Photovoltaic (PV) modules are a general product of the PV industry. PV modules need to be used outdoors. The outdoor reliability of these products is an important quality indicator and light induced degradation (LID) is one of the most intensively discussed reliability indicators.

Mathematically, the relative LID in a crystalline silicon PV product is defined as:

It is also necessary to measure the I-V curves of the module both before and after LID. The accuracy of I-V measurements are affected by the environment temperature, solar simulator spectral correction, etc. Usually, a reference sample calibrated by a third-party certification authority is often applied to generate secondary reference samples. For example, for an I-V calibration by a class AAA solar simulator in TUV Rheinland, the repeatability is ±0.1% and reproducibility is ±0.5% [1]. For module manufactures, if the solar simulator is not calibrated, the measurement uncertainty will be increased to ±2.0% because of spectral mismatch.

Accurate I-V measurements of PV products are a prerequisite for comparing LID.

The detailed analysis for LID data is quite complex.

LID in a narrow sense usually refers to boron-oxygen complex induced LID (BO-LID). BO-LID is often regarded as the main factor of initial LID in crystalline Si solar cells and is often detectable in Si cells made from p-type boron doped Cz silicon. According to IEC 61215, pre-LID is calculated using I-V measurements taken before and after a 5 kWh exposure.

There are three primary categories of LID: BO-LID originated from the boron-oxygen complex, light- and elevated temperature-induced degradation (LeTID), and UVID after UV exposure. This paper will focus on the first two categories.

In 2015, Friederike Kersten et al. published in Solar Energy Material & Solar Cells highlighting a new LID phenomenon discovered on multi PERC modules in the field [2]. As the onset of degradation was found to be more prominent at high temperatures, the degradation was coined light- and elevated temperature-induced degradation, or LeTID. The module back sheet was covered using styrofoam to maintain a high operation temperature. After five weeks (Fig1 CW30), the multi-PERC module sample reached its maximum LeTID extent. LeTID was only detectable if the module temperature was relatively high. Some multi-PERC modules under specific LeTID test conditions reached a maximum LeTID extent beyond 10% relative. With the help of thermally insulating material, the degradation and regeneration process was detectable in the field. The cycle time scale for degradation and regeneration was approximately four months, as shown in Fig 1.

Multi-PERC modules operated in open-circuit condition in the field. The module back sheet was covered using styrofoam to increase the module operation temperature. The EL images and relative efficiency changes were measured over numerous weeks. [2]

Multi-PERC modules operated in open-circuit condition in the field. The module back sheet was covered using styrofoam to increase the module operation temperature. The EL images and relative efficiency changes were measured over numerous weeks. [2]

According to the information above, relative LID data analysis should be made carefully, including:

  • Comparison of LID data is based on measurements with good accuracy and repeatability
  • When the cell temperature is low, BO-LID is the primary initial LID mechanism
  • When the cell temperature is relatively high, for example at 75℃, LeTID begins to occur. LeTID takes a longer time scale to manifest. Its physical mechanism is not yet clear
  • How can we measure LID? What are the proper measurement methods for identifying LID on PERC products in the field?
  • Can LID and LeTID be differentiated?

Moreover, the silicon wafer quality, cell structure, module structure, encapsulating materials and field operation conditions all affect LID results, leading to an increased complexity in understanding defect behaviors.

BO-LID phenomena

The mechanisms behind BO-LID and solution to combat the problem have been studied prior to 2000. In as early as 1997, Professor Jan Schmidt from ISFH discovered that using Ga as an alternative dopant source can solve BO-LID [3]. In 1999, Professor T. Saitoh from Tokyo University of Agriculture and Technology studied light induced lifetime degradation on Ga doped Cz, boron doped Cz,magnetic CZ and Fz [4], as show in Fig 2.

Figure 2 (Above) The influence of Ga doping, boron and oxygen content on BO-LID [3](middle) The influence of low interstitial oxygen content MCZ and different oxygen content on BO-LID [4]

Figure 2 (Above) The influence of Ga doping, boron and oxygen content on BO-LID [3](middle) The influence of low interstitial oxygen content MCZ and different oxygen content on BO-LID [4]

Figure 2 The influence of Ga doping, boron and oxygen content on BO-LID [3](middle) The influence of low interstitial oxygen content MCZ and different oxygen content on BO-LID [4]

Figure 2 The influence of Ga doping, boron and oxygen content on BO-LID [3](middle) The influence of low interstitial oxygen content MCZ and different oxygen content on BO-LID [4]

Figure 2 ( The influence of Ga doping, boron and oxygen content on BO-LID [3](middle) The influence of low interstitial oxygen content MCZ and different oxygen content on BO-LID [4]

Figure 2 ( The influence of Ga doping, boron and oxygen content on BO-LID [3](middle) The influence of low interstitial oxygen content MCZ and different oxygen content on BO-LID [4]

Differences of BO-LID between mono and multi-Si

In cells made from p-type boron doped wafers, BO-LID is regarded as the main reason for initial LID.

Mono-Si ingots are grown in a silica crucible. The ingot pulling process requires melting of poly-Si feedstock. The coating of crucible cannot completely block oxygen out-diffusion from the crucible walls. The oxygen dissolved in the molten silicon segregates into the ingot during cooling, which can lead to an oxygen concentration in the crystallized ingot above 10 ppma.

Compared to mono-Si ingot pulling, there are several advantages in casting multi-ingots. These include a partial melting of the poly Si feedstock, an effective SiNx coating of the crucible, a larger evaporation area and smaller contact area to the crucible. 

Therefore, in a multi-Si ingot,only edge regions in close proximity to the crucible walls have high concentrations of oxygen (approximately 10 ppma) whereas the concentration lower to approximately 1 ppma in the center. Since the multi-ingot is quite large, the average oxygen content is about one-tenth of mono-Si.

The resistivities of p-type mono and multi are close to each other, that is, the concentration of boron dopants can be quite similar. It is well known that mono-cell suffers from BO-LID to a greater extent than multi-cells. Prior to PERC mass production, without BO-LID regeneration process, it is common sense that mono modules have higher LID levels than multi for the first year operation in the field. Usually, the first year LID guarantee of a PV module from mono manufactures is 3% relative, whereas for multi, it is 2.5%. The difference is originated from the BO related initial degradation.

BO-LID solution

PERC structures are more sensitive to BO-related LID. PERC has a more effective near-infrared light absorption integrating rear passivation technology. Compared to traditional aluminum BSF technologies, the efficiency gain from PERC is about 1% absolute. 

However, minority carriers (electrons) generated from the rear-side need to travel across the Si bulk before they can be separated by the front p-n junction effectively. Although the rear passivation greatly improves the PERC efficiency, the initial BO-LID increases to more than 5% relative.

Multiple approaches have been applied to reduce the effects of BO-LID. Reducing boron concentrations is an obvious solution, however, such action will significantly affect final cell performances. Reducing ingot oxygen concentration can also be effective, however, requires expensive alterations to ingot growth. Ga doping is one reasonable solution with a minor cost increase. However, the segregation coefficient of Ga in silicon is five times larger than that of boron, thus the Ga-doped ingot length with an appropriate resistivity range is much shorter.
Fortunately, industrial BO-LID solution has been realized in academic research.

In 2006, Alex Herguth from Konstanz University discovered a BO-LID degradation and regeneration process when sample are exposed to light at elevated temperatures or put under a forward biased (Fig. 3). This was the first publication on the light induced regeneration (LIR) phenomena [5]. Alex Herguth was honored the Junior Einstein Award by Solar World in 2006 for this discovery.

The decay-regeneration phenomena of BO-LID in the case of light and forward bias [5]

The decay-regeneration phenomena of BO-LID in the case of light and forward bias [5]

The decay-regeneration phenomena of BO-LID in the case of light and forward bias [5]

The decay-regeneration phenomena of BO-LID in the case of light and forward bias [5]

In recent years, industrial tools for PERC have been developing rapidly. BO-LID solutions originating from academic publications have also been realized. The LIR process time has been significantly accelerated under intense illumination and elevated temperature.

BO-LID industrial solutions include both light or current injection. Light injection can be supplied by halogen lamp, LED, laser or other light sources. In 2017, Longi, in cooperation with UNSW and DR Laser, released the LIR technology by laser to solve the initial LID of mono-PERC modules [6]. 

Different approaches have different injection levels. Laser have the strongest intensity. The LIR process time reduced to less than 5 seconds. The light intensities of LED, halogen lamp, etc. are relatively weak. The LIR process made from these sources will take 20-60 seconds. Current injection is much cheaper but requires at least ten minutes due to its lower injection levels and inhomogeneous temperatures. Nonetheless, due to short process time and low cost, LIR processes have been widely employed within the industry to mitigate BO-LID. The descending price of mono crystalline silicon wafers and the localization of manufacturing tools and the application of LIR process have brought PV products into the PERC era.

A detailed review on the mechanisms behind the BO-LID phenomena and solutions were published by Herguth Konstanz and Brett UNSW [7].

LeTID phenomenon

LeTID is quite common and not only present in multicrystalline silicon. In the 2017 Silicon PV proceedings, Fabian Fertig et al. published on the LeTID phenomena in Cz [8]. As shown in Fig. 4, the LID extent of cells made on mono Cz-Si is approximately 1-2% when degraded through current injection at 25 °C. “Medium LeTID” was observed in mono and multi using the same process at 75 ℃. The same phenomenon was later thought to also be observed in FZ [9], cast mono-Si [10], and even n-type [11].

Figure. 4 LeTID in mono-SI and multi-Si PERC cell [8]

Figure. 4 LeTID in mono-SI and multi-Si PERC cell [8]

Figure. 5 LeTID in FZ-Si [9]; PL diagrams of boron-doped FZ-Si treated as 900° (a) Before LeTID (b) After LeTID (c) Regeneration from LeTID

Figure. 5 LeTID in FZ-Si [9]; PL diagrams of boron-doped FZ-Si treated as 900° (a) Before LeTID (b) After LeTID (c) Regeneration from LeTID

Figure. 5 LeTID in FZ-Si [9]; PL diagrams of boron-doped FZ-Si treated as 900° (a) Before LeTID (b) After LeTID (c) Regeneration from LeTID

Figure. 5 LeTID in FZ-Si [9]; PL diagrams of boron-doped FZ-Si treated as 900° (a) Before LeTID (b) After LeTID (c) Regeneration from LeTID

LeTID was unexpectedly discovered in FZ silicon

LeTID was unexpectedly discovered in FZ silicon [9]. In Fig. 5, low oxygen content p-type 1 Ωcm FZ wafers were passivated with an Al2O3/SiNx stack and fired at high temperature (900 °C). LeTID was detected after illumination at 75℃. As the firing temperature increases, the maximum LeTID extent increases. Although there are few defects and impurities in FZ, LeTID in FZ is a strong piece of evidence pointing to its universality. The LeTID formation and regeneration cycle was observed to be relatively short in FZ, within several hours. 

It is interesting that there was no obvious LeTID at the edge of FZ sample, as the Al2O3/SiNx stack did not cover the whole wafer. Although the author did not explain the edge abnormalities in detail, it can be speculated that LeTID may be related to the passivation layer. One possible explanation is that after higher temperature firing, more hydrogen may have been released to regions directly beneath the passivating layers, whereas the edge was kept relatively free of hydrogen. It may also appear that metallic impurities and oxygen impurities may be ruled out as a primary contributor to LeTID due to the lower impurity concentrations found in FZ.

Figure 6 SiNx passivated FZ, CZ, mc-Si and cast mono LeTID measure at 2 sun 80℃(a) and 145℃ (b)

Figure 6 SiNx passivated FZ, CZ, mc-Si and cast mono LeTID measure at 2 sun 80℃(a) and 145℃ (b)

LeTID was also studied extensively on cast-mono material [10]. In Fig 6, diverse types of Si wafers were passivated using SiNx and fired. After exposure to LeTID testing conditions, the defect was found in all samples, even in N-type. When comparing the LeTID evolution curves, p-type mono-CZ had the shortest time scales. The defect generation-regeneration cycle time was approximately 100 min. The same cycle was approximately 500 min in cast-mono and over 3000 min in multi-Si. Lifetime degradation was detected in n-type CZ, n-type HP mc-Si and p-type FZ too. If testing is conducted at 145 ℃,the LeTID defect formation-regeneration cycle was accelerated. LeTID degradation trends were related to the Si material, illumination intensity and temperature.

Typical LeTID experimental studies

Multiple experiment approaches were applied to studying the root cause of LeTID. In Google Scholar, searching “LeTID in crystalline silicon solar cells” acquired 817 items since 2015. The main findings are presented in this paper.

Temperature, Si material and cell structure

A new form of LID (prior to being coined “LeTID”) was first discovered by Klaus Ramspeck in 2012 [13]. Fig. 7(1) shows LID measured at 75 ℃ 400 W/m2. These cells were Al-BSF or PERC structures made from different p-type wafers. After 20 hours of illumination, mono-PERC LID reached 3% primarily due to BO-LID. After 300 hours,the degree of LID in multi-PERC was higher than 5%, 3% in Multi-BSF and 2% in cast-mono and 1.5% in mono-PERC. 

Looking at the degradation profile, LID gradually increased in multi-BSF and multi PERC. In mono-BSF and mono-PERC, BO-LID partially regenerated over time. With modern LIR processes, BO-LID in mono can be completely solved. LeTID is more sensitive in cells made from multi wafers and new approaches should be developed to solve LeTID in multi.

The LID trends for multi-PERC measured at different temperatures is shown in Fig 7(2). As the temperature increases, the defect generation rate increases. Severe LID was observed after a short illumination time period at elevated temperature. The maximum LeTID ratio, however, cannot be easily derived by the LID curve measured at different temperatures. Furthermore, it is also not simple to isolate BO-LID and LeTID because both defect generation and regeneration occur simultaneously.

Figure 7 (1)LID evolution of cells made from Cz-BSF,Cz-PERC,mc-BSF, mc-PERC and cast-mono measured at 75℃,400W/m2(2)Multi PERC LID curve measured at temperature 25℃, 50℃, 75℃ and 95℃(All the cells was made without LIR process)[13]

Figure 7 (1)LID evolution of cells made from Cz-BSF,Cz-PERC,mc-BSF, mc-PERC and cast-mono measured at 75℃,400W/m2(2)Multi PERC LID curve measured at temperature 25℃, 50℃, 75℃ and 95℃(All the cells was made without LIR process)[13]

Figure 7 (1)LID evolution of cells made from Cz-BSF,Cz-PERC,mc-BSF, mc-PERC and cast-mono measured at 75℃,400W/m2(2)Multi PERC LID curve measured at temperature 25℃, 50℃, 75℃ and 95℃(All the cells was made without LIR process)[13]

Figure 7 (1)LID evolution of cells made from Cz-BSF,Cz-PERC,mc-BSF, mc-PERC and cast-mono measured at 75℃,400W/m2(2)Multi PERC LID curve measured at temperature 25℃, 50℃, 75℃ and 95℃(All the cells was made without LIR process)[13]

Gettering

Phosphorus diffusion is widely used in cell fabrication for emitter formation. During this process, metallic impurities in the silicon bulk can be segregated into the phosphorus diffusion layer in an effect known as gettering. The concentration of fast-diffusion metal impurities is thus reduced after the diffusion process. Annika Zuschlag et al.  from Konstanz studied the influence of phosphorus diffusion gettering on LeTID [14]. 

The lifetimes of multi-Si wafers were monitored by self-calibrated time resolved photoluminescence imaging. Without phosphorus gettering,the lifetime of specimens degraded significantly under illumination at 75 ℃. A regeneration process then began after 300 hours. The onset point of regeneration is dependent on the effective lifetime of the specimen. LeTID regeneration occurs earlier if the raw lifetime higher, likely duo to a dependence of the excess carrier concentration ∆n. With phosphorus gettering, a suppression of LeTID was observed. It was concluded that phosphorus gettering led to an LeTID mitigation effect. 

Figure 8 Effective lifetime measured by TR-PLI in one single multi wafer. The exposure condition is 1 sun 75 ℃(a)SiNx double side passivated, without phosphorus gettering(b)SiNx double side passivated, with phosphorus gettering [14]

Figure 8 Effective lifetime measured by TR-PLI in one single multi wafer. The exposure condition is 1 sun 75 ℃(a)SiNx double side passivated, without phosphorus gettering(b)SiNx double side passivated, with phosphorus gettering [14]

Figure 8 Effective lifetime measured by TR-PLI in one single multi wafer. The exposure condition is 1 sun 75 ℃(a)SiNx double side passivated, without phosphorus gettering(b)SiNx double side passivated, with phosphorus gettering [14]

Figure 8 Effective lifetime measured by TR-PLI in one single multi wafer. The exposure condition is 1 sun 75 ℃(a)SiNx double side passivated, without phosphorus gettering(b)SiNx double side passivated, with phosphorus gettering [14]

Defect types determined by wafer thickness variation

Dennis Bredemeier et al. from ISFH designed an experiment to identify the amount of LeTID on wafers with different thicknesses. As shown in Fig 9, the total LeTID defect density is linearly proportional to the thickness of the substrate [15]. According to fit, the diffusion coefficient range at 75 ℃ is in the range of (5±2)×〖10〗^(-11) 〖cm〗^2. The diffusion coefficient was found to correlate with those measured on some common impurities, Co, Ni, H. It was concluded that LeTID regeneration may be due to the diffusion of a metal species towards the surface of the Si wafer or a result of H passivation.

Figure 9  (1)LeTID curves of different wafer thicknesses normalized by effective defect density (2) Maximum defect density fitted from LeTID curve (3) Comparison of the fitted diffusion coefficient range with the common defect category [15]

Figure 9 (1)LeTID curves of different wafer thicknesses normalized by effective defect density (2) Maximum defect density fitted from LeTID curve (3) Comparison of the fitted diffusion coefficient range with the common defect category [15]

Figure 9  (1)LeTID curves of different wafer thicknesses normalized by effective defect density (2) Maximum defect density fitted from LeTID curve (3) Comparison of the fitted diffusion coefficient range with the common defect category [15]

Figure 9 (1)LeTID curves of different wafer thicknesses normalized by effective defect density (2) Maximum defect density fitted from LeTID curve (3) Comparison of the fitted diffusion coefficient range with the common defect category [15]

Figure 9  (1)LeTID curves of different wafer thicknesses normalized by effective defect density (2) Maximum defect density fitted from LeTID curve (3) Comparison of the fitted diffusion coefficient range with the common defect category [15]

Figure 9 (1)LeTID curves of different wafer thicknesses normalized by effective defect density (2) Maximum defect density fitted from LeTID curve (3) Comparison of the fitted diffusion coefficient range with the common defect category [15]

Change the firing temperature

Several works have verified that reducing firing temperature can reduce LeTID [16-18]. Catherine Chan [16] and Daniel Chen [17] from UNSW studied the influence of firing temperature on LeTID. With higher firing temperatures, LeTID was exacerbated, as shown in Fig. 10. It was believed that hydrogen released from SiNx:H during firing was the culptit behind for LeTID [18]. 

Hydrogen was also within the range of three possible candidates obtained from ISFH wafer study [15]. A possible solution for LeTID identified by Chan et al. was to refire cells again at a lower temperature [16], or add a gradual low temperature cooling step at the end of firing. This reduced cooling rate was found to be helpful [18].

Figure 10 The influence of firing temperature on LeTID of multi [16]

Figure 10 The influence of firing temperature on LeTID of multi [16]

Figure 10 The influence of firing temperature on LeTID of multi [16]

Figure 10 The influence of firing temperature on LeTID of multi [16]

Change the properties of the dielectric layer

The dielectric layer is the main source of bulk hydrogen. Usually, SiNx and aluminium oxide(Al2O3) are applied on the rear side in the case of PERC structure. The hydrogen content of SiNx is over 10% whereas Al2O3 has approximately 3%.

In 2016, Friederike Kersten found that when LeTID occurs when wafers are fired in the presence of a H containing dielectric film and does not occur when the passivation layer are absent [19], as shown in figure 11(1).

Carlos Vargos from UNSW carried out further research on LeTID combined with firing temperatures in addition to SiNx, deposition temperatures and layer properties [20]. In fig 10, LeTID increased due with higher film deposition temperatures as the film contained a higher fraction of H. While excess hydrogen temporarily increases the lifetime from passivation, it can also provide as a souce of unbound interstitial which may result in a more substantial LeTID/HID effect, as shown in fig 11(2).

Figure 11 (1) Lifetime tested after firing with and without SiNx/Al2O3 dielectrisc[19] (2) The relation of hydrogen fraction released, SiNx deposition temperature and LeTID maximum normalized defect density [20]

Figure 11 (1) Lifetime tested after firing with and without SiNx/Al2O3 dielectrisc[19] (2) The relation of hydrogen fraction released, SiNx deposition temperature and LeTID maximum normalized defect density [20]

Figure 11 (1) Lifetime tested after firing with and without SiNx/Al2O3 dielectrisc[19] (2) The relation of hydrogen fraction released, SiNx deposition temperature and LeTID maximum normalized defect density [20]

Figure 11 (1) Lifetime tested after firing with and without SiNx/Al2O3 dielectrisc[19] (2) The relation of hydrogen fraction released, SiNx deposition temperature and LeTID maximum normalized defect density [20]

At the WCPEC in 2018, Utkarshaa Varshney from UNSW further proved that LeTID defect concentration increases linearly as the SiNx thickness increases [21]. As shown in Fig 12, the normalized LeTID defect density was linearly proportional to the SiNx layer thickness. More hydrogen was presumed to be introduced from thicker layers.

Figure 12 With same deposition condition, LeTID related defect increases as the layer thickness increases [21]

Figure 12 With same deposition condition, LeTID related defect increases as the layer thickness increases [21]

Figure 12 With same deposition condition, LeTID related defect increases as the layer thickness increases [21]

Figure 12 With same deposition condition, LeTID related defect increases as the layer thickness increases [21]

Dark Annealing

For BO-LID, dark annealing can assist in a temporary dissociation of the BO complex. After BO complex annihilation, the sample lifetime can again start to degrade when the sample is illuminated. The “degradation-dark annealing” cycle for BO is shown in Fig 13(a). The cycle periods and amplitudes are remains similar after each cycle.

However, the “degradation-dark annealing” cycle of LeTID is different. After multiple cycles, degree of degradation begins decline. A four states model was introduced to explain the difference. There is a reservoir state of defect prior to precursors state A which releases a given amount of defect precursor during DA and depletes over time (fig 13(b)) [22]. Professor Stuart Wenham proposed the “bucket theory” to simplify this model. In short, reducing hydrogen content in the reservoir state can reduce LeTID/HID [12].

Figure 13 (a) BO-LID degrade-dark annealing cycles (b) LeTID degrade-dark annealing cycles [12]

Figure 13 (a) BO-LID degrade-dark annealing cycles (b) LeTID degrade-dark annealing cycles [12]

Figure 13 (a) BO-LID degrade-dark annealing cycles (b) LeTID degrade-dark annealing cycles [12]

Figure 13 (a) BO-LID degrade-dark annealing cycles (b) LeTID degrade-dark annealing cycles [12]

HID

According to the experiment findings, excess hydrogen introduced into silicon bulk is the main reason of LeTID [11-12]. As shown in fig 13, the bucket theory explains hydrogen transfer in different buckets. After firing, hydrogen in the device can be trapped or stored sometimes with only weak bonds – forming buckets 1 and 2 (B1 and B2). Under LeTID conditions the light and heat breaks these bonds thus freeing the hydrogen to flow into bucket 3 (B3) and causes degradation. 

Initially the degradation is fast as B1 and B2 are very full/high concentration and, the flow into B3 is fast. As B3 fills, a small amount starts to empty as it either moves to passivate another defect or disperses reducing the rate of degradation. As B1 and B2 become emptied (lower concentration), the rates of flow into B3 start reducing, while B3 fills further, such that the rate of hydrogen leaving B3 increases, further reducing the rate of degradation. This continues until the rate of B1 and B2 into B3 equals the rate at which B3 is emptying and the degradation flattens out at a minimum point. As the process continues, B1 and B2 are almost empty and only slowly trickling into B3, while B3 is now emptying much faster so the device starts to recover. The device is fully recovered when all three buckets are empty. All hydrogen is either dispersed or in a stable bond that will not be broken by LeTID testing conditions. 

Usually, the hydrogen is stored in the dielectric layer SiNx. After firing, the hydrogen from SiNx is released into the silicon bulk. Hydrogen is stored everywhere in silicon. For example, Hydrogen is stored in the heavily doped layers, bonded to native dopant and metal impurities, or bonded to vacancies, dislocations, stacking faults and within grain boundaries. Specifically, the hydrogen dimer (H2) is one of the reservoir of H (B1). BH can be considered as one of the defect precursors (B2). BH bonds are easily broken under illumination. Rapid firing changes the hydrogen content in B1 and B2 (accelerate the transfer of H from B1 to B2), and hydrogen flow from B2 to B3 to form LeTID/HID defects.

LeTID/HID Bucket theory of Professor Stuart[12]

LeTID/HID Bucket theory of Professor Stuart[12]

Although the recombination active configuration of LeTID is still unclear, hydrogen control is crucial. One possible way to solve LeTID/HID is to remove all access hydrogen. However, hydrogen is beneficial for device passivation and the device open voltage and efficnency relies on bulk hydrogen passivation, particularly on the cell made from low quality silicon material. Developing new technologies to control hydrogen is the key to making stable PV products.

Silicon wafers

Two main streams for wafers production are Czochralski and ingot casting. Their products are mono and multi.

Mono have a single perfect crystal lattice and few crystallographic defects. In mono, hydrogen can be stored in combination with dopants, vacancies, interstitials and small amounts of metal impurities.
Compared to mono, there are more defects in multi: metal impurities, dislocation networks and grain boundaries. Hydrogen from SiNx therefore has a greater impact in multi wafers. The greater concentration of defect sites means more hydrogen atoms can be stored in the bulk, which may lead to an increase in LeTID/HID more sensitive.

Since all inperfect structures in the Si crystal lattice may lead to hydrogen capture hydrogen, more efforts is needed by wafer manufacturers to reduce the defect concentrations and impurities in silicon.
 

Cells

Hydrogen is released from SiNx during cell fabrication, typically during the metallization firing step. Thermal processes during fabrication should be carefully designed to manipulate hydrogen diffusion, including SiNx deposition temperature, cofiring temperature and the subsequent cooling process etc. In PERC cell, hydrogen sources are located on both sides which provides a more effective means of bulk passivation but a higher risk for HID.

It would be ideal if the cell processing can perform hydrogen passivation without introducing large quantities of hydrogen. Present metallization process uses high temperature firing metal pastes. The frits in these metal pastes melt only above 550 ℃. The peak temperature required for contact formation  is even higher than 750℃. 

The difference in LeTID between mono and multi is related to amount of excess hydrogen involved. HID increase when excess hydrogen concentration increases.  HID regenerates when hydrogen is dispersed. In multi, there are more sinks for hydrogen.  For example, positive metal ions influence the diffusion of negatively charged hydrogen species which are present under illumination. In areas of high crystal defect density, LeTID/HID is hard to  regenerate, as the dark curves shown in Fig 8.

Application

The maximum LeTID can be related to the operation temperatures in field. Usually, the module sits at temperatures lower than 70 ℃. More work is required by the module supplier to reduce the module operation temperature, including new material and technology. 

Higher short circuit currents from improved products in recent years also makes the module hotter. On the user side, attention should be paid to eliminating hot spots during installation. 

Better ventilation atmosphere is essential to reducing the operating temperature of modules and reducing LeTID.

References

[1] Monokroussos C, Etienne D, Morita K, et al. Impact of calibration methodology into the power rating of c-Si PV modules under industrial conditions[C]//Proc. 28th Eur. Photovoltaic Sol. Energy Conf. 2013: 2926-2934.

[2] Kersten, F. , Engelhart, P. , Ploigt, H. C. , Stekolnikov, A. , Lindner, T. , & Stenzel, F. , et al. (2015). Degradation of multicrystalline silicon solar cells and modules after illumination at elevated temperature. Solar Energy Materials & Solar Cells, 142, 83-86.

[3] Jan Schmidt,Armin G. Aberle,Rudolf Hezel, Investigation of carrier lifetime instabilities in CZ-grown silicon, 26th IEEE Photovoltaic Specialists Conference,1997.9.13-18

[4] Saitoh T,A review of Japanese R&D for crystalline silicon solar cells,Ninth Workshop on crystalline silicon solar cell materials and processes, 1999.8:81-86

[5] Herguth A , Schubert G , Kaes M , et al. A New Approach to Prevent the Negative Impact of the Metastable Defect in Boron Doped CZ Silicon Solar Cells[C]// IEEE World Conference on Photovoltaic Energy Conversion. IEEE, 2006.

[6] Longi Release, 2016.5 SNEC

[7] Axel Herguth and Brett Hallam, A generalized model for boron-oxygen related light-induced degradation in crystalline silicon, AIP Conference Proceedings 1999, 130006 (2018);

[8] Fertig F, Lantzsch R, Mohr A, et al. Mass production of p-type Cz silicon solar cells approaching average stable conversion efficiencies of 22%[J]. Energy Procedia, 2017, 124: 338-345.
 
[9] Niewelt T , Selinger M , Grant N E , et al. Light-induced activation and deactivation of bulk defects in boron-doped float-zone silicon[J]. Journal of Applied Physics, 2017, 121(18):185702.

[10] Sio H C , Wang H , Wang Q , et al. Light and elevated temperature induced degradation in p-type and n-type cast-grown multicrystalline and mono-like silicon[J]. Solar Energy Materials and Solar Cells, 2018, 182:98-104.

[11] Daniel Chen, Phillip G. Hamer, Moonyong Kim et al., Hydrogen induced degradation: A possible mechanism for light- and elevated temperature- induced degradation in n-type silicon, Solar Energy Materials and Solar Cells 185 (2018) 174–182

[12] Alison Ciesla née Wenham, Stuart Wenham, Ran Chen et al, Hydrogen-Induced Degradation, 7th World Conference on Photovoltaic Energy Conversion, Waikoloa, Hawaii, 2018

[13] K. Ramspeck, S. Zimmermann, H. Nagel, A. Metz, Y. Gassenbauer, B. Birkmann, and A. Seidl, in Proceedings of the 27th European Photovoltaic Solar Energy Conference (WIP, Munich, 2012), p.861 – 865

[14] Zuschlag, A., Skorka, D., & Hahn, G. (2017). Degradation and regeneration in mc‐Si after different gettering steps. Progress in Photovoltaics: Research and Applications, 25(7), 545-552.

[15] Bredemeier, D., Walter, D. C., & Schmidt, J. (2018). Possible Candidates for Impurities in mc‐Si Wafers Responsible for Light‐Induced Lifetime Degradation and Regeneration. Solar RRL, 2(1), 1700159.

[16] Chan, C. E. , Payne, D. N. R. , Hallam, B. J. , Abbott, M. D. , Fung, T. H. , & Wenham, A. M. , et al. (2017). Rapid stabilization of high-performance multicrystalline p-type silicon perc cells. IEEE Journal of Photovoltaics, 6(6), 1473-1479.

[17] Chen, D. , Kim, M. , Stefani, B. V. , Hallam, B. J. , Abbott, M. D. , & Chan, C. E. , et al. (2017). Evidence of an identical firing-activated carrier-induced defect in monocrystalline and multicrystalline silicon. Solar Energy Materials and Solar Cells,172, 293-300.

[18] Eberle, R. , Kwapil, W. , Schindler, F. , Schubert, M. C. , & Glunz, S. W. . (2016). Impact of the firing temperature profile on light induced degradation of multicrystalline silicon. physica status solidi (RRL) - Rapid Research Letters.

[19] F. Kersten, J. Heitmann, and J. W. M¨uller, “Influence of Al2O3 and SiNx passivation layers LeTID,” Energy Procedia, vol. 92, pp. 828–832, Aug. 2016

[20] Carlos Vargas , Kyung Kim , Gianluca Coletti et al. Influence if silicon nitride and its hydrogen content on carrier-induced degradation in multicrystalline silicon. 2018. 33 EUPVSEC 561-564.

[21] Utkarshaa Varshney, Malcolm D. Abbott, Shaoyang Liu et al, Influence of dielectric passivation layer thickness on LeTID in multicrystalline silicon, 7th World Conference on Photovoltaic Energy Conversion, Waikoloa, Hawaii, 2018

[22] T.H. Fung, M. Kim, D. Chen, C.E. Chan, B.J. Hallam, R. Chen, D.N.R. Payne, A. Ciesla, S.R. Wenham and M.D. Abbott, "A four-state kinetic model for the carrier-induced degradation in multicrystalline silicon: Introducing the reservoir state", Solar Energy Materials and Solar Cells vol. 184, pp. 48-56, 2018

Tags: longi solar, monocrystalline wafer, c-si manufacturing, solar cell, perc, pv celltech, lid, letid, hid

Comments