Fabrication technologies and processes for Solar

Cells: Photolithography
Juan Esteban Villegas Delgado*
Department of Microsystems Engineering, Masdar Institute of Science and Technology, P.O. Box 24224,
Masdar City, Abu Dhabi, UAE
*
jvillegasdelgado@masdar.ac.ae

Abstract: The photolithography process allows the definition patterns

on top of substrates that will later be used to protect specific regions for
either etching or deposition. This is an essential process every time
features need to be create on the substrates and is normally a recurrent
step in any micro and nano-fabrication process flow. Although solar
cells are comprised of relatively large and bulky layers of materials,
traditional designs with metal fingers as front contacts to collect the
electrons are features that rely on the use of photolithography for their
fabrication. This report describes the overall process flow for the
deposition of a photoresist mask that will be later used for etching.
©2017, Masdar Institute of Science and Technology

References and links

1 AG, C.: ‘AZ5214E product datasheet’, in Editor (Ed.)^(Eds.): ‘Book AZ5214E product datasheet’
(2017, edn.), pp.
2 Pease, R.F., and Chou, S.Y.: ‘Lithography and Other Patterning Techniques for Future Electronics’,
Proceedings of the IEEE, 2008, 96, (2), pp. 248-270
3 GmbH, M.P.M.: ‘AZ MIF Developer datasheet’, in Editor (Ed.)^(Eds.): ‘Book AZ MIF Developer
datasheet’ (2017, edn.), pp.
4 GmbH, M.: ‘Baking Steps in Photoresists Processing’, in Editor (Ed.)^(Eds.): ‘Book Baking Steps in
Photoresists Processing’ (2017, edn.), pp.

1. Introduction
The photolithography step is used to pattern features that will later be used to selectively
etch different areas in the solar cell, specifically the layer of ITO, the n+ Silicon region
and partially etch the intrinsic silicon layer. This will allow the electrical isolation of the
device into different solar cells with diverse areas to be used for characterization process.
A second photolithography step will be needed to define the features that will serve as
metal contacts in the solar cell, and its process will to some extent be identical to the one
described below. The process flow followed for the photolithography is:

1. Surface cleaning of the silicon wafer.

2. Dehydration bake.
3. Spin coat of adhesion promoter (HMDS).
4. Spin coat of photoresist.
5. Prebake.
6. UV light exposure.
7. Development.
8. Post bake.

The description of each step as well as the discussion of some of their more relevant
characteristics are presented hereunder.
2. Details of the process

2.1. Surface preparation and adhesion promotion

The first step for a good process is to guarantee a clean surface to work with. For new
Silicon wafer a combination of Piranha etch + RCA cleaning is recommended to create
surface clear for deposition. In our solar cell process, we already deposited layers of doped
and intrinsic silicon and a layer of ITO (that would be etched by the the proposed bath),
therefore only a solvent cleaning of the substrate is necessary, that will remove most of
the organics films and particles while handling them.
A bath in acetone ((𝐶𝐻3 )2 𝐶𝑂) followed by a bath in isopropanol (𝐶3 𝐻8 𝑂), both for 5
min under ultrasonic vibrations, followed by a rinse with deionized water is sufficient for
a good surface preparation. Surfaces that had previously cleaned can be prepared by a
simple rinse of acetone + isopropanol + DI water.
The cleaned surface needs to be dehydrated to reduce the number of hydrogenated
radicals before the application of the adhesion promoter. This bake step is carried at 120C
for 5 minutes.
The dry surface now is coated with the adhesion promoter. HMDS is spinned coated
at 4000 rpm for 45s. HMDS will react with the any OH radicals in the dried surface (such
OH groups would normally form hydrophilic radicals that want to be avoided. As the
HMDS reacts with the surface, release ammonia and leave methyl groups on the substrate,
thus creating an hydrophobic surface [1]. This will enhance the posterior wetting of the
photoresist and its adhesion.

2.2. Photoresist deposition

Photoresist is an organic photosensitive compound and it is the most important
consumable in the photolithography process. Negative and positive photoresist are
available to either protect or uncover a UV light exposed region respectively, this is, a
positive photoresist will leave uncoated the areas that are exposed (after development) and
a negative photoresist will coat the exposed areas. Photoresists are specifically engineered
to be reactive to specific wavelengths in the ultraviolet spectrum, nevertheless the small
power available in this spectrum under normal illumination is sufficient to initiate the
reactions on the resist, therefore a good control of lighting in the working area is necessary,
limiting the uncontrolled UV light to a minimum.

Figure 1. Cell in the spin coater

The photoresist used in our solar cell fabrication process the lithography is AZ 5214.
This is a resist comprised of a novolak resin (phenol formaldehyde) and naphthoquinone
diazide (photoactive compound), with a good spectral sensitivity for wavelength within
310 nm and 420 nm. A agent is added to the resist that allows the resist to create cross
linkage on light exposed areas during a post bake above 110C.The result is a hardened
resist in all the exposed areas that will ultimately act as a mask for the following process
in the solar cell fabrication. This resist can be used as either positive or negative depending
if it is post baked after or before developing respectively.
The viscosity of the unexposed resist is used to create an “homogenous” film on top of
the work substrates. Having such a film is very important to guarantee that during the
photolithography light exposition, the same amount of energy will be input throughout the
substrate (thus having the same reaction in all the photoresist). Through spin coating, a
constant rotational speed is applied to the resist, the resulting viscous force along the resist
will allow only a specific thickness of it to remain on top of the substrate. The final
thickness of the deposition depends on the viscosity of the resist, the rotational speed and
the total time that substrate is spinned.
Note that at a constant speed the thickness of the resulting deposition tends to stabilize
towards a constant value, due to the equilibrium of the apparent centrifugal force and the
viscous force in the resist. This property is used to enable the deposition of thin films that
with a high repeatability by spin coating them for relatively extended time (above 30s,
depending on the viscosity of the fluid). AZ 5214 has a kinematic viscosity of 24 mm2/s,
Table 1 presents different film thicknesses as function of spin angular speeds [1].

Spin speed (rpm) Film thickness

2000 1.98 um
3000 1.62 um
4000 1.4 um
5000 1.25 um
6000 1.14 um
Table 1. AZ 5214 film thickness

The photoresist deposition will not be completely homogenous. At the edges a thicker
layer of resist will be deposited as the resist accumulates during the spin. The negative
impact of this phenomena is however handled easily by fitting the fabricated structures
towards the center of the wafer.
After the deposition of the resist, a soft bake at 110C for 1 minute is carried. This
process will reduce the solvents in the resist by thermally diffusing the photoresist bulk
and allowing the solvents to evaporate. Note that although it shouldn’t be expected to have
crosslinking of the resist in this stage (as it has not been exposed to UV light), a very good
temperature control is important to prevent any residual crosslinking in the resist as the
prebake temperature lies within the limits to start such process.

Figure 2. Lab researcher holding the cell after resist deposition

2.3. Exposure
The resist coated substrate is ready to be exposed to a radiation of 120mJ/cm2 on the SUSS
MA8 mask aligner. A chrome coated glass mask is mounted on the aligner. This mask has
the same features that we want to have coated with photoresist at the end of the process.
The transmission of the features from the mask into the substrate can be designed to be
either through a close contact of the mask and the resist coating (contact lithography) or
transmitted through the mask and into the substrate throughout a medium (e.g. air).
Although contact lithography can give a very high resolution (with features under 10nm),
as it avoids some of the limitations presented below, the mechanical interaction between
the substrate and the mask will necessarily add stress to both, and may render the process
unviable.
A critical step prior to the exposition is to have a good alignment between the mask
and the substrate. This step is however not so relevant during the first stage of
photolithography as there are no previous features to align with. In our fabrication process
the alignment is simplified to only assure that the features in the mask are correctly placed
in the substrate.
Some of the parameters to consider in this process include:
 Ensure that the wavelength of the light emitted is consistent with the range to
which the photoresist is sensitive.
 The total energy radiated to the resist needs to be within the resist specifications
to guarantee a correct and repeatable reaction (this is to control the power and the
total exposure time).
 The process is to be done in a timely manner to avoid uncontrolled light exposures.

Figure 3. Lab researcher adjusting the cell in the mask aligner

2.4. Limitations of photolithography

The exposure to light is an optical process, and therefore is subject to all the limitations of
the optical components that compose the system such as light dispersion, focal shift,
spherical aberration, coma, astigmatism, field curvature and distortion. Theoretically,
using perfect lens with no aberration, light through a point object projection will propagate
beyond the lens with diffraction (the small point feature acts as a source for light to
propagate spherically instead of linearly). The projection of light through the point feature
will be an Airy disk, a set of concentric circles with the wave fronts of the light propagating
form the point source, with a radius given by [2]

𝜆 (1)
𝑟 = 0.61
𝑛 sin(𝛼)
Where 𝜆 is the wavelength of the light, 𝛼 is the convergence semi angle and 𝑛 is the
refractive index. The factor 𝑛 sin(𝛼) is known as the numerical aperture or NA. The 0.61
factor depends on the optical components of the system and is changed to a variable 𝑘1 .
Ultimately for the mask alignment system, this equation leads to the minimum feature size
as equation (2), which has values about 50 nm for standard photolithography processes.
 𝜆 (2)
𝐶𝐷 = 𝑘1
𝑁𝐴
This expression shows that a reduction of the wavelength (increase of frequency) of
the light will signify a proportional reduction of the minimum feature size. This explains
why the technologies for photolithography have evolved to work on lower wavelengths,
and into the 248–193 nm range. This wavelength based feature reduction faces however
other limitations as we get close to the XRay spectrum, where materials including some
metals start being transparent, limiting the possibilities (and increasing the price) for the
fabrication of masks.
Equation (1) is known as the Rayleigh limit of the optical system, but 𝑘1 can have
values lower than 0.61. Through wavelength engineering, features are designed so that the
interference pattern on the resist, and not the shadow, has the desired geometry (in such
case the mask itself may not resemble the final geometry at all). However, in general,
contrast will decrease as the constant 𝑘1 decreases, to a limit in 0.25 where the contrast is
nearly zero. New techniques on light exposition are being developed to overcome this
limit, such as absorption modulated lithography (AMOL), that has reached feature sizes
of 20nm.
Alternative technologies are also available to overcome the minimum feature size such
as x-ray lithography, electron beam and focused ion beam lithography and non-radiative
lithography such as mechanical nanoimprint, micro and nano-printing and chemical
patterning. EBeam lithography for example (available at the Masdar nano fabrication
facilities) consist on the guided projection of a focused electron beam onto a reactive
photoresist, and instead of the use of a mask, the beam is redirected to write on top of the
resist.

2.5. Post baking

An additional baking step can be added after the light exposure when a specific behavior
in the resist wants to be trigger. As described previously, in the case of the is AZ 5214
photoresist, a post-bake at temperatures above 120C will modify the resist behavior and
allow it to work as a negative photoresist.
This is a step particularly sensible as an overbake will activate the cross linkage even
in areas not exposed by light. This photoresist has a standard post bake period of 50s at
130C, that needs to be refined according to the specific conditions of every process (as
recommended in the resist datasheet). The post baking requires a posterior flood exposure
to light, where all the resist is exposed to light without any mask to weaken the resist that
did not cross linked during the postbake.
Note that in our solar cell fabrication process we do not include this step as we want to
use the resist as a positive one.

2.6. Development
The development process is the chemical decomposition of all the photoresist that has not
crosslinked. The general process consists in a chemical bath that will selectively etch away
the organic resist in the areas that were exposed to UV radiation for positive resist or to
the areas that were protected for negative resist. In general photoresist developers are
solutions of tetramethyl-ammonium-hydroxide (TMAH) in concentrations between 2%
and 3% [3].
The developers are the second most important consumable in the photolithography
process after the resist itself, as they need to create a very precise reaction with the resist
and have a controllable process. Carbon contamination (through carbon dioxide in the air
for example) will decrease the activity of the developer; even a completely sealed bottle
will lose a 0.25% of activity per year [3].
For the development, the cell is submerged in a AZ326 bath during 50s. The control of
reaction time is very important. A too short reaction (under-development) will leave
residues of photoresist over the regions that we want to be uncovered, acting as masks for
the subsequent fabrication processes. In the same fashion, a too long exposure (over-
development) will start removing photoresist that we want to remain protecting areas in
the substrate, changing the shapes and size of the features and ultimately leaving a mask
that does not corresponds to the desired features. A bath in DI water is necessary
immediately after finishing the developing time to stop the reactions in the surface.

Figure 4.Develloped cell being dried after DI water bath

2.7. Hard bake

The cell is ultimately subject to a hard bake at 120C for 1 min. This step allows the
stabilization of the properties of the developed photoresist, allowing the cross linkage of
any remaining unlinked resist and creating a more homogenous film. Some of the
parameters that can be controlled trough hard baking include [4]:
 Embrittlement of the resist (this is to be prevented to avoid crack in the resist)
 Reflow (rounding of unlinked resist on features)
 Resist adhesion to the substrate.
 Chemical stability
 Posterior removal of the photoresist.

Figure 5. Image of the cell after the hard bake

3. Conclusions
As discussed, feature fabrication usually includes one or more photolithography processes.
Although many technologies are available to conduct such process, the accuracy remains
largely on a good control if each indent step, rendering every process variable critical. This
necessarily makes the fabrication process in an expensive one, especially for solar cells
where microfabrication techniques are needed to manufacture systems that will ultimately
cover areas of hundreds or thousands of meters squared.
 An alternative system that is being used for the fabrication of the small features needed
for the contacts in solar cells is screen printing, this process is similar to the ink stencil
printing process that has been used for centuries in traditional lithography but depositing
now an aluminum layer. A stencil template is made with the shape of the final features,
placed on top of the cell and metal paste is pressed against it to deposit it in all the
uncovered areas of the stencil. This process is cheap and can be easily adapted for mass
production, but the quality of the features is low and thus is only used on low quality for
poorer quality substrates, where the impact on the overall efficiency of the cell does not
rely on such system.