DNA Translocation through Hydrophilic Nanopore in Hexagonal Boron Nitride

Ultra-thin solid-state nanopore with good wetting property is strongly desired to achieve high spatial resolution for DNA sequencing applications. Atomic thick hexagonal boron nitride (h-BN) layer provides a promising two-dimensional material for fabricating solid-state nanopores. Due to its good oxidation resistance, the hydrophilicity of h-BN nanopore device can be significantly improved by UV-Ozone treatment. The contact angle of a KCl-TE droplet on h-BN layer can be reduced from 57° to 26° after the treatment. Abundant DNA translocation events have been observed in such devices, and strong DNA-nanopore interaction has been revealed in pores smaller than 10 nm in diameter. The 1/f noise level is closely related to the area of suspended h-BN layer, and it is significantly reduced in smaller supporting window. The demonstrated performance in h-BN nanopore paves the way towards base discrimination in a single DNA molecule.

window is about 20-40 mm with a 100-1000 nm sized hole drilled by focused ion beam (FIB). Nanopore was then drilled on the suspended h-BN membrane with highly focused electron beam in a transmission electron microscope (Model: JEM 2010F).
Characterization of h-BN nanopore. Figure 2(a) shows TEM image of a suspended h-BN layer (bright area) on Si 3 N 4 window with an 8 nm nanopore, indicated with a red arrow. Electron energy loss spectra (EELS) were taken on both the suspended h-BN layer (position indicated with blue dashed circle) and the h-BN layer over the Si 3 N 4 membrane (red dashed circle), as shown in Figure 2(b). Clear boron and nitrogen peaks appear in spectrum of suspended h-BN layer (in blue), while strong peaks of silicon and nitrogen are found in spectrum taken over the Si 3 N 4 membrane (in red). The carbon peak in the spectrum of suspended h-BN layer is due to the residual amorphous carbon contaminated during layer transferring. Nanopores with diameter from 2 nm to tens of nanometers are fabricated (see supplementary information, Figure S1) in suspended h-BN layers. Figure 2(c) shows high-resolution TEM image of another nanopore (about 10 nm in diameter). The FFT of such TEM image shows clear six-fold symmetry ( Figure S2), representing the hexagonal structure of the h-BN layer. Raman   spectroscopy has been employed to further characterize the h-BN layers. Figure 2(d) shows typical Raman spectrum of h-BN layer transferred onto a silicon wafer (to minimize the background luminescence from Si 3 N 4 ), excited with a laser wavelength of 532 nm. It is known that the E 2g peak of a single-layer h-BN is centered at 1369 6 1 cm 21 , while that for the bi-layer and the bulk are centered at 1365 6 2 cm 21 and 1366 cm 21 , respectively 23 . The observed Raman peak in our samples is centered at 1369.3 6 0.1 cm 21 ( Figure S3), indicating the presence of single-layer h-BN membrane. The small peaks at 1400 cm 21 and 1600 cm 21 are the third harmonic peaks of the silicon substrate 24 .
Hydrophilicity improvement. The h-BN material is resistant to oxidation and intact after UVO treatment, and such advantage can be used to improve hydrophilicity of h-BN nanopores. We characterized the wetting ability of h-BN layer by measuring the contact angle of a KCl-TE droplet on layer surface and compared it with other materials normally used for fabricating solid-state nanopores. The measurements are taken immediately after 15minute UVO treatment (Jelight Company Inc., model No. 42-220). Figure 3 shows the measured contact angles on Si 3 N 4 surface (50 nm thick), UVO treated Si 3 N 4 surface, h-BN layer on Si 3 N 4 substrate, UVO treated h-BN layer on Si 3 N 4 substrate, and graphene on Si 3 N 4 substrate. The UVO treatment was not applicable to graphene due to its low oxidation resistance 25 . Each measurement was carried out with more than 10 samples. The untreated h-BN layer gives a contact angle of 57u, close to that on Si 3 N 4 surface (54u). The contact angle on UVO treated h-BN layer decreases to 26u. Such good wetting property can last for at least 30 minutes in ambient lab environment, long enough for device assembling in a typical nanopore experiment. We note that the UVO treated Si 3 N 4 membrane shows the lowest contact angle of 6u and the graphene surface produce a contact angle of 67u, highest among all materials tested in our experiment.
DNA translocation through h-BN nanopore. To conduct the DNA translocation experiments, the flow cell, PDMS gaskets and the fabricated h-BN nanopore chips were firstly treated with UVO on both the front and the back side for 15 minutes each and then assembled as fast as possible (typically within 2 minutes). Saline solution (1 M KCl, room temperature, pH 7.8) was then added to both sides of the chip. A pair of Ag/AgCl electrodes supplied a bias voltage V to drive ions through the nanopore and the ionic current was monitored in real-time. Fig. 4(a) is a cartoon picture of such device, and Fig. 4(b) presents a typical I-V curve of a 4 nm nanopore in saline solution, through which a resistance of 45 MV is derived by a linear fit. When negative charged double-strand DNA molecules (l-DNA from New England Biolabs) were added to one side of the chip, a series of spikes were observed in conductance traces (Fig. 4c), arisen from single DNA molecule translocation through the nanopore. The driving voltage is 100-250 mV and a 30 kHz 4pole Bessel filter was employed during data acquisition and the sampling rate was 250 kHz.
Typically, at least hundreds of DNA translocation events can be observed in each nanopore device. Fig. 4(d) shows the scatter plot of conductance blockade DG versus time duration Dt of each DNA translocation event under driving voltage of 150 mV. The right panel is the histogram of conductance blockade, which can be fitted by a single Gaussian peak (blue line). The fitted curve is centered at Where N is the normalized coefficient of the peak, v is the mean velocity of the molecules, and D is the diffusion constant. The passage length L 5 L DNA 1 H eff is simplified to the DNA length L DNA (16 mm for unfolded l-DNA) because of the ultra-small thickness of the h-BN membrane. The fitting result is shown as solid red curve in the plot. The mean translocation speed for DNA translocation is 5.5 mm/ ms, similar to that in previous report on l-DNA translocation through graphene nanopore 10 . The diffusion constant D is fitted to be 8 mm 2 /ms, which is significantly larger than the bulk value of l-DNA.

Discussion
The histogram data do not match the fitting curve perfectly. Translocation events with duration time less than 1 ms (as contained in the black circle) may be contributed by DNA fragments, while those events with much longer duration time (contained in the orange circle) are due to DNA-nanopore interaction. For quantitative analysis, we discriminate elongated DNA translocation events as those events longer than the peak duration time by 1.5 s, where s is the full width at the half maximum (FWHM) of the fitted curve. The translocation time of elongated events in Fig. 4(d) is more than 6 ms. The percentage of elongated events adds up to 65% of all data points in the figure. The average translocation time of the elongated events is 108 ms, about 40 times longer than that of normal events. The anomalously long translocation time, as well as the high percentage, of elongated events indicates a pretty strong interaction between DNA molecules and the h-BN nanopore.
To further investigate such interaction in h-BN nanopores, we examined DNA translocation behavior in pores of different sizes. A moderate 10 nm (with l-DNA) pore and a big 30 nm (with 10 kbp DNA) pore are fabricated for the test. Both folded and unfolded translocation events are observed in both pores ( Figure  S4). By analyzing duration time of all unfolded events ( Figure S5), the percentage of translocation events undergoes molecule-pore interaction is about 71% in the 10 nm pore and about 20% in the 30 nm pore. The elongated translocation time is 16 ms (5 times of normal events) for the 10 nm pore and 0.85 ms (3 times of normal events) for the 30 nm pore. Figure 5(a) shows the noise spectrum of a typical h-BN nanopore on a 200 nm Si 3 N 4 window, derived by FFT of real-time current traces. The low frequency noise can be fitted by Hooge's law: S I 5 A*I 2 /f, where S I is the power spectral density, I is the ionic current, f is the frequency, and A is a dimensionless parameter that can be used to characterize the 1/f noise level. The value of A is derived to be 6.7 3 10 27 for the 200 nm window. Interestingly, it is found that 1/f noise in h-BN nanopore is closely related to the size of the suspended membrane area. Figure 5(b) plots the value of A as a function of membrane area, which is obtained by subtracting nanopore area from the Si 3 N 4 window area. For a smaller 180 nm window, the value reduces to 2.6 3 10 27 ; and for a bigger 550 nm window, it increases to 3.75 3 10 26 . We noticed that the 1/f noise in a graphene nanopore is related to the size of the supporting Si 3 N 4 windows as well 12 .
Reducing the window size is not only good for optimizing the 1/f noise, but also help increase the yield of h-BN nanopores. In our practice, the h-BN nanopore device is very fragile if the supporting Si 3 N 4 window is bigger than 500 nm, while most nanopores were stable when supported by Si 3 N 4 window smaller than 200 nm.
In summary, we have demonstrated the nanopore device fabricated on atomic thick h-BN layer. The h-BN nanopores are of good oxidation resistance and are compatible with UVO treatment. The hydrophilicity of h-BN nanopores is improved by UVO treatment with a reduced contact angle of 26u. Abundant DNA translocation events have been readily detected in such nanopore devices with various pore sizes. Strong DNA-nanopore interaction has been revealed in small nanopores, which slows down the DNA translocation significantly. Smaller supporting windows are desired for h-BN nanopores to reduce the 1/f noise and to improve device stability. The demonstrated performance in such hydrophilic ultra-thin h-BN nanopores suggests further promise towards DNA base discrimination based on ionic current analysis.

Methods
Growth of h-BN. Hexagonal boron nitride was grown on a copper foil (25 mm thick, 99.99%, from Alfa Aesar). After cleaned by acetone, ethanol, isopropanol, and deionized water, the copper foil was placed into a quartz tube at 1050uC, where borane ammonia (BH3-NH3, from Sigma Aldrich) was carried by mixed gas of Ar and H 2 (200/200 sccm) under atmospheric pressure. Hexagonal BN layer formed on the surface of copper foil after 30 min of growth.
Transfer of h-BN. PMMA solution (PMMA 495, concentration 5%) was spin-coated on one side of h-BN/copper foils at 3000 rpm for 60 s and cured at 145uC for 5 minutes.  under about 500 pA/cm 2 current density and a magnification of 800 k. After drilling, the electron beam was defocused to take images as quickly as possible. The EELS were taken subsequently with a magnification of 100 k and a diaphragm of 100 nm, under the energy resolution of 0.3 eV and integration time of 1.0 s. For each spectrum, the object astigmatism and zero-peak drifting were carefully calibrated. The pore size can be elaborately controlled. To measure the Raman spectra, the h-BN membrane was transferred onto a silicon wafer in order to avoid strong luminescence from Si 3 N 4 substrate. All Raman spectra were taken under 532 nm laser excitation, with focal size of 2 mm and power of 1 mW (HORIBA, JY6400). Nanopore experiments. The flow cell, PDMS gaskets and nanopore chips were firstly treated with UVO on both the front and the back side for 15 minutes each and then assembled as soon as possible. KCl-TE buffer (contain 1 M KCl, 10 mM Tris-HCl, 1 mM EDTA, pH 5 7.8) was filled into the device immediately. Subsequently, the BN nanopores were characterized by I-V curve, and only those chips providing linear and symmetry I-V curves were selected for DNA translocation experiment. Both double strand l-DNA (48.5 kbp) and 10 kbp DNA were diluted to 1 nM by KCl-TE buffer and pre-warmed to 70u for 1 minute to activate DNA molecules. After injection of DNA buffer, driving voltages between 50 mV and 250 mV were applied across the membranes. The ionic current was detected by an Axopatch amplifier (200B) at acquisition rate of 250 kHz, with a 30 kHz 4-pole Bessel filter. The recorded translocation events were processed in a Matlab GUI program. The program allows us to view the events one by one and only those events with signal-noise-ratio (SNR) bigger than 8 and have a sharp edge are selected for analysis.