Chapter 3: Results and discussion
3.1 Imprint from hard mask composed of deep pores
In our experiments, we use two different kind of hard masks, which were presented in 2.1.1, and we will introduce the soft-NIL application of the first hard mask that has deeper pores in 3.1. In the following, the results of another hard mask, which has shallow pores, will be discussed in 3.2. We also investigate the suitable pattern transfer with specific dry etch conditions that lead to sidewall passivation, to duplicate the mask structure. This dry-etch steps is combined with an anodization process to fabricate deeper structures for achieving the aspect ratio requirements of the epi-free method. In the end, a better control of suitable aspect ratio for these pores is realized by an extra silicon oxide deposition which acts as hard mask instead of single resist to etch silicon.
3.1 Imprint from hard mask composed of deep pores
The first set of hard mask is composed of 440 nm diameter and 700 nm deep pores, and we fabricate a soft stamp to imprint structures with an anti-stick FDTS coating as shown in 3.1.1. In 3.1.2, we intentionally do the same process as above but without FDTS coating, and test different mix ratios of PDMS for imprinting in 3.1.3. In the end (3.1.4), we discuss a model of three different soft-stamp deformation modes including buckling, lateral collapse, and roof collapse to calculate an ideal hard mask more suitable for our process.
3.1.1 Process with coating anti stick layer
37
After fabricating the soft stamp from the mould of silicon with FDTS-coating, we imprint the stamp into 95-nm-thick resist. Figure 3.1 shows an AFM image of the stamp before imprinting, and the depth profile of the features. From the depth profile, the height of the pillar is 400 nm, and the height of soft stamp is lower than the hard mask probably comes from the limitations of AFM. We observe the surface of the soft stamp has a line defect before imprinting, and this is not a coincidence because it is visible on every sample. This suggests the flow of PDMS is not in an even dispersion and it probably comes from the fact that PDMS flow is limited when the mold has too deep pores. Figure 3.2 shows an AFM image of the soft stamp after imprint, with, on average, 800 nm pitch and 500 nm depth along line 1. Note that after imprinting, we cannot observe individual pillars but pillars collapsed against each other.
Figure 3.3 and Figure 3.4 show AFM and SEM images of the imprinted resist. In AFM image, there are 350 nm depth holes, and each hole is formed as elliptical shape instead of round shape. Measured from the depth profile, the depth of the patterns is around 500 nm. SEM images from top view show these structures have 750 nm pitch and 700 nm diameter, and confirm that etch pillar is closely attached to the adjacent one.
Compared to Figure 3.3, the depth profile of AFM is matched to the SEM measurements.
From the results above, we can confirm that the stamp features are deformed upon imprinting, and we can suppose that the main problem is that the aspect ratio of the stamp is too high, and we will explore what are the limitations of the hard mask by applying mechanical models in 3.1.4.
3.1.2 Process without coating anti stick layer
For fabricating of a low aspect ratio structure, we intentionally not coat FDTS on the hard mask when the soft stamp is fabricated, to avoid PDMS getting deep into the pores. Therefore, soft stamps with decreased pillars’ height can be fabricated, and these short pillars have a better mechanical property avoid bending when imprinting.
Figure 3.5 shows atomic force microscopy (AFM) images of the stamp made from the hard mask. Measuring the stamp features, pillars have 500 nm diameter and only 50 nm depth.
38
In Figure 3.6, the AFM image shows the patterns of resist which is 95-nm-thick on silicon after imprinting. Holes having 500 nm diameter and 50 nm depth were obtained from the depth profile, showing a good correspondence with the soft stamp. In figure 3.7, images obtained from high-resolution microscopy reveal a bad uniformity of this sample. This likely comes from resist adhered to the stamp when imprinting without FDTS coating.
Since PDMS is an elastomer with a low Young’s modulus, a limited pressure can introduce a large deformation of the imprinted areas. However, such a low toughness PDMS stamp cannot guarantee an overall mechanical stability of high-resolution patterns. For this material, a higher Young’s modulus can be obtained by using a higher curing temperature or a higher ratio of PDMS mixture [3-1, 3-2]. In this experiment, we tested different ratios of mixture (1:5 and 1:15) to improve the uniformity. In Figure 3.8, the Topography of 3-D image shows the patterns which imprinted by soft stamp composed of 1:5, 1:15 and 1:10 mixture ratio. Higher or lower ratios of PDMS mixture lead to degrading of other mechanic properties of the stamps, such as fragility and overall toughness. Therefore, the optimal mixture ratio is 1:10.
Comparing the different results between with and without anti-stick layer, we observe that regular pores only exist in without-coating results. Regular pores will be better for the subsequent pattern transfer. However, we need to concern its bad uniformity and the low resist contrast, which is not sufficient for etching. Therefore, we discuss a model of three different soft-stamp deformation modes including buckling, lateral collapse, and roof collapse to calculate an ideal hard mask more suitable for our process.
3.1.3 Modeling the imprint Deformations
As pointed out by Delamarche et al. [3-3], one obvious limitation of imprint approaches results from the material properties typical for an elastomer. The PDMS has a low Young’s modulus of ~ 1 MPa, but this low modulus poses limitations to the obtainable stamp feature sizes. Thus, the mechanical characteristics of the imprint
39
process are critical to the replication of features and the generation of patterns with high fidelity. There are two main methods to avoid deformation when imprinting.
The first method is mentioned by Schmid et al. [3-4] who use composite stamps of PDMS. They formulate a polymeric composite based on vinyl and hydrosilane end-linked polymers herein referred to as hard-PDMS, which has a high Young’s modulus (~9 Mpa). Their design uses a multilayer stamp including a thin layer of hard-PDMS on a slab of 184-PDMS, which is the same as in our experiment. They combine advantages of soft stamps with hard stamps, and extend the capabilities of soft lithography to sub-100-nm features. A disadvantage of using this multilayer stamp is the complicated preparation of the hard-PDMS layer, which requires three steps rather than one. The difficulty in cleaning the surface of the stamp is also a problem because the hard-PDMS layer cracks easily.
Another method is to design an appropriate structure preventing stamp deformation, as mentioned in [3-5]. With an optimal design of hard mask, the fabricated soft stamp will have a better mechanical property to prevent deforming. Here, we consider the stamp with the features of Figure 3.9, where 2a, 2w, and h are the dimensions of the features’ width, spacing, and height, respectively. There are three undesirable consequences of stamp deformation: 1) buckling 2) lateral collapse and 3) roof collapse, and the geometry of these three failure modes is drawn in Figure 3.10.
The mechanism of deformation is mentioned in detail by Hui et al [3-6]. To prevent the three failure modes mentioned above, the stamp conditions must satisfy the following conditions:
40
−4σ∞W
πE∗h (1 +wa) cosh−1�sec �2(w+a)wπ �� < 1 {m}
(Roof collapse)
Here, σ∞ is the uniform stress applied to the top of the stamp and E∗ = E/(1 − v2) is the plane strain of the stamp material, where E is the Young’s modulus and v is the Poisson’s ratio. In the case of PDMS, the Poisson ratio is around 0.5, for incompressive elastomers, and the Young’s modulus is about 1 MPa [3-7]. The Young’s modulus of PDMS depends on the mix ratio and the thickness of sample. For our soft stamp, the value of 1 Mpa is more suitable according to the experimental process. The surface energy of the material (γs), and it can be calculated by measuring the detachment length of a fixed-end cantilever beam. According to [3-8], PDMS has a γs of 0.3 N/m.
Regarding buckling, it mainly results from a high aspect ratio (h/2a). Pillars can collapse when loaded or even under their own weight. When the inequality in equation {k} is not satisfied, it will cause this kind of deformation. Concerning lateral collapse, when the applied forces are sufficiently large to bend the pillars, the pillars might make contact between each other. Once contact occurs, pillars may adhere to each other due to surface adhesive forces. Therefore, lateral collapse of neighboring pillars will happen. To establish a criterion for lateral stability, inequality in equation {l} is considered. Regarding roof collapse, it results from low aspect ratio which causes insufficient relief exists on the surface, and the stamp has to withstand this compressive force. To avoid the roof collapse, the conditions should satisfy formula {m}.
In the case of the first hard mask, both conditions of buckling and lateral collapse are not satisfied. From the above formulas, the depth of pore should be below 400 nm for the desired diameter of 450 nm. Therefore, we fabricate a new hard mask which has a lower aspect ratio avoiding these three deformation problems. The results of this new hard mask’s application will be shown in the next section.