WO2011097745A1

WO2011097745A1 - Method for manufacturing a solar panel

Info

Publication number: WO2011097745A1
Application number: PCT/CH2011/000021
Authority: WO
Inventors: Oliver Kluth; Hanno Goldbach
Original assignee: Oerlikon Solar Ag, Trübbach
Priority date: 2010-02-10
Filing date: 2011-02-08
Publication date: 2011-08-18
Also published as: CN102741451A; TW201201396A

Abstract

A method for manufacturing a solar panel relies on a sequence of steps for manufacturing a silicon layer in a vacuum process chamber, said steps comprising: a) Providing a substrate covered at least partially with an electrode material such as ZnO, b) Introducing said substrate into a vacuum process chamber capable of generating a plasma therein, c) Depositing a first layer of silicon furnished with a first doping agent, d) Removing said substrate from said process chamber, e) Applying an oxidizing plasma to the plasma chamber, g) Reintroducing said substrate into said process chamber and Depositing further layers of silicon. Said first layer of silicon comprises preferably at least microcrystalline silicon.

Description

METHOD FOR MANUFACTURING A SOLAR PANEL

The present invention refers to an improved thin film silicon solar cell and a respective method for manufacturing it by reducing the influence of process chamber contamination when using Zinc oxide (ZnO) coated glass as a base substrate for the solar panel.

FIELD OF THE INVENTION

Thin film silicon solar cells or modules are photovoltaic converter devices for converting light, e. g. sunlight, into electrical energy. Thin film silicon solar cells are hereby understood as cells where at least the absorber layer is being deposited by a physical or chemical vapour deposition technology (PVD, CVD, PECVD, APCVD) . BACKGROUND OF THE INVENTION

Prior Art Fig. 5 shows a basic, simple photovoltaic cell 40 comprising a transparent substrate 41, e. g. glass with a layer of a transparent conductive oxide (TCO, e. g. ZnO, Sn0₂) 42 deposited thereon. This layer is also called front contact and acts as first electrode for the photovoltaic element. The combination of substrate 41 and front contact 42 is also known as superstrate. The next layer 43 acts as the active photovoltaic layer and exhibits three "sublayers" forming a p-i-n junction. Said layer 43 comprises hydrogen- ated microcrystalline , nanocrystalline or amorphous silicon or a combination thereof. Sub-layer 44 (adjacent to TCO front contact 42) is positively doped, the adjacent sub- layer 45 is intrinsic, and the final sub-layer 46 is negatively doped. Finally, the cell includes a rear contact layer 47 (also called back contact) which may be made of zinc oxide, tin oxide or ITO and a reflective layer 48. Alterna- tively a metallic back contact may be realized, which can combine the physical properties of back reflector 48 and back contact 47.

For illustrative purposes, arrows indicate impinging light.

Thus, the basic concept of a thin film silicon solar cell comprises at least a truly intrinsic (dopant-free) or essentially intrinsic silicon absorber layer 45, sandwiched between a p-doped 44 and an n- doped 46 silicon layer, thus forming a p-i-n junction. Depending on the crystallinity of the absorber layer amorphous (a-Si) , micro- (μα-Si) or nanocrystalline (nc-Si) or (fully) crystalline (c-Si) solar cells are being distinguished. In order to discharge the electrical current generated during operation each of the p- and n- layers are in electrical contact with an electrode. In case a dual or triple junction solar cell is designed, two or three p-i-n junctions are stacked electrically in series and the respective outermost p- and n- layer forms the contact with the electrode.

In a thin film solar module in superstrate configuration the solar cell layers are commonly deposited on a glass substrate coated with a transparent conductive metal oxide layer (TCO) . In case of sili- con thin film solar cells it is well known that these TCO layers can be strongly reduced by the PECVD plasma process used for the silicon deposition. Especially the strong hydrogen containing plasmas used for the deposition of microcrystalline silicon (μο-Ξί) leads to strong reduction of the TCO material. In case of Sn0₂-coated glass this effect leaves an opaque metallic Sn film on the TCO surface which strongly reduces the photoelectric current during operation and hence the efficiency of the cell. Due to this reason c-Si p contact layers are problematic for cells on Sn0₂ substrates.

In case of ZnO coated substrates the mechanism is comparable, but usually there is no opaque metallic Zn layer found on the TCO surface even after applying hydrogen containing PECVD plasmas as typically used for deposition of μσ-βί.

The following invention addresses the problem of the metallic Zn, which is created during the plasma exposure of a ZnO electrode layer deposited on a base substrate and what problems can occur due to Zn contamination of the process chamber and how these problems can be overcome . Zinc oxide ZnO has been found to be a very suitable TCO material for the application in thin film solar cells. It is more transparent and conductive compared to Sn0₂ covered substrates and allows lower mate- rial costs than Sn0₂ or ITO. Moreover, it is reported to be more resistant against strong hydrogen containing plasmas as for example used for the deposition of uc-Si. In contrast to Sn0₂ where a good TCO p-contact can easily be achieved with an amorphous p-layer 44, in case of ZnO it is widely reported that a p-doped μα-Βϊ -Ξϊ double contact layer is mandatory in order to achieve low series resistance, high FF (Fill Factor) and V_oc (voltage open circuit) . In other words, p-layer 44 is composed of a p-doped μο-Si layer adjacent to the TCO front contact 42 and a subsequent a-Si layer.

Good values for series resistance, Fill Factor and V_oc are indicators for a good TCO/p contact behaviour. Only in very view examples good cell performance could be achieved on laboratory scale using simple amorphous contact layers on ZnO. In these cases special treatments of the TCO or TCO/p interface were applied in order to achieve high Voc and FF. Alternatively good cell results have been shown only on very small cell areas of 1cm² or less.

PROBLEM TO BE SOLVED

Fig. 1 shows two experiments to demonstrate the problem. On the left_. side two 50x50mm² samples of ZnO coated glass are placed on a 1.4m² (uncoated) carrier glass for cell deposition in a PECVD reactor designed to deposit silicon layers (p-i-n structure) on 1.4m² substrates. For comparison in a second run shown on the right side the same cell structure is deposited on a 1.4m² ZnO coated glass. After all deposition steps two 50x50 mm² pieces are taken from the same positions compared to the positions of the ZnO samples from the first experiment. On all of these four pieces eight 1cm² cells are prepared by means of a laser (laser scribed) . The cell design comprises a uc-Si/a-Si double p-contact layer and all Si layers are de- posited in one sequence in the same PECVD deposition chamber. In short words, the main difference between both experiments lies in the effective area of ZnO layer exposed to the silicon deposition plasma. Fig. 2 shows representative IV curves of cells obtained from experiment one and two. It can clearly be seen that the cell of experiment 2 cut from the 1.4m² ZnO substrate has a severe problem compared to the cell deposited with the same cell recipe on two small ZnO covered substrate samples .

Table 1 above gives an overview over the cell IV parameters of the cells shown in Fig. 1 plus, for comparison, a cell containing only a simple a-Si p-layer.

It can be derived that cells with pc-Si/a-Si double p-layer 44 only show superior performance compared to a simple amorphous p-layer 44 configuration if the cell deposition is accomplished on small ZnO covered substrates. If the deposition of the pc-Si p layer is done on a large area substrate, the FF and V_oc are substantially lower and even inferior compared to the cell with a simple a-Si p-layer. It is thus clear that applying a c-Si p-layer on large area ZnO covered substrates causes severe problems that have to be overcome in order to use this single chamber process in large area solar module production. It is however highly desirable to benefit from the potential of the pc-Si p-layer for further increasing the FF and V_oc and hence the cell efficiency. This is underlined by the comparison of the QE curves (quantum efficiency) of the cells from both experiments described above as shown in Fig. 3a and b.

The cell with pc-Si/a-Si double p-layer on the large area substrate exhibits a bulk i- layer problem which can be derived from the strong drop of the QE under different forward bias (as indicated as parameters on the right) observed over the whole visible wavelength range. It has been found that for a-Si cell deposition on ZnO some special effect occur which cause a drop of the maximum power point of the IV curve and a decrease of the overall QE under forward bias leading to electrically poor performing cells. For the commonly used pc-Si/a-Si double p- layer this effect makes it impossible to get solar module with satisfactory performance on large area LPCVD-deposited ZnO covered substrates. Based on the knowledge that ZnO is strongly reduced in strong hydrogen containing plasmas as used for the μο-Ξϊ silicon deposition we attribute the above described effect to a Zn contami- nation of the PECVD deposition chamber during the plasma deposition process of the c-Si p-layer followed by a subsequent cross- contamination of the silicon i-layer which is accomplished in the same PECVD deposition chamber. This contamination results in a severe increase of defects in the silicon i-layer and hence promotes the recombination of photo-generated charge carriers during operation of a respective solar panel. In case the first doped Si layer and the intrinsic Si layer are done in separate PECVD chambers the cross-contamination will not occur. In a single chamber approach however, which is desirable for shortening process times, the ef- fectt will take place.

This concept was directly proven by the experiment shown in Fig. 4a and b. A bare silicon wafer was placed on a 1.4m² LPCVD ZnO covered substrate. After jointly depositing an a-Si layer on both substrates the Zn concentration in the a-Si layer/Si wafer stack was measured with SIMS. The Zn signal is shown in the right graph (Fig. 4b) . It is clear that there exists an at least two orders of magnitude high- - er Zn concentration in the a-Si layer compared to the silicon wafer. The most probable explanation is that the a-Si layer has been con- taminated during deposition with Zn coming from the surrounding ZnO coated glass substrate. The homogenous incorporation of Zn into the i-layer nicely explains the overall drop of the QE and also the observed drop of the maximum power point of the IV curve. The contamination effect is stronger in case of the cell with μσ-Ξϊ p-layer be- cause the hydrogen rich plasma used for μσ-Ξϊ deposition more strongly reduces the ZnO to Zn and oxygen and creates thereby a much higher amount of metallic Zinc. RELATED ART

US 4,873,118 describes an improvement to a manufacturing process for solar cells with one ore more hydrogenated thin film silicon layers upon a zinc oxide film by applying a glow discharge containing oxygen prior to deposition of the first thin film silicon hydrogen alloy layer onto the zinc oxide film to improve the ZnO/p contact.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 shows an experimental setup for demonstration of the underlying problem

Fig. 2 shows the results of the setup according to Fig. 1.

Fig. 3a and b show the effect on Quantum efficiency for experiments according to Fig. 1 and 2.

Fig. 4 shows the Zn contamination in a silicon layer.

'Fig. 5 shows a basic thin film photovoltaic cell.

Fig.6 shows the influence of an oxygen plasma treatment on the IV curve of an a-Si cell with a-Si p layer.

Fig.7 Influence of an oxygen plasma treatment on the IV curve of an a-Si cell with uc-Si/a-Si p layer.

SUMMARY OF THE INVENTION

A method for manufacturing a photovoltaic converter stack comprises the steps of providing a substrate covered at least partially with an electrode material such as ZnO; introducing said substrate into a process chamber capable of generating plasma therein; depositing a first layer of silicon furnished with a first doping agent and applying oxidizing plasma to the plasma chamber.

In an alternative embodiment said first layer of silicon furnished with a first doping agent is being deposited applying softer plasma conditions thus minimizing the reduction of ZnO to metallic Zn and avoid the Zn cross contamination of the subsequently deposited i- layer . DETAILED DESCRIPTION OF THE INVENTION

Different solutions were tested to reduce or avoid the cross contamination of the i- layer with Zn: 1. Application of an oxygen plasma on the ZnO substrate prior to the cell deposition to oxidize excess Zinc on the ZnO surface and/or make the ZnO more resistive against reduction in the following hydrogen plasma.

2. Application of oxygen plasma to the empty PECVD chamber after deposition of a p- layer on ZnO in this chamber. The oxygen plasma can be applied after a single a-Si p-layer, after the μα-Ξί/Ά-Ξϊ double p layer and/or after the μο-Ξχ part of the double p-layer.

3. Use softer plasma conditions with lower power and hydrogen content during the deposition of the first Si layer onto the ZnO in order to minimize the reduction of ZnO to Zn and subsequent Zn contamination of the deposition chamber. This strategy can be applied either to ο-εϊ or a-Si p-layer. Since the plasma conditions used for a-Si are in general much softer compared to the one required for μο-Si, another solution can be to use a single a-Si p-layer as first Si -layer on ZnO instead of the commonly used μα-Si/a-S double p-layer.

Experiments have been made to test the above described methods according to the invention. Cells with either a simple amorphous p- layer or a μο-Ξϊ / a-Si double p-layer deposited on large area ZnO covered substrates without and with applying different plasma treatment steps .

Figure 6 and 7 show the effect of an 0₂ plasma treatment of the LPCVD ZnO substrate in the PECVD system prior to the a-Si cell deposition for a cell with only a-Si p layer and a cell with the commonly on ZnO used μα-Ξί/Ά-Ξχ double p layer. Table 2 gives an overview over the IV parameter of the corresponding cells.

Table 2: IV parameter of a-Si p-i-n cells with a-Si and μα-Si/a-S p-layer and different 0₂ plasma treatments shown in Fig 6 and Fig

Findings for solution 1:

According to the inventors' findings, 0₂ plasma treatment of the ZnO substrate prior to p-layer deposition alone does not result in a sufficient improvement independent of the type bf p-layer.

Figure 6 demonstrates that the 0₂ plasma treatment of the ZnO leads to an inferior IV characteristic compared to the standard cell with- out any 0₂ treatment. We conclude that exposure of the ZnO substrate to 0₂ plasma as described in US 4,873,118 does not provide a solution for a low FF, Voc and high series resistance induced by Zn contamination. This is confirmed also by Figure 7 for the case of a-Si cells providing a μα-β /^'a-Si double p-layer which leads to an even stronger Zn contamination effect.

Findings for solution 2 :

μο-Ξϊ/ -Ξ double p-layer on ZnO results in improved FF and VOC only, if a 0₂ plasma treatment of the process chamber is applied after the ο-Ξί/^-εί p-layer deposition. Figure 6 shows that the 0₂ plasma treatment of the empty PECVD chamber considerably improves the IV characteristics, particular the FF and V_oc over the one observed for the cell without treatment as well as over the one with 0₂ plasma treatment of the ZnO substrate .

Findings for solution 3 :

From comparison of Fig. 6 and 7 it is obvious that the Zn contamination problem is much more severe in case of the cell comprising a μο-Si p-layer as the first layer deposited on ZnO. The case of the solar cell with a-Si p layer can be considered as an example for using softer PECVD plasma conditions to avoid the reduction of ZnO to metallic Zn and hence minimizing the Zn cross contamination effect. The same strategy can also be applied to the deposition of the μο-Ξΐ p-layer using a combination of lower power, higher pressure, lower hydrogen content in the gas mixture and higher excitation frequencies at least in first growth phase until the ZnO surface is completely covered. In conclusion: ZnO is reduced to Zn and O in the strong hydrogen containing plasma of the μο-Si player deposition. The Zn is dispersed in the process chamber and later contaminates the i- layer in the subsequent deposition, resulting in defect formation in the bulk of the i- layer. By applying oxidizing plasma to the process chamber after the p- layer deposition this contamination effect can be avoided leading to superior FF and Voc compared to the cell with simple a-Si p-layer.

The oxidizing plasma as described herein can be realized in a parallel-plate PECVD plasma reactor (as e. g. commercially available as Oerlikon Solar KAI) by applying a plasma power of 300 to an electrode system capable of handling 1.4 m² substrates, which equals about 21mW/cm². Further, a flow of 100 seem oxygen and a plasma duration of at least 60-120s is suggested. These values can be varied to adopt the inventive principle to other systems and the amount of Zn contamination. It is important that the amount of Energy / cathode area is being kept essentially the same as well as the oxygen flow relative to the cathode area or process chamber volume.

The oxidizing plasma has the main purpose to re-oxidize Zn remnants which had been released and/or reduced by the hydrogen containing plasma used in earlier Si-deposition step(s). It is thus partially immobilized and/or converted such it can be pumped from the process volume and thus removed. The oxygen plasma can be applied after a single a-Si p-layer, after the /.c-Si/a-Si double p layer and/or after the μο-Si part of the double p-layer. Based on the detailed understanding of the mechanism of the Zn contamination effect we want to address different solutions to the problem described above.

A method for manufacturing a silicon layer in a vacuum process cham- ber will therefore comprise:

1. Providing a substrate covered at least partially with an electrode material such as ZnO, 2. Introducing said substrate into a process chamber capable of generating a plasma therein

3. Depositing a first layer of silicon furnished with a first doping agent

4. Removing said substrate from said process chamber

5. Applying an oxidizing plasma to the plasma chamber

6. Reintroducing said substrate

7. Depositing further layers of silicon

In a first embodiment, said first layer of silicon comprises a stack of μο-εί and a-Si.

In a second embodiment, said first layer of silicon comprises a-Si only. Amorphous silicon is conventionally deposited with less hydrogen and therefore less aggressive hydrogen plasma is effected.

In a third embodiment said first layer of silicon comprises σ-Ξϊ and the layer deposited in step 7 is initially a a-Si p- layer.

A method for manufacturing a silicon layer in a vacuum process chamber will comprise:

• Providing a substrate covered at least partially with an electrode material such as ZnO,

• Introducing said substrate into a process chamber capable of generating a plasma therein

• Depositing a first layer of silicon furnished with a first doping agent, said first layer comprising c-Si while

• Controlling at least one plasma parameter to start with at a value lower than the one for highest deposition rate foreseen, said value being one or more of: plasma power, hydrogen content or flow, pressure; and

• Changing said value stepwise or continuously to create a graded or gradient profile of said value in the resulting layer

• Optionally: Removing said substrate from said process chamber and subsequently applying an oxidizing plasma to the plasma chamber

In an embodiment this could be realized as: Starting with a low plasma power value and increasing it, starting with a low hydrogen flow and increasing it, starting with a high chamber pressure and reducing it.

A further method for manufacturing a silicon layer in a vacuum process chamber will comprise:

• Depositing a first layer of silicon furnished with a first doping agent, said first layer comprising μο-3ί while

• Choosing and applying first process settings which allow a high rate layer deposition for a first period of time;

• Choosing optimum process settings to result in the best balance of deposition time and layer quality

• Reducing said deposition rate stepwise or continuously from said first settings to said optimum settings

Any of the methods mentioned above may be advantageously combined with an increased pumping effort during p- layer deposition.

Of course different combinations of the above solutions can also be applied.

All the described solutions are capable of increasing the long-term stability of a module production process on large area ZnO substrates and can help to further improve efficiency.

The invention is applicable also for other types of solar cells and modules .

Claims

CLAIMS :

1. A method for manufacturing a silicon layer in a vacuum process chamber comprising:

a) Providing a substrate covered at least partially with an electrode material such as ZnO,

b) Introducing said substrate into a vacuum process chamber capable of generating a plasma therein

c) Depositing a first layer of silicon furnished with a first dop- ing agent

d) Removing said substrate from said process chamber

e) Applying an oxidizing plasma to the plasma chamber

f) Reintroducing said substrate into said process chamber

g) Depositing further layers of silicon

2. A method according to claim 1, wherein said first layer of silicon comprises microcrystalline silicon.

3. A method according to claim 1, wherein said first layer of sili- con comprises a stack of microcrystalline (μο-) silicon and amorphous (a-) silicon.

4. A method according to claims 1-3, wherein said first doping agent is a p-dopant such as boron.

5. A method according to claim 1-4, wherein the deposition step c) comprises

• Controlling at least one plasma parameter to start with at a value lower than the one for highest deposition rate fore- seen, said value being one or more of: plasma power, hydrogen content or flow, pressure; and

• Changing said value stepwise or continuously to create a graded or gradient profile of said value in the resulting layer A method according to claim 1-4, wherein the deposition step c) comprises

• Controlling at least one plasma parameter to start with at value lower than the one for highest deposition rate foreseen, said value being one or more of: plasma power, hydrogen content or flow, pressure; and

A method according to claims 1-6, wherein in step e) a plasma power of 300 W and an oxygen flow of lOOsccm during 60-120s are being established, in relation to a 1.4m² substrate.