Cryo-FIB-SEM visualization and radiation damage of a water-THF phase-separated mixture and in situ formed organic nanoparticles.

A dual beam FIB (Focused Ion Beam) system which provides the ion beam (i-beam) and electron beam (e-beam) function are widely used in semiconductor manufacture for construction analysis and failure cause identification. Although FIB is useful for defect or structure inspection, sometimes, it is still difficult to diagnose the root cause via FIB e-beam image due to resolution limitation especially in products using nano meter scale processes. This restriction will deeply impact the FA analysts for worst site or real failure site judgment. The insufficient e-beam resolution can be overcome by advanced TEM (Transmission Electron Microscope) technology, but how can we know if this suspected failure site is a real killer or not when looking at the insufficient e-beam images inside a dual beam tool? Therefore, a novel technique of device measurement by using C-AFM (Conductive Atomic Force Microscope) or Nano-Probing system after cross-sectional (X-S) FIB inspection has been developed based on this requirement. This newly developed technology provides a good chance for the FA analysts to have a device characteristic study before TEM sample preparation. If there is any device characteristic shift by electrical measurement, the following TEM image should show a solid process abnormality with very high confidence. Oppositely, if no device characteristic shift can be measured, FIB milling is suggested to find the real fail site instead of trying TEM inspection directly.

The mechanical behaviour of metallic materials that have a microstructure composed of a brittle phase embedded in a ductile matrix is dictated by a complex interplay of factors such as local phase properties, cohesive properties, geometrical characteristics, and specific damage mechanisms. An important example of such two-phase materials are Al-Si casting alloys, which are widely used in automotive applications. In the micromechanics of these alloys, fracture of the particulate brittle phase plays a dominant role. Namely, when the alloy is deformed, the brittle silicon particles within the aluminium matrix start fracturing, leading ultimately to fracture of the alloy. In this thesis, micromechanical methods to measure local fracture toughness or strength of individual brittle microscopic particles within alloys and metal matrix composites (MMCs) are developed. The methods are based on coupling experimental techniques such as Scanning Electron Microscopy (SEM), Focused Ion Beam (FIB) milling and nanoindentation with finite element modelling (FEM). Special attention is put in probing portions of the particles that are left unaffected by the FIB micromachining process, and are thus representative of the particles’ intrinsic properties. These novel methods are also used to study the fracture of silicon particles within Al-Si casting alloys. To measure fracture toughness at a small scale, a microscopic chevron notch test is developed and demonstrated on benchmark materials. The main advantage of this method with respect to most existing small-scale fracture toughness test methods is that in chevron-notched samples the crack growth resistance is measured on a real crack instead of a pre-notch. The main difficulty, on the other hand, is achieving crack initiation at applied loads low enough to allow for subsequent stable crack growth. This was found to be particularly challenging in silicon. Local strength measurements on individual microscopic silicon particles within Al-Si alloys were achieved through two different, novel, methods. The first is a microscopic 3-point bending test by which the large facets of plate-like particles extracted from an Al-Si alloy can be probed. In the second approach, particles that are only partially exposed by deep-etching from an Al-Si sample are carved by FIB milling into such a shape that by compressing its top, bending is produced on a well-defined portion of the particle. The main finding of the local strength measurements is that silicon particles within Al-Si alloys can achieve extremely high strength values, yet fracture early when an Al-Si alloy is deformed because most of them feature stress-concentrating defects on their surfaces. The most important stress-limiting defects are found to be grooved interfaces between contacting silicon crystals, followed by surface pinholes, this being a defect identified here. Using FIB cross-sectioning and EDX examination it was revealed that these pinholes originate from alloy impurities. The insights gained on the intrinsic strength of silicon particles unveil the great potential of silicon as a reinforcing phase in Al-Si alloys. This, together with the acknowledgement of particle strength-limiting defects, may be used to devise strategies that, through the avoidance of those defects, should lead to improved alloy mechanical properties.

Nanofabrication is an integral part of the realization of advanced functional devices ranging from optical displays to memory devices. Focused-ion beam (FIB) milling is one of the most widely used nanofabrication methods. Conventionally, FIB milling has been carried out for patterning single-phase stable thin films. However, the influence of FIB milling on the phase separation of metastable alloy films during subsequent treatments has not been reported. Here, we show how FIB milling of Ag–Cu thin films influences the separation process and microstructure formation during post-milling annealing. The phase-separated microstructure of the film consists of fine, randomly distributed Ag-rich and Cu-rich domains, whereas adjacent to milled apertures (cylindrical holes), we observe two distinctly coarser rings. A combination of imaging and analysis techniques reveals Cu-rich islands dispersed in Ag-rich domains in the first ring next to the aperture, while the second ring constitutes mostly of Ag-rich grains. Copper silicide is observed to form in and around apertures through reaction with the Si-substrate. This substrate interaction, in addition to known variables like composition, temperature, and capillarity, appears to be a key element in drastically changing the local microstructure around apertures. Thus, the current study introduces new avenues to locally modulate the composition and microstructure through an appropriate choice of the film-substrate system. Such an ability can be exploited further to tune device functionalities for possible applications in plasmonics, catalysis, microelectronics, and magnetics.

We have developed and demonstrated new techniques for failure analysis based on focused ion beam (FIB) cross-sectioning and inspection by atomic force microscopy (AFM). Normally, inspection after FIB cross-sectioning is done by scanning electron microscopy (SEM). As features of interest shrink below limits detectable by SEM, often the next method chosen is transmission electron microscopy (TEM). However, sample preparation for site-specific TEM is difficult and time-consuming, even using newer methods based on FIB milling. AFM offers higher resolution imaging than SEM, and relaxes many of the sample preparation constraints of TEM. The AFM/FIB technique has been demonstrated on GaAs-AlGaAs and GaAs-InGaP heterojunction bipolar transistors (HBTs), including devices which have been electrically stressed to failure.

We present a fabrication method to obtain freestanding optical microcavities in Single Crystal Diamond (SCD), based on a combination of Reactive Ion Etching (RIE) and multidirectional Focused Ion Beam (FIB) milling, and we report for the first time experimental optical characterization of freestanding diamond optical microdisk resonators obtained by this fabrication method. Patterning of the optical microcavities is achieved by contact photolithography on single crystal CVD diamond plates (3 mm x 3 mm x 0.15 mm), using a SiO2 hard mask and optimized O2 diamond plasma etching, resulting in multiple circular pillars in a single etch step. Individual pillars are subsequently undercut by multi-directional FIB milling from two orthogonal directions, shaping the anchor to the bulk substrate. Sequential FIB thinning and smoothing of the disks allows obtaining freestanding optical microcavities. During FIB milling, an Al/Cr layer (50 nm/75 nm) is used to ground the diamond substrate, simultaneously limiting ion implantation and reducing FIB induced edge rounding. We experimentally probe the cavities by a tunable laser, coupled to the resonator by a tapered single mode fiber. The spectral response of a typical microdisk (diameter 5.9 μm, thickness 800 nm) in transmission over the tuning range of the laser (1485 nm to 1550 nm) reveals multiple optical resonances with a Free Spectral Range of 52.5 nm and optical Q-factors attaining up to 1500 (at 1496 nm). To our knowledge, this is the first time that freestanding optical microdisk resonators are demonstrated in Single Crystal Diamond by a combination of RIE and multidirectional FIB milling, providing a path for high-Q optical cavities in diamond.

Due to diamond extreme properties, it is very hard to prepare a TEM sample from diamond using traditional methods of preparation including mechanical thinning, ion milling or chemical etching. At the same time diamond can be relatively easily micro‐machined using focused ion beam (FIB) technique. Using this technique a cross‐sectional TEM sample of diamond can be prepared in a few hours. Also, combination of ion implantation and FIB milling allows device fabrication in diamond at micro and nano scales levels. In the last decade FIB milling became essential tool for TEM sample preparation as well as for nanofabrication in diamond. However, Ga FIB milling has an unavoidable result in formation of the damage layers which can significantly reduce the device working areas and limit the applications of the FIB technique for nanofabrication of diamond. The damage layers in the FIB prepared TEM diamond samples can also significantly aggravate the quality of high‐resolution imaging. So, the knowledge of the extent of damage induced in diamond during FIB milling is critical for nanofabrication as well as for TEM imaging. In this work the damage layers after FIB milling of the synthetic single crystal diamond at different ion beam energies were studied using high‐resolution and analytical electron microscopy. TEM image of cross‐section of TEM lamella prepared using 30 keV Ga FIB milling is shown in Fig. 1. Amorphous layer with thickness ~ 16 nm are clearly visible on both sides of TEM lamella. EELS measurements of the carbon K‐edge in the amorphous region shows a prominent feature at 285eV, the p * peak associated with the presence of sp 2 bonding. This indicates the conversion of diamond sp 3 bonds to sp 2 in the amorphous damage area. Electron energy loss spectrum image was taken in STEM mode from central part of cross‐section of TEM lamella shown in Fig.1. Using Gatan Digital Micrograph software the chemical maps for sp 2 and sp 3 bonded carbon were obtained. Fig.2 shows maps for features of carbon K‐edge at 285 eV (sp 2 ) and 290 eV (sp 3 bonding). It is visible from Fig. 1, 2 that TEM lamella prepared from diamond using 30 keV FIB milling contained ~ 20 % of amorphous sp 2 bonded carbon. In case of thinner lamellas (60 nm) prepared using 30 keV FIB milling the fraction of amorphous sp 2 bonded carbon increases to ~50%. The thickness of damage layers with sp 2 bonded carbon in TEM diamond samples can be reduced by using low voltage FIB milling. Fig.3 and Fig. 4 show the HREM images of damage layers in diamond after 30 keV and 2 keV FIB milling. The thickness of amorphous damage layers reduced from 16 to less than 2 nm. Thus, very thin TEM diamond samples with low fraction of amorphous sp 2 bonded carbon could be prepared by using 2 keV FIB milling at final stage of TEM lamella preparation.

T.M. Moore Omniprobe, Inc., Dallas, TX moore@omniprobe.com In 1965, Gordon Moore forecast that the microprocessor industry would continually scale to smaller feature sizes and the number of transistors would double every 18 months. Scaling below the 100nm node, combined with the implementation of copper and low dielectric constant insulators to increase the processor speed, has produced the situation in which SEM inspection no longer offers suitable resolution to image key artifacts and structures. The transmission electron microscope (TEM), once considered more of a development tool, is now in the forefront for process control and failure analysis, especially for measurements such as the thickness of semiconductor device non-planar barrier and seed layers. The use of focused ion beam (FIB) microscopes has become the method of choice for site-specific TEM sample preparation. Originally, the FIB was used as a final thinning step for mechanically prepared ribbons of semiconductor material adhered to modified TEM grids, known as the “H-Bar” method. More recently, the method for performing the entire TEM sample preparation process within the FIB is known as “in-situ lift-out” and is based on the use of a chamber-mounted nanomanipulator and beam-induced material deposition.[1-5] The use of the FIB offers advantages over conventional mechanical TEM sample preparation. The dual-beam FIB offers the ability to locate the lift-out site with SEM resolution and then use the ion beam to excise the sample without sacrificing the wafer, followed by thinning the extracted sample to the thickness required for TEM inspection. This is especially attractive for 300 mm processing where the value of each wafer in the flow can exceed $100,000. The risk to the quality and reliability of the process wafer due to gallium contamination from the ion beam is considered manageable.[6] In-situ lift-out also enables the return of the mostly abandoned practice of including informative test die on product wafers. The Total Release method for in-situ lift-out is designed to maximize throughput of a TEM sample preparation process.[4-5] The method can be simplified into three successive steps (see Figure 1). The first is the excision of the lift-out sample using FIB milling and extraction of the sample from its trench with two rapid ion milling steps, or “cuts”. The first cut is “U”-shaped and partially surrounds the target. The second is a straight cut that intersects the first cut beneath the target and produces a wedge-shaped sample. Then the probe is fixed to the released sample, typically with ionbeam metal deposition, and the sample is removed from the wafer by the nanomanipulator. The second step is the “holder-attach” step, during which the wedge is translated on the probe tip to the TEM sample holder (the lift-out grid). Then the sample is attached to the TEM holder (again, typically with ion beam-induced metal deposition) and later detached from the probe tip point using FIB milling. The third and final step is the thinning of the wedge into an electron-transparent thin section using FIB milling. The use of a simple probe tip for lift-out in the FIB has throughput and efficiency advantages over alternative methods. For example, lift-out can also be accomplished by a method referred to as ex-situ lift-out, in which ion beam-thinned samples, still attached to the wafer, are removed from the FIB and then detached from the wafer with a statically charged glass needle. These samples are then permanently deposited onto the suspended areas on a polymer membrane-coated TEM grid. Ex-situ lift-out can be faster than the in-situ method, and it requires less time in the FIB. However, in-situ lift-out offers a much higher overall throughput through

It is shown that the focused ion beam in the Physical Electronics SMART-Tool can be used to create the conditions needed to locate interconnect defects by voltage-contrast analysis. The SMART-Tool is designed for the analysis of small defects on full wafers by Auger electron spectroscopy. These defects are typically located using a defect coordinate map from a light-scattering based inspection tool. The SMART-Tool can be equipped with a focused ion beam for cross-sectioning defects. Stand-alone focused ion beam tools have been used to locate defects by voltagecontrast analysis. Unlike stand-alone tools, the ion beam in the SMART-Tool is situated off the surface normal. This does not hinder its ability to ground interconnect parts to the substrate, creating the conditions for passive voltagecontrast imaging. A defective via chain, identified by high resistivity on parametric test, was grounded to the substrate by focused ion beam milling in the SMART-Tool. The defective via was then identified by voltage contrast in images. The defect was marked and cross-sectioned by the focused ion beam and analyzed by Auger electron spectroscopy, all in the SMART-Tool without breaking or repositioning the wafer. Studies suggest that unbroken wafers can be returned to the manufacturing line to complete processing after focused ion beam milling without compromising unaffected die. Thus, this type of interconnect defect analysis can be performed on defective die without sacrificing non-defective die on the same wafer.

Inorganic hardmasks are routinely employed in reactive ion etching (RIE) processes due to their excellent etch resistance. However, since pattern definition is commonly performed using organic resist materials, the enhanced etch resistance provided by the inorganic hardmasks comes at the expense of added process complexity. In this work, the authors introduce the method of direct patterning of hard masks (DPHM) utilizing milling and gas assisted deposition (GAD) with a focused ion beam (FIB). DPHM by FIB allows to structure hardmask materials, which are otherwise not accessible with standard processes. Further, it reduces the high number of (typically seven) processing steps required for resist based patterning down to only three using FIB milling of hardmasks or even two using FIB GAD for patterning. The authors found that by FIB milled hard masks made of oxide such as aluminum zinc oxide exhibited excellent pattern clarity. For other materials, effects such as ion beam induced dewetting were found to affect the patterning result and must be considered in the choice of hardmask materials. Comparing DPHM and RIE to pure FIB milling of bulk material a speed enhancement of at least 755 times has been achieved. DPHM by FIB milling offers the highest versatility in material choice while FIB GAD enables faster patterning of selected hardmask materials.

A typical dual-beam platform combines a focussed ion beam (FIB) microscope with a field emission gun scanning electron microscope (FEGSEM). Using this platform, it is possible to sequentially mill off > ~ 50 nm slices of a material by FIB and characterise, at high resolution, the crystallographic features of each new surface by electron backscatter diffraction (EBSD). The successive images can be combined to generate 3-D crystallographic maps of the microstructure. This paper describes various aspects of 3-D FIB tomography in the context of understanding the microstructural evolution of metals during deformation and annealing. The first part of the paper describes the influence of both metal type and milling parameters on the quality of EBSD patterns generated from a surface prepared by FIB milling. Single crystals of some face centred cubic metals were examined under varying FIB milling parameters to optimise EBSD pattern quality. It was found that pattern quality improves with increasing atomic number with the FIB milling parameters needed to be adjusted accordingly. The second part of the paper describes a useful technique for FIB milling for the reliable reconstruction of 3-D microstructures using EBSD. There is an initial procedure involving extensive milling to generate a protruding rectangular-shaped volume at the free surface. Serial sectioning is subsequently carried out on this volume. The technique was used to investigate the recrystallization behaviour of a particle-containing nickel sample, which revealed a number of features of the recrystallizing grains that are not clearly evident in 2-D EBSD micrographs.

The applications of focused ion beam (FIB) technology in micromachining has advantages over other micromachining technologies, such as high feature resolution, capable markless process, rapid prototyping and adaptive for various materials and geometries. FIB direct-writing techniques are explored for their excellent abilities in micromachining. In addition to FIB technology and its principles for imaging, milling and deposition, a typical FIB system is presented. The key to FIB direct-writing technology is to operate a FIB with a proper beam size, shape, current and energy to remove or add a required amount of material from a pre-defined location in a controlled manner. In this way, high-precision and complicated three-dimensional structures with controlled profiles can be fabricated. Several examples of using milling technique for making high-quality microdevices or high-precision microcomponents for optical and other applications are given. The demonstration of milling a narrow readout gap at an oblique angle on a microaccelerometer shows a FIB's application on a small but accurate post-processing step on a micromechanical device. The diffractive optical element (DOE) with continuous relief and submicron feature size fabricated by FIB milling is also presented to prove high resolution and accurate relief control. Furthermore, FIB milling is used to shape a variety of cutting tools with extremely precise dimensions and complex tool face shapes.

Focused ion beam (FIB) technology is a reliable tool for the defined local surface modification on the nanoscale and therefore a promising technique to “write” a predefined texture on a point‐by‐point basis. FIB applications allow flexible scaling of surface patterns both laterally as well as in their height and have therefore the potential to become a versatile tool for purpose‐tailored roughness standards. For surface texture creation by FIB, both real measurement data of actual surfaces and artificially defined surface models can be used. Before FIB is applied to create the desired surface roughness, the input topography data needs to be converted into control commands of the FIB instrument. For this purpose, an automated procedure has been developed. This includes the conversion of the resolution of the given surface topography data into the resolution of the FIB patterning engine and the vertical segmentation of this surface topography into equidistant height layers that are later written by FIB milling (subtractive) or deposition (additive). In addition, this software allows the projection of the data on parametric surfaces (i.e. cylindrical surfaces or spheres). Finally, all data are integrated together with supporting tasks, like orientation marks, finder grid and identifier into a patterning script, which allows the automated “writing” of the structure by FIB. While FIB deposition is not material‐dependent, the milling rate strongly depends on the specimen material. The correct milling dose has therefore to be determined experimentally before the creation of the roughness fields. In addition, the pixel size of the FIB patterning layers has to be chosen with respect to the diameter of the used focused ion beam. In the initial tests, the actual roughness data used as input data for replication by FIB were real AFM (Atomic Force Microscopy) measurements with a nearly symmetric distribution of height values; Sq (root‐mean‐square roughness) values around 100 nm and Sz (peak‐to‐peak) values around 1000 nm. The FIB reproductions were then investigated by AFM in the same way, thus allowing best possible high‐resolution comparison of input and result. Tests were successfully performed both with a higher‐frequency and a lower‐frequency roughness to check the FIB limits. While in the beginning, both deposition and milling were tested with success, milling was chosen in the following, as it can be applied faster and thereby reduces machine hours and thus costs. The figures show a polished Si‐specimen with a FIB‐milled roughness area of about 180 µm x 180 µm. The similarity of the FIB‐created roughness and the model data is apparent, while a closer look reveals that the highest frequency components are not reproduced by FIB. As these frequencies of rather small amplitudes are only on top of the dominating surface texture, their absence does not alter the key roughness values significantly, that agree with those of the model data within 10 % in this case. The AFM investigations showed that FIB structuring allows the reproduction of a given surface texture on different substrates, resulting in a homogeneous, isotropic roughness. Tests with a more precise milling depth calibration proved that amplitudes are reproduced with only 1 % to 3 % deviation from the chosen model data. Furthermore, these works prove that FIB is a unique tool to rescale a given roughness in all three directions, opening a wide range of applications, both to the smaller and to the larger sizes, over a rather broad range of dimensions: Towards lower vertical scales, tests with amplitudes downscaled down to 1/30 (i.e. Sq ~ 3 nm) were performed successfully. In the opposite direction, lateral scales were enlarged by a factor of 5 so the roughness is composed of spatial frequencies that can well be measured by most optical surface measurement techniques. Larger roughness fields require the application of advanced stitching techniques; roughness structures with a size up to 290 µm x 290 µm were already created successfully. For many practical applications, the performance of topography measurement techniques on curved surface needs to be characterized. In order to address this issue, defined roughness fields again of 290 µm x 290 µm were transferred successfully onto cylinders and spheres (both with a diameter of about 1 mm) by FIB and applied for characterization of a broad scope of instruments, including AFM, WLI (White Light Interferometry) and optical 3D microscopes. The authors like to thank André Felgner and Peter Krebs (both PTB) for extensive AFM and CLSM measurements. This work is partly supported by the EMRP JRP IND59 “Microparts” jointly funded by the EMRP participating countries within EURAMET and the European Union.

Micro/nanoscale diamond cutting tools used in ultra-precision machining can be fabricated by precision grinding, but it is hard to fabricate a tool with a nanometric cutting edge and complex configurations. High-precision geometry accuracy and special shapes for microcutting tools with sharp edges can be achieved by FIB milling. Because the FIB milling method induces much smaller machining stress compared with conventional precision grinding methods. In this study, the FIB milling characteristics of single-crystal diamond were investigated, along with methods for decreasing the FIB-induced damage on diamond tools. Lift-off process method and Pt(Platinum) coating process method with FIB milling were investigated to reduce the damage layer on diamond substrate and quadrilateral-shaped single-crystal diamond cutting tool with cutting edge width under 500 nm were obtained.

A focused ion beam (FIB) was applied for cross-sectional sample preparation with both transmission electron microscopes (TEM) and scanning electron microscopes (SEM). The FIB sample preparation has the advantage of high positioning accuracy for cross sections. On the other hand, a broad ion beam (BIB) has been conventionally used for thinning TEM samples. Although both FIB and BIB use energetic ion beams, they are essentially different from each other in many aspects such as beam size, beam current density, incident angle of the beam with respect to cross sections, and beam scanning (i.e., dynamic or static beam). In this study, FIB cross-sectioning is compared with BIB thinning. We review inherent characteristics such as positioning accuracy and uniformity of cross section, radiation damage, and beam heating. Discussion is held from a view-point of ion beam and sample interaction.

Focused ion beam (FIB) technology is one of the most widely used methods for fabricating crosssectional analysis specimens because of its high precision and characteristics that minimize the occurrence of defects. Demand for large cross-sectional area analysis is increasing to improve product reliability in various industries, but is limited by the low milling speed of FIB. Other potential techniques such as Ar ion milling and plasma FIB have been adopted, but low milling speed for large areas still remains a problem. A promising solution to this issue involves laser machining prior to FIB milling. In laser machining a laser beam is irradiated to remove materials from the target. This technique can provide several orders of magnitude higher material removal rate than FIB, however, tapering of the machined surface and laser induced damage can occur. Removing these defects leads to increased FIB milling time. In this study, the laser parameters including angle of incident (AOI) were optimized to achieve a vertical like sidewall and minimize laser induced defects. Before applying AOI, laser machining parameters were optimized to reduce the angle of the machined sidewall. The taper angle of 2.5° was fabricated using the optimized parameters and application of AOI. Raman spectroscopy, SEM, and EDS analysis were used to measure not only the geometry of the laser machined sidewalls, but laser induced residual stress and defects. These results were then used to calculate the volume of FIB milling required to remove the laser induced damages and achieve vertical sidewalls. The application of AOI can significantly reduce the processing time in the FIB milling compared to the processing time when AOI is not applied.