Charging suppression in focused-ion beam fabrication of visible subwavelength dielectric grating reflector using electron conducting polymer

High reflectivity and broadband mirror with polarization selectivity can be achieved by utilizing subwavelength grating (SWG). Recently, the SWG has been demonstrated as reflector for vertical-cavity surface-emitting laser,^1–9 quantum cascade laser (QCL),¹⁰ photodetector,¹¹ solar cell,¹² and optical waveguide.¹³ Depending on the operating wavelength of interest, SWG reflector can be patterned using micro- and nanofabrication methods, such as optical lithography,¹⁴ laser interferometric lithography,¹² holographic lithography,⁶ nanoimprint lithography,⁹ and electron beam lithography (EBL),^2–5,7,8 or focus ion beam (FIB).¹⁰ 

SWG patterning using FIB has thus far demonstrated the SWG antireflector for QCL at infrared (IR) spectrum,¹⁰ plasmonic collimator for QCL at IR spectrum,¹⁵ and membrane SWG for GaN at visible spectrum.¹⁶ The first two examples are demonstrated at IR spectrum. As for application in the visible spectrum, the GaN membrane SWG is still patterned using EBL while FIB is only used for direct milling of the membrane layer. All these studies are based on III–V compound semiconductor materials. For dielectric materials, realizing SWG structure using the FIB method at visible wavelength is inherently challenging. This is due to the nanoscale patterning dimension (subwavelength requirement) and poor dielectric conductivity (sample charging) under the FIB environment. The ion charging affects the feature size and uniformity even more so for visible photonic devices due to the nanoscale dimension requirement for shorter wavelength operation. To counter the charging effect, several charge neutralization schemes have been reported, such as the simultaneous electron and beam irradiation,¹⁷ thin conductive film coating,¹⁸ microprobing,¹⁹ electron flood gun,²⁰ ultraviolet (UV) flood gun,²¹ and lower energy irradiation.²² However, these anticharging schemes introduced complexity in terms of mathematical calculation, additional sample processing steps and require hardware/software modification to the FIB equipment. In the simultaneous electron and beam irradiation scheme, mathematical calculation is required to couple both electron and ion beam spots. The charging mitigation using the thin film coating and microprobing schemes require conductive material deposition (using evaporation or sputtering) on the sample and later this material need to be removed (using wet- or dry-etching) for subsequent sample processing. For the electron- and UV-flood guns, both anticharging schemes require FIB hardware modification to insert the tungsten filament (for the electron gun scheme) and UV light-emitting-diodes (for the UV gun scheme). As for the lower energy irradiation scheme, the FIB operation require both hardware and software modifications to monitor and program the patterning using memory device.

We present here a simpler charging suppression scheme in FIB fabrication of visible dielectric SWG reflector using electron conducting polymer. The electron conducting polymer scheme has been recently used for charging suppression in EBL,^23–25 but has not been demonstrated for FIB yet. In this scheme, we spin-coated the electron conducting polymer on the front and back surfaces of the sample prior to FIB patterning. We compared the FIB patterning using the electron conducting polymer scheme with the thin conductive film coating scheme. Using the electron conducting polymer scheme, we obtained nanoscale precision with high uniformity and sharp contrast for the FIB patterning of visible dielectric SWG reflector. We demonstrate that the electron conducting polymer scheme can provide straightforward charging control for FIB patterning, eliminating complicated processing steps or equipment modification.

Figure 1 shows the schematic of the dielectric SWG reflector on top of GaN substrate with sapphire template. It is composed of binary grating of Si₃N₄ layer as high refractive index (RI) material embedded in between two low RI layers of SiO₂ layer (bottom) and air (top). We employed rigorous-coupled wave analysis (RCWA) method to design the grating parameters for the visible dielectric SWG reflector. The grating parameters consist of period (Λ), high RI layer thickness (HI_d), low RI layer thickness (LI_d), and fill-factor (FF), which is defined as ratio of grating width to period.

We assumed that the incident light propagate bottom up to the surface-normal plane through the SWG structure (z-direction). The transverse electric (TE)- and transverse magnetic (TM)-polarization are considered parallel to (y-direction) and perpendicular to (x-direction) the grating lines, respectively. The dielectric SWG reflector is designed to maximize reflectivity at visible wavelength of 425 nm with TE-polarization. The Si₃N₄ and SiO₂ dielectrics layers were deposited on a generic GaN-sapphire template substrate using plasma-enhanced chemical vapor deposition (PECVD) technique. From ellipsometry measurement, the refractive index at 425 nm wavelength for each layer is Si₃N₄ = 1.92, SiO₂ = 1.47, GaN = 2.52, and sapphire = 1.77, respectively. These refractive index values are used in the RCWA computation.

The FIB patterning was performed using a FEI Quanta 3DFEG dual beam system that consisted of both electron- and ion-columns. Prior to the FIB patterning, we spin-coated approximately 20 nm electron conducting polymer (Espacer 300Z from Showa Denko) on the surface and backside of the dielectric deposited sample to suppress surface and subsurface charging. The electron conducting polymer was spin-coated at 2000 rpm for 60 s. As for the thin conductive coating film sample, we sputter-deposited approximately 30 nm of Cr on the top of another dielectrics deposited sample. Using FIB, we patterned the dielectric SWG reflector on a square area of 30.6 × 30.6 μm² (consisted of 75 grating lines) for both samples. The FIB parameters were fixed at Ga⁺ ion source accelerated voltage of 30 kV, probe current of 30 pA (beam diameter of approximately 17 nm), dwell time of 100 ns, beam overlap of 50%, and using serial scan mode (left to right direction). The patterning time was around 4 h per sample. After FIB patterning is completed, the dielectric sample with the electron conducting polymer was immersed in deionized (DI) water for 60 s, followed by DI water rinsing for 120 s and finally N₂ drying. We purposely extended the electron conducting polymer removal processing where we immersed and rinsed with DI water for total 150 s longer time instead of only rinsing with DI water for 30 as described in the manufacturer technical datasheet for Espacer 300Z. From the optical microscopy, we observed no significant residue existence on the sample surface due to the extended electron conducting polymer removal processing. Thus, FIB patterning in our case should be straightforward without residue hindrance. Furthermore, our patterning resolution is larger (>200 nm). Using the EBL technique, the electron conducting polymer has succeeded in demonstrating around 20 nm resolution patterning.²⁵ As for the Cr-coated sample, we removed the Cr by dipping the sample in Cr etchant (from Sigma Aldrich) at room temperature for 20 s. In addition, we performed a cross-section milling for the dielectric SWG reflector using the FIB to investigate the grating cross-section profile. We utilized the scanning electron microscope (SEM) capability in the FEI Quanta 3DFEG FIB/SEM dual beam system to measure the grating dimensions. All samples were sputter-coated with 3 nm Ir to neutralize charging during the SEM imaging.

Figures 2(a)–2(d) exhibit the TE-polarized reflectivity contour plot for all grating parameters (Λ, HI_d, LI_d, and FF) as a function of wavelength computed by the RCWA. By varying these grating parameters, we can design the optimal dielectric SWG reflector at 425 nm operating wavelength. In addition, we can investigate the fabrication tolerance (defined at ≥99% reflectivity) for all grating parameters. For each of the result in Figs. 2(a)–2(d), we computed the corresponding grating parameter and fixed the remaining grating parameters at the respective optimal values. Figure 2(a) indicates that high reflectivity (>90%) is obtained whenever the Λ parameter is at the subwavelength of the designated operating wavelength. By designing the SWG period at subwavelength, all diffraction orders except for the zero-order are suppressed, resulting in the incident light to be either reflected or transmitted in the normal direction.²⁶ We chose the Λ = 408 nm for the dielectric SWG reflector, which cross-over the operating wavelength of 425 nm, thereby resulted in ±30 nm fabrication tolerance for ≥99% reflectivity. In Fig. 2(b), the effect of varying the HI_d parameter to the dielectric SWG reflectivity is illustrated. To achieve ≥99% reflectivity, we proposed the HI_d = 165 nm with the fabrication tolerance approximately ±30 nm. A high reflectivity and symmetric SWG reflector is obtained for the FF parameter as shown in Fig. 2(c). We suggested FF = 0.4 for the optimal dielectric SWG reflector design to accommodate high fabrication tolerance (±0.19 for ≥99% reflectivity). This result in grating width of 163 nm and air separation of 245 nm for the fabrication of each grating period. Figure 2(d) exhibits that LI_d parameter has negligibly effect on the reflectivity spectrum of the dielectric SWG reflector. We selected the LI_d = 100 nm for broad reflectivity spectrum and to reduce the PECVD deposition duration. Furthermore, thinner LI_d layer resulted in broader reflectivity bandwidth.²⁷ Based on the grating parameters results, the SWG operating wavelength is controlled using the Λ parameter whereas the reflectivity intensity is regulated using the HI_d and FF parameters.

Figure 3 shows the computed reflectivity spectrum for the optimized visible dielectric SWG reflector for both TE- and TM-polarizations. It exhibits high reflectivity >99.9% at 425 nm wavelength for the TE-polarization while the TM-polarization was suppressed below 40% reflectivity, indicating the SWG reflector capability for polarization selectivity. The optimized TE-polarized dielectric SWG reflector also exhibited 65 nm broadband for reflectivity of >90%.

Table I summarized all optimized grating parameters design with the respective fabrication tolerance (≥99% reflectivity) for the FIB patterning of the dielectric SWG reflector. The period and fill-factor can be precisely controlled using FIB patterning whereas the grating- and low RI-thicknesses can be controlled through accurate FIB milling and PECVD deposition.

To demonstrate the severity of charging in realizing nanoscale structure for the visible dielectric SWG reflector, we intentionally patterned a dielectric deposited sample using a basic charging mitigation scheme. In this basic scheme, we simply used Ag tape and Ag paste on the sample edges and back surfaces. The sample consists of similar layers with the Cr-coated sample and the electron conducting polymer-coated sample. Figure 4 exhibits the SEM image after the completion of FIB patterning. No uniform and continuous grating lines can be patterned under the influence of charging even at low current FIB operation (30 pA). We notice a high charges built-up at the beginning of the patterning (top-left of the image) from the bright contrast. As the milling continues, the pattern is severely deformed and distorted toward the bottom-right from the designated patterning area due to the extreme charging effect and charge accumulation.

Figure 5 shows the SEM image for the Cr-coated sample (for the thin conductive film coating anticharging scheme). The Cr film managed to improve the FIB patterning by dissipating most of the charges. However, pattern distortions are observed at the end of the patterning (from left to right) due to charge accumulation. Furthermore, although the patterned SWG demonstrated decent contrast, the patterned Λ and FF parameters are not uniform across the whole patterning area. The insets in Fig. 5 exhibited examples of the grating period variations at several spots on the FIB patterned area.

To investigate the effectiveness of the Cr-coated sample in dissipating the charges, we measured Λ and FF parameters for every single patterned grating lines (from left to right, 75 grating lines in total). These measurements are baselined against the fabrication tolerance for Λ and FF parameters, as plotted in Fig. 6. For the Λ parameter, more than 50% of the patterned dielectric SWG is beyond the fabrication tolerance of ±30 nm. To ensure the dielectric SWG reflected the light at the designated 425 nm operating wavelength and only allowed fundamental mode propagation (the subwavelength physic), we require precise FIB patterning for the Λ parameter. However, we observed that more than 20% of the grating lines exhibit Λ > 420 nm, especially for the gratings patterned near the end. Thus, these gratings are no longer considered as a subwavelength structure, and the reflected wavelength will be red-shifted. As for the FF parameter, although we designed a large fabrication tolerance (±0.19 or 48%), almost 20% of the patterned grating exceeded the fabrication tolerance. This will reduce the reflectivity intensity. As a result, the dielectric SWG reflector will not uniformly reflect light above 99%. For both Λ and FF parameters, we observed that the patterning intolerance occurred at several spots of the patterned area (grating lines 10–30 and 40–60) and significantly increased as the FIB patterning approaching completion (grating lines 68–75). This observation confirms that charging is still occurs across the Cr-coated sample. It will accumulate substantially as the FIB patterning is nearing the end, resulting in pattern distortion.

Figure 7 shows the SEM image of the FIB patterning on the electron conducting polymer-coated sample. We achieved a high uniformity and high contrast patterning for the dielectric SWG reflector. No significant pattern distortion is observed on the electron conducting polymer-coated sample. The inset in Fig. 7 indicates high contrast with nanoscale precision gratings were fabricated for the dielectric SWG reflector.

Similar to the Cr-coated sample, we also measured Λ and FF parameters for every single patterned grating lines on the electron conducting polymer-coated sample. In Fig. 8, we plotted the results against the fabrication tolerance for each grating parameter. In comparison to the Cr-coated sample, we achieved a highly uniform dielectric SWG reflector patterning across the whole patterned area. Both Λ and FF parameters indicate nanoscale precision patterning within the optimal design dimensions (408 nm for Λ, and 0.40 for FF, respectively) for all grating lines. For both grating parameters, no single grating line is patterned beyond the specified fabrication tolerance. Based on the constant pattern preservation within the whole range of patterned area, the electron conducting polymer anticharging scheme is efficient for charging suppression in FIB patterning. The patterned dielectric SWG will reflect uniformly with >99% reflectivity at the designated 425 nm operating wavelength across the entire patterned area.

Our results demonstrate that the electron conducting polymer scheme is more efficient in suppressing charging and patterning constantly compared to the thin conductive film coating scheme. This finding is similar to the reported patterning using EBL,²⁵ where the utilization of electron conducting polymer anticharging scheme realized the finest nanopatterning resolution within larger dose range compared to thin metal coatings of Al and Cr. We attributed the ineffectiveness of the thin conductive film coating scheme for FIB patterning due to charge existence and broadening of the ion beam by elastic scattering from metallic material. For the charge existence factor, we assumed that certain charges still trapped on the grainy surface (due to sputtering-deposition process) of the coated Cr. This is supported by similar observation for the Cr sputter-deposited anticharging scheme for EBL patterning²⁵ (referred to Cr islands as in the paper). As for the ion beam broadening factor, we considered the same scattering effect as occurred in EBL patterning.¹⁸ In the paper, Monte Carlo simulation of point spread functions of electrons in Al, Cr, and Cu reveals that heavier material coating for anticharging scheme would scatter the electrons broader in EBL, resulting in larger patterning variation. As the atomic number of the coating material increased, it would increase the elastic scattering of electrons.²⁵ Besides atomic number, the coating material density also would increase the electrons scattering.¹⁸ In our case, the incoming ions and the secondary electrons generated during FIB patterning were scattered widely, thus distorted the patterned profile when heavier and denser material coating using Cr is employed as the anticharging scheme. Using a lighter and less dense material coating like polymer would generally result in less scattering than the heavier and denser metallic coating. For this reason, we achieved a constant patterning of Λ and FF parameters when using the electron conducting polymer as the anticharging scheme. It has also been demonstrated that electron conducting polymer anticharging scheme would provide constant patterning preservation (±40 nm) for low energy exposure in EBL at nanoscale regime.²⁴ In contrast, larger patterning variation (±150 nm) is reported for the thin conductive film scheme (Al in this case). Therefore, this supports our constant patterning achievement for the electron conducting polymer anticharging scheme at low current (30 pA) FIB operation.

In addition to the fabrication tolerance investigation for Λ and FF parameters, we also performed cross-section milling to investigate the HI_d parameter (grating thickness). Figure 9 shows the cross-section microscopy for the electron conducting polymer-coated sample. The patterned structure exhibits a filleted profile that is typical for a direct writing process. We obtained approximately HI_d of 167.5 nm from the FIB patterning. This is within the nanoscale precision of the designated thickness (165 nm) and comply with the fabrication tolerance (±25 nm). The cross-section examination also shows that the dielectric SWG reflector sidewalls are fabricated almost vertically. The slight angled slope variation obtained is expected because of the fabrication imperfection.

In this paper, we presented a charging suppression scheme for FIB patterning using the electron conducting polymer for the first time. By fabricating dielectric SWG reflector, we demonstrated the effectiveness of the electron conducting polymer as an anticharging scheme for nanoscale FIB. Prior to the FIB experiment, we designed the optimal structure and the fabrication tolerance for all grating parameters (period, grating thickness, fill-factor, and low RI layer thickness) using the RCWA computation. Our results show that the FIB patterning of dielectric SWG reflector has nanoscale precision with high contrast and uniform gratings formed. The patterned gratings are also constant within the fabrication tolerance across the entire patterned area. These findings indicate that the electron conducting polymer is effective in suppressing potential charging built up in FIB patterning. This will lead to a simpler anticharging control for future FIB processing (deposition or milling) without the need of additional processing steps or equipment modification.

The authors gratefully acknowledge the funding support from KAUST and King Abdulaziz City for Science and Technology Technology Innovation Center (TIC) for Solid-State Lighting at KAUST.