**4.1 Minimization of pressure and dimension control**

In order to achieve an accurate pattern replication, it is important to control the eventual broadening and deformations of the replicated structures. During T-NIL and UV-NIL processes, a parallel and uniform surface contact between the hard mold and the substrate is obtained by applying high pressure, and the replicated structures preserve a high accuracy when compared to the original ones in the master mold. During the Soft UV NIL process, the flexibility of the PDMS stamp ensures a conformal adhesion with the substrate at low pressure. On the other hand, the low elastic module of PDMS can produce deformations of high aspect ratio structures when too much pressure is applied. If it is not possible to completely avoid these deformations, it is important to control and reduce them under appropriate values, depending on the specific application. Because soft UV-NIL is a very recent variant of NIL,

**Standard deviation ч 5% (or ч 10 nm)**

and measurement on the master mold and measurement in the Amonil resist

**1 2 3 4 5 6 7 8 9 10 <sup>200</sup>**

**Nanodot position in different fields exposed with different doses**

**250**

Fig. 8. Control dimension fidelity after soft UV-NIL at optimized pressure: (a) SEM image of 310 nm-diameter pillar array recorded in the Amonil resist and (b) comparison of the

diameter values measured on the master mold and in the Amonil thickness contrast, with the

A second important point of the nanoimprint process concerns the removal of the residual layer of the resist. Because T-NIL and UV-NIL use rigid mold and high pressure, a thin residual layer of resist is normally left between the protrusions of the mold and the substrate. It acts as a soft cushion layer that prevents direct impact of these fragile nanostrucures and the substrate. This residual layer is normally removed by reactive ion etching (RIE) which can largely affect the original shape and size of the pattern. Soft UV NIL uses flexible stamps and this residual layer can be reduced as much as possible by simply adapting the initial resist thickness to the depth of the stamp (height of patterns). Figure 9(b) shows a SEM cross-section image of the imprinted structures at a optimized pressure of 0.7 bars, with a 25 nm thick residual layer, obtained by adjusting the initial thickness of resist. These results

Finally, in order to validate the ultra-high resolution HSQ process for EBL and our improved hard-PDMS/PDMS bilayer stamp fabrication process, we have carried out Soft UV NIL in the sub-100 nm scale. Figure 9 (b-c) shows the thickness contrast in the Amonil resist after Soft UV-NIL, at 10 PSI and at room temperature, by using the bi-layer mold hard-PDMS/PDMS for 50 nm dots with a pitch of 100 nm and 20 nm dots with a pitch of 60 nm. The high and uniform contrast thinknesses obtained on the whole pattern area demonstrate the ability of Soft UV NIL with hard-PDMS/PDMS stamps to replicate nanostructures in the 20-nm scale. We believe than our current limitations in achieving higher resolution, specifically for the replication of highly dense structures, are essentially due to the replication of the bilayer stamp. It will be necessary to work on this stamp replication process, in particular by further reducing the viscosity of the hard-PDMS prepolymer and by controlling the effects of the thermal curing process, which results in a isotropic volume shrinkage of the hard-PDMS that

can be compared to the recent ones from AMO that are presented in Figure 9(a).

**300**

average 310nm

**Diameter of nanodot (nm)**

**(a) (b)**

A Versatile Tool for Nanostructuration at the 20nm Scale

relative calculated diameter standard deviation.

**4.2 Minimization of the thickness of the resist residual layer**

becomes more and more important at the sub-20 nm scale.

**350**

<sup>151</sup> Soft UV Nanoimprint Lithography:

average 215nm

**diameter standard deviation (%)**

**0**

**2**

**4**

**6**

**8**

Fig. 7. Schematic of the fabrication process of a hard-PDMS/PDMS bilayer composite stamp and a pictures of a Silicon master mold and of a bilayer stamp after peeling off.

such a resolution study has been conducted by only a few groups. For instance, KarlSuss GmbH has recently demonstrated the replication of nanoholes with a diameter of 340 nm (±5%) and a minimized residual layer thickness of 36 nm in the Amonil resist with a pitch uniformity of 2 nm over a surface area of 6 inches (Ji et al. (2010)).

The pressure applied during imprinting directly influences the flow of the resist and determines the accuracy of the imprinted nanoscale structures. The first step consists of minimizing the imprinting pressure to reduce local distortions while ensuring a complete resist filling in the stamp nanocavities. We have shown that this pressure can be largely reduced to 0.7 bar (10 PSI) (Cattoni et al. (2010); Shi et al. (2009)). Combined with a UV exposure of 10 min (dose of 2 Jcm2) at 365 nm wavelength, patterns with a high quality shape can be obtained (Figure 8(a)). To quantify the accuracy of the control dimension, we investigated the changes in pattern size on a two-inch wafer scale under this optimum pressure (10 PSI). Nanodot patterns with two different diameters (215 nm and 310 nm) have been chosen for this study. Each field size is 200 um x 200 um on the wafer. The critical dimension of the nanodot patterns on the mold and replicated in the resist were then measured using a high-resolution scanning electron microscopy (Fig. 8(b)). Black and blue points correspond to the measured diameters on the EBL master mold whereas red points correspond to measured diameter in the Amonil resist after nanoimprinting. We observe a broadening of the dot diameter, which never exceeds 10 nm, corresponding to a standard deviation of less than 5%.

12 Will-be-set-by-IN-TECH

h-PDMS / PDMS bilayer stamp

Fig. 7. Schematic of the fabrication process of a hard-PDMS/PDMS bilayer composite stamp

such a resolution study has been conducted by only a few groups. For instance, KarlSuss GmbH has recently demonstrated the replication of nanoholes with a diameter of 340 nm (±5%) and a minimized residual layer thickness of 36 nm in the Amonil resist with a pitch

The pressure applied during imprinting directly influences the flow of the resist and determines the accuracy of the imprinted nanoscale structures. The first step consists of minimizing the imprinting pressure to reduce local distortions while ensuring a complete resist filling in the stamp nanocavities. We have shown that this pressure can be largely reduced to 0.7 bar (10 PSI) (Cattoni et al. (2010); Shi et al. (2009)). Combined with a UV exposure of 10 min (dose of 2 Jcm2) at 365 nm wavelength, patterns with a high quality shape can be obtained (Figure 8(a)). To quantify the accuracy of the control dimension, we investigated the changes in pattern size on a two-inch wafer scale under this optimum pressure (10 PSI). Nanodot patterns with two different diameters (215 nm and 310 nm) have been chosen for this study. Each field size is 200 um x 200 um on the wafer. The critical dimension of the nanodot patterns on the mold and replicated in the resist were then measured using a high-resolution scanning electron microscopy (Fig. 8(b)). Black and blue points correspond to the measured diameters on the EBL master mold whereas red points correspond to measured diameter in the Amonil resist after nanoimprinting. We observe a broadening of the dot diameter, which never exceeds 10 nm, corresponding to a standard

and a pictures of a Silicon master mold and of a bilayer stamp after peeling off.

uniformity of 2 nm over a surface area of 6 inches (Ji et al. (2010)).

Silicon master mold fabricated by EBL

Replicated h-PDMS / PDMS bilayer stamp

1) Spin-coating of h-PDMS on Si master & degassing

2) Casting of PDMS, degassing and soft baking

Si master mold

PDMS

PDMS

deviation of less than 5%.

hard-PDMS

hard-PDMS

3) Demolding and anti-sticking treatment

Fig. 8. Control dimension fidelity after soft UV-NIL at optimized pressure: (a) SEM image of 310 nm-diameter pillar array recorded in the Amonil resist and (b) comparison of the diameter values measured on the master mold and in the Amonil thickness contrast, with the relative calculated diameter standard deviation.

#### **4.2 Minimization of the thickness of the resist residual layer**

A second important point of the nanoimprint process concerns the removal of the residual layer of the resist. Because T-NIL and UV-NIL use rigid mold and high pressure, a thin residual layer of resist is normally left between the protrusions of the mold and the substrate. It acts as a soft cushion layer that prevents direct impact of these fragile nanostrucures and the substrate. This residual layer is normally removed by reactive ion etching (RIE) which can largely affect the original shape and size of the pattern. Soft UV NIL uses flexible stamps and this residual layer can be reduced as much as possible by simply adapting the initial resist thickness to the depth of the stamp (height of patterns). Figure 9(b) shows a SEM cross-section image of the imprinted structures at a optimized pressure of 0.7 bars, with a 25 nm thick residual layer, obtained by adjusting the initial thickness of resist. These results can be compared to the recent ones from AMO that are presented in Figure 9(a).

Finally, in order to validate the ultra-high resolution HSQ process for EBL and our improved hard-PDMS/PDMS bilayer stamp fabrication process, we have carried out Soft UV NIL in the sub-100 nm scale. Figure 9 (b-c) shows the thickness contrast in the Amonil resist after Soft UV-NIL, at 10 PSI and at room temperature, by using the bi-layer mold hard-PDMS/PDMS for 50 nm dots with a pitch of 100 nm and 20 nm dots with a pitch of 60 nm. The high and uniform contrast thinknesses obtained on the whole pattern area demonstrate the ability of Soft UV NIL with hard-PDMS/PDMS stamps to replicate nanostructures in the 20-nm scale. We believe than our current limitations in achieving higher resolution, specifically for the replication of highly dense structures, are essentially due to the replication of the bilayer stamp. It will be necessary to work on this stamp replication process, in particular by further reducing the viscosity of the hard-PDMS prepolymer and by controlling the effects of the thermal curing process, which results in a isotropic volume shrinkage of the hard-PDMS that becomes more and more important at the sub-20 nm scale.

hard-PDMS/PDMS stamp

<sup>153</sup> Soft UV Nanoimprint Lithography:

**(a) (b)**

**(c)**

**Si master Mold (EBL)**

**LSPR biosensor (Soft UV NIL)**

1. Soft UV NIL of Amonil

2. RIE of residual layer, Ge, PMMA

"tri-layer" Soft UV NIL

A Versatile Tool for Nanostructuration at the 20nm Scale

3. Deposition of Au and lift-off

Silicon master mold fabricated by EBL.

Amonil Ge PMMA Au

Fig. 10. (a) Schematic of the tri-layer Soft UV NIL process, (b) SEM image of square gold nanoparticle (size = 200 nm, pitch 400 nm) realized with this method and lift-off, (c) the LSPR biosensor fabricated by Soft UV NIL, integrated in a microfluidic channel with its original

The basic element of the nanocavities array (not shown) is composed by lower thick gold film acting on a glass substrate, a thin dielectric layer forming the gap of the optical antenna and an upper gold nanoparticle realized by "tri-layer" Soft UV imprint lithography and standard lift-off. In figure (a) is shown the concept of the tri-layer system on a generic substrate: typical UV NIL resists (like Amonil) are not soluble in solvents and a lift-off process is possible by using a PMMA layer under the UV NIL resist. In our process a further 10 nm Ge layer is insert between the thick PMMA layer and the thin UV NIL resist (Amonil) to improve the selectivity of the former one over the PMMA layer. After imprinting with a UV light and separation, the top layer structure is transferred into the bottom layer by a sequential reactive ion etching. The high aspect ratio tri-layer so obtained can be used directly as etching mask or for the lift-off of metals. In Figure 10 (b) is shown a SEM image of square gold nanoparticle (size = 200 nm, pitch 400 nm) realized by "tri-layer" Soft UV imprint lithography. Figure 10 (c) shows the LSPR biosensor based on *λ*3/1000 plasmonic nanocavities fabricated by Soft UV Nanoimprint Lithography and integrated in a glass/PDMS/glass fluidic chamber for the

Fig. 9. SEM images of: (a) imprinted structures into AMONIL resist, the structure depth is 170 nm and the residual layer is 36 nm. Reprinted with permission from (Ji et al. (2010)). Copyright 2007 Elsevier. (b) imprinted structures into AMONIL resist, the structure depth is 160 nm and the residual layer is 25 nm. (c-d) High resolution replication at the 50 nm and at the 20 nm range in AMONIL resist.
