Characteristics of solid-state nanometre pores fabricated using a transmission electron microscope

Solid-state nanopores can be used to detect nucleic acid structures at the single molecule level. An e-beam has been used to fabricate nanopores in silicon nitride and silicon dioxide membranes, but the pore formation kinetics, and hence its final structure, remain poorly understood. With the aid of high-resolution TEM imaging as well as TEM tomography we examine the effect of Si3N4 material properties on the nanopore structure. In particular, we study the dependence of membrane thickness on the nanopore contraction rate for different initial pore sizes. We explain nanopore formation kinetics as a balance of two opposite processes: (a) material sputtering and (b) surface-tension-induced shrinking. (Some figures in this article are in colour only in the electronic version) Nanometre-sized pores can be used to detect and characterize unlabelled biopolymers such as DNA, RNA and polypeptides, with single-molecule resolution. Initial experiments performed with the 1.5 nm pore α-Hemolysin (α-HL) demonstrated that single-stranded DNA and RNA molecules can be electrophoretically threaded through the pore, and that the ion current flowing through the pore contains information on the biopolymer sequence, its type, length and secondary structure (Kasianowicz et al 1996, Akeson et al 1999, Meller et al 2000). In addition, the α-HL nanopores have been used to study the unzipping kinetics of DNA hairpin molecules under stationary or time-varying forces, to detect DNA hybridization kinetics and to study the interaction of DNA with bound proteins using nanopore force spectroscopy (Meller et al 2001, Nakane et al 2002, Meller 2003, Bates et al 2003, Wang et al 2004, Mathe et al 2006, Hornblower et al 2007). Solid state nanopores can be fabricated in free standing Si3N4 and SiO2 films, using Ar+ ion beam (Li et al 2001) or electron beam sculpturing (Storm et al 2003). Solid-state 5 Authors to whom any correspondence should be addressed. nanopores offer several advantages over phospholipidembedded protein channels: their size can be tuned with nanometre precision and they exhibit an increased mechanical, chemical and electrical stability. Recent studies using solid state pores have begun to emerge, demonstrating the detection of double-stranded and single-stranded DNA conformations (Li et al 2003, Heng et al 2004, Storm et al 2005a, 2005b). E-beam fabrication involves contraction or expansion of nanoscale pores in thin Si3N4 films (10–50 nm) using an intense electron beam in a field emission TEM (Kim et al 2006). One of the main advantages of this method of fabrication is that the nanopores are constantly imaged, allowing the user to stop the contraction/expansion process when the desired size has been reached. Despite progress made in nanopore fabrication the reproducibility of their size and shape has remained poor, partly due to the fact that the mechanism for the pore formation is not well understood, and the geometries have not been fully characterized nor controlled (Li et al 2001, Chen et al 2004, Storm et al 2003). In this paper we seek to enhance our understanding of the three-dimensional 0957-4484/07/205302+05$30.00 1 © 2007 IOP Publishing Ltd Printed in the UK Nanotechnology 18 (2007) 205302 M J Kim et al Figure 1. (a) An image of DuraSiNTM film/chip (courtesy of Protochips, Inc.), (b) a TEM image of 50 nm thick silicon nitride membrane. (c) A live fast Fourier transform (FFT) of silicon nitride membrane, and (d) a typical solid-state nanopore. The inset in (d) is a schematic of the silicon nitride membrane supported by silicon. The scale bars are 5 nm. Figure 2. A sequence of TEM images displaying the dynamics of drilling (a) and contraction (b) in the 50 nm thick silicon nitride membrane. The scale bar is 5 nm. structures of the solid state pores by employing transmission electron microscope (TEM) tomography, and characterizing the dynamics of nanopore formation at various thicknesses of the silicon nitride membranes. These studies are used to develop a model for nanopore dynamics, that involves nanopore expansion through molecule sputtering and surfacetension-induced contraction. Fabrication begins with the formation of a DuraSiNTM silicon nitride membrane, 20 or 50 nm thick (Protochips Inc., Raleigh, NC) deposited across a 300 μm thick silicon wafer. The silicon nitride is deposited using low-pressure chemical vapour deposition (LPCVD) at a temperature of 825 ◦C using ammonia and dichlorosilane gases. The flow rate of the ammonia to dichlorosilane is approximately 1:5. This results in a silicon-rich nitride film, with a tensile stress in the range of 50–150 MPa. This stress is low enough to allow the formation of free standing membranes. A 50 μm×50 μm window is then fabricated on the silicon substrate wafer using photolithography and KOH wet etching. 50 nm thick membranes were further etched using a reactive ion etching (RIE) system (Nexx Systems, Inc., Billerica, MA) to generate 10, 30 and 40 nm thick membranes. Each silicon nitride membrane (10, 20, 30, 40 and 50 nm) has a thickness tolerance of 10% measured by Woollam spectroscopic ellipsometry (J A Woollam Co., Inc., Lincoln, NE). The nanopores can then be fabricated in this solid state material using a JEOL 2010F field emission TEM with an acceleration voltage of 200 keV. The intense e-beam (108–109 e nm−2) can be used to directly fabricate nanopores in the range of 4–8 nm diameter as previously reported (Storm et al 2003, Krapf et al 2006, Wu et al 2005, Kim et al 2006). For example, figure 1 shows an 8 nm diameter nanopore. Depending on the thickness of the membrane (20–50 nm), the pore formation time will vary between 5 and 60 s (thicker membrane, longer times). Nanopore contraction is achieved by slightly defocusing the ebeam, effectively reducing the peak intensity to ∼106 e nm−2. This intensity is sufficient to fluidize the thin SiN membrane, inducing rapid nanopore contraction. Figure 2 shows a