This page describes the complete Imprint Cell Process.

The flatness of incoming wafers  determines the processing strategy, for more details got to Substrates

There are simple tests for cleanliness, if they fail, wafers will need to be cleaned. For more details go to

Most wafers need to be surface treated  before the imprint material is dispensed to ensure wetting.  HMDS
is the industry standard for pretreating wafers before polymer coating. A very thin spin on polymer layer can
also be used.

Planarization is  required for  non flat  or device wafers and is accomplished either by spin on or by
imprinting with a reference flat. Spin on planarization works well up to a maximum device feature size of
around 20 um. For larger features, imprint planarization is required. In principle, any imprinter can planarize  
using an unpatterned mold. One challenge is to ensure that the materials in the different layers are
compatible. The other challenge is to obtain the correct combination of planarization, and conformality over
the different length scales of wafer and device structure.

It is worth noting that there is an alternative, post imprint planarization process (SFIL-R) that achieves the
same objective, and is described in the appropriate section below.

For more details go to
Conformality, Planarization and Separation.

There are 4 distinct dispense strategies, the illustration on the right shows three of them - spin on, drop
dispense; and dispense on the mold.

Spin on – is the most common used for UV and thermal imprint.  

Drop dispense – used for very low viscosity materials that cannot be spin coated  as in MII’s  SFIL process,  
and for filling deep optical device structures with high viscosity liquids as reported by Suss and  CSEM.  
Drop dispense must be  part of the imprint tool.

On Mold / Reverse Imprint – spin coating the mold with a solvent and  polymer solution, drying and then
transferring the layer on the mold  to the substrate allows a number of unique materials and devices to be
fabricated. There are many possible variants including, patterning both sides before transfer. Also,self
assembled materials can be deposited on the template.

Inking – in the transfer inking process, the material is placed in a ink pad. The mold is contacted onto the
pad to pick up the material and then transferred to the substrate.   

The mold is pushed into the liquid
. In a classic fluid flow model, the pressure required is directly
proportional to the viscosity and the minimum residual layer gap that the liquid must pass through. Short
imprint times (down to 1 second) have been reported for high molecular weight polymers were the surface
is being embossed so there are no restrictions to flow.

With very low viscosity and very thin layers, it has been reported that thin gaps fill quickest probably because
of capillary forces (Liang 2005). Fill times of 7 seconds have been reported for drop dispense of very low
viscosity liquids and residual layers  less than 100 nm (Shumacker 2005). Repeating structures fill quicker
than varying patterns.

Higher viscosity imprint materials result in residual layers with a thickness variation that depends on
pattern density. For the fastest filling times, drop dispense systems vary the material volume depending on
the pattern density. The effects of residual layer variations can be minimized by post imprint processing as
described below.

Elimination of trapped air is the other limiting factor in molding. Air gets trapped when the non flat surfaces
of the mold and the substrate make contact irregularly. Air also gets trapped when a recess in the template
gets closed off. Finally capillary effects can result in recess being filled last (Liang 2005). There are two
strategies to reduce these effects; either imprint under vacuum which was the traditional contact printing
solution, or use a controlled atmosphere of a gas that diffuses quickly (Hiroshima 2003-1 and  Shumacker

In a device application, the imprint pattern must be overlayed on an existing pattern. Overlay consists of
2 components; align of one location on the mold to the device, and ensuing that the magnification and
distortion of the rest of the field are matched.  The align can occur before the mold touches the liquid. For
the finest align, the mold can be moved while in contact with very low viscosity liquids. Fine overlay also
requires magnification correction (Choi 2004).

IMPRINT SETTING - illustrated in Imprint Essentials
Tg Set – the material is heated far enough above Tg so the viscosity becomes low enough to allow the
material to flow. The mold is pushed into the liquid with pressures up to 50 atmospheres. High pressures
are needed because of the viscosity of  high molecular weight polymers. The material and mold are cooled
below the Tg to set the pattern.

UV Set – the pre-imprint material is a much lower molecular weight, lower viscosity fluid at room
temperature.  The mold is either pushed into a coated liquid film with lower pressures of around 1
atmosphere, or into a  drop dispensed liquid with pressures as low as 1/20 atmosphere. Then the material
is exposed to UV light to crosslink.  The crosslinking produces a material with sufficient mechanical
strength for successful separation. Combined UV and thermal imprinters can be used to crosslink higher
Tg materials.

Thermal Set – thermally initiation can also be used to crosslink low molecular weight liquids.

Transfer from mold imprint – A coated mold is loaded into the imprint tool, and the coating is transferred to
the substrate. The adhesion to the substrate is ensured by a combination of pressure and temperature.

Transfer inks – The mold is first contacted to the ink pad, and then  contacted to the substrate. A very soft
rubber mold made of PDMS is used to ensure complete contact.

Separating the mold from the substrate with a thin layer of a polymer in between is probably the most
challenging mechanical operation. Nominally there are two rigid surfaces separated by relatively thin
flexible layer.  Step and Repeat systems have demonstrated reliable automated separation with an imprint
field size of 25 mm.  Whole wafer imprinters have separated 40 - 200 mm wafers by bowing the mold.

For more details go to
Conformality, Planarization and Separation.

The result is a patterned wafer. In a functional application the process is now complete.  In a application
were the imprint acts as a resist, there are two types of material; silicon containing and organic (non-
silicon materials .

The properties of the post imprint material can be improved by further thermal and UV processing.  An
organic polymer imprinted pattern can also be coated with a silicon containing material as part of an image
reversal process, also known as SFIL-R, illustrated on the right. There are three interesting side effects;  
the image is reversed, the first patterned  imprint step also planarizes underlying topography, and
eliminates any effects due to residual layer variation from imprint pattern density (Shreenivasan 2005).   

For more details go to
Conformality, Planarization and Separation.

The illustration on the right shows the silicon containing layers in green and the organic layers in yellow.  
The etching of bi layer silicon containing resist systems has been the documented in publications from the
mid 1980’s onward. To etch the silicon containing layer, a source of fluorine is added to increase the etch
rate relative to the organic layer. An  O
2 reactive ion etch (RIE) is used for the planarization layer.

The illustration also shows the etch steps in SFIL-R. The spun on silicon containing layer is etched to clear
the raised imprint, the etch is changed to oxygen that etches the exposed organic imprint and the tone is

There are several methods of pattern transfer;

  • Etch - use the resist pattern to etch the underlying layers. This is the most common process
    examples shown in Resnick 2004. Difficult to etch materials are patterned by etching an
    intermediate "hard mask", for example using an intermediate Cr layer to etch SiO2 in making molds.

  • Lift off - material deposited from a point source over a undercut pattern. The underlying layer must
    be solvent soluble or have  triggered adhesion failure to lift off material on top of the resist.  A
    popular solution  when surface quality is critical such as Shottky gate devices. Lift off reported by
    Hyogo 200, Tan 2004 and Lee 2004.

  • Implant  - a standard part of IC processing  

  • Self assembly  of materials on the imprinted surface (Chou 1999 and Sharf 2002)

Depends on the type of resist material and the underlying layers.

  • Wet strip is the classic technique for non - oxidizable substrate patterns such as Si, Si O2, SiN4 etc.  
    Will work on crosslinked and uncrosslinked materials

  • Dry strip using O2 mixtures is  the standard for oxidizable substrate patterns such as metals.

  • Lift off or solvent strip. Requires non crosslinked imprint materials or an underlying soluble layer.

For more on the Imprint Step go to Mold, Material and Tools

Or use the tool bars.
Imprint Cell Process