 6 indicate a substantial shift to lower relative permittivity across the frequency spectrum that is consistent with binding of the lower permittivity protein within device.  sequence for forming the G quadruplex structure that is necessary for binding, and therefore cannot bind to the human alpha thrombin. Experiments were carried out in exactly the same manner as above. The black curve in Figure 7 is the relative permittivity as a function of frequency for the case of a bare (non-functionalized) nanogap device in the presence of the 0.5xSSC buffer. The same device was measured, again in a 0.5xSSC buffer, but after the functionalization process with the control aptamer (red curve). The data demonstrates an overall shift to lower permittivity, similar to that observed in Figure 6, indicating successful surface functionalization. The same device was then measured (blue curve) after a 2 hour exposure of the device to a 0.5xSSC buffer solution containing the human α-thrombin (concentration of 1μM) and following a wash cycle with just pure 0.5xSSC buffer. In contrast to Figure 6, there is no shift to lower permittivity in this case, indicating the protein has not bound to the control aptamer. Interestingly there is a slight shift to higher permittivity in the higher frequency part of the spectrum. The cause of this is not yet fully understood but will be examined in future studies.  13 concentration of the ions present in the SSC buffer (Na + , Cl -, OH -, H + ) as well as those in the following reactions (using the Henderson-Hasselbach equation).
The molar concentrations of the ions are used to determine the total ionic strength of the solution using: (1) The EDL thickness ( -1 ), or Debye length (or Debye-Huckel parameter), can then be determined using: where  r is the dielectric constant (or relative permittivity),  0 is the permittivity of free space, k B is the Boltzmann constant, T is the temperature, N A is the Avogadro number, e is the elementary charge and I is the ionic strength (note: mol/m 3 ). In the case of water ( r =78) at 25C this can be simplified to: where  -1 is expressed in nanometer and I is in mol/L. The graph shown in Figure 8 depicts the typical decay of the EDL thickness with increasing ionic strength and indicates that at low 0.001xSSC concentrations the EDL layers overlap since the EDL thickness on just one of the electrodes is  20 nm, whilst in the case of buffer solutions used in typical impedance spectroscopy measurements (and in this work i.e. 0.5xSSC concentration), the EDL is very small,  1 nm, and would importantly, appear to not overlap. Implications of this in relation to the effectiveness of the device and dielectric response, which surprisingly still appears flat at low frequency, are detailed in the discussion section.

Nanogap Bio-sensing
Our results demonstrate successful operation of the device as a biosensor for the detection of small proteins with high specificity and sensitivity. Figure 6 shows that binding of human α-Thrombin to the aptamer functionalized surface leads to a 32% decrease in the measured relative permittivity at 1 Hz. In contrast, when a non-binding control aptamer was used, there was no decrease in the relative permittivity. Importantly, the figure indicates the device is most sensitive at low frequencies, which can be attributed to almost complete elimination of electrode polarization effects, which would otherwise normally swamp the device response and masking all important information in this frequency range. Low frequency detection measurement are particularly important since it can be used to better detect large-scale changes in molecules structures, such as binding events and conformation changes of DNA. In this study the reduction in the relative permittivity after binding of the protein is largely attributed to replacement of the high relative permittivity solution (0.5xSSC) in the nanogap region with the low relative permittivity protein. However, as discussed further below, the nature of the dielectric response of both water and high ionic strength solutions when contained in nanogaps is poorly understood, and this is made particularly more complex due to the presence of the aptamer and protein molecules at the electrode surfaces, whose additional effect on the electrical double layer and possible ordering of the solution molecules is unknown. In this article, we have not investigated the detection limits and sensitivity of the device, which will be reported in future work. From a theoretical perspective, we expect the ultimate detection limit is in the vicinity of the attomolar range. This is based on a total sensor area of 1600 μm 2 (active area of the 100 capacitors including top and bottom electrodes) with coating of the aptamer molecules 32 at approximately 10 4 mol/μm 2 to give the number of aptamer probes per device as 1610 6 mol. For the situation in which all of the aptamer probes have bound protein molecules the device sensitivity would be 1610 6 mol / 6.02210 23 = 3010 -18 mol.
The nanogap biosensors, once fabricated, have good repeatability (80 -90 %) between different devices and also with repeated measurements. A greater challenge is the yield of successful devices during fabrication. Currently this is only about 15 %. A number of improvements can be made to greatly increase the yield of working devices, such as improving the adhesion of the electrodes to the underlying substrate and increasing the robustness of the top electrode. Optimization of these will allow more rigorous studies of the upper limit of the device sensitivity and investigations of different aptamer/protein types as well as their use in biological relevant fluids. These will be examined in future work.

Nature of the Dielectric Response and Electrical Double Layer in Nanogap Devices
The nanogap capacitor device shows almost complete absence of electrode polarization, which indicates very good applicability of the device in bio-sensing. However, the permittivity results raise a number of interesting issues concerning the nature of the electrical double layer and the dielectric relaxation mechanisms of water and ionic solutions when confined in nanoscale dimensions.
The first issue concerns the nature of the dielectric response in the case of when only de-ionized water is present within the nanogap. The real permittivity shows a flat response at low frequency but undergoes a broad transition to low values from about 200 Hz onwards, reaching a value approaching 1 at around 100 kHz. This is very different to the accepted relative permittivity values of water in macroscopic systems, which almost match the ideal Debye relaxation model and have instead a step to lower relative permittivity at much high frequency, typically in the vicinity of 2-20 GHz. 27,28 The occurrence of the step at much lower frequency is possibly because of the inability of the water molecules' dipole moment to sufficiently polarize in synchronization with the alternating electric field. It has been previously suggested 18,19 that water in the nanogap may be in an ice-like phase, and indeed the dielectric response has many similarities with relative permittivity values of ice, which has almost the same real and imaginary permittivity features of liquid water, 33 but are shifted to lower frequency by 10 6 Hz. Ice has as many as 17 known crystalline forms, each having different frequency dependent permittivity. 34,35,36,37 A comparison of our data with frequency dependent relative permittivity measurements on pure ice 38 over a range of temperature, -57C to -0.1C, indicates that our data, although having a slightly broader transition ranging from 200 Hz to 100 kHz, is most similar to that of ice at a temperature of -20C to -0.1C. The most common phase transition is hexagonal crystalline ice (phase Ih), which occurs when liquid water is cooled to 0C at standard atmospheric pressure. The frequency response of the relative permittivity of ice Ih at 0C and it's static relative permittivity, 30,31  s 91.5, are both in agreement 33 with that of water in the nanogap device.
The question of whether water molecules can truly retain an ice-like structure over distances greater than a few atomic length scales is however still a topic of debate. Recent new experiments have given much insight to the adsorption of water monolayers on metals and the structure of the first few molecular layers. The work has challenged the commonly held theory of a bi-layer structure, 39 consisting of a buckled hexagonal overlayer resembling the basal plane of ice. However, beyond the first few layers, and especially at ambient temperatures, there is still limited understanding on the structure and dynamics of both multilayer water adsorption and liquid-water thin-films on surfaces.
In the case of the nanogap device the presence of strong electric fields within the electrical double layer must also be considered as these can easily reach up to 10 8 -10 9 V/m in the Helmholtz layer directly adjacent to the electrode. Electric fields orient the water molecules depending on its direction; a positively charged electrode orients water so that its hydrogen atoms are directed away from the surface whilst a negatively charge electrode orients the hydrogens to be instead directed toward the surface. Molecular dynamics simulations of water lamina embedded between two platinum electrodes 40 have shown that without the presence of an electric field, the first adsorbed layer has a solid-like state whilst the second and third layers are liquid but with an orientation distribution favouring a hexagonal ice formation. However, with an electric field of 210 10 V/m present, simulations have shown the water molecules reorient, aligning their dipoles along the field. A large change in the density profile is predicted and the water undergoes crystallisation process to give a cubic ice structure (phase I C ). A similar point of view was also found in another study that examined the dielectric response, 41 where it was shown the polarization underwent a non-linear transition starting from 0.110 9 V/m, whilst at higher electric fields, above 3010 9 V/m, a phase transition occurred to a completely ordered ice-like structure. Experimental verification of results in this area is challenging however, and it is likely that other effects are also important, since impurities, defects and surface roughness may stimulate or inhibit crystallization and high electric fields are known to modify electrochemical reaction rates as well as promote the ionization 42 of water. In this context, the nanogap devices could prove to be a useful probe to better understand such effects and improve the understanding of how molecules order and interact at liquid-metal interfaces. The second issue concerns the increase in the relative permittivity of the SSC buffer with increasing concentration. The higher relative permittivity indicates greater polarizability of the solution with increasing SSC concentration, which contrasts with that of macroscopic measurements on ionic solutions whereby the relative permittivity instead reduces with increasing salt concentration due to the effect of the electric field of the ions inhibiting the external electric field. 43 For example, the accepted model in the case of a simple ionic solution, such as NaCl, is that the polar water molecules align themselves with the electric fields of the dissociated sodium and chloride ions, forming a hydration shell around individual ions. The water molecules are then less able to respond to the external electric field, which has the effect of lowering the dielectric constant. In a dilute solution with (<1.5 M) the relative permittivity decreases linearly with salt concentration with the relationship 44 ε r = ε w -c, where c is the salt concentration,  is an ion specific constant and ε w is the dielectric constant of pure water. The absence of this relationship in our data, which also supports results from other studies, 18,19 is intriguing and requires further work to understand its cause.
The third area requiring further understanding is whether the electrical double layer (EDL) is really overlapping in the case of when the biosensor is used with the relatively high concentration SSC buffers. In the case of pure water it is straightforward to show the electrical double layers overlap. In contrast to classical electrolyte theories of dilute electrolytes, which indicates diminished screening at higher ionic concentrations, the measurements reported instead showed an increase in screening length with electrolyte concentration. This study has created much excitement in the field and a plethora of reports of long range screening lengths on a wide range of ionic liquids and salt solutions, 46,47,48,49 as well as attempts to explain these puzzling observations. 50,51,52,53 Although the theory is not well understood, it is apparent from these recent studies that long range double layers can possibly be present in such systems with lengths of 10 nm or more. This effect could help to explain why increasing permittivity values were found with increasing ionic strength, and the lack of correlation between the large observed changes in the measured static relative permittivity ( Figure   5) and the relatively small change in the EDL length for concentrations between 0.2SSC and 1.0SSC (Figure 8).
Lastly, we would like to emphasize the need for further work in the area so that a better understanding can be gained of one of the most important, yet unique and puzzling materials on this planet. 54 Water plays a central role in life, biology, science and technology. It dominates a variety of solvation and charge transfer phenomena. However, because of its peculiar properties imbibed from its molecular structure, which enables a strong dipole moment and weak van der Waals interaction, many of its properties are still not fully understood. The understanding of water in confined regions is hugely important for numerous fields, including biological systems (e.g. the interaction of proteins in cells), materials science (e.g. clays), chemistry (e.g. colloids) and the development of new technologies (e.g. electrolytic batteries and hydrogen fuel cells). Further studies on the origin of the dielectric relaxation in nanogap systems may help better understanding of the long running debate on the molecular origins of the Debye relaxation in water, which is a topic that is still under considerable debate. 55

Conclusions
We report on a horizontal thin-film nanogap capacitive sensor with electrode separation of 40 nm, fabricated using only standard photolithography techniques. The confined geometry of the active sensing region of the device causes overlap of the two electrical double layers when a low ionic strength solution is present within the 40 nm separated electrodes. Frequency dependent dielectric spectroscopy results indicates no electrode polarization effects are present when measured with either water or ionic buffer solutions, which greatly enhances the application of the devices since it enables low frequency measurements to be used for detection. By functionalizing the device electrodes with thiol-immobilized single strand DNA aptamers we have successfully fabricated a working nanogap biosensor targeted for the detection of human alpha thrombin. The wide range of aptamers variants available along with the simplicity of the device fabrication should enable use of the biosensor for the early detection of many diseases types. We as well discuss the interesting dielectric response of water and high ionic strength solutions contained within the nanogap device.

19
There are no conflicts of interest to declare.