The role of polarization-induced reorientation of DNA strands on organic field-effect transistor-based biosensors sensitivity at high ionic strength Free

The possibility of an electronic readout of chemical and biological reactions has attracted a rising interest in the last decades as a suitable alternative to classical analytical techniques. In particular, field-effect transistor-based biosensors (bioFETs) have been widely studied for detecting and transducing several kinds of biological reactions related to a variation of the charge and/or charge distribution in the close proximity of the sensing area of the device.

By definition, a bioFET consists of a field-effect transistor integrated with a layer of electrically active biomolecules that act as receptors. Several transduction mechanisms can be implemented according to the way biomolecules are integrated in the device and to the employed technology. Starting from the basic working principle of the Ion-Sensitive FET (ISFET,¹), several examples of bioFETs in silicon technology have been proposed.² More recently, several examples of bioFETs fabricated with the organic technology have also been proposed.³ Organic technology allows the fabrication of devices on large areas with relatively low costs, thus resulting particularly promising for the fabrication of disposable biosensors on flexible substrates as plastic and paper. Moreover, in addition to the ISFET working principle,^4,5 different transduction mechanisms can be implemented. For instance, the direct functionalization of the organic semiconductor with DNA single strands has been widely employed as a sensing strategy for DNA hybridization detection in dry^6,7 and in wet conditions.^8,9

Independently of the employed technology and the implemented transduction mechanism, the performances of bioFETs are strictly related to the features of the receptors at the nanoscale. Biological molecules chemically grafted onto a surface form Self-Assembled Monolayers (SAMs), whose properties are strongly influenced by parameters as density, chain length, orientation, and tilt angle with respect to the normal of the surface; as a consequence, all these features play a significant role in the transduction mechanism of the device. Moreover, all these characteristics have to be considered in relation with the environment where the tests are carried out, especially for operations in liquids, that is a typical condition for the majority of biological reactions. More in detail, the detection ability of field-effect devices is heavily affected by the ionic strength of the measurement environment:¹⁰ the higher is the ionic strength, the larger is the screening effect on the intrinsic charge of the molecules by the free ions in solution. The screening effect is generally described in terms of Debye length, κ⁻¹, given by

where ε₀ is the electrical permittivity of the vacuum, ε_r is the relative dielectric constant of the solution, k_B is the Boltzmann constant, T is the absolute temperature, N_A is the Avogadro number, e is the elementary charge, and I is the ionic strength of the solution. According to the definition, only the molecular charge located within the Debye length from the sensing surface of the device can contribute to the field-effect modulation, while the charge that lies beyond the Debye length is completely screened. As a consequence, the charge associated with large macromolecules can be electrically detected only in very low ionic strength solutions, i.e., in conditions that are very different from those of reactions occurring in vivo. This is therefore considered as a strong limitation to the potential of these devices for a reliable detection. As an example, for a 1:1 electrolyte with sodium chloride, Equation (1) (Ref. 11) can be approximated by

Thus, for a 1 M NaCl electrolyte, the Debye length is 0.304 nm, which is approximately the distance between adjacent bases in a DNA chain. As a consequence, only the first bases of hybridized DNA are unscreened and only their charge could be detected by a field-effect device. As DNA hybridization typically occurs in solutions with salt concentration between 100 mM and 1 M,^12–14 the employment of bioFETs for hybridization detection at high ionic strengths seems not feasible. On the other hand, lowering the ionic strength of the measurement solution (after hybridization has occurred) is not advisable as, in these conditions, a denaturation of the hybridized double-strands may occur.¹⁵ 

In the recent past, we proposed a device structure for DNA hybridization sensing based on floating gate transistors, which can be implemented both in CMOS (Charge-Modulated FET, CMFET¹⁶) and organic technology (Organic CMFET, OCMFET^17,18). The structure of the OCMFET is based on a floating gate structure; a part of the floating gate, namely, the sensing area, is used for detection in liquid environment without needing of a reference electrode in solution.¹⁹ In particular, in Lai et al.,¹⁸ a new implementation of the device capable to operate at low voltages was presented; such a feature was obtained thanks to an innovative, hybrid organic-inorganic insulating film which can be fabricated with an highly reliable process.²⁰ The transduction mechanism of the OCMFET is based on the modulation of the floating gate voltage due to the variation of the charge related to biomolecules anchored onto the sensing area. As thoroughly explained elsewhere,²¹ this turns into a modulation of the threshold voltage ΔV_TH of the device, proportionally to the charge on the sensing area Q_S according to the relationship

where C_SUM is the sum of all the capacitances in the device. The shift of the threshold voltage is recorded both after the anchoring of single-stranded DNA probes (functionalization) and after their hybridization with fully complementary target sequences, as a consequence of the negative charge of DNA backbones.

Such device, though based on a field-effect transistor, reproducibly showed record values of sensitivity and selectivity at relatively high ionic strengths of the measurement buffer ([NaCl] = 50 mM), while, according to Equation (2), the effect of the hybridization should be, in this case, completely screened. These results suggest that the relationship between the device working state, the DNA strand's features at the nanoscale and the ionic strength of the measurement liquid must be considered in a more exhaustive way.

The aim of the present paper is therefore to shed light upon the mechanisms that give rise to these characteristics, enhancing the relation between the “macro” properties derived by the electrical characterization of the device, and the micro- and nano-features of the receptor layer that allow achieving the desired chemo-sensitivity. As it will be shown in the following, the recorded sensitivity seems to be motivated by the effect of device polarization on the orientation of DNA single strands immobilized on the sensing surface of the device. Such a phenomenon was thoroughly investigated by Rant and co-workers^22,23 through the direct application of a voltage drop between a functionalized electrode and the solution. By switching the voltage, a persistent tilting of the oligonucleotides was obtained; in addition, the distribution of ions inside the solution, and so the screening length, was affected by a repulsion/attraction mechanism according to the sign of the voltage and the charge of the ions. The combinations of these effects lead to an increase of the effective Debye length on one hand, and to an increase of the portion of the molecule included within the Debye length on the other hand. Such effects may also be invoked for explaining the results obtained with the OCMFET; however, in this case, the bias effect on the receptor orientation is indirect as both the sensing area and the solution are floating. Thus, a direct evaluation of the effective voltage drop between them is impossible, thus making tests and considerations mandatory. In the following, a key-experiment is described and discussed.

At first, the effect of the sole ionic strength on the sensor response was considered. In order to decouple the effect of hybridization efficiency from the effect of ionic strength on the effective Debye length, the tests were carried out on the thiol-modified, single-stranded DNA (HS-ssDNA) probes onto the sensing area. Indeed, the ionic strength of the hybridization solution strongly affects the efficiency of the hybridization process. Thus, by varying the ionic strength in an experiment on double-stranded DNA (dsDNA), a variation in the sensor response could be observed as an effect of the hybridization efficiency variation rather than, purely, to a variation of the Debye length. A 24-base long probe sequence was employed, P₂₄ = HS 5′-(T)₆-GGT TTC CGC CCC TTA GTG-3′;¹⁹ beyond this length, ssDNA molecules typically starts to fold.²⁴ 

The functionalization was verified by extracting the threshold voltage variation from the transfer characteristic curves taken before and after the process, according to a basic model developed by Braga and Horowitz.²⁵ In particular, differential measurements were performed: a reference (i.e., not functionalized) device underwent to the same biological and chemical treatments and to the same storing and measurement conditions of the sensor, in order to evaluate and compensate any unspecific threshold voltage variation in the sensor response. The Debye length of the measurement buffer was modulated by employing different NaCl concentration ([NaCl] = 50 mM, 10 mM, 1 mM); consequently, estimated Debye lengths κ⁻¹ of 1.36 nm, 3.04 nm, 9.62 nm, respectively, were obtained.

The length of P₂₄ is about 7.9 nm, so, accordingly to the estimated Debye length in the experimental conditions, the sensor response has been recorded when the probe should be almost completely screened ([NaCl] = 50 mM), when almost one half is unscreened ([NaCl] = 10 mM), and when it is totally unscreened ([NaCl] = 1 mM). This scenario is described in the cartoon in Figure 1(a). In Figure 1(b), the net variation of the threshold voltage recorded in the different salt concentrations is reported. In particular, the values have been obtained by averaging upon three different samples for each value of the salt concentration. It is possible to notice that the net threshold voltage variation due to the electrical charge of the probes decreases as the salt concentration increases, as expected, thus demonstrating that a coherent response of the sensor can be obtained in “quasi-static” conditions.

Subsequently, in order to further explore the mechanisms that give rise to the observed sensitivity of the device in hybridization detection, the tests carried out in Ref. 18 were here repeated by modulating the probe length for a fixed Debye length in the measurement buffer. In particular, three probes were employed: P₂₄, P_31, and P₄₂. The only difference between the sequences was the initial timing spacer, which is, respectively, 6, 13, and 24 bases long, thus setting the distance between the first base of the hybridized segment at 1.98 nm, 4.3 nm, and 7.92 nm from the anchoring point of the DNA chain, respectively. The P₃₁ sequence was the same employed in the original tests reported in Ref. 18; consequently, P₂₄ and P₄₂ allow considering a lower and a higher hybridization distance. Hybridization was performed in ambient conditions and at room temperature.¹⁹ The electrical response to the hybridization was measured by monitoring in real-time the output current of the sensors. In the inset of Figure 2, the measurement setup is shown: a constant source-to-drain voltage drop V_DS was set, while the control gate was biased with a square-wave voltage V_GS in order to reduce the bias stress effect on the organic semiconductor that could lead to a continuous current reduction possibly masking the expected current increase related to the hybridization effect. The amplitude of the gate voltage was chosen in the different devices (that may have different threshold voltages) in order to operate them in the same over threshold conditions. V_GS frequencies from 10 Hz to 100 Hz were chosen. The efficiency of the hybridization was evaluated by measuring the relative variation in the output current, ΔI_DS/I_DS,baseline, where I_DS,baseline is the output current before the spotting of the target sequences. The results are reported in Figure 2(a). By reducing the length of the spacer, hybridization can be better detected by the sensor, as expected. Interestingly, the hybridization signal is almost completely lost (ΔI_DS/I_DS,baseline of about 0.1%) only with the P₄₂ chain, i.e., with a 24-base long spacer. According to Equation (2), with the ionic strength chosen for this experiment, the expected Debye length should be 1.4 nm; therefore, the screening effect of the ions in solution should affect the device response for any chain length, not only for P₄₂. As this did not happen, we must assume that the dsDNA molecules are not lying perpendicularly to the surface but are tilted to a certain extent such that, at least for P₂₄ and P₃₁, a certain portion of the molecule is included within the Debye length; a plausible scenario is shown in Figure 2(b).

Tilt angles of 30° have been reported for the DNA molecules in the monolayer formed on gold coated silicon with a similar biochemical procedure.²⁶ According to the obtained results, the tilting should be substantially higher than this value (about 80°, estimated as the arc sine of the ratio between the Debye length and the spacer length). Such large difference with the previously extracted value of the tilt angle could be partially explained with the roughness of the sensing area, which is larger than that reported for gold on silicon as the device is entirely fabricated on plastic.²⁷ In addition, it could also be an effect of the polarization of the sensing area on the DNA chains, as suggested in Ref. 22. Moreover, as in this case a redistribution of the screening ions induced by the charged surface may occur, a larger Debye length can be assumed, and an actual tilt angle smaller than estimated could be consequently derived.

A final, key-experiment was then done, in order to prove the possible effect of the device polarization on the specific tilt/repulsion effect observed in the previous experiments. Variations in the DNA strands' tilt angle due to the device's polarization have been investigated by means of fluorescence quenching tests. P₃₁ ssDNA strands modified with Cyanine-3 fluorescent dyes were employed in the functionalization process; the chains were later hybridized using the complementary sequence according to the previously reported procedures. Quenching was evaluated by analyzing the photographs of the sensing area taken by means of a digital camera directly connected to the fluorescence microscope. As for the hybridization measurements previously presented, V_GS was switched from 0 to −2 V using a waveform generator. Several seconds are needed to observe a switch of DNA in the experiments reported in Ref. 22; in our measurements, a period of 60 s was waited in order to allow a complete reorientation of the molecules after switching the potential. This slow response is probably related to the fact that the voltage drop is not directly imposed, thus being affected by the device dynamic. By evaluating red (R), green (G), and blue (B) levels obtained by an average on the whole picture by a custom-made Matlab^® script, the luminance of the image was calculated

The results are reported in Figure 3: it is possible to observe that, as a response to the negative voltages applied to the control capacitor and to the drain, a reduction of the luminance was obtained with respect to the zero-voltage condition. Moreover, when the device was grounded again, the original luminance was perfectly restored. In Figure 3(b), the wavelengths in the emission band of the Cy-3 dye (570–650 nm), which comprises red, yellow (Y), and magenta (M) colors, are separately analyzed and compared with the green and blue levels, which, on the contrary, lie outside the emission band. It is clearly noticeable that the luminance variation is mainly related to the variation of RYM levels during the application of the polarization, while the G and B levels are almost constant. These results prove a fluorescence quenching, which occurs when Cy-3 dyes are close to a metal surface that absorbs part of their energy, thus leading to a reduction of the fluorescence. Consequently, the observed quenching can be reasonably ascribed to a physical displacement of the fluorescent sites in the DNA molecules that occurs as a response to the voltage switching. This result demonstrates that the conditions for inducing a tilt of the DNA molecules are obtained in the OCMFET structure when it is in the ON state. As a consequence, the charge to be sensed is closer to the surface, and, at the same time, the Debye length in the solution is increased as a consequence of the repulsion of the positive ions from the surface. Moreover, the attraction exerted by the surface potential on the targets compensates the electrostatic repulsion with the probes, determining a faster and more efficient hybridization. The combination of these effects is therefore suggested for explaining the capability of OCMFETs of detecting DNA hybridization with a sensitivity that is well beyond the limits normally accepted as due to screening effects in liquids. The proposed analysis, never performed before for FET-based sensors, is of general interest, suggesting that several biological reactions, in addition to DNA hybridization, as antigen-antibodies interactions and enzyme activity, can be detected by field-effect devices in liquids at relatively high salt concentrations, i.e., in conditions comparable with an in vivo environment.

S. Lai gratefully acknowledges Sardinia Regional Government for the financial support of his Ph.D. scholarship (P.O.R. Sardegna F.S.E. Operational Programme of the Autonomous Region of Sardinia, European Social Fund 2007–2013-Axis IV Human Resources, Objective l.3, Line of Activity l.3.1.).