TECHNICAL FIELD

The present invention relates to a top contact grid design for thin film solar cells and a method of producing an optimized top contact grid by patterning a surface with a metal by controlling the generation of cracks in a deposited metal layer.

BACKGROUND

A well-recognized challenge in the design and production of thin film solar cells is to provide a top contact that is to a large extent transparent and at the same time offers sufficient conductivity. A top contact typically comprises a Transparent Conductive Oxide, TCO, such as ZnO: Al, MgO or ITO or combinations thereof. In order to keep the TCO relatively thin it has been suggested and tested to complement the TCO with a metallic grid on top of the TCO. Grids have been provided by for example evaporation and printing. Grid materials has been typically been high conductivity metals such as gold, silver copper and aluminum. In the case that the grid has been provided by printing the grid typically comprises metal particles in a binder, for example curable resin. It is further recognized that such a grid must be optimized with regards to the shadowing effect of incoming rays, the collection of charges from the underlying photovoltaic layers and the conductivity in the grid. http : //pyeducation . or / pycdrom/ desi n/metal- rid-pattern disclose a grid that have a tree shaped design with a central busbar and a plurality fingers extending out in a perpendicular direction from the busbar, the busbar and the fingers spanning the top surface of the solar cell. The fingers and the busbar are tapered so that they are significantly broader at their base than at their extremity. Although outlining the requirements, the depicted design is not optimal. Additionally methods for producing an optimal or near optimal grid design on an industrial scale have been lacking. Primarily, the lack of precision and reproducibility has hindered all but the simplest grid designs.

One common patterning technique is lift-off, where the surface is covered with a photoresist which is then developed to expose parts of the surface. Then, a layer of the metal is deposited, generally by evaporation or sputtering, and excess metal is removed by removing the photoresist through lift-off, leaving only the desired metal pattern on the surface.

A problem associated with this technical field is that the deposited metal layer makes it difficult to gain access to the photoresist in order to perform the lift-off, since the solvent used for the lift-off is not able to penetrate the metal and contact the photoresist beneath. One solution to this problem is to apply a second layer of photoresist that extends slightly beyond the pattern edges of the first layer of photoresist to protect these edges from the metal and enable contact with the lift-off solvent. This does solve the problem, but causes additional costs and extends the time required for the process. Another solution, as mentioned by the article“Microfabrication and Integration of a Sol-Gel PZT Folded Spring Energy Harvester” by Lueke et al. (published in Sensors 2015, 15(6), 12218-12241), is to use only one single layer of photoresist (positive photoresist profile) but to apply harsh means before lift-off in order to penetrate the deposited metal and access the photoresist beneath. Ultrasonic cleaning is given as one example of such harsh means. The problem with this approach is that the means for gaining access to the photoresist must be able to damage the metal layer in order to be successful, and that this will result in unwanted damage to the metal (or underlying layers) not removed by the subsequent lift-off. Especially in applications such as solar cells where metal is applied as a front contact and where the conductivity of the metal is important, such damage will result in losses in the finished product.

There is therefore clearly a need for a more time and cost efficient high precision method of providing a top contact grid for thin film solar cells, where the drawbacks described above can be avoided.

SUMMARY OF THE INVENTION

The object of the present invention is to eliminate or at least to minimize the problems mentioned above and to provide a solar cell with a top grid contact that is optimized to a higher degree than the prior art top grid contacts. It is a further object to provide a method of industrial scale production of such a solar cell that is cost effective and use as little chemicals as possible. This is achieved by the solar cell as defined in claim 1 and in claim 17 and by the method of production as defined in claim 11.

A solar cell according to the invention comprises a top contact grid provided on top of a transparent front contact of the solar cell. The top contact grid comprises a plurality of branches, wherein at least one branch has a first side surface and a second side surface extending upwards from the surface of the transparent front contact, a fractured top surface parallel to the surface of the transparent front contact and a first branch edge joining the first side surface and the fractured top surface and a second branch edge joining the second side surface and the fractured top surface. The solar cell according to has a top contact grid comprising branch that are near an ideal structure with regards to the shape of its cross section. Each cross section of the branch has a maximum height, /?, from the transparent front contact, the maximum height, /?, provided at the fractured top surface. The first side surfaces and the second side surface extends upwards from the surface of the transparent front contact and join the respective branch edges at a distance d measured from the transparent front contact surface in the direction of the normal of the surface, wherein the distance d is at least 70% of the maximum height /?, more preferably at least 80% of the maximum height h and even more preferably at least 90% of the maximum height h.

Alternatively the ideal structure may be envisaged as a rectangle with a with the height corresponding to the maximum height h of the branch and a width w of the branch at the position of the cross section The branch has a cross sectional area that is at least 70% of the area of an ideal structure, more preferably at least 80% of the area of the ideal structure and even more preferably at least 90% of the area of the ideal structure.

According to one aspect of the invention the first side surface and the second side surface are columnar structures.

According to one aspect of the invention the first side surface and the second side surface forms an angle with the surface of the transparent front contact that deviates at the most 30°, preferably at the most 20° from the normal of the surface the transparent front contact.

According to one aspect of the invention the top contact grid comprises a busbar and the plurality of branches extends out from the busbar, so that junctions are formed there a branch joins the busbar. According to one aspect of the invention the busbar is tapered and widest at its base connecting to an interconnect of the solar cell and most narrow at the top end and at least one of the branches is tapered with the smallest width at its extremity and the larges width at the junction. According to one aspect of the invention the busbar is increased in size after the junction with a branch as seen going from the top end towards the base so that the busbar after the junction has a cross section area that is in the order of the sum of the cross section areas of the branch and the busbar before the junction.

According to one aspect of the invention the junction comprises two branches joining the busbar at the same position of the busbar, the width of the busbar after the junction is the sum of the widths of the busbar before the junction and the widths of the two branches just before the junction.

Thanks to the invention a solar cell may be provided with a top contact grid which is much better optimized than prior art grids. The optimization may be with regards to one or a combination of the factors:

-Charge collection from the photovoltaic layers,

-Charge transport in the grid,

-Shading effects due to the area of the TFC directly covered by the top contact grid,

-Shading effects at incident light with an inclination to the top surface due to the top grid contact extending in the normal direction of the top surface of the solar cell,

-The conductivity of the grid material(s),

-Material consumption,

-Possibility to realize the design of the top contact grid industrially and economically.

The different aspect of the grid design represents different degrees of optimization, but all with a substantial degree of optimization higher than what was possible with prior art techniques. The skilled person will, given the presented options select a design suitable for a particular application.

The method according the invention for providing a top contact grid on a transparent front contact of a multilayered photovoltaic structure comprises the steps of: a) Providing a photovoltaic multilayered structure with a transparent front contact as the uppermost layer; b) applying at least one layer of photoresist on the surface of the transparent front contact; c) baking the photoresist for removing excess solvent; d) creating a pattern defining the top contact grid in said photoresist to expose at least one portion of the surface, said pattern comprising at least one edge between the photoresist and the exposed at least one portion of the surface; e) depositing a metal on said photoresist and said at least one portion of the surface to form a metal layer, said metal having a temperature at deposition that is higher than a temperature of the substrate; f) cooling the deposited metal through contact with the exposed surface and the

photoresist layer, thereby causing a contraction of the deposited metal and generating at least one crack or gap through the metal layer at the edge; g) applying a solvent through the at least one crack or gap for removing the

photoresist and thereby removing the metal layer deposited thereon but leaving the metal deposited on the exposed surface, the deposited metal forming the top contact grid.

According to one aspect of the invention the temperature of the substrate (during the step of depositing the metal, step e), is less than l40°C. Thereby, the photoresist is maintained at a suitable temperature and is able to accommodate the contraction of the metal layer during cooling. The photoresist may be deposited in a layer having a thickness that is equal to or lower than a thickness of the metal layer deposited in the step of depositing the metal, step e). By applying the photoresist in a thin layer the process is made more cost efficient and the generated cracks can be placed in a part of the edge close to the exposed portion of the surface, resulting in a desirable shape of edges of the metal left on the surface after lift-off. Preferably, the layer of photoresist is thinner than the deposited metal layer.

According to one aspect of the invention the materials are selected so that the thermal expansion coefficient of the deposited metal is larger than the thermal expansion coefficient of the substrate or the at least one deposited layer. Thereby, the contraction of the deposited metal during cooling will be larger than a contraction of the substrate or deposited layer(s) on the substrate, enabling the generation of cracks even when the substrate has a temperature close to the deposited metal during deposition.

According to one aspect of the invention the surface of the substrate has a uniform

temperature during the baking of step c), said uniform temperature being such that a difference in temperature across the surface (is less than 7°C, preferably less than 4°C and more preferably less than 2°C. Thereby, the photoresist layer will have essentially the same properties across the surface of the substrate and is able to contract and generate cracks in a uniform way. This avoids areas of the substrate where the photoresist is harder and less able to contract and also areas where the photoresist contains excessive amounts of solvent that would prevent a controlled deposition of metal.

According to one aspect of the invention the substrate is cooled after the baking of step c) and wherein the surface of the substrate has a uniform temperature during cooling, said uniform temperature being such that a difference in temperature across the surface is less than 7°C, preferably less than 4°C and more preferably less than 2°C. This also serves to keep the layer of photoresist uniform so that the properties of the photoresist are the same across the substrate. By cooling at a uniform temperature, unevenness in the baking is avoided so that the deposition of metal and the following generation of cracks are made in a controlled way, thereby improving the results of the method according to the invention.

The solar cell according to the invention comprises a top contact grid and is produced with the method described above.

Many additional benefits and advantages of the invention will become readily apparent to the person skilled in the art in view of the detailed description below.

DRAWINGS

The invention will now be described in more detail with reference to the appended drawings, wherein

Fig. la discloses schematically a substrate with a surface on which a metal is to be patterned;

Fig. lb discloses schematically a substrate with at least one deposited layer and a

surface on said layer on which a metal is to be patterned; Fig. 2 discloses schematically the substrate and deposited layer of Fig. lb with a layer of photoresist on said surface;

Fig. 3a discloses schematically the substrate, deposited layer and photoresist of Fig. 2 where a pattern has been created in the photoresist; Fig. 3b discloses a schematic top view of a substrate on which an exemplary pattern has been created in a layer of photoresist;

Fig. 4 discloses schematically the substrate, deposited layer and patterned photoresist of Fig. 3a with a deposited metal layer;

Fig. 5 discloses schematically the substrate, deposited layer, patterned photoresist and deposited metal layer of Fig. 4 with cracks or gaps generated at edges of the pattern;

Fig. 6a discloses the schematic view of the substrate, deposited layer and deposited metal according to the pattern, after removal of the photoresist;

Fig. 6b discloses a schematic top view of the exemplary pattern of Fig. 3b with metal deposited according to said pattern; Fig. 6c discloses a schematic view of an enlargement of a feature in Fig. 3a where edges of the photoresist are shown in more detail;

Fig. 7a _{l 2} discloses a cross-section made with focused ion beam of an enlarged view of a deposited aluminium layer with cracks along the sides of a feature, according to Example 1, in a photograph and in a crosshatched version;

Fig. 7b ! ₂ discloses an enlarged view of the deposited aluminium of Fig. 7a after lift-off in a photograph and in a crosshatched version;

Fig. 8a _{l 2} discloses an enlarged view of a deposited aluminium layer with cracks along the sides of a feature, according to Example 2 in a photograph and in a crosshatched version;

Fig. 8b ! ₂ discloses an enlarged view of the deposited aluminium of Fig. 8a after lift-off in a photograph and in a crosshatched version; Fig. 9a _{l 2} discloses an enlarged view of a photoresist layer with features after developing, according to Example 3 in a photograph and in a crosshatched version;

Fig. 9b _{! ,2} discloses an enlarged view of deposited aluminium before lift-off with cracks along the sides of a feature according to Example 3 in a photograph and in a crosshatched version; and

Fig. 10 discloses a photograph of an enlarged view of a metal portion deposited

according to the method of the invention and showing a columnar structure in a fractured surface.

Fig. 1 la-d illustrates schematically a thin film solar cell (a) according to the present

invention, a cross section of a branch of the top contact grid according to the invention (b), a cross section of a branch of the top contact grid according to prior art (c) and a cross section of a branch of the top contact grid according to the invention (d);

Fig- 12 illustrates schematically a top contact grid for a thin film solar cell according to one embodiment of the invention;

Fig- 13 illustrates schematically a top contact grid for a thin film solar cell according to one embodiment of the invention;

Fig- 14 illustrates schematically a top contact grid for a thin film solar cell according to one embodiment of the invention; Fig- 15 illustrates schematically a top contact grid for a thin film solar cell according to one embodiment of the invention;

Fig- 16 is a flowchart of the method according to the invention.

DETAILED DESCRIPTION

The present invention relates to an optimized top contact grid design for thin film solar cells and a photolithographic method for patterning a surface of a solar cell module with a metal, referred to as a lift-off method, to produce the optimized grid design. The method according to the invention will firstly be described in a general way.

There is a general prejudice within the art that cracks in a metal layer are not desirable due to the risk of discontinuity and defects in the metal pattern. Due to this requirement, experiments and developments in the field have not been made in this direction; rather, efforts have been made to facilitate lift-off by applying additional layers of photoresist or by developing more aggressive methods for penetrating the metal to allow access for the solvent to the photoresist underneath. The inventive method developed by the present inventors provides a way to control the generation of cracks so that they appear where desired, allowing a solvent to access the photoresist and allow the lift-off to take place as described herein. The present invention thereby provides an advantageous method that can be used to facilitate and improve the patterning of metal on a surface.

Also, according to the general knowledge of the person skilled in the art, a photoresist layer needs to be 2-3 thicker than a metal layer when performing lift-off. Surprisingly, however, a much thinner photoresist in relation to the metal layer can be used, giving the significant advantage of a more cost efficient and environmental friendly process.

Thus, Fig. la discloses in a schematic view from the side a substrate 101 having a surface 103 on which a metal is to be patterned, for instance in printed circuit boards, address plates in the flat panel display industry or a substrate collector grid for thin film solar cells. Fig. lb discloses a substrate 101 having at least one deposited layer 102 formed thereon, with a surface 103 on the layer(s) 102 on which the metal is to be patterned. The deposited layer or layers 102 may be simply an oxide grown on the substrate, but may also comprise a plurality of deposited layers 102 that form a more complex structure. In one advantageous

embodiment, the layers 102 are a back contact and a thin- film of a solar cell. The thin film may be a CIGS stack, but can also be another type of solar absorber stack or thin film.

The substrate 101 may be a silicon wafer or a glass sheet, but can alternatively also be another material suitable for use in photolithographic processes and for patterning with a metal.

In one preferred embodiment, the patterning according to the invention is performed on thin- film solar cells, preferably CIGS cells that are deposited on a substrate made from soda-lime glass with a thickness of about 1-3 mm and with a back contact in the form of a molybdenum layer. The CIGS layer is created on the back contact, and on top of the CIGS layer is generally provided a transparent conductive oxide (TCO) layer (preferably ZnO) that forms the surface 103. In this preferred embodiment, the patterned metal layer as described below will together with the TCO serve as a front contact for the solar cells. The manufacturing process of thin-film solar cells in general will not be described in detail but is already well- known within the art.

The method for patterning a surface with a metal according to a preferred embodiment of the invention will now be described with reference to Fig. 2-6, and will also be described in more detail below with reference to the enclosed examples.

In the following, the method will be described with reference to the substrate and deposited layer(s) disclosed in Fig. lb, but it is to be noticed that what is said with regard to this will also apply equally to the method using the substrate disclosed in Fig. la.

After providing the substrate 101 and deposited layer(s), at least one layer of photoresist 104 is applied on the surface 103. In the preferred embodiment, the photoresist has a thickness that is equal to or thinner than the metal layer and be provided over the entire surface 103. After application, the photoresist layer 104 is baked to remove excess solvent. Depending on the type of photoresist selected, the time and temperature for soft baking may differ and are selected so that the solvent content of the photoresist after baking is 0.5 percent by weight or higher. If excessive solvent is left after baking, the photoresist would still be wet and would not be able to maintain its shape after patterning. If on the other hand too little solvent is left after baking, the photoresist will be too hard and inflexible and will make it more difficult to control the generation of cracks through the metal. Depending on the photoresist used, some experimentation may be required to arrive at the most beneficial properties of the photoresist after baking, as will be readily understood by the person skilled in the art.

It is advantageous to closely control the temperature of the surface 103 and thereby also of the photoresist layer 104 during baking to prevent the photoresist layer 104 from becoming unevenly baked or with varying hardness and flexibility in different parts of the photoresist layer 104. It is especially advantageous to maintain a uniform temperature of the photoresist layer 104 on the surface 103 during baking since this will ensure that properties of the photoresist are the same across the entire photoresist layer 104 after baking, improving control of a generation of cracks in a deposited metal layer as will be described in detail further below. A uniform temperature is defined herein as a temperature where a difference between a highest temperature and a lowest temperature is small. For the present application, that difference should preferably be less than 7°C in order to improve controlling the generation of cracks, but it is beneficial to have an even smaller difference such as 4°C where especially good results have been achieved during testing. For a smaller difference yet, where only 2°C differ between a highest and lowest temperature of the photoresist layer 104 during baking, excellent results have been achieved. It is to be noted also that the temperature of the photoresist layer 104 and the temperature of the surface 103 will generally be the same or differ only a little, since the photoresist layer 104 is thin.

After baking, the substrate 101 and the photoresist layer 104 on the surface 103 of the substrate are cooled down and it is advantageous to maintain a uniform temperature also during cooling. This is especially beneficial immediately after baking when the temperature is still high, but as the cooling progresses it becomes less important. It has been shown in experiments that temperatures above 40°C or preferably above 50°C and more preferably above 60°C should be uniform, whereas lower temperatures can be allowed to differ more. A uniform temperature for the cooling is defined in the same way as a uniform temperature for the baking, stating that a temperature difference of less than 7°C is acceptable but that better results are achieved with a temperature difference of less than 4°C, with even better results for a temperature difference of 2°C or less. Cooling is preferably performed using an air blade, but other means for cooling may also be suitable. It is advantageous to avoid cooling under an ambient temperature to prevent condensation. Since the baking aims to remove solvent from the photoresist, the removal will continue as long as the photoresist is held at a temperature where solvent can evaporate. In order to achieve a photoresist layer 104 with uniform properties, especially regarding flexibility and hardness, it is therefore advantageous to ensure that no unevenness has arisen during baking and cooling. Preferably, the baking is a prebake or soft bake. The term“soft baking” is used to refer to a baking performed after a deposition of photoresist and before exposure of said photoresist.

The aim is to remove solvent from the photoresist. Soft baking is well known within the field of photolithography and is also referred to as a prebake.

In the following, any thickness of the photoresist layer given is the thickness after the soft baking step. Following the soft baking, a pattern 110 is created in the photoresist layer 104 in order to expose portions of the surface 103. The patterning comprises exposing defined areas of the photoresist to light. The exposure could be performed by application of a mask showing the pattern, exposure to light and developing to remove the photoresist layer 104 according to the pattern and to expose the surface 103 at the at least one feature 105. It can also be patterned by a laser without mask. The photoresist may be positive or negative, with the pattern in the mask being adapted accordingly but preferably positive resist is used since it generally allows for higher pattern resolution. The feature 105 has edges 109 that mark the border between the photoresist 104 and the exposed portions 103, and these edges 109 have an edge width w defined as a distance from a region with a full thickness of the photoresist layer 104 to a region where the photoresist layer 104 is completely removed and the surface 103 is exposed. Thus, the edge 109 can be said to have a tapering photoresist layer 104 after developing. Preferably, the edge width w is smaller than or equal to five times the thickness of the photoresist layer 104, preferably less than 2 times the thickness of the photoresist layer, and more preferably less than the thickness of the photoresist layer, thereby creating a sharp decrease in thickness of the photoresist layer along the edge width. It is advantageous to have such a sharp decrease in order to control the generation of cracks, as will be described in more detail further below. It is advantageous to use a light source with a collimation of ±10 degrees to give a high degree of precision and control when the photoresist layer 104 is exposed to the light.

After the pattern 110 has been created, a solvent is used for removing undesired photoresist by development as is well known within the art. Especially in applications where thin film solar cells are patterned using the method according to the present invention, it is important to monitor the pH of the solvent during development to avoid damaging the oxide layer (generally ZnO) on the solar cells. A pH of less than 13 will result in the development taking much longer time or not occurring at all, whereas a pH of more than 13.1 risks damaging the ZnO. Similarly, a shorter time of exposure to the solvent may result in development being incomplete or not taking place at all, but a longer time of exposure again risks damaging the ZnO. Fig. 3b discloses a plurality of features 1 lOa, 1 lOb, 1 lOc that together form the pattern 110 and that may have different shapes and sizes. A feature is also denoted by 105 in the following and refers to any one of the features 1 lOa, 1 lOb, 1 lOc forming the pattern 110 as a whole. With a suitable solvent content after baking, the photoresist will be able to maintain the pattern and will also be flexible enough to allow a contraction of a metal deposited thereon, as will be explained in detail in the following.

The distance between features 105 is denoted by D _f whereas the thickness of the metal layer is denoted by T _M and the thickness of the photoresist layer by TR. The thickness T _M of the metal is preferably in the range of l-lOpm, more preferably in the range of 2-4.5 pm and the thickness T _R is preferably in the range of 0.8-0.5pm. It is advantageous from a cost perspective that T _M ³ T _R.

The minimum distance D _f between features should be large enough that the accumulated tensile stress over that distance is sufficient to generate the crack in the metal. In order to achieve the best results, the minimum distance should be larger than the thickness of the metal layer so that D _f > T _M.

Fig. 3a shows the pattern in the photoresist from the side, whereas Fig. 3b shows the resist pattern from above. It is to be noted that the pattern shown in Fig. 3a is only part of the pattern of Fig. 3b, and that these are also merely examples of patterns that can be created through the method according to the invention. The actual shape and dimensions of the pattern will in practice vary depending on the shape and properties of an end product, such as for example a front contact on a thin-film solar cell, as is readily understood by the person skilled in the art. After patterning the photoresist layer 104, a metal layer 106 is deposited and covers the photoresist layer 104 and the exposed surface 103 to cover all available surface in metal. Preferably, the metal is deposited through an evaporation process in an evaporation chamber, but alternatively the metal could also be applied through sputtering or through another process suitable for the same purpose. The metal is preferably selected among metals having good electrical conductive properties. Among these aluminium, copper, and silver are especially advantageous. They each have high thermal expansion properties and are able to contract during cooling in such a way that cracks 108 are formed through the entire thickness of the metal layer 106 and expose the photoresist layer 104. During deposition, the deposited metal has a temperature that is higher than a temperature of the substrate 101. This allows the metal layer 106 to cool through contact with the layer(s) already deposited on the substrate 101, and also ensures that the metal will contract more during cooling than the substrate 101.

To further increase this contraction difference, the metal can be selected to have a higher thermal expansion coefficient than the substrate. In some embodiments, it the metal layer 106 may comprise a plurality of sub-layers comprising the same or different metals that are deposited one at a time during the step of metal deposition.

The temperature of the evaporated metal is generally above 700°C but can be as high as l500°C when it leaves the evaporation boat. The temperature is related to the kinetic energy of the evaporated metal and fades with distance from the evaporation boat due to gravity and radiation and thus the heating effect when hitting the substrate also fades with distance.

Temperature measurement in vacuum is challenging and often yields contradictory results.

For example, when measuring the temperature with a 0,5mm thermocouple at the same distance from the evaporation boats and with a process speed as a normal substrate it may show anything in the range of 400-800°C. A 3mm glass substrate will act as a heat sink preventing overheating of the photoresist (typically photoresist starts to take damage over l40°C). Measuring the temperature with temperature stickers at the photoresist surface shows temperatures around 80°C. When measuring the width of the crack and calculating backwards how large the temperature difference at least must have been to generate those cracks it indicates temperatures of 400-800°C. Some tension might arise from stress induced by the quickly grown film as is understood by a person skilled in the art.

Thus, after deposition of the metal layer 106, the metal is cooled to a lower temperature through contact with the substrate 101 and the layer(s) 102 and photoresist layer 104 previously deposited thereon. During cooling, the metal layer 106 will contract and thanks to the flexibility of the photoresist and to the exposed portions of the surface 103 according to the pattern 105, tension is created in the metal layer 106 that eventually cause the metal to crack through its entire thickness from a surface of the metal layer and down to a bottom of the metal layer. Due to the arrangement of the pattern 105, the areas of photoresist 104 and the placement and shape of the edges 109 between them, the cracks 108 or gaps can be controlled to appear at the edges 109.

After cooling of the metal layer 106, the photoresist layer 104 and the metal deposited on top of the photoresist is removed through lift-off by applying a solvent that is able to dissolve the photoresist and cause it to be released from the substrate 101 or at least one layer 102 on the substrate. The solvent is applied on top of the metal layer 106 and penetrates the cracks or gaps 108 at the edges 109 of the pattern 105, thereby accessing the photoresist layer 104 beneath. The metal deposited directly on the surface 103 will be unaffected by the lift-off step and will remain in place.

The layers of photoresist used in similar photolithographic processes in the prior art have been comparatively thick. It is well known within the industry that the resist should be 2-3 times thicker than the metal layer to facilitate the lift-off. For the present invention, however, the photoresist layer 104 can be thinner than the metal layer 106. A very thin photoresist layer 104 has the advantage of being cost efficient, since less photoresist and lift-off solvent is used, but also that more importantly the generation of cracks can be controlled so that the cracks 108 appear on a side of the edges 109 that is closest to the exposed surface 103. This results in less metal left on each side of the metal left on the surface 103 according to the pattern 105 after lift-off.

Another significant advantage of the thin photoresist layer enabled by the present invention is that a significantly smaller volume of solvent is used for performing lift-off. This renders the method highly cost efficient and also makes it possible to achieve a more accurate control of a pH value of the solvent during lift-off. Generally within the art, a solvent such as DMSO is used and potassium hydroxide (KOH) is added when the pH becomes too low due to the photoresist affecting the solvent.

It is therefore advantageous that a thin layer of photoresist 104 as compared with

photolithographic methods according to the prior art is used, since the effect on the pH of the solvent is significantly decreased. This allows for a higher degree of control over the lift-off and also over the development described above. Furthermore, it also decreases the risk of damage to sensitive layers underneath the photoresist, such as the ZnO layer provided on thin film solar cells.

The pattern 110 comprises a plurality of features 105, such as features or patterned sections of other shapes that are separated from each other. It is to be noted that the generation of random cracks between features would in most cases not be a problem, since the lift-off can still be performed through these cracks where the solvent can contact the photoresist layer 104, but if the randomly generated cracks would appear too close to an edge 109 there is a risk that the ability of the metal to crack at that edge 109 would be decreased. The steps of the method are preferably performed in an in-line continuous process where a substrate is transported through a process chamber and subjected to the steps of the method before exiting the process chamber when the patterning is complete.

Especially the deposition of the metal layer 106 on the photoresist layer 104, is

advantageously performed in connection with such a process since it limits the time in which the substrate 101 is subjected to the deposition of the metal. Thereby, the rate of transport and the flux from the evaporation sources can be adjusted to prevent additional deposition of metal after the cracks have appeared. Another way of achieving this would be with a stationary substrate and a shutter that allows for an abrupt stop of the deposition. To be able to prevent additional metal from being deposited on the surface is beneficial, since the deposition of metal into a newly formed crack 108 can therefore be prevented.

The cracks will appear in the metal layer at a moment during the evaporation where the metal layer has reached sufficient thickness and the metal has been cooled a sufficient amount, often in the area of 2-4pm for aluminium. If the evaporation zone is large, the evaporation continues after cracks have appeared, resulting in a continuing deposition of metal in the cracks and thereby creating smaller cracks which prolong the amount of time needed for preventing the lift-off in the next step of the method as well as the residue metal on the sides of the metal on the exposed part of the surface. With a smaller evaporation zone however, the evaporation rate is dropped sharply after cracks have appeared and no continued deposition of metal in the cracks can take place, giving access to the photoresist layer beneath during the lift-off.

The metal used with the present invention preferably has an electrical conductivity of at least 36* 10 ⁶ siemens/m and is preferably aluminium, copper or silver. This enables control of the generation of cracks in the metal layer 106 as well as good electrical conductive performance of the metal after the patterning. In a preferred embodiment, the surface to be patterned is a surface of a solar panel comprising thin-film solar cells, preferably CIGS cells, deposited on a substrate, and either aluminium, copper or silver would in this embodiment be advantageous for use as part of a front contact.

By using the method described above, at least one metal portion is formed on the substrate or on the previously deposited layers on the substrate. This metal portion is characterized by at least one part of its surface being a fractured surface, generated through the crack that separates the metal deposited on the surface of the substrate or deposited layers from the metal deposited on the photoresist and removed by the subsequent lift-off. The fractured surface has a columnar structure as shown in Fig. 10 and may also have microvoid nucleation where intergranular fractures in the metal have appeared.

Fig. 10 discloses a fractured surface 201 where the columnar structure is clearly visible after the metal has cracked to form the surface, and in contrast also discloses a facetted surface 202 where the metal surface has been formed through evaporation without further chemical or mechanical changes.

In contrast, a metal portion deposited through evaporation according to the prior art and not subjected to any further chemical or mechanical changes such as the generation of cracks has a smooth surface with a facetted structure. For a thin metal portion deposited in this way, the surface would also be smooth and following a surface roughness of the material underneath.

Thus, by a close inspection of the surface of a metal portion resulting from the herein described inventive method the fractured surface can be recognized by its columnar structure and such a metal portion is easily distinguishable from a metal portion deposited according to known prior art methods.

A thin film solar cell 1111 according to the invention is schematically illustrated in Fig. 1 la. The device comprises, from back to front, a substrate 1101, a back contact 1121, the photovoltaic layers 1122, a transparent front contact, TFC, 1123 and a metal layer, corresponding to metal layer 105 of Figs. 1-10 forming a top contact grid 1105 comprising of a plurality of branches 1106. As appreciated by the skilled person various other layers may exist in the structure, for example buffer layers. A variety of photovoltaic layers 1122 suitable for thin film applications are known in the art, for example Copper Indium Gallium Selenide (CIGS). The other layers are often selected to work well with the photovoltaic layers of choice. For a CIGS cell the back contact 1121 is often a Mo-layer and the transparent front contact 1123 is often made of ZnO, ZnO: Al. The layers below the top contact grid 1105 are well-known for the skilled person and will be referred to as the multilayered photovoltaic structure 1112. Materials for the top contact grid 1105 are typically Al, Ni or Ag or alloys thereof. Multilayered structures of different highly conducting metals are also known in the art. As described in the background section and in references cited wherein, a top contact grid suitable for thin film solar cells, in particular CIGS, should satisfy a number of design requirements in order to provide an optimal functionality as a complement to the transparent oxide top contact typically associated with thin film solar cells. Preferably the top contact grid 1105 should be optimized with respect to at least the following:

-Charge collection from the photovoltaic layers,

-Charge transport in the grid,

-Shading effects due to the area of the TFC directly covered by the top contact grid,

-Shading effects at incident light with an inclination to the top surface due to the top grid contact extending in the normal direction of the top surface of the solar cell,

-The conductivity of the grid material(s),

-Material consumption,

-Possibility to realize the design of the top contact grid industrially and economically.

Thanks to the new and inventive process it is possible to produce a top contact grid 1105 that can be better optimized with regards to the above factors than was previously possible. The novel process provides a thickness of the top contact grid 1105 of 2- 4.5 pm. The top contact grid discloses the fractured surface 201 discussed above, with the columnar structure is clearly visible after the metal has cracked to form the surface, and also discloses a facetted surface 202 where the metal surface has been formed through evaporation without further chemical or mechanical changes.

Fig. 1 lb illustrates schematically a cross sectional view of the top contact grid 1105 according to the invention on top of the multilayered photovoltaic structure 1112. An ideal structure 1 l06i in cross section is indicated with a dashed line, forming a rectangle. Each branch 1106 of the top contact grid 1105 according to the invention has a first side surface 1 l06a and a second side surface 1 l06b extending upwards from the surface of the transparent front contact 1123. The columnar structure ensures that the side surfaces 1 l06a, 1 l06b will be generally perpendicular to the transparent front contact 1123 surface. However, as appreciated by the skilled person slight deviations and unevenness will occur, which is visible in Fig. 10. The first side surface 1 l06a joins the top surface 1 l06c, which is essentially parallel to the transparent front contact 1123 surface, forming a first branch edge 1 l06d. The second side surface 1 l06b joins the top surface 1 l06c, forming a second branch edge 1 l06e. The cross section of top contact grid 1105 may be rounded at the branch edges 1 l06d, 1 l06e and hence the top surface 1 l06c may have a slight convex curvature. However the cross section is still substantially rectangular as compared to the cross section of a prior art grids, illustrated in Fig. 1 lc, provided by printing, for example, which will have a pronounced curvature and no distinguishable edges. For a given maximum height h and width w the cross section of the branch 1106 will have a substantially larger cross sectional area than the cross section of a branch of a prior art grid 1 l06p.

An ideal structure 1 l06i in each cross section of the branch 1106 may be defined as a rectangle, indicated with a dashed line, with the height h corresponding to the maximum height of the branch 1106 in the cross section and a width w of the branch 1106 at the position of the cross section, and with an ideal area that is h x w. Preferably the cross sectional area of the branch 1106 is at least 70% of the ideal structure, more preferably at least 80% of the ideal structure and even more preferably at least 90% of the ideal structure 1 l06i. The near ideal structure as seen in a cross sectional view may alternatively be described as the side surfaces 1 l06a, 1 l06b extending upwards from the surface of the transparent front contact 1123 and joining the respective branch edges 1 l06d, 1 l06e at a distance d measured from the transparent front contact 1123 surface in the direction of the normal of the surface. The distance d should be at least 70% of the maximum height /?, more preferably at least 80% of the maximum height h and even more preferably at least 90% of the maximum height h.

Depending on the process parameters protrusions 1 l06f may occur in the corners of the cross section of the top contact grid 105, which is schematically illustrated in Fig. 1 ld, the size of the protrusions being exaggerated for illustrative purposes. Such protrusions will not impede the function of top contact grid 1105. The top contact grid comprises in all of the below described embodiments the fractured surface with the columnar structure is clearly visible after the metal has cracked to form the surface, the facetted surface 202 and the distinct rectangular cross section.

Fig. 12 illustrates schematically a solar cell 1111 comprising a top contact grid design according to one embodiment of the invention. The top contact grid 1105 has a general tree shaped structure with a stem or busbar 1126 and a plurality of branches 1106 extending out from the busbar 1126, so that junctions 1108 are formed there a branch 1106 joins the busbar 1126. The busbar 1126 and branches 1106 are referred to as grid elements. The branches 1106 are uniform in width, with a width in the range 10- 20 pm, and a length from the busbar 1126 to the end of the branch 1106 in proximity to the edge of the cell, of preferably at least 10 mm. The branches 1106 are spaced apart by 100-500 pm. The busbar 1126 has a width of at least 10 pm and spans the cell from an interconnect to, or close to, a scribe defining the end of the cell. The combination of a central busbar 1126 and comparably long branches 1106 provides a large cell, which may be operated at a lower voltage than a smaller cell, which is a great advantage as more cells may be connected in a string and thus lower the balance of the system cost. The cell 1111 is defined by a scribe which reaches to the back contact 1102. The top contact grid 1105 has been provided by the lift-off process described below and the cell scribed after the provisioning of the top contact grid 1105. The combination of the novel lift off process and the scribing to give the cell defining scribe(s) after application of the top contact grid 1105 provides for high precision and a very good alignment of the top contact grid with the edges of the cell 1111. This is an advantage for making larger cells, wherein just a small deviation or mismatch between the top contact grid 1105 and the cell defining scribes would lead to a fatal malfunction.

One embodiment of the invention is schematically illustrated in Fig. 13 wherein the busbar 1126 is at its base 1 l26a connected to an interconnect of the solar cell (not shown). The busbar 1126 has a top end 1 l26b extending to and ending at, or close to, the end surface of the cell. The busbar 1126 is tapered and widest at its base 1 l26a and smallest at the top end 1 l26b, the increase in width towards the base reflecting that the busbar 1126 needs to carry an increasing number of charges towards its base. In a similar manner the branches 1106 are tapered with the smallest width at their extremity and the larges width at the respective junction 1108 with the busbar 1126. The branches 1106 has a width ranging from a few pm to up to around 20 pm, and a length from the busbar 1126 to the end of the branch 1106 in proximity to the edge of the cell, of at least 10 mm. The branches 1106 are spaced apart by 100-500 pm.

One embodiment of the invention is schematically illustrated in Fig. 14, wherein the busbar 1126 and the branches 1106 have similar tapered shape as in the embodiment illustrated in Fig. 13. To further optimize the top contact grid design the cross section of the busbar 1126 is increased after the junction with a branch 1106, as seen going from the top end 1 l06b towards the base 1 l06c, so that the busbar 1126 after the junction 1108 has a cross section that is in the order of the combined cross sections of the branch 1106 and the busbar 1106 before the junction 1108. The effect is exaggerated in the figure. Given a constant thickness of the top contact grid 1105 this is preferably realized by increasing the width of the busbar 1126, so that the width of the busbar 1126 after the junction 1108 is the sum of the width of the busbar 1126 before the junction 1108 and the width of the branch 1106 just before the junction 1108 as seen from the top of the busbar 1 l26b towards the base of the busbar 1 l26a, i.e. the direction of the electron flow in the top contact grid 1105 in use. If the junction 1108 comprises two branches 1106 joining the busbar 1126, the width of the busbar 1126 after the junction should approximately be the sum of to the width of the two branches 1106 and the busbar 1126 before the junction 1108. Alternatively, the busbar 1126 and optionally also the branches 1106 are not tapered, as described with reference to Fig. 12 and the increase in width of the busbar 1126 is confined to be at the junctions 1108 only. Hence, in this slightly simplified design the busbar 1126 has a uniform width in-between the junctions 1108.

One embodiment of the invention is schematically illustrated in Fig. 15, wherein the width of the busbar 1126 and/or optionally the branches 1106 are symmetrical with regards to a central line and the width of each half of the busbar 1126 and/or branches 1106 follows the expression: a+bx ^a (1) wherein x is the distance from the top end 1 l26b of the busbar 1126 or the farthest end from the junction (branch 1106), a is the smallest width of the branch or busbar and b is a constant. The exponent a is between 1.5 and 3. The above expression (equation 1) takes into account how the grid elements should be increased in width given a constant thickness in order to keep the charge density constant as the total amount of collected charges increases. The expression is an approximation of a relatively complex scenario and the skilled person would understand that other expressions giving approximately the same increase in width of the grid elements would also represent an optimized or near optimized top contact grid design. The effect is exaggerated in the figure.

Above embodiments can be combined in various ways. For example, the branches 1106 could be of uniform width, for example to simplify production, and only the busbar 1126 exhibit an increase in width towards its base 1 l26a, generally continuous as in Fig. 13 or stepwise as in Fig. 14 and 15.

The top contact grids have been illustrated as having grid elements that are straight and junctions that are perpendicular. This should be seen as an example, primarily for the ease of illustration and understanding. As appreciated by the skilled person a busbar and a branch may have a curvature and a junction may be non-perpendicular. Given the above detailed design alternatives the skilled person may provide a top contact grid design that is optimized for a particular solar cell and still, thanks to the novel lift-off process, is suitable for industrial scale production.

The method of producing the solar cell 1111 according to the invention will be described with reference to the flowchart of Fig. 16 and the above general description referring to Fig. 1-10. The method comprises the steps of: a) Providing a photovoltaic multilayered structure 1112 with a transparent front contact 1123 as the uppermost layer, b) applying at least one layer of photoresist on the surface of the transparent front contact 1123, c) baking the photoresist for removing excess solvent d) creating a pattern defining the top contact grid 1105 in said photoresist to expose at least one portion of the surface, said pattern comprising at least one edge between the photoresist and the exposed at least one portion of the surface, e) depositing a metal on said photoresist and said at least one portion of the surface to form a metal layer, said metal having a temperature at deposition that is higher than a temperature of the substrate, f) cooling the deposited metal through contact with the exposed surface and the photoresist layer, thereby causing a contraction of the deposited metal and generating at least one crack or gap through the metal layer at the edge, g) applying a solvent through the at least one crack or gap for removing the

photoresist and thereby removing the metal layer deposited thereon but leaving the metal deposited on the exposed surface, the deposited metal forming the top contact grid 1105.

The method for producing a solar cell 1111 may incorporate all the versions and alterations described with reference to the general improved lift-off process described with reference to Fig. 1-10.

According to one embodiment the temperature of the substrate 101 during the step of depositing the metal, step e), is less than l40°C. According to one embodiment the photoresist 104 is deposited in a layer having a thickness that is equal to or lower than a thickness of the metal layer 106 deposited in the step of depositing the metal, step e). According to one embodiment the materials are selected such that the thermal expansion coefficient of the deposited metal 106 is larger than a thermal expansion coefficient of the substrate 101 or the at least one deposited layer 102.

According to one embodiment the surface 103 of the substrate 101 has a uniform temperature during the baking of step c), said uniform temperature being such that a difference in temperature across the surface 103 is less than 7°C, preferably less than 4°C and more preferably less than 2°C.

According to one embodiment the substrate 101 is cooled after the baking of step c) and wherein the surface 103 of the substrate 101 has a uniform temperature during cooling, said uniform temperature being such that a difference in temperature across the surface 103 is less than 7°C, preferably less than 4°C and more preferably less than 2°C.

The invention will now be described with reference to a number of examples.

Example 1

A glass substrate with a thin film CIGS solar cell stack was used and coated with a layer of photoresist (AZ 1529) having a thickness of 2pm. The photoresist was baked at 1 l0°C during 1 minute and exposed using a 405 nm laser to create features, and subsequently developed using AZ400K:H20 at 1 :4 during 90 seconds.

Next, an aluminium layer was deposited by evaporation at around 200A/s, ending when the thickness of the aluminium was 3.5 pm. Substrate to evaporation distance was 30 cm.

Acetone was used for lift-off to remove the photoresist. Fig. 7a discloses Example 1 after deposition of aluminium, with cracks clearly appearing along each side of a center feature. The photoresist underneath the aluminium on the sides of the feature is stretched out due to the accumulated tensile stress in the metal. In Fig. 7b, the same example is shown after lift-off, with the string of aluminium extended where the feature was located and patterned on the surface beneath. In this example, additional aluminium was deposited after the formation of cracks which created residues in the shape of wings on the fractured surfaces that are visible in Fig. 7b along the upper side edges of the string of aluminium.

Example 2

A glass substrate with a thin film CIGS solar cell stack was used and coated with a layer of photoresist (TFR2900) having a thickness of 2-3 pm. The photoresist was baked during 1 minute at 1 lO°C and exposed with a 405 nm laser to create features. The photoresist was developed using AZ400K:H20, 1 :9 for 25 seconds.

Next, an aluminium layer was deposited by evaporation at ~200A/s, ending when the thickness of the aluminium was 3.5 pm. Substrate to evaporation distance was 30 cm. Acetone was used for lift-off to remove the photoresist.

Fig. 8a discloses Example 2 after deposition of aluminium, with well-defined cracks along each side of the feature. Thanks to the more controlled evaporation, ending when the cracks had appeared, the finished string of metal in Fig. 8b lacks the additional material disclosed in Fig. 7b of Example 1 and is instead has an essentially rectangular shape in a cross-sectional view.

Example 3

A glass substrate with a sputtered ZnO layer on its surface was used and was coated with a layer of photoresist (AZ1529) having a thickness of 5 pm. The photoresist was baked during 1 minute at 1 lO°C and exposed with a stationary Hg lamp to create features with a spacing of 10 pm. In this example, a plurality of features was created running both in parallel and crossing each other. The photoresist was developed using AZ400K:H20, 1 :4 for 90 seconds.

Next, an aluminium layer was deposited using evaporation to create a layer having a thickness of 3.5 pm. Again, acetone was used for the lift-off.

Fig. 9a discloses Example 3 after developing of the photoresist. The features created run in different directions but are still well-defined with walls slanted to make the diameter of each feature smaller at its bottom than at a top of the photoresist layer. In Fig. 9b, another section of the substrate is shown after the deposition of aluminium through evaporation. The figure discloses a portion having one feature running from top to bottom of the figure, but also another feature that intersects that feature, and it can clearly be seen that cracks have been generated along the sides of the features regardless of their orientation and the fact that they are connected to each other.

Example 4

A glass substrate with a thin film CIGS solar cell stack was used and coated with a layer of photoresist (TFR2900) having a thickness of 2-3 pm. The photoresist was baked during 1 minute at 1 lO°C and exposed with a 405 nm laser to create features. The photoresist was developed using AZ400K:H20, 1 :9 for 25 seconds. Next, a copper layer was deposited by evaporation at ~30A/s. The evaporation was blocked by a shutter when the thickness of the copper was ~3pm. Substrate to evaporation distance was 40 cm.

Example 5

A glass substrate with a thin film CIGS solar cell stack was used and coated with a layer of photoresist (TFR2900) having a thickness of 2-3 pm. The photoresist was baked during 1 minute at 1 l0°C and exposed with a 405 nm laser to create features. The photoresist was developed using AZ400K:H20, 1 :9 for 25 seconds. Next, a silver layer was deposited by evaporation at ~30A/s. The evaporation was blocked by a shutter when the thickness of the silver was ~3pm. Substrate to evaporation distance was 40 cm.

Discussion

The examples show that the process is stable and works with Cu, Al and Ag. Al is by far the cheapest electrically conductive metal there is and is therefore preferable. Cu is the second most interesting metal since it is not as expensive as Ag. Other metals with good electrical conductivity are at present not suitable for cost efficiency reasons, but would also be technically feasible within the scope of the present invention. Also, alloys of conductive metals would be suitable for use with the present invention. In some cases it could be beneficial to use Ni between the substrate and the metal (Cu, Al or Ag) or other means of promoting adhesion to the substrate as is known in the art. Several sub-layers of different metals can also be used as barriers in order to prevent diffusion.

Another beneficial property of the process is that the cross section of the metal feature is rectangular and not tapered as commonly obtained with undercut lift off. Rectangular cross sections will conduct electricity better than a tapered or bell shaped cross section at the same width and height which is desirable especially for solar cells to reduce shadowing. The examples also show that it is possible toreadily control where the cracks form and how to avoid feature widening due to defects as“wings”,“feet” or“fences”. Wings are created when evaporation is allowed in the already formed crack. Feet are created when the resist is overheated and then releases from the substrate as the Al is contracting, while the substrate is still in line of sight to the evaporation allowing metal to be deposited under the photoresist. Fences are created when the resist is too thick and the crack appears at the top edge of the resist edge. If the metal layer is very thick it is also possible that a crack is generated on one side of the feature only.

It is worth noting that Example 3 above discloses an experiment with a photoresist layer that is thicker than the subsequently deposited metal. The risk for fences being created is significantly higher with a thicker layer of photoresist, but in some applications it would still be possible to use the proportions given in the current example.

The shape of the top resist edge affects when the cracks in the aluminium appear during the evaporation. A sharp edge allows for the aluminium to crack early during the evaporation since a sharp edge creates a high stress in the aluminium in a similar way as a sharp knife when cutting. A rounded/dull resist edge requires a higher stress in the aluminium to crack, just as a dull knife requires more force when cutting. The thicker aluminium layer that is required in order to obtain enough stress in this case, causes the aluminium to crack later during evaporation. A rounded resist edge can be preferred since it decreases the risk of evaporating metal into an already formed crack.

When using the method according to the present invention, the photoresist can be as thin as 0.8 pm when the substrate is clean and particularly good results in controlling the generation of cracks have been achieved with a photoresist layer of 1 pm. If the process does not take place in an environment where the substrate can be kept very clean, the photoresist layer would generally have to be thicker but very good results have been achieved at such conditions with a photoresist layer having a thickness of 2.8 pm or even lower. This can be compared with prior art methods where a photoresist layer of about 12 pm is considered normal.

It is worth noting that Example 3 above discloses an experiment with a photoresist layer that is thicker than the subsequently deposited metal. The risk for fences being created is significantly higher with a thicker layer of photoresist, but in some applications it would still be possible to use the proportions given in the current example. The shape of the top resist edge affects when the cracks in the aluminium appear during the evaporation. A sharp edge allows for the aluminium to crack early during the evaporation since a sharp edge creates a high stress in the aluminium in a similar way as a sharp knife when cutting. A rounded/dull resist edge requires a higher stress in the aluminium to crack, just as a dull knife requires more force when cutting. The thicker aluminium layer that is required in order to obtain enough stress in this case, causes the aluminium to crack later during evaporation. A rounded resist edge can be preferred since it decreases the risk of evaporating metal into an already formed crack.

It is to be noted that elements of different embodiments described herein may freely be combined with each other unless such a combination is expressly stated as unsuitable, as will be readily understood by the person skilled in the art.