The IBP process is able reduce surface roughness to 30%~40% of initial diamond turned surfaces of NiP after a single step for all samples, irrespective of spatial wavelength (from 1.5 μm up to 25 μm) and depth of diamond turning marks (amplitudes from 10 nm to 60 nm), which has been presented previously [7]. The degree of roughness reduction is mainly limited by the planarizing behavior of the resist (see Ref. [7] and later discussion in section 3.2).
In general, the driving force for surface levelling is a combination of surface tension and gravity, which is opposed by the viscosity of the liquid (resist) [8]. Based on Ref. [8] the exponential decay time T for each Fourier components of surface roughness depends on material properties (surface tension γ, viscosity η, resist thickness d) and the respective spatial frequency f (or spatial wavelength λ). For very thick layers of resist, i. e., d >> λ/2π, the decay time is independent on resist thickness. In contrast, for very thin layers (d < < λ/2π) the decay time is proportional to 1/d
3. If one looks at the spatial frequency range covered by the AFM measurements and under consideration of typical resist thicknesses, it can be seen that the transition range between both levelling regimes is covered within the experiments. Consequently it is not clear how the levelling changes with resist thickness. Therefore former investigations [7] were extended to test experimentally the potential effect of the thickness of coated resist and repeated coating/etching steps.
Roughness of different resist thickness
Various resist of thickness (240 nm, 380 nm, 700 nm) were spin coated on diamond turned NiP sample D. The surface roughness of diamond turned surface is Rq 6.56 nm (RMS). After coating, the roughness drops down to 3.09 nm, 2.83 nm and 2.50 nm, respectively (Fig. 1). The surface roughness decreases slightly with the increase in thickness of resist, which is readily understandable according the discussion above. Although thicker resist is marginally superior to thinner ones in terms of surface roughness, the etching time to remove resist will significantly increase. For example, coating 700 nm resist will decrease surface roughness from ~ 3.10 nm to 2.50 nm compared to 240 nm resist, 19% reduction in surface roughness; however, the etching time to remove resist will increase from ~ 22 min to ~ 64 min, 191% increase in etching time. Coating thicker resist can reduce surface roughness indeed, though it is not an advisable way to reduce surface roughness by virtue of making resist thicker in that it is not efficient.
Repeating coating/etching (IBP) processes
It has been shown that IBP is able to decrease the surface roughness of diamond turned NiP surface to the ~ 30% of the initial surfaces after one step. Here we repeated IBP processes to see if the roughness could further decrease by re-applying the IBP processes (coating/etching) to smoothed surface.
Sample D and A was selected in our experiments. The resist thickness was controlled as the same as possible in each step (~300 nm). The results for sample D and A are plotted in Figs. 2 and 3, respectively. After first IBP process, the roughness for sample D and A was reduced from 6.56 nm and 6.49 nm to 2.79 nm and 1.92 nm, respectively, 30% ~ 40% reduction rate. The surface roughness after coating for sample A (2.12 nm) is smaller than that of sample D (3.2 nm), so after ion beam etching the roughness of sample A is better than sample D, which is attributed to different spatial wavelength of D (6 μm) and A (1.5 μm). The surface roughness after IBP continues decreasing as repeating IBP process, but the reduction rate gradually decreases. After repeating IBP process several times, it is found that both the sample D and A behaves similarly and surface roughness drops exponentially (Figs. 2 and 3). The surface of sample D (1.29 nm) has been smoothed to 20% of initial diamond turned surface after 3 times IBP processes and the surface roughness of sample A (0.70 nm) has dropped to 10% of initial surface roughness after 5 times IBP processing. Based on Figs. 2 and 3 it is once again clarified that the roughness reduction is mainly controlled by the resist coating.
The mid spatial frequency is also able to be mitigated by means of multiple IBP processes. Shown in Fig. 4 are the PSD curves of sample A after each IBP step. It is clear that the intensity of the peak located at spatial frequency 1.7 × 10−4 nm−1 abates with increasing the repetition number of IBP processes.
As seen from Fig. 5, small depressions evolve during the IBP process. These are caused by particles agglomerates in the photo resist, which are caused by aging effect. For optimized resist formulations, e.g. with respect to viscosity, further improvement of the IBP process can be expected.