Direct surface patterning of amorphous chalcogenide layers with high- energy H⁺ and He⁺ ion beams

In our experiments the largest efficiency of ion beam (proton and He⁺) patterning was found for amorphous selenium (Fig. 3), that basically correlates with the data on light-induced surface relief formation at equal average exposure [14, 18]. At the same time, from technology and stability points of view application of pure amorphous selenium as material for optical devices seems to be questionable, because amorphous, glassy Se is rather instable, undergoes structural transformations, crystallization. After comparison of surface relief recording efficiency at the same absorbed optical energy, we established that almost the same high optical recording parameters may be achieved on glasses with small, few at% addition of As, like the As₆Se₉₄. Also very good is As₂₀Se₈₀ which is stable, has higher softening temperature and was successfully used in our optical and e-beam recording experiments [12]. Much lower efficiency was found for ion beam recording in As₂S₃ and As₃₀Ge₂₀S₅₀ thick plates (Fig. 3). Their advantage is the sensitivity in the green spectral region for optical recording, wider optical transparency range in visible spectral region and higher softening temperatures, stability.

The rigidity of structure and the softening temperatures T_g, as well as the heat and electrical conductivities for these compositions seems to be among the first parameters, which influenced efficiency of ion beam recording.

Really, Se and As–Se compositions with small amount of As have chainy-layered floppy structure (average coordination number between 2 and 2.2). The structure of As₂S₃ is similar but more rigid, with stronger interatomic bonds. The increase of rigidity is characteristic for Ge-containing glasses, comprising tetrahedral structural units and increased average coordination number above 2.4. In our Se rich compositions bonding energies for Se–Se (1.9 eV), As–As (1.38 eV), As–Se (1.8 eV) bonds are also lower compared to the predominant As–S (2 eV), Ge–S (2.4 eV) and S–S (2.2 eV) bonds in As₂S₃ and As₃₀Ge₂₀S₅₀ glasses [24].

First, the role of possible heating by ion beam and simple decrease of viscosity with temperature may be supposed, which results in faster mass-transport and increased relief heights, like it was shown in [25, 26]. If so, even a small local increase of temperature in Se sample, where the softening temperature near 310 K is the smallest among our investigated compositions [27], will enhance the mass-transport. At the same time, the overheating of the sample to T_g temperatures should increase the reverse effect of thermal erasing [25]. These effects may be less important in high T_g glasses like As₂S₃ or As₃₀Ge₂₀Se₈₀ even at small heating.

Since it is hard to measure the local temperature during the recording, we modelled, calculated the thermal field in the chalcogenide layer induced by ion beam and light irradiation.

In the case of the He⁺ ion beam irradiation of Se layer (cross section of resulting hump shown on Fig. 2a) at beam current of 80 pA and 2 MeV ion energy the absorbed ion beam power is 0.16 mW. The spot size (diameter) was ≈ 2.5 μm and stopping depth for the ions was ≈ 8 μm.

In the case of laser beam irradiation of As₂₀Se₈₀ layer (Fig. 2d) the power of the focused light beam was ≈ 0.5 mW. Close to 60% of light energy (at λ = 633 nm) was absorbed (and ≈ 17% of light power was reflected, ≈ 20% of light power was transmitted). Therefore, approximately 0.3 mW of the power was absorbed. The diameter of the laser light spot was close to 2–3 μm (Fig. 2d).

For r >r_s ≈ 10 μm, (r_s is the minimal size of thermal field, where the spherical symmetry model is applicable, see Fig. 4) the thermal field in the chalcogenide layer may have a (semi) sphere symmetry, if the thickness H_g of the glass substrate (1 mm) is much larger than the ChG layer thickness, and r_s is larger than the penetration depth and diameter of the ion or light beam. This approximation can apply to the region from r_s to the metal sample holder (its temperature equals to room temperature), which has much higher thermal conductivity as compared to the glasses.

In a spherical coordinate system the dependence of the temperature on the radius r (r > r_s) can be approximated by the differential equation:

where k_g(T) is the thermal conductivity of the glass, which can depend on the temperature (T), P_a is the absorbed power of the laser light.

To simplify the calculations we neglect the temperature dependence of thermal conductivity of glasses (substrate or chalcogenide) and use one coefficient of thermal conductivity for the system.

For the calculation, we can use the thermal conductivity of glass substrate: 0.8–1 W·m⁻¹·K⁻¹ or we can correct this calculation by using the thermal conductivity of the chalcogenide glass layer (~ 0.7 W·m⁻¹·K⁻¹ for Se and ~ 0.8 W· m⁻¹ K⁻¹ for As₄₀Se₆₀ [28, 29]. The boundary conditions of temperature field can be expressed as: T → T_M at R → H_g, where T_M—temperature of the metal holder (≈ room temperature).

The solution of the differential Eq. (1) gives an increase of temperature ΔT (ΔT = T(r) − T_room) as a function of r (at r_s < r≪H_g):

For r = 10 μm: in the case of Se layer (2 MeV He⁺ ion beam, I = 80 pA, power 0.16 mW) ΔT ≈ 2.5 K and in the case of As₂₀Se₈₀ layer (0.3 mW, 633 nm laser beam) ΔT ≈ 5 K. The lower thermal conductivity of chalcogenide glass in comparison with thermal conductivity of the glass substrate can increase the heat effect approximately in 1.5 times. So from our calculations for the given experimental conditions (beam intensity, radius) no significant temperature rise in the sample should occur, i.e. it can be neglected or have small influence in compositions with T_g higher than in a-Se.

The rise of temperature ΔT should be higher at r < 10 μm. It depends on the intensity distribution in the beam (ions or light) and their focusing. At r ≪ 10 μm the temperature can be close to T_g of the chalcogenide glass or higher. It is more likely to occur in Se, as it was described in [30].

The increase of temperature (to T⪅ T_g) increases the light induced mass transport efficiency [26] and decreases it at temperatures above T_g [25]. So, if the light induced mass transport and ion beam patterning have similar nature then the ion beam patterning can be more effective in case of Se layer than in As₂₀Se₈₀ or As₂S₃ layers at similar parameters of irradiation just due to the heating effects. (Figure 3).

Although, very good beam focusing can decrease ion beam patterning efficiency if heating is strong, the local temperature is larger than T_g of the chalcogenide glass, because the hump pattern will be erased by the heat, and even dip may be produced [30]. At extreme cases even evaporation, ablation, hole formation can be produced.

Some differences between patterning process by He⁺ and H⁺ beams, at the same energy exposure may depend on their different beam parameters: current, spot form, focusing (Gaussian beam intensity distribution), penetration depth, linear energy transfer. In our case, the efficiency of ion beam patterning of ChGs (Fig. 3) shows that He⁺ ion beam patterning of chalcogenide glasses is more effective than H⁺ ion beam patterning. This difference ranges from ~ 2 times in case of Se to ~ 30 times in the case of As₂S₃ (Fig. 3). For Se glass, the 2 MeV He⁺ ion penetration depth is around 7–8 μm and for the same energy H⁺ ion it is 41–42 μm. Linear energy transfer (deposited energy per unit depth) for He⁺ ions was 5–10 times larger compared to H⁺ ions according to SRIM simulations in our materials. The linear energy transfer in Se for He⁺ ions is 236 keV/μm and 31 keV/μm for H⁺ ions. Similar results can be obtained for the other investigated compositions too. This higher deposited energy per unit depth may cause greater changes in the material, which could explain the higher patterning efficiency with He⁺ ions, especially in As–Se compositions compared to sulphides, because of the former’s lower viscosity, also connected with structure [31].

Further challenging question related to the recording mechanism in the investigated glasses may be connected with pore, bubble formation, especially at elevated temperatures. The last were excluded by the selection of exposition conditions and supported by direct FIB and SEM observations of hump and layer cross-sections in compositions like As₂₀Se₈₀, where no phase-separated regions were found (see Fig. 5).

Local EDX measurements on As₂₀Se₈₀ layer cross-sections gave no measurable composition changes between the irradiated and non-irradiated area. There may be still some small, near 1% local expansion–contraction even at short ion beam irradiation due to the local amorphous structure transformations towards more disordered state, like it was observed in Se and many amorphous chalcogenide layers [3, 6, 32] under light irradiation. At our experimental conditions (layer thickness, interaction volume) it can give us maximum a few hundred nanometers of surface relief height, which is much smaller than observed. Still it can be a component of any (light, beam) recording in amorphous chalcogenide layers connected with unavoidable structural transformations, bond breaking-rearrangement within amorphous phase.

In our cases of ion beam recording the charging of irradiated spots and appearance of electrical, polarization forces at the edges of Gaussian beam intensity distribution also may have essential influence, since the volume diffusion of charged glass components, small clusters depends on the built-in electric field, as it was described in [12, 15, 21].

This effect should be and was higher in our layer samples deposited on dielectric substrates.

Therefore, the influence of substrate conductivity on hump height and volume also have been investigated. We compared height and volume of the hump on the surface of Se and As₆Se₉₄ layers irradiated by 2 MeV of He⁺ ions. In both cases the influence of substrate conductivity on height and volume of the produced hump were observed. For chalcogenide glasses deposited on the surface of vitreous glass without ITO layer the humps had larger height and volume (Fig. 6). This effect can be explained by the role of Coulomb force repulsion –attraction effect in irradiated and surrounding surface areas, in the surface deformation process and volume expansion.

Penetrating the chalcogenide glass layer, the ions create an electric charge around their stopping region. The electric field magnitude (E) around the charged zone can be evaluated, but the magnitude of the electric field must be limited by avalanche multiplication threshold in ChG. In case of amorphous Se it is 7*10⁷ V/m [33, 34]. Here, the avalanche multiplication in ChG may be stimulated by ion beam. Therefore, we think that the extra patterning of layers may happen at the electric field magnitude higher than ~ 7*10⁷ V/m, which is possible at high exposures because the high electrical resistivity of ChG (for example ρ _selenium ≈ 10¹⁴ Ω∙m [35]) does not allow to reduce the charge around the beam stopping zone without avalanche process. This built-in electric field can also stress the chalcogenide glass and cause additional surface deformation. This effect is supported by that with an insulator substrate surface humps are higher than in samples with the conductive ITO covered substrate (Fig. 6), where the ITO layer grounds the charges. Besides the influence on charge related processes, the interface can influence the heat transfer and adhesion between chalcogenide and substrate glasses. In our approximation of the thermal model we neglect these, because the adhesion on the boundary of chalcogenide and substrate is usually good, the transition layers are thin and therefore the thermal contact must be good between the two glasses.

Further comparisons may be made with efficiency of optical recording, but here again additional effects like optical fields [36], influence of polarized illumination or just the experimental conditions of holographic recording with polarized laser beams [7, 8] introduce additional variables which make difficult calculated comparisons. For example, we also compared efficiency of micro lens hump patterning by He⁺ ion beam and e-beam beam on amorphous selenium. In the case of He⁺ (2 MeV) the efficiency is ~ 6 nm/μJ (Fig. 3a). For electron beam (30 keV and current 5 nA, total energy per hump 750 μJ and height ~ 1.1 μm [37]) the maximum of efficiency (efficiency depends on scanning frequency) is ~ 1.5 nm/μJ. That shows 4 times higher efficiency of ion beam patterning in comparison with efficiency of electron beam recording. This difference can be caused by charge multiplication and also by larger penetration depth of ions than electrons to ChG.

Comparisons of our present data with light- and e-beam induced giant relief formation effects support the most general conclusion about the basic similarities of relief formation at ion or photon, electron irradiations, due to the stimulated structural transformations and mass-transport processes. These start with electron–hole excitation, charged defect creation, local structural changes and continues in directed shift of charged elements (chalcogens, pnictides, structural chain fragments) due to the gradients of chemical potentials, electric fields, capillary effects, like it was described in [21]. At the same time, one difference is the absence of possible light polarization effects on recording, which are well separated in laser-recording processes on chalcogenide glasses. Another difference is the increased efficiency of excitation by ions in a deep interaction volume, which makes the ion beam prototyping favorable in thick layers or bulk glass elements.