1 Introduction

DNA, as a potential molecular wire with semi-conductor properties, continues to attract significant research interest in the nanoelectronics community. This has been mainly driven by DNA’s self-assembly and molecular recognition, its small size (2 nm diameter), and reported semi-conductor properties (Triberis and Dimakogianni 2009; Hassan and Asghar 2010). Over the past decades, several groups have investigated the conductivity of DNA and the influence of various parameters, including number of bases, DNA sequence, temperature, and metal deposition, both theoretically and experimentally (Porath et al. 2000; Braun et al. 1998). A summary of the key and relevant researches in the literature that have reported often dissimilar conductive properties of DNA molecular wires was provided in our earlier work (Vahidi et al. 2014). Briefly, however, some of the important works in this area are reviewed here as an introduction. Porath and Shoseyov investigated DNA polymers as conducting wires and demonstrated—contrary to earlier beliefs that the conductivity of double stranded DNA molecules is negligible—that measurements of high currents under controlled experimental conditions that rely on avoiding nonspecific molecule–substrate interactions, could have considerable conductivity (Porath et al. 2000; Shoseyov and Levy 2008; Di Felice and Porath 2008). Parallel efforts explored mechanisms of enhancing DNA conductivity by using DNA molecules as templates for self-assembled conductive metallic nano-wires by gold coating DNA (Braun et al. 1998). Seeman and colleagues had investigated construction of structures using DNA molecules as scaffolds, which may ultimately serve as frameworks for nanoelectronics devices (Seeman 1982; 1991). Along these lines, Mirkin et al. (1996) and Alivisatos et al. (1996) were the first to describe self-assembly of gold nanoclusters into periodic structures using DNA. Mirkin et al. (1996) described a method of assembling colloidal gold nanoparticles into macroscopic aggregates using DNA as linking elements. Mucic et al. (1998) had also described the construction of binary nanoparticle networks composed of colloidal gold. In a series of publications, Braun and his colleagues demonstrated the use of DNA as a template for the fabrication of nanowires (Braun et al. 1998). They reported DNA bridge between two thin-film gold electrodes using thiol attachment. Attachment techniques presented by Braun et al. allow for precise localization of conductive materials (e.g. gold or silver particles) on DNA, converting the biological molecule into functionalized wires.

Table 1 DNA immobilization protocol
Full size table

However, in order to utilize DNA as a molecular wire, biosensor, or even processor, DNA must be easily immobilized onto a substrate. So far, almost all of the reported work in the literature in testing of the various DNA functions involved immobilizing thiol-terminated DNA onto a thin-film of metal (typically gold) on silicon or glass substrate (Braun et al. 1998; Hassan and Asghar 2010). However, thiol-termination of DNA first requires reduction of di-sulfide bonds, a quite involved procedure, and has a tendency to result in salt accumulation on the substrate and electrodes which may cause interaction with the substrate or the DNA itself, thereby significantly affecting the outcome of conductivity experiments. Since the electrical properties of DNA molecular wires continues to be challenged by skeptics—despite some encouraging results—it is important to eliminate any experimental condition that could affect this outcome such as salt build-ups. Therefore, while there exist some alternative methods of immobilizing DNA, we believe that there is an unmet research need for a bionanoelectronics platform that (1) is lithographically patternable, (2) has high aspect-ratio microelectrodes that enable the suspension of DNA away from the substrate (for electrical characterization), and (3) is resistant to non-specific binding of salts, and (4) easily immobilizes DNA on the microelectrodes.

Consequently, this current research introduces a new bionanoelectronics platform consisting of high aspect-ratio GC-MEMS electrodes to which aminated oligonucleotides and λ-DNA are attached. Specifically, we introduce GC-MEMS microelectrodes which are (1) patternable, (2) have high-aspect ratio (Kassegne et al. 2008; Wang et al. 2004), and (3) are resistant to salt build-up, as a platform for efficiently immobilizing DNA for the purposes of DNA molecular wire characterization (Kassegne et al. 2003), DNA computing, or biosensing. The GC-MEMS microelectrodes are functionalized through oxygen plasma etching (forming carboxyl groups) and are carbodiimide crosslinked for oligonucleotide attachment and DNA hybridization for double stranded DNA immobilization. To the best of the authors’ knowledge, this work is the first demonstration of DNA molecular wire bridge between non-metallic glassy carbon microelectrodes.

2 Materials and methods

In this section, we describe geometry and fabrication processes of the GC-MEMS based bionanoelectronics platform and the microelectrode structure used in the study. Figure 1 shows a conceptual illustration of this bionanoelectronics platform with DNA molecular wire attached to high aspect-ratio carbon electrodes using oxygen plasma functionalization of the GC-MEMS electrodes and their carbodiimide crosslinking with amine-terminated oligonucleotides. λ-DNA was used primarily due to its typical dimensions of 2 nm diameter and 10–15 μm length which corresponds to the gap between the microelectrode sets (Vahidi et al. 2014). This type of DNA is derived from a bacteriophage called lambda which typically inhabits E. coli bacteria and is easy to isolate in large quantities and cut to fragments.

Fig. 1
figure 1

Schematics of 3-D bio-nanoelectronics platform where amine-terminated oligonucleotides A and B are attached to the microelectrodes followed by their hybridization with λ-DNA. Four pads are used in the chip for four-point probe test

Full size image

2.1 Microelectrode fabrication

GC-MEMS electrodes were designed with a 10–15 μm gap and four bump pads and fabricated using SU8-10 (Microchem), a negative acting photosensitive epoxy on both bare silicon and silicon with an oxide coating. The lithography of microelectrodes that form component of a bionanoelectronics chip consists of the usual well-known steps (with slight variations for the various grades for the negative-tone photoresist) (Vahidi et al. 2014). This is then followed by pyrolyzing the microelectrodes to alter the structure of the photoresist from a non-conductive photoresist polymer to conductive glassy carbon. Pyrolysis of SU-8 was done under nitrogen flow and using the heating protocol described our previous work (Hirabayashi et al. 2014). Then, the microelectrode structure was finally exposed to oxygen plasma etching to functionalize it for attachment through the generation of carboxyl groups.

2.2 Microelectrode functionalization

Due to the inert nature of GC-MEMS carbon, modification of the surface is necessary to generate reactive functional groups in order to immobilize DNA. Our previous studies demonstrated that carboxyl groups could be generated on the surface using either of the following functionalization approaches: (1) 4-ABA (amino benzoic acid) treatment, (2) nitric acid treatment, (3) sulfuric acid treatment, and (4) oxygen plasma treatment (Hirabayashi et al. 2014). Of these functionalization methods investigated, oxygen plasma treatment at higher energies was found to be most efficient for generating functional groups without damaging the electrode. Thus, to functionalize the surface, plasma treatment was done for at least 60 s at 50 W.

2.3 Reactive ester formation

1-Ethyl-3-[3-dimethylaminopropyl] carbodiimide hydrochloride (EDC) is a crosslinking carbodiimide often used to react amine (–NH2) and carboxyl (–COOH) groups. As shown in Fig. 2, EDC reacts with a carboxyl molecule located at the surface of plasma treated GC-MEMS microelectrode to form an unstable O-acylisourea molecule. At this stage the molecule has three pathways it can take, i.e., (1) if the O-acylisourea molecule encounters an amine molecule, a stable amide bond can form, (2) the reaction runs backward and the carbonyl regenerates, or (3) EDC reacts with NHS (N-hydroxysuccinimide) to form a more stable ester (Choi et al. 2010). Since the goal is to maximize the probability of generating an amide bond between the carboxylated electrode and aminated DNA, NHS is added to generate a more stable reactive ester which can be used to form an amide bond between the oligonucleotide and the electrode (Fig. 3). To use this crosslinking property, EDC and NHS were added to phosphate buffer solution (PBS) to generate a solution with a 40 mM concentration of EDC and 100 mM concentration of NHS. The GC-MEMS electrode was stirred in the solution for 2 h, removed from solution, and rinsed with PBS.

Fig. 2
figure 2

Reactive ester formation

Full size image
Fig. 3
figure 3

Amide bond formation

Full size image

2.4 Oligonucleotide attachment

Oligonucleotides (Oligo A: 5′-GGG CGG CGA CCT/Amine/-3′ and Oligo B: 5′-AGG TCG CCG CCC/Amine/-3′) (Integrated DNA Technologies, Coralville, Iowa), were customized to be amine terminated and complementary to linearized λ-DNA (New England Biolabs, Ipswich, MA). Lyophilized oligonucleotides were diluted to a 500 μM solution using TE solution (10 mM Tris–HCl, 1 mM EDTA), 2 μL of these oligonucleotides were added to the center of the GC-MEMS microelectrode. To increase the movement of the oligonucleotides, a bias was applied between the electrode and substrate at 5 V for 2 min using a Keithley 2400 Source Meter. Oligonucleotides were then left to react with the surface of the electrode for 18 h before rinsing with PBS.

2.5 DNA hybridization

λ-DNA (New England Biolabs, Ipswich, MA) suspended in TE solution was dephosphorylated using FastAP, a thermosensitive alkaline phosphatase (Fisher Scientific, Pittsburgh, PA). 10 μL of λ-DNA, 50 μL of FastAP buffer, 20 μL of FastAP, and 420 μL of PBS solution were vortexed in a vial for 1 min. The vial was heated to 42 °C in a water bath to activate the phosphatase and cleave off the phosphate cap on the DNA. After 10 min the solution was heated to 80 °C for 5 min to stop the reaction and help linearize the DNA. The solution was then vortexed again. 250 μL of solution was placed in a beaker with a stir bar and the microelectrode. Over the course of 4 h, the DNA was allowed to stir with the microelectrode at 40 °C allowing the DNA to hybridize with the oligonucleotides immobilized on the microelectrode (Table 1).

3 Results and discussion

In this section, we will present the characterization of the ensuing microelectrode structures to which DNA is attached to.

3.1 Optical characterization

Fluorescent imaging was used to confirm DNA attachment to GC-MEMS electrodes. For each electrode, 0.5 μL of DNA safe stain (Invitrogen, Pittsburgh, PA) was diluted in 2 mL of de-ionized (DI) water. The microelectrode was placed in the diluted stain, agitated for 10–30 min, removed from the solution. Each GC-MEMS microelectrode was pressure rinsed to remove DNA not covalently bonded. Then, the GC-MEMS electrodes were placed on a microscope slide and imaged with Leica fluorescent microscope. Since random particles can cause noise in the fluorescent image, control images were taken as shown in Fig. 4. Figure 4 shows GC-MEMS microelectrode with no DNA but stain applied to give a reference of noise (random fluorescence not caused by the DNA safe stain binding to DNA). Figure 4b and c shows what the DNA and stain look like on a glass slide.

Fig. 4
figure 4

Control images a stain on GC-MEMS, no DNA b vortexted DNA with stain on a glass slide c DNA with stain on a glass slide

Full size image

In Fig. 5, DNA attachment under various conditions is demonstrated. In Fig. 5a, attachment was done according to the methods described in Sect. 2, resulting in DNA immobilization over the surface of the electrode. Figure 5b shows the effects of electrical-biasing on the electrode during oligonucleotide attachment. These results confirm that biasing generates mixing which in turn results in more oligonucleotide attachment, and thus more DNA attachment than the unbiased case (Fig. 5a). However, these results also show that biasing—while helpful in maximizing attachment—is not a necessary step. Figure 5c shows the effects of no functionalization on the surface. In this case, the protocol conditions were identical to Fig. 5a with the exception of the omission of the plasma treatment step. Since GC-MEMS has an inert surface, very little attachment on the surface of the electrodes was expected. The image in Fig. 5c confirms that the effects of no plasma treatment results in absence of attachment at the surface, but—interestingly—also demonstrated attachment at the edges and on some unorganized pyrolyzed carbon on the substrate. Though, ideally the substrate would be completely clean of carbon, incomplete development of the SU8 after the lithography process during fabrication can leave remnants of SU8 on the substrate resulting in unorganized carbon on the substrate after pyrolysis. Presence of DNA in these areas suggests that the DNA is immobilizing on these small carbon bits because carboxyl groups tend to exist in these areas due to the unorganized nature of these carbon fragments. The attachments on the edge suggest that carboxyl functional groups also occur naturally along the edges of the features.

Fig. 5
figure 5

a Using the standards protocol above DNA attachment was accomplished on the electrode surface. b DNA attachment on surface with electrical-biasing on the electrode during oligonucleotide attachment, c DNA attachment occurs at edges when plasma etching is not preformed, d DNA molecular bridge between two GC-MEMS electrodes

Full size image

Figure 5a and d shows two different types of attachment that can occur when running the same protocol. In Fig. 5a, the oligonucleotides and DNA bind to the surface of the electrode evenly but not in the gap. In Fig. 5d, the DNA binds slightly to the surface but mostly in the gap. The difference in the binding is likely due to the difference in the size of the gap. Due to the shrinking of the photoresist during the pyrolysis process, the microelectrode gap ended up being larger in many of the electrodes.

3.2 Factors affecting attachment

The protocol stated above demonstrates an optimized technique of attaching λ-DNA to GC-MEMS. However, our investigation shows that key factors such as pH as well as the amount of biasing time can have significant effect on attachment. Though many standard protocols for EDC/NHS crosslinking report typical concentration of 2 mM of EDC and 5 mM of NHS, our several attachment experiments to GC-MEMS, however, indicate that a much higher concentration of both EDC and NHS were required. We carried out attachment experiments with 2 mM EDC and 5 mM NHS, 4 mM EDC and 10 mM NHS, 8 mM EDC and 20 mM NHS, and 40 mM EDC and 100 mM NHS concentrations. Of all these protocols, those involving concentrations greater than 8 mM EDC and 20 mM NHS showed substantial attachment.

In addition, the protonation of carboxyl group on the surface of the GC-MEMS and amine group on the oligonucleotide could potentially interfere with the attachment chemistry. To address this issue of unwanted protonation, therefore, it is required to modify the pH of each reaction such that the solution deprotonates the carboxyl or amine group. For GC-MEMS electrodes, the surface of the carbon was assumed to deprotonate at a pH below 6. Our experiments showed that when the reaction was carried out at a pH 7.2, very little or no attachment occurred, but when the experiment was carried out at pH 6.0 or below, attachment did occur, supporting our assumption that the ionization pH for GC-MEMS was below pH 6. DNA attachment at this pH, therefore, shows that formation of the reactive ester was optimum at pH 6. However, the second step in the EDC/NHS reaction that generates the amide bond with the oligonucleotide requires a basic condition with pH value higher than, 8 to deprotonate the amine. Consequently, the attachment protocol was modified. First, GC-MEMS is allowed to interact with the EDC for 10 min. Then the pH of the solution with the EDC and GC-MEMS was raised to a pH of 8.5 by adding NaOH before NHS was added to the mix. Results from this protocol showed more attachment than just running the whole protocol at a pH 6.

As stated previously, though biasing was not necessary for attachment, it helped increase the amount of attachment. In order to determine how much biasing was needed to affect attachment, various amounts of time were tested at 5 V. The results showed that anything over 2 min actually decreased the attachment. This was likely due to some heating from the biasing that would affect the EDC/NHS crosslinking reaction.

4 Conclusion

The results of this study establish an efficient means of immobilizing DNA to high aspect ratio GC-MEMS. These GC-MEMS offer the advantages of high throughput pattern-ability making the shape repeatable and customizable. The GC-MEMS also has properties that make the carbon resistant to salt build-up, giving it advantages over the previous gold and platinum platforms used. Furthermore, this study demonstrated the ability to build features with a high aspect ratio on oxide coated silicon, giving the platform the ability to minimize the electrical effects of the substrate by suspending the DNA above the substrate an insulating the substrate. Fluorescent microscopy revealed that through oxygen plasma carboxyl functionalization and carbodiimide (EDC/NHS) crosslinking between amine terminated oligonucleotides, DNA can be immobilized onto these GC-MEMS electrodes. Investigation within the immobilization protocol demonstrated that (1) biasing at 5 V can promote the crosslinking when done for 2 min or less, (2) altering pH and EDC and NHS concentrations in the crosslinking protocol can induce more DNA immobilization, and, (3) while plasma etching is necessary to generate carboxyl groups on the surface, the presence of carboxyl functional groups along the edges occur naturally and that a bridge across a small gap can be obtained for DNA wire tests.