Piezoresistivity in single DNA molecules

Piezoresistivity is a fundamental property of materials that has found many device applications. Here we report piezoresistivity in double helical DNA molecules. By studying the dependence of molecular conductance and piezoresistivity of single DNA molecules with different sequences and lengths, and performing molecular orbital calculations, we show that the piezoresistivity of DNA is caused by force-induced changes in the π–π electronic coupling between neighbouring bases, and in the activation energy of hole hopping. We describe the results in terms of thermal activated hopping model together with the ladder-based mechanical model for DNA proposed by de Gennes.

text, the tip retraction is stopped when a molecular plateau is detected in the current. The current is then monitored for 100ms while a sinusoidal modulation is applied to the tip position. During this process, due to the finite nature of single molecule junctions, the junction can spontaneously break down, resulting in a sudden decrease in current, to the amplifier background.
Supplementary Figure 1 shows two separate molecular junctions, Supplementary Figure 1a and S1d, in which the tip modulated STM-BJ technique is used to measure piezoresistance. The junction in Supplementary Figure 1a-c shows a typical molecular junction in which the molecular plateau is maintained through the entire measurement. Supplementary Figure 1d-f, on the other hand shows a molecular junction which spontaneously breaks down during the tip modulated STM break junction measurement. The sudden decrease in current, which is associated with the breaking of a single molecule junction, is accompanied by a sudden increase in the derivative response (1⁄ ⁄ ) which is a result of the change in transport mechanism from single molecule to tunneling through solvent. Considering the case when the molecular junction has broken down, the tip modulated STM measurement will be probing the tunneling decay constant of the solution environment, in this case phosphate buffer. The decay constant in these solvents have been shown to be on the range of ~10nm -1 , in agreement with the derivative signal after the breakdown (see Supplementary Figure 1e and f).

Supplementary Note 2 | Sequence dependent measured and calculated values
The measured conductance and piezoresistance values are presented below. The conductance values differ from those previously measured through STM break junction technique because of differing linker group and differing sequence. However, the order of magnitude is roughly similar and the differences are predictable. For example, the present conductance is smaller than that of Xu et al., which is understandable by the difference in sequence between these two experiments. Particularly, the addition of an A-T base pair at the ends is expected to decrease molecular conductance. Additionally, the present conductance is smaller than that of Xiang et al., who use very similar sequences, but use modified thiamine bases to directly contact the DNA base pairs to the electrodes, while the present work uses thiol linkers contacting the phosphate backbone. This difference in linker group is expected to increase the number of bonds between bases and electrode, thus increasing contact resistance. (1)

Supplementary Note 3 | Coherent correction fitting
Here the first term (R 0 ) is the contact resistance, which is determined using the linear fitting of the inter-strand purine sequence, and the second term is the coherent corrected hopping through the intra-strand purine sequence. The fitting parameters include gives the hole mass and energy, m and E respectively. To use the coherent correction, first a linear fitting of the resistance vs. molecular length was used to determine the y-intercept, the value of R 0 above, for both inter-strand and intra-strand purine sequences (R 0 = 13.1 MΩ and 20.7 MΩ respectively). Next the coherent correction was used to fit the resistance vs. molecular length data for both sequences. The fitting parameters for the inter-strand sequence were − −1 = 5.29±0.38 MΩ/bp, B = 0.923±0.38 nm, and C = 4.06±0.43eV. The fitting parameters for the intra-strand sequence were: − −1 = 2.95±0.14 MΩ/bp, B = 0.67±0.08 nm, and C = 2.09±0.14eV. As can be seen in Figure S4, comparison of the linear (pure hopping) and coherent correction fitting sheds light on the most appropriate model for each sequence. One way to quantify this is to consider the R 2 value obtained from fitting to each model. For the inter-strand purine stacked sequence (Supplementary Figure 4a) fitting with the hopping model (linear fitting) resulted in R 2 = 0.910 while fitting with the coherent correction resulted in R 2 = 0.759, suggesting that the most appropriate model for transport in inter-strand purine stacked sequences is hopping between neighboring guanines. Alternatively, fitting the intra-strand purine stacked sequence to the hopping model resulted in R 2 = 0.599 while fitting to the coherent correction resulted in R 2 = 0.914, supporting the previously observed delocalization in stacked guanine sequences and suggesting that there is a coherent portion of the hopping transport.
In addition to indicating that hopping in intra-strand stacked sequences involves delocalized, coherent hopping sites, the fitting data above also supports the assumption of small contact transfer rate compared to the base to base transfer rate. We can see this by comparing the contact resistance (R 0 = 13.1 MΩ and 20.7 MΩ for inter-stand and intra-strand respectively) with the slope of molecular resistance vs. molecular length (2.36 and 4.21 MΩ/bp respectively) that the limiting factor in transport through DNA molecular junctions is the contact resistance. This observation has been seen previously and the present resistance vs. length decay are similar to those of Xu et al. 3

Supplementary Note 4 | Guanine-adenine interaction
It has been reported earlier that addition of multiple adenine-thymine base pairs between guanine-cytosine tracts creates a tunneling barrier 3 . This tunneling barrier is seen as an exponential length dependence of conductance with the addition of adenine-thymine base pairs, in multiples of 2. However, as we note in the main text, the addition of individual adeninethymine base pairs, between guanine-cytosine base pairs, does not appear to substantially alter the conductance or hopping transport mechanism of single DNA molecules. This is seen when comparing the conductance values in Supplementary Table 1. We attribute this observation to the similarity in coupling between neighboring guanine-cytosine base pairs and neighboring guanine-cytosine and adenine-thymine base pairs. As can be seen in Supplementary Table 2, the calculated coupling between neighboring guanine and adenine bases can be of the same order of magnitude as that between neighboring guanine bases, depending on the order of the sequence. Similar observations can be made about the coupling between neighboring guanine and thymine bases. To further illustrate this, Supplementary Figure 6, shows the calculated highest occupied molecular orbitals (HOMO) of a guanine-cytosine base pair surrounded by adenine-thymine base pairs, in either TGT or AGA order. When purines are stacked inter-strand (TGT) the HOMO is isolated on the guanine, while the HOMO-1 is degenerate on the adenine bases due to mismatch in orbital energy of guanine and adenine. However, for intra-strand stacked purine sequences (AGA), the HOMO is delocalized from the central guanine base to partially reside on the neighboring adenines. This is a similar effect as that seen for neighboring guanine bases, as discussed in Figure 5 and in the main text.

Supplementary Note 5 | DNA sample structure characterization
In order to ensure that the DNA samples measured were indeed double helical B-form

Circular dichroism measurments:
Circular dichroism measurements of the DNA samples were used to confirm the B-form structure of the DNA samples measured. The CD measurements were performed in a Jasco (Easton, MD) J-815 Spectropolarimeter from 320 nm to 220 nm with a scanning rate of 50 nm/min to determine the structure of DNA samples. The spectra were average results from 5 scans, taken in 100mM Na + phosphate buffer solution with 5 uM of dsDNA at room temperature to mimic STM-BJ experimental conditions. The spectra, Supplementary Figure 8, are in good agreement with previously reported spectra of similar sequences 4-6 . Notably, inter-strand purine sequences are in exact B-form structure, while intra-strand purine sequences have subtle structural changes from increased coupling between neighboring bases which result in slightly modified B-form structures.

Melting temperature:
Melting temperature experiments were performed in a Varian Cary 300 Bio UV spectrophotometer with a Peltier thermal controller to determine melting temperature. 5 uM dsDNA were prepared with 100mM Na + phosphate buffer and annealed as for STM-BJ measurements, then heated at a rate of 0.2C/min from 10C to 80C with the absorbance at 260 nm