The dual beam is a combination of a focused ion beam and scanning electron microscope (FIB-SEM). The FIB and SEM columns are angled typically at 52° from each other (Fig. 1A), enabling the FIB to mill into the bulk of the sample, and the subsequent use of the SEM to immediately analyse the exposed internal structure. Dual beams are able to perform all of the functions of the single-beam systems, while enhancing the capabilities by operating the columns concurrently. The conventional dual beam's main applications include material deposition (Tao et al., 1990), transmission electron microscopy (TEM) sample preparation (Kirk et al., 1989), defect characterizations, mask-less lithography (Kubena et al., 1989) and circuit edits and repairs. Most dual-beam work has concentrated on conventional FIB materials such as silicon. This is because of FIB's origins, which lie mainly in the semiconductor industry, from the late 1970s to early 1980s.
FIB columns follow principles similar to conventional SEM, in which the ion column can be divided into four separate sections. The ion sources of FIBs, known as liquid metal ion sources (LMIS; e.g. a tungsten coil dipped in liquid gallium), are positioned at the top of the column and resistively heated. The ion gun emits ions over a small spatial volume, with small angular spread and selectable energy. The ions then go through the optical system, consisting of several ion lenses, which direct the ions towards the specimen surface, before being deflected by the scan unit, which moves the beam in a raster pattern. The beam is typically focused to diameters of approximately 7 nm.
A major advantage of using the finely focused beam of ions is that it allows the site-specific milling and deposition of materials in a very controllable way (Tseng, 2004). The site-specific deposition of material is achieved by introducing a gas that chemisorbs onto the sample, and the FIB causes the gas to decompose into volatile and non-volatile components. The non-volatile components remain on the sample and are deposited precisely. FIB deposition has served as a simpler and faster, resistless (non-lithographic) technique (Marzi et al., 1999; Ziroff et al., 2003) for producing patterned material with sub-micrometre resolution (Vila et al., 2006). Traditionally, these depositions are used to connect or isolate wires when conducting device edit operations.
Platinum, which has a thermal conduction of 72 Wm−1 K1 and an electrical resistivity of 10.0 μΩ cm−1 at 300 K, can also be deposited on insulating surfaces (Tao et al., 1990; Marzi et al., 1999; Telari et al., 2002). A thin layer of platinum is applied with a low beam current. A low beam current enables the deposition of material onto insulating surfaces without causing any major charge-induced artefacts. However, low beam currents mean long deposition times. Hence, a further thicker layer of platinum is applied on top of the initial layer with a higher beam current. This ensures a ‘flat uniform surface’. This two-step process is important for rough insulating surfaces. A layer of platinum is able to protect insulating surfaces that are to undergo milling and enhances charge and heat transfer during the process.
Sputtering is the physical process by which material is removed by FIB. Site-specific atoms in a solid target material are ejected into the gas phase due to bombardment by energetic focused ions, largely driven by momentum exchange between the ions and the atoms in the material. The sputtering yield, defined as the number of atoms ejected per incident ion, is essentially a measure of the efficiency of material removal. For conventional dual-beam materials, the yield is normally in the range of 1–50 atoms per ion and is a function of the mass of ion and target atoms, ion energy, temperature and ion flux (Tseng, 2004). The yield is controlled usually by altering the energy of the beam. Initially, sputtering yield increases as ion energy increases, but the yield starts to decrease as the energy is increased past a certain level, for which ions can penetrate deeper into the material (Brodie & Murray, 1992). If the energy is too high, it can cause a variety of unwanted interactions including swelling, deposition, implantation, backscattering and nuclear reaction (Brodie & Murray, 1992). The milling yield dependence on incident angle, defined as the angle between the beam of ions and the surface of the target material, is optimal at 90° (Tseng, 2004).
The inclusion of a precursor gas during milling increases the volatility of the products significantly, reducing the degree of re-deposition. Material re-deposition has been envisioned as a process in which initially momentum is transferred from the ion to the surface, with the subsequent ejection of surface and near-surface species that form a crater (milled section in which the walls are slanted) (Lindquist et al., 1993). As milling proceeds, the slant of the walls increases, and the whole milled section starts to resemble a cosine distribution (Lindquist et al., 1993), as shown in Fig. 2. When the ion beam is applied in a raster pattern over a small area, that is, when the milling depth is greater than the width (known as aspect ratio >1), re-deposition is very likely to occur in all materials.
The advantages of using a precursor gas include enhancement of the FIB etch rate, hence a reduction in processing times, a reduction of unwanted implantation from primary ion species onto the substrate and a reduction of re-deposition on side walls (Young et al., 1993; Harriott, 1994). Traditionally, halogen-based chemical etching has been typically used for materials in the semiconductor industry; these include the use of Cl2 with GaAs, Si and InP (Young et al., 1990, 1993) and XeF2 for SiO2, W (Harriott, 1994) and diamond (Russell et al., 1998; Datta et al., 1999). Very briefly, the reaction of the irradiated surface species with absorbed halogen produces volatile species and re-deposition of surface material is essentially quenched. However, although halogen-based chemical etching is useful for many material systems, halogens do not universally enhance removal rates for most carbon-containing specimens (Stark et al., 1995). It has been observed that the introduction of water vapour increases the removal rate of carbon-containing materials such as polymethyl methacrylate (PMMA) and other resists by a factor of 20 (relative to physical sputtering) (Stark et al., 1995). Selective carbon mill SCM™ (FEI, Eindhoven, the Netherlands) is a novel special water-based chemical (MgSO4·7H2O) that has been developed for removing carbon-based materials effectively.
As stated earlier, the FIB was originally developed in the 1970s and 1980s for the requirements of the semiconductor industry, in which it was mainly used for applications such as circuit edit and repair. Today, when milling conventional FIB materials such as silicon, the ion beam is able to precisely mill cross-sections, without any electrostatic discharge damage. However, this is of course not the case with insulating material; the higher the ion current, the greater the risk of FIB-induced charging. Under irradiation, ions are implanted into the sample and electrons and ions are emitted from the surface. Because the emission yield of secondary electrons is approximately 10 times greater than secondary ion emission for insulators (Komoda et al., 2005), the sample surface is charged up positively, which in turn deflects the positively charged ion beam.
The use of the FIB in conjunction with a measured ‘defocused’ low-energy primary beam of electrons has permitted milling through insulating structures to be achieved in a controlled way, exposing the inner structure, without introducing charge damage into the sample (Stokes et al., 2007) (see Fig. 3A). Experimentally, it has been found that if specimen current ISP is measured during ion irradiation, it measures roughly three times the primary ion beam current. This has led to the suggestion that for each positive charge implanted, there are two positive charges remaining as a result of secondary electron emission δ, that is, Q=QPI+ 2Qδ. This has provided the criteria for choosing the electron beam current for neutralization IPE∼3IPI (Stokes et al., 2007).
Furthermore, during charge neutralization, the electron beam is totally ‘unblanked’, that is, a constant flux of electrons is directed onto the sample surface. Also, by raising the focus plane of the electron beam, the electron beam is spread over the sample. The incident surface area of the electrons (Fig. 4) is controlled and measured by vertically raising the focus plane of the electron beam by distance l in millimetres and is calculated via the formula (Stokes et al., 2007):
where r is the radius of the electron beam calculated from a =πr2, where a is the area of the mill.
Once uniform milling of the heterostructure is achieved, the use of the environmental scanning electron microscope (ESEM) (Danilatos, 1988) for imaging the revealed milled section surfaces then enables insulating materials to be imaged without charging problems (see Fig. 3B). ESEM employs a gas in the specimen chamber (Meredith et al., 1996; Thiel et al., 1997) in which water vapour is typically used at pressures of 1–2 torr (133–266 Pa). As in conventional SEM, the electron source is kept in high vacuum, and for this reason, a pressure gradient along the column is necessary. The column is separated into a series of zones, each of which is pumped individually and is separated from its neighbours by small limiting apertures (PLAs) that allow the electron beam to pass through but maintain the pressure difference between each zone. The vacuum arrangement along the column is shown in Fig. 5.
The emission of electrons in ESEM is specimen-dependent, and subsequent collisions between these electrons and gas molecules result in the generation of daughter electrons, that is, a cascade amplification effect (Meredith et al., 1996; Thiel et al., 1997), and positive ions. The signal is amplified as a result of the cascade, which is detected by a gaseous secondary electron detector (GSED). The positive ions produced drift towards the specimen surface and compensate the negative charge built up on the surface (Meredith et al., 1996).
In this study, we mill a cross-section of a model polymer system of 900-nm-thick spun cast film of phase-separated polystyrene–polybutadiene blends stained with osmium tetroxide (OsO4) with the aim of developing and optimizing dual-beam methodologies for application to polymeric samples generally. The three-dimensional morphology of polystyrene–polybutadiene thin films cast from a toluene solution has been fully characterized previously (Geoghegan et al., 1994) by combining results from three different techniques: nuclear reaction analysis, neutron reflectometry and ultramicrotomy/TEM. An asymmetric mixture of polystyrene and polybutadiene (molecular weights [MW] 5050 and 540 000 for polystyrene and polybutadiene, respectively) forms a blend in which polystyrene-rich phases wet both air and substrate interface and are separated by a polybutadiene-rich central phase. These layers are very well defined, and the interfaces between them are sharp, with a width of <20 Å. The blends also show regions in which isolated polybutadiene domains are surrounded by polystyrene shells (Geoghegan et al., 1994).