Yield Performance of Standard Multicrystalline, Monocrystalline, and Cast-Mono Modules in Outdoor Conditions

Abstract

On the journey to reduce the cost of solar modules, several silicon-growing techniques have been explored to grow the wafers the cells are based on. The most utilized ones have been the multicrystalline silicon (mc-Si) and the monocrystalline ones, with monocrystalline grown by the Czochralski (Cz) technique being the current winner. Cast-mono (CM-Si) was also largely employed during the last decade, and there are several gigawatts (GWs) of modules on the field, but no data were shared on the performance of those modules. In this study, we put three small installations next to each other in the field consisting of 12 modules each, with the only difference being in the wafers technology employed: mc-Si, CM-Si, and CZ-Si. The first two systems have been manufactured with the same equipment and had their field performance closely monitored for three years, while the CZ-Si one has been monitored for 17 months. The performance data shared show that CM-Si performance on the field is better than mc-Si and is very similar to CZ-Si, with no abnormal degradation. CM-Si requires less energy than CZ-Si to be manufactured, and high efficiencies have been reported; the field performance suggests that it is a very valid technology that deserves further exploration.

1. Introduction

The Photovoltaic (PV) industry passed through a very steep learning curve and a spectacular cost reduction [1] over the last five decades based on mass production and technological evolution [2].

Silicon-based modules are the vast majority of modules produced today, representing more than 95% of the market [3]. The whole silicon-based modules manufacturing value chain experienced and continues to experience significant advances from polysilicon manufacturing, crystal growth, and wafers manufacturing to cell and module production.

In the crystal growth segment of the value chain, the main technologies employed today are monocrystalline wafers using the Czochralski route (CZ) and the multicrystalline (mc-Si) silicon casting technology. CZ-based modules are of higher efficiencies due to lower impurities levels in the substrate but are more expensive to produce than the mc-Si wafers. An alternative technology to produce silicon wafers is the so-called cast-mono (CM-Si) or mono-like silicon that basically consists of utilizing the casting process of mc-Si to grow monocrystalline silicon instead. This is achieved by introducing a seed or several seed crystals at the bottom of the casting crucible and making small adjustments in the casting process, resulting in a cost very similar to mc-Si casting but with a performance similar to the monocrystalline silicon [4]. Interestingly, the mechanical properties of CM-Si wafers were proven to be higher than those of mc-Si [5], also quite comparable to CZ’s. Furthermore, it was suggested that direct solidification processes, more cost-effective than the CZ method, are particularly suitable for the use of less pure silicon feedstock such as that of upgraded metallurgical-grade silicon [6].

CZ-based modules are the clear winners [3], being the main technology employed today. Nevertheless, cast-mono was also largely implemented during the last decade, with several GWs of modules on the field currently. However, to the knowledge of these authors, no data have been reported so far on the field performance of cast-mono modules.

In this study, we report the field performance of a CM-Si-based PV system for the first time and compare it with similar technologies. To do so, we put three small installations next to each other in the field consisting of 12 modules each; all of them used the same manufacturing technology for cells (PERC-passivated emitter and rear cell technology) and modules, manufacturing equipment, and raw materials, with the only difference between the systems being the wafers technology employed. Two of the systems are manufactured with the same casting technology (direct solidification system, DSS [7,8,9]) and the same equipment, but we grew mc-Si in one of them and CM-Si in the other. The third system uses standard Cz silicon wafers. The facilities have been closely monitored for three years for the casting technologies and 17 months for the Cz system; the performance data are shared and analyzed in this study.

2. Materials and Methods

2.1. Description of the Installation

The 3 generators used for this experiment consisted of 12 glass-backsheet PV modules (Model CS3U) manufactured by Canadian Solar, Guelph, ON, Canada, each mounted on a typical two-row portrait-fixed racking system. One of the systems used CS3U-P (P3) mc-Si modules for a total of 4.06 kWp (Array mc), another used CS3U-P (P5) CM-Si modules for a total of 4.50 kWp (Array CM), and the last one used CS3U-MS Cz-Si modules for a total of 4.54 kWp (Array CZ). All systems were located next to each other in a Canadian Solar testing facility in Suzhou, China (coordinates N 31.3 E 120.8), south oriented, 25-degree tilt, and without shadowing, as shown in Figure 1. Each generator was connected to the grid through a Huawei Sun 2000-10 KLT inverter of 10 kW.

2.2. Data Collection

Detailed measured module electrical parameters under Standard Test Conditions (STC) after 120 kWh/m² outdoor exposure are shown in

Table 1. Module electrical information summary.

Array mc	SN	Pm [W]	Vm [V]	Im [A]	Voc [V]	Isc [A]
1	Y1019206450004	337	38.3	8.79	46.1	9.26
2	Y1019206450005	338.4	38.4	8.81	46.2	9.27
3	Y1019206450006	337.2	38.4	8.79	46.1	9.25
4	Y1019206450007	337.2	38.4	8.79	46.1	9.25
5	Y1019206450008	339	38.4	8.83	46.2	9.3
6	Y1019206450009	338	38.4	8.81	46.1	9.28
7	Y1019206450010	339.6	38.4	8.85	46.1	9.31
8	Y1019206450012	337.5	38.4	8.79	46.1	9.27
9	Y1019206450013	338.5	38.4	8.81	46.1	9.28
10	Y1019206450015	339	38.4	8.82	46.2	9.3
11	Y1019206450016	338.7	38.4	8.81	46.2	9.27
12	Y1019206450017	338.6	38.4	8.82	46.2	9.27
Sum		4058.7
Array CM	SN	Pm [W]	Vm [V]	Im [A]	Voc [V]	Isc [A]
1	Y1019206450004	337	38.3	8.79	46.1	9.26
2	Y1019206450005	338.4	38.4	8.81	46.2	9.27
3	Y1019206450006	337.2	38.4	8.79	46.1	9.25
4	Y1019206450007	337.2	38.4	8.79	46.1	9.25
5	Y1019206450008	339	38.4	8.83	46.2	9.3
6	Y1019206450009	338	38.4	8.81	46.1	9.28
7	Y1019206450010	339.6	38.4	8.85	46.1	9.31
8	Y1019206450012	337.5	38.4	8.79	46.1	9.27
9	Y1019206450013	338.5	38.4	8.81	46.1	9.28
10	Y1019206450015	339	38.4	8.82	46.2	9.3
11	Y1019206450016	338.7	38.4	8.81	46.2	9.27
12	Y1019206450017	338.6	38.4	8.82	46.2	9.27
Sum		4058.7
Array CZ	SN	Pm [W]	Vm [V]	Im [A]	Voc [V]	Isc [A]
1	Y1019206450004	337	38.3	8.79	46.1	9.26
2	Y1019206450005	338.4	38.4	8.81	46.2	9.27
3	Y1019206450006	337.2	38.4	8.79	46.1	9.25
4	Y1019206450007	337.2	38.4	8.79	46.1	9.25
5	Y1019206450008	339	38.4	8.83	46.2	9.3
6	Y1019206450009	338	38.4	8.81	46.1	9.28
7	Y1019206450010	339.6	38.4	8.85	46.1	9.31
8	Y1019206450012	337.5	38.4	8.79	46.1	9.27
9	Y1019206450013	338.5	38.4	8.81	46.1	9.28
10	Y1019206450015	339	38.4	8.82	46.2	9.3
11	Y1019206450016	338.7	38.4	8.81	46.2	9.27
12	Y1019206450017	338.6	38.4	8.82	46.2	9.27
Sum		4058.7

.

Table 2. Temperature coefficients.

Temperature Coefficient	γ (Pmax) (%/°C)	β (Voc) (%/°C)	α (Isc) (%/°C)
CS3U-335P(P3) (mc-Si)	−0.38	−0.31	0.05
CS3U-370P(P5) (CM-Si)	−0.37	−0.29	0.05
CS3U-370 MS (Cz-Si)	−0.37	−0.31	0.05

shows the temperature coefficient of power (γ), voltage (β), and current (α) for the three mc-Si-, CM-Si-, and CZ-Si-based modules provided by the manufacturer and measured in accordance with standard procedures.

Arrays CM and mc were measured from 26 July 2019 to 19 July 2022, while array CZ was measured from 20 August 2019 to 31 December 2020. System performance (DC and AC voltage and current and AC power) and ambient data (ambient and module temperature measured at four different points for each array, wind speed and direction, inclined plane irradiance and horizontal irradiance, and rain) were measured every minute using the equipment described in

Table 3. Measurement equipment list.

Equipment	Vendor	Model	Tolerance
DC Meter	GMC-I	V604s-20A	Voltage: ±0.2% Current: ±0.2%
Wind speed sensor	Met one	034b	V < 10.1 m/s: ±0.1 m/s V > 10.1 m/s: ±1.1% × isplay value
Wind direction	Met one	034b	±4°
Rain sensor	Intell-sun	PHYL	±4%
Ambient temperature sensor	Rotronic	HC2S3	±0.1 °C at 23 °C config. setting
Humidity sensor	Rotronic	HC2S3	±0.8% × display value
Pyranometer	Kipp & Zonen	CMP10	Yearly instability < 0.5%
Module temperature sensor	SUYI	PT100	±0.2 °C at 25 °C config. setting

. There was no difference in the equipment employed or the location of the sensors on the arrays.

3. Results

After the respective years of testing in outdoor conditions, the total irradiation measured in the system amounted to 3600 kWh/m² for arrays mc and CM and 1715 kWh/m² for array CZ. During the testing period, the facility did not provide data for a total of 13 out of 156 weeks of the three-year testing period (which also affected array CZ) for several reasons (system downtime or failed measurement from any sensor or measurement equipment in either of the arrays). Thanks to the long testing period, we can assume that the data reflect any climatic condition in the location.

The modules’ appearance through visual inspection showed no significant aging, degradation, hot spots, or any other performance phenomena.

The performance of the systems on a given day is compared in these results, as well as the energy production (yield) during the testing period.

Figure 2 plots the site weather information over the testing period and ambient temperature and irradiance at the array plane.

3.1. Daily Performance

Figure 3 shows the behavior of the systems on a standard winter and summer day. The maximum power vs. the time of the day is represented and compared to the pyrometer of array (POA) measurement. Arrays CM and CZ show almost identical results, with a higher Pmax than mc since the systems have more Wp installed.

The daily yield curves are also very similar, but array CM remains consistently higher on a typical winter and summer day as shown in Figure 4. The temperatures of the modules are also shown in the same figure.

Using the cloudy day Figure 3 graph, the kick-in and kick-out times are examined in Figure 5, and they are almost simultaneous events in all systems. In fact, in a sample of 30 consecutive days, CM activated 17 times later and 13 times earlier or equal to mc, concluding that both systems have a very similar or identical activation energy. The result is very similar when compared to the array CZ.

3.2. Energy Yield

Related to the energy production of the systems, we collected a total of 143 weeks of production of the mc and CM systems and 61 weeks for the CZ system under the same conditions; a total of 156 (mc and CM) and 74 (CZ) weeks were measured, but, as already mentioned, 13 of those presented measurement problems on different equipment or events of default, such as loose cables, inverters defaults, etc., that invalidated the data.

Figure 6 shows the monthly energy yield of the three systems in kWh/kWp and the differences in the energy yield of the CM system with the other two.

We can see that in every single month, the performance of the CM system is better than the mc, with a, on average, 1.5% higher energy yield. Unexpectedly, the energy yield obtained from CM is also 1% higher than the CZ one, despite it being well documented that it should be the other way around. This would require further study. Nevertheless, the systems perform all of them in a very similar way; this is consistent over time, which signals no abnormal degradation of the MC system.

3.3. Performance Ratio and Degradation

The Performance Ratio (PRSTC) is calculated as the relation between the actual energy produced, according to the power at the maximum power point Pmp exhibited, and the final energy generated as a result of the received irradiation, considering its nominal (rated) peak power. We adjusted it using the average temperature of the system as indicated below, where $\gamma$ is the module temperature power coefficient and $T$ is the average temperature of the module in the period considered.

$$PR={PR}_{STC}\times \left({\displaystyle \frac{1}{1-\frac{\gamma }{100}\times (T-25)}}\right)$$

Figure 7 shows that the CM array systematically has a higher $PR$ than the other two, which seems strange for the CZ system that should behave better, as mentioned before. Again, the behavior of the CM system’s $PR$ is consistent over time, showing no sign of significant degradation. Considering that the first years are the most relevant ones when it comes to degradation [10], the data analyzed during the 156 weeks should cover the most critical period. The results do not show any evidence that could result in a concern about the reliability of the silicon when the CM route is the one used; it shows quite the opposite.

4. Discussion

4.1. Comparison of CM-Si vs. mc-Si

According to the results above, CM-Si-based modules provide systems that generate more energy than mc-Si-based ones at a very similar cost; a 2% higher average yield was measured over a period of 3 years in this experiment. Both systems started and stopped delivering energy at the same time of the day, but the CM-Si had better behavior, as shown by the higher yield and higher PR.

Figure 8 plots the yield difference between both systems under different radiation conditions. We can see that the yield difference is higher at lower radiation, meaning that the CM-Si-based system could behave better than the mc-Si-based one at low light. However, it is not very conclusive based on the results obtained.

While the system also behaves better with temperature, as shown by the measured coefficient temperature in Table 2, we believe low light has a higher effect since the main difference between the CM-Si- and the mc-Si-based modules is the defect (dislocation and grain boundary) density. The mc-Si module with a larger defect density is slower to react to the light energy of the low energy state, and even though we cannot see this with an earlier kick, we can see it in the weaker performance under the low-light conditions of the mc-Si-based system. Recent research has demonstrated that temperature coefficients are illumination-dependent [11], with variations moving from 30% to double, depending on the cell technology and conditions; this could also explain the further improvement under low irradiance.

Table 4. Crystal main parameters.

Parameter	CM-Si	mc-Si
Oxygen content (ppma)	5–8	5–8
Dislocation density (pcs/cm2)	500–1000	104–106
Grain boundary density (pcs/cm2)	~0	1–20
Resistivity (Ω.cm)	1–1.2	1–1.2

shows the main parameters measured at the crystal level of samples of both modules, and Figure 9 shows electroluminescence (EL) and metalloscopy sample pictures of both types of modules; the higher defect density can be noticed on the mc-Si-based one.

Even though light-induced degradation analysis was not performed in this work, there is no sign of significant performance difference over time between the two systems, as shown in Figure 8.

4.2. Comparison of CM-Si vs. Cz-Si

In Figure 10, we again plot the yield difference between both systems under different radiation conditions, and we can now see that the yield difference is almost constant, with no difference in performance for low or high light, which makes us think that the main difference in the yield is due to the better performance under the temperature of the CM system.

5. Conclusions

There are several GWs of mono-cast modules in the field, and to our knowledge, this is the first time that the field performance of cast-mono modules over a long period of time has been reported. The results obtained from this study demonstrate that it is a very valid technology that does not show signs of relevant degradation after several years of operation and whose performance is better than mc-Si modules and similar to CZ-Si modules. Both the energy yield and the PR over time have been quantified and compared among the three technologies.

As the energy required to grow cast-mono modules is significantly lower than the energy required to grow CZ-Si ones [12], resulting in a smaller carbon footprint and high efficiencies [13], we conclude that it is a technology with potential that deserves to be further explored. There is equipment that was manufacturing several GWs a year of cast-mono wafers that is being scrapped when it instead could be employed in manufacturing solar cells for the utility-scale PV market since it is a very valid technology, as this study shows. This equipment sitting idle adds to the CO₂ footprint due to the electricity required to manufacture unnecessary new equipment.