- Email: [email protected]
Observing the evolution of regular nanostructured indium phosphide after gas cluster ion beam etching
Applied Surface Science 459 (2018) 678–685
Contents lists available at ScienceDirect
Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc
Full Length Article
Observing the evolution of regular nanostructured indium phosphide after gas cluster ion beam etching
T
⁎
Anders J. Barlowa, , Naoko Sanob, Billy J. Murdochc, Jose F. Portolesb, Paul J. Pigrama, Peter J. Cumpsonb a
Centre for Materials and Surface Science (CMSS), Department of Chemistry and Physics, La Trobe University, Melbourne, Victoria 3086, Australia National ESCA and XPS Users’ Service (NEXUS), School of Engineering, Newcastle University, Newcastle upon Tyne, Tyne and Wear NE1 7RU, UK c School of Science, RMIT University, Melbourne, Victoria 3001, Australia b
A R T I C LE I N FO
A B S T R A C T
Keywords: Helium ion microscopy Gas cluster ion beam Indium phosphide Nanostructure XPS SAN
Indium phosphide (InP) surfaces develop a pronounced nanostructured texture upon irradiation by energetic ion beams. We have observed the mechanism of nanostructure evolution of InP under irradiation by an Ar gas cluster ion beam (GCIB) using helium ion microscopy (HIM). Initially, metallic indium nanoparticles form on the surface after removal of the top-most oxide layer. These nanoparticles form a mask which shadows the underlying InP. As the ion dose is increased, the masking effect results in substantial nanostructured topography in the form of pillars or nanocones, oriented along the axis of the incident GCIB. The surface sensitivity and high resolution of the HIM facilitates the direct observation of the metallic indium cap at the top of the pillars.
1. Introduction The impact upon a solid surface by a beam of energetic ions can result in significant topographical and chemical changes along with the ejection of surface atoms and fragments, and implantation effects in the near-surface region. This is put to practical use in surface analysis by secondary ion mass spectrometry (SIMS) where the mass of the ejected fragments is analysed to deduce composition and molecular structure, and in X-ray photoelectron spectroscopy (XPS), Auger electron spectroscopy (AES) and SIMS for the removal of unwanted contamination or residues from a surface prior to analysis, or for depth-profiling through a layer or interface by successive etching cycles of surface material during analysis. Prolonged exposure to an ion beam, such as during a sputter depth-profile, can lead to surface structures that have a specific morphological evolution based on an equilibrium between surface roughening (ion erosion) and surface diffusion [1]. This has been demonstrated a number of times on compound semiconductors, and in particular with indium phosphide (InP) [2–5]. Homma et al. [3] observed the preferential sputtering of phosphor from InP under irradiation by monatomic argon ions, coupled with the formation of surface features of the order of 10–100 nm in size depending on beam conditions. The surface morphology that results from such exposure has consequences in analytical depth profiling, where interface depth resolution in a multi-layered stack that includes InP can
⁎
be degraded [4]. Nanostructured ‘cones’ were observed by MacLaren et al. and attributed to a surface diffusion effect whereby indium metal is enriched at the surface leading to cone formation due to differential sputter rates of the metallic indium versus InP [5]. This was later supported by Homma who observed a masking effect from particles that exist on the InP surface prior to ion beam etching [6]. Due to the regular and nano-sized structures that could be obtained, these surfaces were subsequently found to be useful for the determination of tip profiles in atomic force microscopy (AFM) [7]. Recently, atom probe tomography has been used to study individual structures on ion-irradiated InP and demonstrated indium enrichment at the surface with very high spatial resolution [8]. Arrays of geometrically optimised InP nanostructures should, in principle, offer highly efficient and cost effective alternatives to planar photovoltaic cells. Employing InP is advantageous as it offers efficient absorption of light in very thin layers, with a favourable match between its bandgap and the peak energy in solar radiation. Indeed Sanatinia et al. demonstrated the applicability of a surface of InP “nanopillars” generated by a nitrogen ion beam [9]. Nanostructured InP surfaces could be a means towards generating efficient nanoparticle-based solar cells. In practice however their large surface area leads to decreased quantum yields due to non-radiative charge recombination pathways via surface defect states [10]. While adequate short circuit currents can be achieved using highly doped nanostructure-array photovoltaics [11],
Corresponding author. E-mail address: [email protected] (A.J. Barlow).
https://doi.org/10.1016/j.apsusc.2018.07.195 Received 22 March 2018; Received in revised form 6 July 2018; Accepted 28 July 2018 Available online 30 July 2018 0169-4332/ © 2018 Elsevier B.V. All rights reserved.
Applied Surface Science 459 (2018) 678–685
A.J. Barlow et al.
be on the order of 10 nm under ideal conditions. Moreover, high aspect ratio features in the vertical direction, such as those observed on InP, prevent imaging of finer surface details since the tip sidewalls impinge upon neighbouring features. Scanning electron microscopy (SEM) is often another popular choice for studying surface topography on the nano-scale, though it too can be limited in resolution at very high magnification. Helium ion microscopy (HIM) is a relatively new scanning ion microscopy technique that offers a number of advantages over SEM, the most important for the present work being greater resolution at high magnification and increased sensitivity to nanoscale surface topography [22,23]. In this work we investigate the evolution of nanostructures on InP surfaces following argon GCIB depth profiling using HIM as the observation technique. We utilise the high spatial resolution coupled with the surface topographic sensitivity of the HIM to investigate the formation mechanism of these structures, and to extract meaningful metrics about the GCIB etching of the InP.
achieving high open circuit voltages (VOC) has remained a fundamental issue, with Fermi-level splitting for inorganic semiconductor nanowires typically < 900 meV. Recently we have demonstrated the plasmonic properties of arrays of nanostructured InP, with results showing improvements in the surface enhanced Raman scattering effect from biomolecules, due to the close proximity of the metallic indium nanoparticles. This represents an opportunity to move away from traditional noble-metal SERS substrates to more cost-effective materials [12,13]. The introduction of the gas cluster ion beam (GCIB) into surface analysis represents a step-change in capability, with enhanced sputter yields that are greater for organics versus inorganics enabling selective etch rates [14], and reduced chemical damage that exists in a shallower surface layer than previously [15], much shallower than the information depth of XPS (< 10 nm) for example. Polymer and organic materials can be chemically profiled with high yields and minimal damage [16–18] and the same can be true for inorganics, albeit with markedly lower sputter yields [19–21]. Yet we have observed that for some compound semiconductors preferential sputtering can still occur, and in particular with indium-containing compounds this leads to the formation of an ultra-thin metallic indium layer at the surface [19]. When studying the surface topography of a sputter-etch crater after an XPS or SIMS experiment, AFM is often the first choice. This not only gives a visual representation of the surface, but also metrics on feature size, distribution, height and roughness can be obtained. It is inherently limited, however, when looking at increasingly smaller surface topography by the shape and dimensions of the AFM tip: larger tips simply cannot resolve features much smaller than the tip radius, which might
2. Materials and methods XPS and ion beam experiments were performed in a Theta Probe XPS instrument (Thermo Scientific, East Grinstead, UK) maintaining a base pressure of 5 × 10−10 mbar. The GCIB was a monomer and gascluster ion source (MAGCIS™) capable of delivering monatomic Ar+ ions at energies of 1–4 keV and Arn+ gas clusters at energies 1–10 keV for n = 75–2000 where n is the centre of the nominal distribution of atoms per cluster. For the present study, Ar300 clusters (i.e. 300 atoms
Fig. 1. HIM images of GCIB etched InP surfaces at incidence angles of (a) 90°, (b) and (c) 45°, to the surface plane (indicated by the diagrams to the left of the images). Images were collected at two angles, 0° (top down) and 54°. 679
Applied Surface Science 459 (2018) 678–685
A.J. Barlow et al.
3. Results
per cluster) at 8 keV were used for depth profiling, giving an energyper-atom of 26.7 eV. A total beam current of 5 nA was measured for this particular mode of the ion gun, and the beam was rastered across an area of 1 × 2 mm2. Spectra were analysed in CasaXPS [24] (version 2.3.18). Substrates of undoped InP(100) were mounted with a beam incidence angle of 90° (normal incidence) and 45° to the surface plane. The routine geometry for XPS analysis in the Theta Probe has the beam incidence angle at 45°, thus this angle was considered the most critical for this study. Scanning Auger microscopy was performed on a PHI 710 Scanning Auger Nanoprobe (SAN, Physical Electronics, Chanhassen, USA). The SAN utilises a thermal field emission electron source (operated at 20 keV, 10 nA) mounted coaxially with a cylindrical mirror analyser (CMA) for chemical analysis. Secondary electron images identified regions of interest and the electron beam was positioned at points on the imaged area for microanalysis. Auger spectra were collected at each point and then smoothed (9 point Savitzky-Golay), differentiated (5 point) and normalised using PHI MultiPak (version 9.8). HIM was performed in an ORION NanoFab (Carl Zeiss, Peabody, MA, USA) using an Everhart-Thornley (ET) secondary electron detector. The accelerator voltage was 30 kV with the gun pressure maintained at 2 × 10−6 mbar of helium with a 10 μm beam-defining aperture, resulting in a beam current of 0.2 pA. The images were collected at 2048 × 2048 pixels, and each line-averaged over 32 lines with a dwell time of 5 μs. Beam-current fluctuations were generally absent from these images, but those that did occur were small and removed from images in post-processing using fast-Fourier transform processing as previously described by the authors [25]. Measurements of particle radius and feature height were performed manually using ImageJ [26] (version 1.48). Images at a 5 μm field-of-view such as the one presented in Fig. 2(a) were used for these measurements. Particle radii were measured by drawing ellipses around each particle and taking the average of the radius along the X and Y axes. Feature heights were measured using a vertical line drawn from the top of a feature to the best judgement of its base at the surface level. As many measurements as possible were made on each image however the number of reliable measurements that could be made declined with increased nanostructure density.
Here we will focus on the topographical changes on InP following substantial ion beam etching by Ar gas clusters. Previous work suggested that there may be chemical damage due to indium reduction. As mentioned above, it is well-known in the literature that InP is prone to topography formation under monatomic ion bombardment, but to our knowledge until now no studies on surface topography have been performed following GCIB etching of InP. Fig. 1 presents HIM images of an area of the InP surface near the centre of an etch crater after 20,000 s of exposure to an 8 keV Ar300 GCIB at an incidence angle of 90° in Fig. 1(a) and 45° in Fig. 1(b) and (c). The field-of-view (FOV) for the images in (a) and (b) is 5 μm and the scale bar is 2 μm. The images in Fig. 1(c) are high magnification images of those in (b) to highlight the nanoscale topography that results after GCIB etching. Images were collected at two observation angles, 0° and 54°, by tilting the stage in the HIM, as indicated by the inset text within each image. The InP surface is covered with nodule-like nanostructures that are aligned with the angle of incidence of the GCIB, as demonstrated by the features appearing vertical in Fig. 1(a) and at 45° to the surface plane in Fig. 1(b) and (c). The images in Fig. 1 were collected from the centre of the defined etch position, but images from adjacent areas indicated that the topography extends across the entire crater for both ion beam incidence angles. The formation of the nanostructures appears to originate from the apex of the individual nodules. The high magnification images in Fig. 1(c) show that the features have a conical shape, increasing in size from the tip to the base of the structures. An investigation of the surface surrounding the main crater area, i.e. areas that have received substantially lower ion dose, reveals the initial origin of these conical nanostructures. The surface around the crater, as shown by the HIM image in Fig. 2(a), has well-distributed nanoparticles extending across a broad area. An analysis of the particle size given in Fig. 2(b) indicates that the average radius is 23 nm (histogram generated from 90 measurements, see Table S1 in the Supplementary Information). These nanoparticles have been seen to emerge from the material shortly after the oxide is removed during etching, and are attributed to a surface diffusion effect [9]. Briefly, the preferential etching of P atoms from the surface leads to an enrichment of In atoms, and these atoms migrate across the surface within a certain diffusion length that increases with temperature and the input of energy from the ion beam. Indium atoms gather at
Fig. 2. (a) HIM image of a region of the InP on the outer edge of the GCIB etch crater showing a broad coverage of nanoparticles, (b) a histogram of measured particle radii showing the average radius to be 23 nm across the image, generated from 90 measurements. 680
Applied Surface Science 459 (2018) 678–685
A.J. Barlow et al.
Fig. 3. XPS spectra of the (a) In3d5/2 and P2p (inset) regions and (b) InMNN X-ray induced Auger region for the GCIB etched InP and reference In foil from the XPSSurfA database [27].
Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.apsusc.2018.07.195. XPS provides insight into the chemical state of the surface across a large area (400 × 800 µm2 in the current work), and is thus an average analysis of many nanoparticles and the underlying InP surface. Interrogation of the chemical state of individual nanoparticles requires an instrument capable of both high resolution microanalysis with chemical state sensitivity. In this study a scanning Auger nanoprobe (SAN) was employed for this purpose. A SAN derived secondary electron image (2 µm FoV) is presented in Fig. 4(left). The figure shows the nanoparticles immediately after removal of the oxide layer at the InP surface. Electron-induced InMNN spectra were collected at six points, three directly on nanoparticles (red targets) and three in regions of bare InP surface (yellow targets). These spectra are presented in Fig. 4(right) as normalised and differentiated spectra. Also presented are InMNN spectra from the unetched InP surface (black dotted line) and from a sputter cleaned In reference foil (black solid line). The spectra collected from regions devoid of nanoparticles show the same peak position as the unetched InP surface at
nucleation sites forming small clusters, and additional atoms preferentially join nearby clusters leading to the formation of the nanoparticles. No particles were observed on the InP surface prior to the GCIB exposure, as shown in HIM images in the Supplementary Information (Fig. S1), and the surface features were not induced by the He+ ion beam during HIM imaging. XPS analysis of the GCIB etched InP surface is presented in Fig. 3. The P2p peak (Fig. 3a inset) was fitted with a doublet of separation 0.85 eV, and the P2p3/2 component was observed at 128.6 eV, as expected for a metal phosphide. The In3d5/2 peak (Fig. 3a) could be fitted with two components, a broad symmetric component at 444.4 eV for InP, and a narrower asymmetric component at 443.7 eV attributed to metallic indium (further details about the fitting is provided in the Supplementary Information). The X-ray induced InMNN Auger feature (Fig. 3b) shows sharp features at 402.9 and 410.5 eV (kinetic energy). A reference InMNN spectrum of metallic In foil obtained from the XPSSurfA database [27] is also presented, and validates that the sharp features are from a metallic In signal. The In3d5/2 position from this reference dataset is 443.5 eV, in good agreement with the fitted metallic component in Fig. 3a.
Fig. 4. Secondary electron image (left) of nanoparticles on InP formed immediately after removal of the oxide layer, and Auger spectra (right) of three individual nanoparticles in red and three bare InP in yellow. Reference spectra of cleaned InP and pure metallic In foil are provided in black. A peak shift is observed from 411.1 eV to 412.5 eV off and on the nanoparticles, directly indicative of an enrichment of metallic indium at the nanoparticles.
681
Applied Surface Science 459 (2018) 678–685
A.J. Barlow et al.
Fig. 5. (a) Stitched microscope image of the GCIB etch crater on InP, and (b)–(g) HIM images at three magnifications showing the variation in nanostructures at positions of 0 µm, 1000 µm, 1500 µm, 2000 µm, 2500 µm, and 3500 µm respectively across the crater edge. The white arrows highlight the formation of the tail structures that ultimately lead to the nanostructured surface. HIM images collected at 54°.
sputter craters made with a variation in total ion dose (or equivalently etch time). Thus from a single etch crater we can gain an insight into the formation mechanism of the nanostructures for varying total ion dose. The outer-most region of the crater is exposed to the lowest total ion dose. Consequently, in Fig. 5(b) where the HIM is imaging the outside of the crater a low density of nanoparticles is observed. At the highest magnification some roughening of the surface surrounding the nanoparticles is evident. In Fig. 4(c), representative of an increased ion dose, the density of particles is increased. This is a direct result of the combined effect of greater cumulative preferential sputtering of P, and increased diffusion of In atoms at the surface that results from this sputtering coupled with the continued energy input from the ion beam. Also visible are small features leading away from the particles as indicated by the white arrows in the figure. It is worth stressing at this point that the GCIB incident angle is 45° to the normal and from right to left across the surface in the images. In Fig. 3(d) the surface is increasing in roughness, and the features to the left of the particles now become clearer, and appear as ‘tails’ to the particles as indicated by the white arrows. This feature is evident on all particles in the image and are all aligned with the incident beam direction. The particle density continues to increase in Fig. 5(e), as does the tail length and height. In some cases, the tails from one particle approach the feature of another
411.1 eV. On the nanoparticles, however, this peak has shifted to a higher kinetic energy at 412.5 eV, much closer to the pure metallic indium spectrum at 413.8 eV. These data, coupled with the XPS data already presented, confirm that the nanoparticles are composed of metallic indium. Fig. 5(a) shows an overview microscope image of the GCIB etch crater on the InP sample. It is a 1 × 2 mm2 area, visible as the lightest area in the image. The mark at the bottom of the image was simply used as a reference point for position registration and correlative analysis between instruments. A series of HIM images were collected across the crater edge, with the stage shifted by a set distance between each image. The images in Fig. 5(b)–(g) were collected at the indicated points on the microscope image. The lateral separation of the images in (b)–(c) was 1000 µm, (c)–(f) was 500 µm, and (f)–(g) 1000 µm. Images at three magnifications are presented, with scale bars inset within the images. Since it is expected that the spot size of the GCIB is quite broad for the XPS instrument used here (as opposed to in SIMS instruments for example), and the beam is rastered over an area of similar size to the FWHM of the spot, the collection of images over the crater edge therefore represents a variation in the total GCIB ion dose that the InP surface received. It follows then that a set of measurements made over a broad crater edge is an analogous representation of a series of GCIB 682
Applied Surface Science 459 (2018) 678–685
A.J. Barlow et al.
Fig. 6. The mechanism for nanostructure formation during GCIB etching of InP. Total ion dose is increasing from left to right. Initially the nanoparticles present on the surface create a shadow that masks a region of the InP from the beam, leading to masked material on the far side of the particle from the beam. As etching continues, or ion dose increases, this feature grows in scale, ultimately leading to the nanostructured formations observed in the last panel.
as the ones used in this work are expected to collect secondary electron yield from surfaces with energies up to 50 eV. To test this hypothesis, low kinetic energy spectra from 10 to 50 eV were collected within the SAN and are presented in the Fig. S4. We observe that the intensity of electron signal in this energy range is consistently greater at the nanoparticles than that of the surrounding InP areas, demonstrating that the material contrast is the result, at least in part, of increased metallic content at the tip. This increased material contrast allowing metallic In to be identified is not observed in the literature for SEM images of nanostructured InP, and is an advantage afforded by the HIM. The mechanism suggests that the nanoparticles have lesser sputter yield compared to the bulk InP. If the sputter yield was greater, the nanoparticles would erode more quickly than the formation of the nanostructures could occur leading to, at most, a surface with enhanced roughness. In the case where the sputter yield of the nanoparticles is similar to or less than the InP the particle would remain intact for a greater length of time, allowing for the nanostructure formation to become the dominant effect. The mechanism presented here contradicts some previously discussed mechanisms of texture formation during sputtering of InP whereby the nanostructures are proposed to result from diffusiondriven growth of filaments from the bulk, capped by a metallic In shell [7,28,29]. It does, however, agree with the proposed mechanisms of Homma [6] and Sanatinia [9], where preferential sputtering leads to diffusion and particle growth initially, then removal of material becomes the dominant mechanism and masked sputtering occurs. The experimental conditions for previous studies were typically vastly different to those presented here. Ion beams used were monoatomic, high beam energy and thus presumably high beam current (µA). Much of the literature also discusses filament growth at elevated temperatures, where surface and bulk diffusion is enhanced. Here the nanostructure formation is at room temperature, with a beam current of only 5 nA and an energy per atom of 26.7 eV, both orders of magnitude less than previously reported experiments. It is proposed here that the formation of the nanostructures in this regime, i.e. the low beam current and energy-per-atom of the Ar GCIB, is dominated by physical sputtering effects and contrasting sputter yields between the indium nanoparticles and the bulk InP. This effect has also been observed by the authors on polyvinyl chloride (PVC) surfaces as presented in Supplementary Fig. S2. Two competing processes are present in this mechanism. Diffusion of mobile In atoms is the mechanism behind the formation of the initial nanoparticles. The limiting factor for the diffusion process ultimately
particle. In Fig. 5(f) and (g) we are now imaging the base of the GCIB etch crater, representing the greatest total ion dose, and the density of the features is the highest. It can be seen at the highest magnification that the tail features are now taller and fin-like, and in some cases multiple features have coalesced into single features. In Fig. 5(g) the characteristic nanostructured InP surface is visible. It is proposed here that the mechanism of the nanostructure formation is a ‘self-masking’ effect during the etching of the InP, directly caused by the presence of the nanoparticles emerging early in the etching process. This is described schematically in Fig. 6. In the first panel the nanoparticles are present on the surface, with the GCIB incident upon them and the InP surface at 45°. The initial nanoparticle formation is due to a surface diffusion effect resulting from the preferential sputtering of P from the surface after the removal of the topmost oxide layer [9]. In the absence of sample rotation the nanoparticles create a shadowed region that the GCIB cannot reach and thus cannot etch. The result is shown in the second panel, where the shadow created by the particle leads to a region of masked material that now protrudes from the newly exposed surface that is the crater base. The height of this protrusion could be considered the etch depth, since it is a measure of how much material has been removed from the surface in areas devoid of nanoparticles. In the third and fourth panels in Fig. 6 this process continues as the total ion dose is increased. In the case presented here this is equivalent to moving further across the crater edge. As more material around the particles is removed, the masked material increases in height above the new surface. The features grow and are inherently aligned with the incident beam angle. The ultimate result is shown in the fifth panel in Fig. 6, where the surface is now covered in nanostructures. The nanoparticle itself experiences some amount of ion beam etching. This can be seen from the highest magnification images in Fig. 5. The particles appear somewhat spherical initially, at least symmetric in shape. As the ion dose increases, the particles become elongated along the axis of the ion beam. The side of the particle facing the beam experiences the greatest dose, decreasing around the particle until the opposite face experiences effectively no dose. This has the effect of undercutting the particle and simultaneously eroding the sides to produce an elongated particle. This enhances the production of the nanostructures making the features appear to protrude even further from the surface. It is also worth noting the apparent material contrast at the tip of the protrusions. This could be due to the increased metallic content at the tip, and thus increased secondary electron emission. ET detectors such 683
Applied Surface Science 459 (2018) 678–685
A.J. Barlow et al.
Fig. 7. (a) Etch depth relative to the original surface level as a function of position across the GCIB crater edge. The dotted curve is a fit to the data using a Gaussian function. (b) The same data as in (a), but converted to show etch depth versus total ion dose.
with published data of etch depths versus ion dose from similar Ar GCIB conditions on other materials [30].
becomes the length across which diffusion can occur. By increasing the process temperature one could increase the diffusion length and promote growth of features outward from the surface. All sputtering in the present work was performed at constant ambient temperature and thus diffusion and the length scales across which it can occur were expected to be constant. During ion bombardment both diffusion and sputtering is occurring. However, once a nanoparticle reaches a certain size greater than the diffusion length then particle growth ceases since liberated In can no longer migrate far enough to reach the particle before sputtering removes material around the particle. At this point sputtering becomes the dominant mechanism and the particles begin to act as masks for the underlying material, leading to nanostructure formation through masked sputtering. Feature heights from the full images used to generate Fig. 5 were measured using the processing software ImageJ (see Section 2), and were corrected for the viewing angle. As many measurements as possible were made on each image, provided the complete feature was visible allowing an accurate indication of the height however the number of measurements possible declined across the crater. Table S1 provides detailed particle analysis and Fig. S3 presents a plot of these results including the particle radius. While the feature height is observed to increase rapidly across the crater, the nanoparticle radius is largely constant, supporting the mechanism described above whereby the diffusion length limits the maximum particle size and ion beam sputtering becomes the dominant mechanism. The measured height was then adjusted for the diameter of the nanoparticles and the resulting figures were considered to be the “etch depth”, i.e. the depth below the initial surface level. These are presented in Fig. 7(a) as Relative Etch Depth, i.e. the etch depth relative to the surface level in nanometres. Moving across the crater edge the etch depth increases as would be expected given the increase in total ion dose at the surface. This reaches a maximum (i.e. the deepest part of the crater) around 3000 µm from the first measurement (i.e. the farthest edge of the crater). If it is assumed that the profile of the GCIB is largely Gaussian in shape, then we can fit the edge of the crater to a Gaussian profile, shown as the dotted grey line in Fig. 7(a). This gives a beam FWHM of 1.5 mm, which is not unexpected for the GCIB used in this work. These beams are non-analytical in XPS, and thus do not have the same rigorous constraints as in SIMS. Since the beam current is measured, and the total etch time was set, given this newly fitted Gaussian profile the abscissa can be scaled and converted to show, instead of position, a total ion dose for each depth measurement. Thus a plot of etch depth versus total ion dose can be generated, and is given in Fig. 7(b). The resulting plot indicates a linear relationship between the total ion dose of the GCIB and the resulting InP etch depth. This linear relationship compares favourably
4. Conclusion Helium ion microscopy (HIM) has been used to observe the formation of nanoscale topography on indium phosphide (InP) during argon gas cluster ion beam (Ar GCIB) etching. The technique is extremely well-suited to studying the surfaces of etch craters, and reveals nanoscale features that other commonly-used techniques such as atomic force and scanning electron microscopies would have greater difficulty observing at such high magnification. Shortly after GCIB etching begins the native oxide is removed and metallic indium nanoparticles form on the InP surface. It is proposed that these particles mask the underlying surface from the GCIB, and over time a surprisingly regular array of nanostructured InP results, with the metallic nanoparticles at the tip. By studying the surface across the edge of the GCIB etch crater, insight can be gained into how the surface is varied with total GCIB ion dose. As the dose increases, the density of the nanoparticles increases, as does the resulting nanostructure density. The height of the features, or equivalently the depth of the crater, is increased with total ion dose. The relationship between InP etch rate and the total GCIB ion dose is found to be linear. Acknowledgements X-ray photoelectron spectroscopy (XPS) data were acquired at the National ESCA and XPS Users’ Service (NEXUS), Newcastle University, UK. Helium ion microscopy (HIM) was also performed at NEXUS, and the instrument was purchased as part of a package funded by EPSRC’s ‘Great Eight’ capital funding grant EP/K022679/1 and Newcastle University. This work incorporates data from the Victorian node of the Australian National Fabrication Facility (ANFF), a company established under the National Collaborative Research Infrastructure Strategy to provide nano and microfabrication facilities for researchers in Australia, through the La Trobe University Centre for Materials and Surface Science. Data are reproduced under a Creative Commons licence (CC BY-NC 4.0 International). References [1] R.M. Bradley, J.M.E. Harper, Theory of ripple topography induced by ion bombardment, J. Vac. Sci. Technol. A 6 (1998) 2390, https://doi.org/10.1116/1. 575561. [2] F. Frost, A. Schindler, F. Bigl, Roughness evolution of ion sputtered rotating InP surfaces: pattern formation and scaling laws, Phys. Rev. Lett. 85 (2000) 4116–4119,
684
Applied Surface Science 459 (2018) 678–685
A.J. Barlow et al.
4791669. [17] D. Rading, R. Moellers, H.-G. Cramer, E. Niehuis, Dual beam depth profiling of polymer materials: comparison of C60 and Ar cluster ion beams for sputtering, Surf. Inter. Anal. 45 (2012) 171–174, https://doi.org/10.1002/sia.5122. [18] B.R. Strohmeier, R.G. White, T.S. Nunney, P. Mack, A.E. Wright, Enhanced surface and in-depth characterisation of organic and inorganic materials using XPS and soft depth profiling with argon cluster ion beams, Microsc. Microanal. 18 (Suppl. 2) (2012) 918, https://doi.org/10.1017/S1431927612006447. [19] A.J. Barlow, J.F. Portoles, P.J. Cumpson, Observed damage during Argon gas cluster depth profiles of compound semiconductors, J. Appl. Phys. 116 (2014) 054908, https://doi.org/10.1063/1.4892097. [20] J.D.P. Counsell, A.J. Roberts, W. Boxford, C. Moffitt, K. Takahashi, Reduced preferential sputtering of TiO2 using massive argon clusters, J. Surf. Anal. 20 (2014) 211–215. [21] N. Sano, A.J. Barlow, C. Tsakonas, W. Cranton, P.J. Cumpson, Optimal conditions for gas cluster ion beams in studying inorganic interface species: improved chemical information at a ZnO interface, Surf. Inter. Anal. (2016), https://doi.org/10.1002/ sia.5949 (ECASIA Special Issue). [22] B.W. Ward, J.A. Notte, N.P. Economou, Helium ion microscope: a new tool for nanoscale microscopy and metrology, J. Vac. Sci. Technol. B 24 (2006) 2871–2874, https://doi.org/10.1116/1.2357967. [23] R. Hill, J.A. Notte, L. Scipioni, Scanning helium ion microscopy, Adv. Imaging Electron. Phys. 170 (2012) 65–148, https://doi.org/10.1016/B978-0-12-394396-5. 00002-6. [24] Neal Fairley, CasaXPS: Processing Software for XPS, AES, SIMS and More, Casa Software Ltd, Teignmouth, UK, 2016. [25] A.J. Barlow, J.F. Portoles, N. Sano, P.J. Cumpson, Removing beam current artefacts in helium ion microscopy: a comparison of image processing techniques, Microscop. Microanal. 22 (2016) 939–947, https://doi.org/10.1017/S1431927616011673. [26] C.A. Schneider, W.S. Rasband, K.W. Eliceiri, NIH Image to ImageJ: 25 years of image analysis, Nat. Methods 9 (2012) 671–675. [27] A.J. Barlow, R.T. Jones, A.J. McDonald, P.J. Pigram, XPSSurfA: an open collaborative XPS data repository using the CMSShub platform, Surf. Interface Anal. 50 (2018) 527–540, https://doi.org/10.1002/sia.6417. [28] H.W. Gries, Radiation-induced sample modification in surface analysis: InP as an extreme example, Surf. Inter. Anal. 14 (1989) 611–618, https://doi.org/10.1002/ sia.740141009. [29] F. Okuyama, J. Kato, Novel fiber growth on Ar+-sputtered InP, J. Vac. Sci. Technol. B 12 (1994) 3054, https://doi.org/10.1116/1.587559. [30] I. Yamada, J. Matsuo, N. Toyoda, A. Kirkpatrick, Materials processing by gas cluster ion beams, Mater. Sci. Eng. R. 34 (2001) 231–295, https://doi.org/10.1016/S0927796X(01)00034-1.
https://doi.org/10.1103/PhysRevLett. 85.4116. [3] Y. Homma, H. Okamato, Y. Ishii, Surface microtopography and compositional change of cesium-ion-bombarded semiconductor surfaces, Jpn. J. Appl. Phys. 24 (1985) 934, https://doi.org/10.1143/JJAP.24.934. [4] T. Ogiwara, S. Tanuma, Y. Nagasawa, N. Ikeo, Auger depth profiling analyses of InP/GaInAsP multilayer thin films, Microbeam Anal. 2 (1993) 133–138. [5] S.W. MacLaren, J.E. Baker, N.L. Finnegan, C.M. Loxton, Surface roughness development during sputtering of GaAs and InP: evidence for the role of surface diffusion in ripple formation and sputter cone development, J. Vac. Sci. Technol. A 10 (1992) 468–476, https://doi.org/10.1116/1.578173. [6] Y. Homma, Formation mechanism of indium microcrystals on ion-bombarded InP surfaces, J. Surf. Anal. 3 (1997) 641–645. [7] M.P. Seah, S.J. Spencer, P.J. Cumpson, J.E. Johnstone, Cones formed during sputtering of InP and their use in defining AFM tip shapes, Appl. Surf. Sci. 144 (1999) 151–155, https://doi.org/10.1016/S0169-4332(98)00794-6. [8] T. Radny, H. Gnaser, Self-organizing nanodot structures on InP surfaces evolving under low-energy ion irradiation: analysis of morphology and composition, Nanoscale Res. Lett. 9 (2014) 403, https://doi.org/10.1186/1556-276X-9-403. [9] R. Sanatinia, et al., Wafer-scale self-organized InP nanopillars with controlled orientation for photovoltaic devices, Nanotechnology 26 (2015) 415304, https://doi. org/10.1088/0957-4484/26/41/415304. [10] J. Wallentin, et al., InP nanowire array solar cells achieving 13.8% efficiency by exceeding the ray optics limit, Science 339 (2013) 1057–1060, https://doi.org/10. 1126/science.1230969. [11] T.-T.D. Tran, et al., High brightness InP micropillars grown on silicon with Fermi level splitting larger than 1 eV, Nano Lett. 14 (2014) 3235–3240, https://doi.org/ 10.1021/nl500621j. [12] B.J. Murdoch, J.F. Portoles, S. Tardio, A.J. Barlow, I.W. Fletcher, P.J. Cumpson, Visible wavelength surface-enhanced Raman spectroscopy from In-InP nanopillars for biomolecule detection, Appl. Phys. Lett. 109 (2016) 253105, https://doi.org/10. 1063/1.4972558. [13] B.J. Murdoch, A.J. Barlow, I.W. Fletcher, P.J. Cumpson, The plasmonic properties of argon cluster-bombarded InP surfaces, Appl. Phys. Lett. 111 (2017) 081603, https://doi.org/10.1063/1.4993535. [14] P.J. Cumpson, J.F. Portoles, N. Sano, A.J. Barlow, M. Birch, Depth profiling organic/inorganic interfaces by argon gas cluster ion beams: sputter yield data for biomaterials, in-vitro diagnostic and implant applications, Surf. Int. Anal. 45 (2013) 1859, https://doi.org/10.1002/sia.5333. [15] N. Winograd, Molecular depth profiling, Surf. Inter. Anal. 45 (2013) 3–8, https:// doi.org/10.1002/sia.4913. [16] P.J. Cumpson, J.F. Portoles, N. Sano, Material dependence of argon cluster ion sputter yield in polymers: Method and measurements of relative sputter yields for 19 polymers, J. Vac. Sci. Technol. A 31 (2013) 020605, https://doi.org/10.1116/1.
685
Contents lists available at ScienceDirect
Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc
Full Length Article
Observing the evolution of regular nanostructured indium phosphide after gas cluster ion beam etching
T
⁎
Anders J. Barlowa, , Naoko Sanob, Billy J. Murdochc, Jose F. Portolesb, Paul J. Pigrama, Peter J. Cumpsonb a
Centre for Materials and Surface Science (CMSS), Department of Chemistry and Physics, La Trobe University, Melbourne, Victoria 3086, Australia National ESCA and XPS Users’ Service (NEXUS), School of Engineering, Newcastle University, Newcastle upon Tyne, Tyne and Wear NE1 7RU, UK c School of Science, RMIT University, Melbourne, Victoria 3001, Australia b
A R T I C LE I N FO
A B S T R A C T
Keywords: Helium ion microscopy Gas cluster ion beam Indium phosphide Nanostructure XPS SAN
Indium phosphide (InP) surfaces develop a pronounced nanostructured texture upon irradiation by energetic ion beams. We have observed the mechanism of nanostructure evolution of InP under irradiation by an Ar gas cluster ion beam (GCIB) using helium ion microscopy (HIM). Initially, metallic indium nanoparticles form on the surface after removal of the top-most oxide layer. These nanoparticles form a mask which shadows the underlying InP. As the ion dose is increased, the masking effect results in substantial nanostructured topography in the form of pillars or nanocones, oriented along the axis of the incident GCIB. The surface sensitivity and high resolution of the HIM facilitates the direct observation of the metallic indium cap at the top of the pillars.
1. Introduction The impact upon a solid surface by a beam of energetic ions can result in significant topographical and chemical changes along with the ejection of surface atoms and fragments, and implantation effects in the near-surface region. This is put to practical use in surface analysis by secondary ion mass spectrometry (SIMS) where the mass of the ejected fragments is analysed to deduce composition and molecular structure, and in X-ray photoelectron spectroscopy (XPS), Auger electron spectroscopy (AES) and SIMS for the removal of unwanted contamination or residues from a surface prior to analysis, or for depth-profiling through a layer or interface by successive etching cycles of surface material during analysis. Prolonged exposure to an ion beam, such as during a sputter depth-profile, can lead to surface structures that have a specific morphological evolution based on an equilibrium between surface roughening (ion erosion) and surface diffusion [1]. This has been demonstrated a number of times on compound semiconductors, and in particular with indium phosphide (InP) [2–5]. Homma et al. [3] observed the preferential sputtering of phosphor from InP under irradiation by monatomic argon ions, coupled with the formation of surface features of the order of 10–100 nm in size depending on beam conditions. The surface morphology that results from such exposure has consequences in analytical depth profiling, where interface depth resolution in a multi-layered stack that includes InP can
⁎
be degraded [4]. Nanostructured ‘cones’ were observed by MacLaren et al. and attributed to a surface diffusion effect whereby indium metal is enriched at the surface leading to cone formation due to differential sputter rates of the metallic indium versus InP [5]. This was later supported by Homma who observed a masking effect from particles that exist on the InP surface prior to ion beam etching [6]. Due to the regular and nano-sized structures that could be obtained, these surfaces were subsequently found to be useful for the determination of tip profiles in atomic force microscopy (AFM) [7]. Recently, atom probe tomography has been used to study individual structures on ion-irradiated InP and demonstrated indium enrichment at the surface with very high spatial resolution [8]. Arrays of geometrically optimised InP nanostructures should, in principle, offer highly efficient and cost effective alternatives to planar photovoltaic cells. Employing InP is advantageous as it offers efficient absorption of light in very thin layers, with a favourable match between its bandgap and the peak energy in solar radiation. Indeed Sanatinia et al. demonstrated the applicability of a surface of InP “nanopillars” generated by a nitrogen ion beam [9]. Nanostructured InP surfaces could be a means towards generating efficient nanoparticle-based solar cells. In practice however their large surface area leads to decreased quantum yields due to non-radiative charge recombination pathways via surface defect states [10]. While adequate short circuit currents can be achieved using highly doped nanostructure-array photovoltaics [11],
Corresponding author. E-mail address: [email protected] (A.J. Barlow).
https://doi.org/10.1016/j.apsusc.2018.07.195 Received 22 March 2018; Received in revised form 6 July 2018; Accepted 28 July 2018 Available online 30 July 2018 0169-4332/ © 2018 Elsevier B.V. All rights reserved.
Applied Surface Science 459 (2018) 678–685
A.J. Barlow et al.
be on the order of 10 nm under ideal conditions. Moreover, high aspect ratio features in the vertical direction, such as those observed on InP, prevent imaging of finer surface details since the tip sidewalls impinge upon neighbouring features. Scanning electron microscopy (SEM) is often another popular choice for studying surface topography on the nano-scale, though it too can be limited in resolution at very high magnification. Helium ion microscopy (HIM) is a relatively new scanning ion microscopy technique that offers a number of advantages over SEM, the most important for the present work being greater resolution at high magnification and increased sensitivity to nanoscale surface topography [22,23]. In this work we investigate the evolution of nanostructures on InP surfaces following argon GCIB depth profiling using HIM as the observation technique. We utilise the high spatial resolution coupled with the surface topographic sensitivity of the HIM to investigate the formation mechanism of these structures, and to extract meaningful metrics about the GCIB etching of the InP.
achieving high open circuit voltages (VOC) has remained a fundamental issue, with Fermi-level splitting for inorganic semiconductor nanowires typically < 900 meV. Recently we have demonstrated the plasmonic properties of arrays of nanostructured InP, with results showing improvements in the surface enhanced Raman scattering effect from biomolecules, due to the close proximity of the metallic indium nanoparticles. This represents an opportunity to move away from traditional noble-metal SERS substrates to more cost-effective materials [12,13]. The introduction of the gas cluster ion beam (GCIB) into surface analysis represents a step-change in capability, with enhanced sputter yields that are greater for organics versus inorganics enabling selective etch rates [14], and reduced chemical damage that exists in a shallower surface layer than previously [15], much shallower than the information depth of XPS (< 10 nm) for example. Polymer and organic materials can be chemically profiled with high yields and minimal damage [16–18] and the same can be true for inorganics, albeit with markedly lower sputter yields [19–21]. Yet we have observed that for some compound semiconductors preferential sputtering can still occur, and in particular with indium-containing compounds this leads to the formation of an ultra-thin metallic indium layer at the surface [19]. When studying the surface topography of a sputter-etch crater after an XPS or SIMS experiment, AFM is often the first choice. This not only gives a visual representation of the surface, but also metrics on feature size, distribution, height and roughness can be obtained. It is inherently limited, however, when looking at increasingly smaller surface topography by the shape and dimensions of the AFM tip: larger tips simply cannot resolve features much smaller than the tip radius, which might
2. Materials and methods XPS and ion beam experiments were performed in a Theta Probe XPS instrument (Thermo Scientific, East Grinstead, UK) maintaining a base pressure of 5 × 10−10 mbar. The GCIB was a monomer and gascluster ion source (MAGCIS™) capable of delivering monatomic Ar+ ions at energies of 1–4 keV and Arn+ gas clusters at energies 1–10 keV for n = 75–2000 where n is the centre of the nominal distribution of atoms per cluster. For the present study, Ar300 clusters (i.e. 300 atoms
Fig. 1. HIM images of GCIB etched InP surfaces at incidence angles of (a) 90°, (b) and (c) 45°, to the surface plane (indicated by the diagrams to the left of the images). Images were collected at two angles, 0° (top down) and 54°. 679
Applied Surface Science 459 (2018) 678–685
A.J. Barlow et al.
3. Results
per cluster) at 8 keV were used for depth profiling, giving an energyper-atom of 26.7 eV. A total beam current of 5 nA was measured for this particular mode of the ion gun, and the beam was rastered across an area of 1 × 2 mm2. Spectra were analysed in CasaXPS [24] (version 2.3.18). Substrates of undoped InP(100) were mounted with a beam incidence angle of 90° (normal incidence) and 45° to the surface plane. The routine geometry for XPS analysis in the Theta Probe has the beam incidence angle at 45°, thus this angle was considered the most critical for this study. Scanning Auger microscopy was performed on a PHI 710 Scanning Auger Nanoprobe (SAN, Physical Electronics, Chanhassen, USA). The SAN utilises a thermal field emission electron source (operated at 20 keV, 10 nA) mounted coaxially with a cylindrical mirror analyser (CMA) for chemical analysis. Secondary electron images identified regions of interest and the electron beam was positioned at points on the imaged area for microanalysis. Auger spectra were collected at each point and then smoothed (9 point Savitzky-Golay), differentiated (5 point) and normalised using PHI MultiPak (version 9.8). HIM was performed in an ORION NanoFab (Carl Zeiss, Peabody, MA, USA) using an Everhart-Thornley (ET) secondary electron detector. The accelerator voltage was 30 kV with the gun pressure maintained at 2 × 10−6 mbar of helium with a 10 μm beam-defining aperture, resulting in a beam current of 0.2 pA. The images were collected at 2048 × 2048 pixels, and each line-averaged over 32 lines with a dwell time of 5 μs. Beam-current fluctuations were generally absent from these images, but those that did occur were small and removed from images in post-processing using fast-Fourier transform processing as previously described by the authors [25]. Measurements of particle radius and feature height were performed manually using ImageJ [26] (version 1.48). Images at a 5 μm field-of-view such as the one presented in Fig. 2(a) were used for these measurements. Particle radii were measured by drawing ellipses around each particle and taking the average of the radius along the X and Y axes. Feature heights were measured using a vertical line drawn from the top of a feature to the best judgement of its base at the surface level. As many measurements as possible were made on each image however the number of reliable measurements that could be made declined with increased nanostructure density.
Here we will focus on the topographical changes on InP following substantial ion beam etching by Ar gas clusters. Previous work suggested that there may be chemical damage due to indium reduction. As mentioned above, it is well-known in the literature that InP is prone to topography formation under monatomic ion bombardment, but to our knowledge until now no studies on surface topography have been performed following GCIB etching of InP. Fig. 1 presents HIM images of an area of the InP surface near the centre of an etch crater after 20,000 s of exposure to an 8 keV Ar300 GCIB at an incidence angle of 90° in Fig. 1(a) and 45° in Fig. 1(b) and (c). The field-of-view (FOV) for the images in (a) and (b) is 5 μm and the scale bar is 2 μm. The images in Fig. 1(c) are high magnification images of those in (b) to highlight the nanoscale topography that results after GCIB etching. Images were collected at two observation angles, 0° and 54°, by tilting the stage in the HIM, as indicated by the inset text within each image. The InP surface is covered with nodule-like nanostructures that are aligned with the angle of incidence of the GCIB, as demonstrated by the features appearing vertical in Fig. 1(a) and at 45° to the surface plane in Fig. 1(b) and (c). The images in Fig. 1 were collected from the centre of the defined etch position, but images from adjacent areas indicated that the topography extends across the entire crater for both ion beam incidence angles. The formation of the nanostructures appears to originate from the apex of the individual nodules. The high magnification images in Fig. 1(c) show that the features have a conical shape, increasing in size from the tip to the base of the structures. An investigation of the surface surrounding the main crater area, i.e. areas that have received substantially lower ion dose, reveals the initial origin of these conical nanostructures. The surface around the crater, as shown by the HIM image in Fig. 2(a), has well-distributed nanoparticles extending across a broad area. An analysis of the particle size given in Fig. 2(b) indicates that the average radius is 23 nm (histogram generated from 90 measurements, see Table S1 in the Supplementary Information). These nanoparticles have been seen to emerge from the material shortly after the oxide is removed during etching, and are attributed to a surface diffusion effect [9]. Briefly, the preferential etching of P atoms from the surface leads to an enrichment of In atoms, and these atoms migrate across the surface within a certain diffusion length that increases with temperature and the input of energy from the ion beam. Indium atoms gather at
Fig. 2. (a) HIM image of a region of the InP on the outer edge of the GCIB etch crater showing a broad coverage of nanoparticles, (b) a histogram of measured particle radii showing the average radius to be 23 nm across the image, generated from 90 measurements. 680
Applied Surface Science 459 (2018) 678–685
A.J. Barlow et al.
Fig. 3. XPS spectra of the (a) In3d5/2 and P2p (inset) regions and (b) InMNN X-ray induced Auger region for the GCIB etched InP and reference In foil from the XPSSurfA database [27].
Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.apsusc.2018.07.195. XPS provides insight into the chemical state of the surface across a large area (400 × 800 µm2 in the current work), and is thus an average analysis of many nanoparticles and the underlying InP surface. Interrogation of the chemical state of individual nanoparticles requires an instrument capable of both high resolution microanalysis with chemical state sensitivity. In this study a scanning Auger nanoprobe (SAN) was employed for this purpose. A SAN derived secondary electron image (2 µm FoV) is presented in Fig. 4(left). The figure shows the nanoparticles immediately after removal of the oxide layer at the InP surface. Electron-induced InMNN spectra were collected at six points, three directly on nanoparticles (red targets) and three in regions of bare InP surface (yellow targets). These spectra are presented in Fig. 4(right) as normalised and differentiated spectra. Also presented are InMNN spectra from the unetched InP surface (black dotted line) and from a sputter cleaned In reference foil (black solid line). The spectra collected from regions devoid of nanoparticles show the same peak position as the unetched InP surface at
nucleation sites forming small clusters, and additional atoms preferentially join nearby clusters leading to the formation of the nanoparticles. No particles were observed on the InP surface prior to the GCIB exposure, as shown in HIM images in the Supplementary Information (Fig. S1), and the surface features were not induced by the He+ ion beam during HIM imaging. XPS analysis of the GCIB etched InP surface is presented in Fig. 3. The P2p peak (Fig. 3a inset) was fitted with a doublet of separation 0.85 eV, and the P2p3/2 component was observed at 128.6 eV, as expected for a metal phosphide. The In3d5/2 peak (Fig. 3a) could be fitted with two components, a broad symmetric component at 444.4 eV for InP, and a narrower asymmetric component at 443.7 eV attributed to metallic indium (further details about the fitting is provided in the Supplementary Information). The X-ray induced InMNN Auger feature (Fig. 3b) shows sharp features at 402.9 and 410.5 eV (kinetic energy). A reference InMNN spectrum of metallic In foil obtained from the XPSSurfA database [27] is also presented, and validates that the sharp features are from a metallic In signal. The In3d5/2 position from this reference dataset is 443.5 eV, in good agreement with the fitted metallic component in Fig. 3a.
Fig. 4. Secondary electron image (left) of nanoparticles on InP formed immediately after removal of the oxide layer, and Auger spectra (right) of three individual nanoparticles in red and three bare InP in yellow. Reference spectra of cleaned InP and pure metallic In foil are provided in black. A peak shift is observed from 411.1 eV to 412.5 eV off and on the nanoparticles, directly indicative of an enrichment of metallic indium at the nanoparticles.
681
Applied Surface Science 459 (2018) 678–685
A.J. Barlow et al.
Fig. 5. (a) Stitched microscope image of the GCIB etch crater on InP, and (b)–(g) HIM images at three magnifications showing the variation in nanostructures at positions of 0 µm, 1000 µm, 1500 µm, 2000 µm, 2500 µm, and 3500 µm respectively across the crater edge. The white arrows highlight the formation of the tail structures that ultimately lead to the nanostructured surface. HIM images collected at 54°.
sputter craters made with a variation in total ion dose (or equivalently etch time). Thus from a single etch crater we can gain an insight into the formation mechanism of the nanostructures for varying total ion dose. The outer-most region of the crater is exposed to the lowest total ion dose. Consequently, in Fig. 5(b) where the HIM is imaging the outside of the crater a low density of nanoparticles is observed. At the highest magnification some roughening of the surface surrounding the nanoparticles is evident. In Fig. 4(c), representative of an increased ion dose, the density of particles is increased. This is a direct result of the combined effect of greater cumulative preferential sputtering of P, and increased diffusion of In atoms at the surface that results from this sputtering coupled with the continued energy input from the ion beam. Also visible are small features leading away from the particles as indicated by the white arrows in the figure. It is worth stressing at this point that the GCIB incident angle is 45° to the normal and from right to left across the surface in the images. In Fig. 3(d) the surface is increasing in roughness, and the features to the left of the particles now become clearer, and appear as ‘tails’ to the particles as indicated by the white arrows. This feature is evident on all particles in the image and are all aligned with the incident beam direction. The particle density continues to increase in Fig. 5(e), as does the tail length and height. In some cases, the tails from one particle approach the feature of another
411.1 eV. On the nanoparticles, however, this peak has shifted to a higher kinetic energy at 412.5 eV, much closer to the pure metallic indium spectrum at 413.8 eV. These data, coupled with the XPS data already presented, confirm that the nanoparticles are composed of metallic indium. Fig. 5(a) shows an overview microscope image of the GCIB etch crater on the InP sample. It is a 1 × 2 mm2 area, visible as the lightest area in the image. The mark at the bottom of the image was simply used as a reference point for position registration and correlative analysis between instruments. A series of HIM images were collected across the crater edge, with the stage shifted by a set distance between each image. The images in Fig. 5(b)–(g) were collected at the indicated points on the microscope image. The lateral separation of the images in (b)–(c) was 1000 µm, (c)–(f) was 500 µm, and (f)–(g) 1000 µm. Images at three magnifications are presented, with scale bars inset within the images. Since it is expected that the spot size of the GCIB is quite broad for the XPS instrument used here (as opposed to in SIMS instruments for example), and the beam is rastered over an area of similar size to the FWHM of the spot, the collection of images over the crater edge therefore represents a variation in the total GCIB ion dose that the InP surface received. It follows then that a set of measurements made over a broad crater edge is an analogous representation of a series of GCIB 682
Applied Surface Science 459 (2018) 678–685
A.J. Barlow et al.
Fig. 6. The mechanism for nanostructure formation during GCIB etching of InP. Total ion dose is increasing from left to right. Initially the nanoparticles present on the surface create a shadow that masks a region of the InP from the beam, leading to masked material on the far side of the particle from the beam. As etching continues, or ion dose increases, this feature grows in scale, ultimately leading to the nanostructured formations observed in the last panel.
as the ones used in this work are expected to collect secondary electron yield from surfaces with energies up to 50 eV. To test this hypothesis, low kinetic energy spectra from 10 to 50 eV were collected within the SAN and are presented in the Fig. S4. We observe that the intensity of electron signal in this energy range is consistently greater at the nanoparticles than that of the surrounding InP areas, demonstrating that the material contrast is the result, at least in part, of increased metallic content at the tip. This increased material contrast allowing metallic In to be identified is not observed in the literature for SEM images of nanostructured InP, and is an advantage afforded by the HIM. The mechanism suggests that the nanoparticles have lesser sputter yield compared to the bulk InP. If the sputter yield was greater, the nanoparticles would erode more quickly than the formation of the nanostructures could occur leading to, at most, a surface with enhanced roughness. In the case where the sputter yield of the nanoparticles is similar to or less than the InP the particle would remain intact for a greater length of time, allowing for the nanostructure formation to become the dominant effect. The mechanism presented here contradicts some previously discussed mechanisms of texture formation during sputtering of InP whereby the nanostructures are proposed to result from diffusiondriven growth of filaments from the bulk, capped by a metallic In shell [7,28,29]. It does, however, agree with the proposed mechanisms of Homma [6] and Sanatinia [9], where preferential sputtering leads to diffusion and particle growth initially, then removal of material becomes the dominant mechanism and masked sputtering occurs. The experimental conditions for previous studies were typically vastly different to those presented here. Ion beams used were monoatomic, high beam energy and thus presumably high beam current (µA). Much of the literature also discusses filament growth at elevated temperatures, where surface and bulk diffusion is enhanced. Here the nanostructure formation is at room temperature, with a beam current of only 5 nA and an energy per atom of 26.7 eV, both orders of magnitude less than previously reported experiments. It is proposed here that the formation of the nanostructures in this regime, i.e. the low beam current and energy-per-atom of the Ar GCIB, is dominated by physical sputtering effects and contrasting sputter yields between the indium nanoparticles and the bulk InP. This effect has also been observed by the authors on polyvinyl chloride (PVC) surfaces as presented in Supplementary Fig. S2. Two competing processes are present in this mechanism. Diffusion of mobile In atoms is the mechanism behind the formation of the initial nanoparticles. The limiting factor for the diffusion process ultimately
particle. In Fig. 5(f) and (g) we are now imaging the base of the GCIB etch crater, representing the greatest total ion dose, and the density of the features is the highest. It can be seen at the highest magnification that the tail features are now taller and fin-like, and in some cases multiple features have coalesced into single features. In Fig. 5(g) the characteristic nanostructured InP surface is visible. It is proposed here that the mechanism of the nanostructure formation is a ‘self-masking’ effect during the etching of the InP, directly caused by the presence of the nanoparticles emerging early in the etching process. This is described schematically in Fig. 6. In the first panel the nanoparticles are present on the surface, with the GCIB incident upon them and the InP surface at 45°. The initial nanoparticle formation is due to a surface diffusion effect resulting from the preferential sputtering of P from the surface after the removal of the topmost oxide layer [9]. In the absence of sample rotation the nanoparticles create a shadowed region that the GCIB cannot reach and thus cannot etch. The result is shown in the second panel, where the shadow created by the particle leads to a region of masked material that now protrudes from the newly exposed surface that is the crater base. The height of this protrusion could be considered the etch depth, since it is a measure of how much material has been removed from the surface in areas devoid of nanoparticles. In the third and fourth panels in Fig. 6 this process continues as the total ion dose is increased. In the case presented here this is equivalent to moving further across the crater edge. As more material around the particles is removed, the masked material increases in height above the new surface. The features grow and are inherently aligned with the incident beam angle. The ultimate result is shown in the fifth panel in Fig. 6, where the surface is now covered in nanostructures. The nanoparticle itself experiences some amount of ion beam etching. This can be seen from the highest magnification images in Fig. 5. The particles appear somewhat spherical initially, at least symmetric in shape. As the ion dose increases, the particles become elongated along the axis of the ion beam. The side of the particle facing the beam experiences the greatest dose, decreasing around the particle until the opposite face experiences effectively no dose. This has the effect of undercutting the particle and simultaneously eroding the sides to produce an elongated particle. This enhances the production of the nanostructures making the features appear to protrude even further from the surface. It is also worth noting the apparent material contrast at the tip of the protrusions. This could be due to the increased metallic content at the tip, and thus increased secondary electron emission. ET detectors such 683
Applied Surface Science 459 (2018) 678–685
A.J. Barlow et al.
Fig. 7. (a) Etch depth relative to the original surface level as a function of position across the GCIB crater edge. The dotted curve is a fit to the data using a Gaussian function. (b) The same data as in (a), but converted to show etch depth versus total ion dose.
with published data of etch depths versus ion dose from similar Ar GCIB conditions on other materials [30].
becomes the length across which diffusion can occur. By increasing the process temperature one could increase the diffusion length and promote growth of features outward from the surface. All sputtering in the present work was performed at constant ambient temperature and thus diffusion and the length scales across which it can occur were expected to be constant. During ion bombardment both diffusion and sputtering is occurring. However, once a nanoparticle reaches a certain size greater than the diffusion length then particle growth ceases since liberated In can no longer migrate far enough to reach the particle before sputtering removes material around the particle. At this point sputtering becomes the dominant mechanism and the particles begin to act as masks for the underlying material, leading to nanostructure formation through masked sputtering. Feature heights from the full images used to generate Fig. 5 were measured using the processing software ImageJ (see Section 2), and were corrected for the viewing angle. As many measurements as possible were made on each image, provided the complete feature was visible allowing an accurate indication of the height however the number of measurements possible declined across the crater. Table S1 provides detailed particle analysis and Fig. S3 presents a plot of these results including the particle radius. While the feature height is observed to increase rapidly across the crater, the nanoparticle radius is largely constant, supporting the mechanism described above whereby the diffusion length limits the maximum particle size and ion beam sputtering becomes the dominant mechanism. The measured height was then adjusted for the diameter of the nanoparticles and the resulting figures were considered to be the “etch depth”, i.e. the depth below the initial surface level. These are presented in Fig. 7(a) as Relative Etch Depth, i.e. the etch depth relative to the surface level in nanometres. Moving across the crater edge the etch depth increases as would be expected given the increase in total ion dose at the surface. This reaches a maximum (i.e. the deepest part of the crater) around 3000 µm from the first measurement (i.e. the farthest edge of the crater). If it is assumed that the profile of the GCIB is largely Gaussian in shape, then we can fit the edge of the crater to a Gaussian profile, shown as the dotted grey line in Fig. 7(a). This gives a beam FWHM of 1.5 mm, which is not unexpected for the GCIB used in this work. These beams are non-analytical in XPS, and thus do not have the same rigorous constraints as in SIMS. Since the beam current is measured, and the total etch time was set, given this newly fitted Gaussian profile the abscissa can be scaled and converted to show, instead of position, a total ion dose for each depth measurement. Thus a plot of etch depth versus total ion dose can be generated, and is given in Fig. 7(b). The resulting plot indicates a linear relationship between the total ion dose of the GCIB and the resulting InP etch depth. This linear relationship compares favourably
4. Conclusion Helium ion microscopy (HIM) has been used to observe the formation of nanoscale topography on indium phosphide (InP) during argon gas cluster ion beam (Ar GCIB) etching. The technique is extremely well-suited to studying the surfaces of etch craters, and reveals nanoscale features that other commonly-used techniques such as atomic force and scanning electron microscopies would have greater difficulty observing at such high magnification. Shortly after GCIB etching begins the native oxide is removed and metallic indium nanoparticles form on the InP surface. It is proposed that these particles mask the underlying surface from the GCIB, and over time a surprisingly regular array of nanostructured InP results, with the metallic nanoparticles at the tip. By studying the surface across the edge of the GCIB etch crater, insight can be gained into how the surface is varied with total GCIB ion dose. As the dose increases, the density of the nanoparticles increases, as does the resulting nanostructure density. The height of the features, or equivalently the depth of the crater, is increased with total ion dose. The relationship between InP etch rate and the total GCIB ion dose is found to be linear. Acknowledgements X-ray photoelectron spectroscopy (XPS) data were acquired at the National ESCA and XPS Users’ Service (NEXUS), Newcastle University, UK. Helium ion microscopy (HIM) was also performed at NEXUS, and the instrument was purchased as part of a package funded by EPSRC’s ‘Great Eight’ capital funding grant EP/K022679/1 and Newcastle University. This work incorporates data from the Victorian node of the Australian National Fabrication Facility (ANFF), a company established under the National Collaborative Research Infrastructure Strategy to provide nano and microfabrication facilities for researchers in Australia, through the La Trobe University Centre for Materials and Surface Science. Data are reproduced under a Creative Commons licence (CC BY-NC 4.0 International). References [1] R.M. Bradley, J.M.E. Harper, Theory of ripple topography induced by ion bombardment, J. Vac. Sci. Technol. A 6 (1998) 2390, https://doi.org/10.1116/1. 575561. [2] F. Frost, A. Schindler, F. Bigl, Roughness evolution of ion sputtered rotating InP surfaces: pattern formation and scaling laws, Phys. Rev. Lett. 85 (2000) 4116–4119,
684
Applied Surface Science 459 (2018) 678–685
A.J. Barlow et al.
4791669. [17] D. Rading, R. Moellers, H.-G. Cramer, E. Niehuis, Dual beam depth profiling of polymer materials: comparison of C60 and Ar cluster ion beams for sputtering, Surf. Inter. Anal. 45 (2012) 171–174, https://doi.org/10.1002/sia.5122. [18] B.R. Strohmeier, R.G. White, T.S. Nunney, P. Mack, A.E. Wright, Enhanced surface and in-depth characterisation of organic and inorganic materials using XPS and soft depth profiling with argon cluster ion beams, Microsc. Microanal. 18 (Suppl. 2) (2012) 918, https://doi.org/10.1017/S1431927612006447. [19] A.J. Barlow, J.F. Portoles, P.J. Cumpson, Observed damage during Argon gas cluster depth profiles of compound semiconductors, J. Appl. Phys. 116 (2014) 054908, https://doi.org/10.1063/1.4892097. [20] J.D.P. Counsell, A.J. Roberts, W. Boxford, C. Moffitt, K. Takahashi, Reduced preferential sputtering of TiO2 using massive argon clusters, J. Surf. Anal. 20 (2014) 211–215. [21] N. Sano, A.J. Barlow, C. Tsakonas, W. Cranton, P.J. Cumpson, Optimal conditions for gas cluster ion beams in studying inorganic interface species: improved chemical information at a ZnO interface, Surf. Inter. Anal. (2016), https://doi.org/10.1002/ sia.5949 (ECASIA Special Issue). [22] B.W. Ward, J.A. Notte, N.P. Economou, Helium ion microscope: a new tool for nanoscale microscopy and metrology, J. Vac. Sci. Technol. B 24 (2006) 2871–2874, https://doi.org/10.1116/1.2357967. [23] R. Hill, J.A. Notte, L. Scipioni, Scanning helium ion microscopy, Adv. Imaging Electron. Phys. 170 (2012) 65–148, https://doi.org/10.1016/B978-0-12-394396-5. 00002-6. [24] Neal Fairley, CasaXPS: Processing Software for XPS, AES, SIMS and More, Casa Software Ltd, Teignmouth, UK, 2016. [25] A.J. Barlow, J.F. Portoles, N. Sano, P.J. Cumpson, Removing beam current artefacts in helium ion microscopy: a comparison of image processing techniques, Microscop. Microanal. 22 (2016) 939–947, https://doi.org/10.1017/S1431927616011673. [26] C.A. Schneider, W.S. Rasband, K.W. Eliceiri, NIH Image to ImageJ: 25 years of image analysis, Nat. Methods 9 (2012) 671–675. [27] A.J. Barlow, R.T. Jones, A.J. McDonald, P.J. Pigram, XPSSurfA: an open collaborative XPS data repository using the CMSShub platform, Surf. Interface Anal. 50 (2018) 527–540, https://doi.org/10.1002/sia.6417. [28] H.W. Gries, Radiation-induced sample modification in surface analysis: InP as an extreme example, Surf. Inter. Anal. 14 (1989) 611–618, https://doi.org/10.1002/ sia.740141009. [29] F. Okuyama, J. Kato, Novel fiber growth on Ar+-sputtered InP, J. Vac. Sci. Technol. B 12 (1994) 3054, https://doi.org/10.1116/1.587559. [30] I. Yamada, J. Matsuo, N. Toyoda, A. Kirkpatrick, Materials processing by gas cluster ion beams, Mater. Sci. Eng. R. 34 (2001) 231–295, https://doi.org/10.1016/S0927796X(01)00034-1.
https://doi.org/10.1103/PhysRevLett. 85.4116. [3] Y. Homma, H. Okamato, Y. Ishii, Surface microtopography and compositional change of cesium-ion-bombarded semiconductor surfaces, Jpn. J. Appl. Phys. 24 (1985) 934, https://doi.org/10.1143/JJAP.24.934. [4] T. Ogiwara, S. Tanuma, Y. Nagasawa, N. Ikeo, Auger depth profiling analyses of InP/GaInAsP multilayer thin films, Microbeam Anal. 2 (1993) 133–138. [5] S.W. MacLaren, J.E. Baker, N.L. Finnegan, C.M. Loxton, Surface roughness development during sputtering of GaAs and InP: evidence for the role of surface diffusion in ripple formation and sputter cone development, J. Vac. Sci. Technol. A 10 (1992) 468–476, https://doi.org/10.1116/1.578173. [6] Y. Homma, Formation mechanism of indium microcrystals on ion-bombarded InP surfaces, J. Surf. Anal. 3 (1997) 641–645. [7] M.P. Seah, S.J. Spencer, P.J. Cumpson, J.E. Johnstone, Cones formed during sputtering of InP and their use in defining AFM tip shapes, Appl. Surf. Sci. 144 (1999) 151–155, https://doi.org/10.1016/S0169-4332(98)00794-6. [8] T. Radny, H. Gnaser, Self-organizing nanodot structures on InP surfaces evolving under low-energy ion irradiation: analysis of morphology and composition, Nanoscale Res. Lett. 9 (2014) 403, https://doi.org/10.1186/1556-276X-9-403. [9] R. Sanatinia, et al., Wafer-scale self-organized InP nanopillars with controlled orientation for photovoltaic devices, Nanotechnology 26 (2015) 415304, https://doi. org/10.1088/0957-4484/26/41/415304. [10] J. Wallentin, et al., InP nanowire array solar cells achieving 13.8% efficiency by exceeding the ray optics limit, Science 339 (2013) 1057–1060, https://doi.org/10. 1126/science.1230969. [11] T.-T.D. Tran, et al., High brightness InP micropillars grown on silicon with Fermi level splitting larger than 1 eV, Nano Lett. 14 (2014) 3235–3240, https://doi.org/ 10.1021/nl500621j. [12] B.J. Murdoch, J.F. Portoles, S. Tardio, A.J. Barlow, I.W. Fletcher, P.J. Cumpson, Visible wavelength surface-enhanced Raman spectroscopy from In-InP nanopillars for biomolecule detection, Appl. Phys. Lett. 109 (2016) 253105, https://doi.org/10. 1063/1.4972558. [13] B.J. Murdoch, A.J. Barlow, I.W. Fletcher, P.J. Cumpson, The plasmonic properties of argon cluster-bombarded InP surfaces, Appl. Phys. Lett. 111 (2017) 081603, https://doi.org/10.1063/1.4993535. [14] P.J. Cumpson, J.F. Portoles, N. Sano, A.J. Barlow, M. Birch, Depth profiling organic/inorganic interfaces by argon gas cluster ion beams: sputter yield data for biomaterials, in-vitro diagnostic and implant applications, Surf. Int. Anal. 45 (2013) 1859, https://doi.org/10.1002/sia.5333. [15] N. Winograd, Molecular depth profiling, Surf. Inter. Anal. 45 (2013) 3–8, https:// doi.org/10.1002/sia.4913. [16] P.J. Cumpson, J.F. Portoles, N. Sano, Material dependence of argon cluster ion sputter yield in polymers: Method and measurements of relative sputter yields for 19 polymers, J. Vac. Sci. Technol. A 31 (2013) 020605, https://doi.org/10.1116/1.
685