A Review of the Techniques to Measure the Hermeticity of Glass Frit Encapsulated Solar Cells

Emerging 3 generation photovoltaic technologies such as perovskite and dyesensitized solar cells are very attractive for commercialization mainly due to their low-cost materials and fabrication processes. The main drawback of these devices is their poor long-term stability. To increase the long-term stability of these devices, a hermetic encapsulation is required. The hermeticity of encapsulated devices are measured and characterized using hermeticity tests according to standard test procedures. A review of the several techniques to measure the hermeticity is presented, addressing the test methods, limitations and applicability to perovskite and dye-sensitized solar cells glass frit encapsulated devices. Author


Introduction
Dye-sensitized solar cells (DSSCs) and perovskite solar cells (PSCs) are included in the emerging third generation of PV technologies. These technologies use low-cost and abundant materials, and can be used for building integrated applications. The mentioned advantages make these PV devices very attractive for commercialization. However, their main disadvantage is the poor long-term stability (IRENA 2017). Conventional DSSCs are prepared on conductive glass substrates and consist of a photoanode, a dye, an electrolyte (i.e. liquid), and a counter electrode. The leakage of the liquid electrolyte is the main source instability for DSSCs (Hamann et al. 2008). PSCs are typically fabricated on conductive glass substrates and comprise several layers: an electron transport layer (ETL), a mesoporous scaffold, a perovskite light absorber, a hole transport layer (HTL) and a back contact (Leijtens et al. 2015). The perovskite light absorber is highly unstable in the presence of moisture and oxygen (Asghar et al. 2017). DSSCs and PSSCs must be properly encapsulated to avoid the leakage of electrolyte and to prevent the ingress of moisture and oxygen, thus increasing the long-term stability of the devices. The sealant material must be inert, to avoid the degradation of the components of the cell, and to provide airtight encapsulation through the lifetime of the devices (Wang et al. 2016).

Understanding Leakage Mechanisms
The molecules of a gas move freely with high velocity interacting with each other and with any surface in their path. In a sealed package, the molecules of a gas bump into the package walls, i.e. the substrates and the sealant perimeter. The number of collisions is dependent of the number of molecules of the gas, and is correlated to the pressure and temperature by the ideal gas law (Greenhouse, Lowry, and Romenesko 2012) where, P is pressure, V is volume, n is number of gas molecules, ℛ is the gas constant and T is the absolute temperature. Two permeation mechanisms can be identified when a gas is entering or leaving a cavity trough a leak channel: viscous flow and Knudsen diffusion. Viscous flow occurs when the gas molecules collide predominantly with each other. Knudsen flow takes places when the gas molecules strike mainly the wall of the leak channel (Sundén and Fu 2017).
The average distance a molecule of a gas travels before colliding with another is called the mean free path (λ). λ is inversely proportional to the pressure of the gas and the diameter of the molecule (Roy et al. 2003). The ratio of λ and the diameter of the leak channel (⌀) (considering the leak channel has a shape of a pipe) is known as Knudsen number (Kn) and is used to define the type of flow (Roy et al. 2003 For Kn ≤ 0.01 the viscous flow is the dominant permeation mechanism and the Knudsen diffusion is the principal permeation mechanism for Kn ≥ 10. For 0.01 < Kn < 10 both permeation mechanisms contribute for the gas permeation (Roy et al. 2003). The gas flow rate, q, through a pipe is given by Greenhouse, Lowry, and Romenesko (2012): where, Pout is the total pressure at the outlet of the pipe, Pin is the total pressure at the inlet of the pipe and C is the flow conductance. The conductance represents the resistance to the mass transport due to the friction between the molecules and the walls (Greenhouse, Lowry, and Romenesko 2012). The conductance in the viscous flow regime is proportional to the average pressure. The viscous flow rate in a long round shape tube is given by the Hagen-Poiseuille equation (Greenhouse, Lowry, and Romenesko 2012): where, d is the diameter of the tube, Pa is the average pressure, η is the viscosity of the gas and ℓ is the length of the tube. Viscous flow cannot occur when the pressure inside a cavity is equal to the external pressure. The pressure gradient is the driving force for viscous flow. Thus, packages encapsulated at 1 atm for normal atmospheric use (i.e. 1 atm) are not subject to gas permeation by viscous flow (Greenhouse, Lowry, and Romenesko 2012). Viscous flow happens when a pressure difference is imposed to packages and when the leak channels of the sealant are significantly larger than the mean free path of the molecules of a gas. Under these conditions, viscous flow is only presented in gross leaks, i.e. leak rates higher than 1 × 10 -5 Pa·m 3 ·s -1 (Greenhouse, Lowry, and Romenesko 2012). In Knudsen diffusion, the conductance is proportional to average velocity of the molecules, thus it is dependent on the molecular mass and the temperature of the gas. The Knudsen diffusion rate in a long round tube is given by Roy et al. (2003): where, d is the diameter of the tube, ℓ is the length of the tube, M is the molecular mass, ℛ is universal gas constant, T is temperature and pout and pin are the partial pressure of a specific gas molecule at the outlet and inlet of the pipe. Knudsen diffusion is present when there is a difference in the partial pressure of a gas between the sealed package and the external environment. The Knudsen diffusion mechanism is dominant when the leaking channel is substantially smaller than the mean free path of the molecules of the gas and therefore when leaks rates are lower than 1 × 10 -7 Pa·cm 3 ·s -1 (Greenhouse, Lowry, and Romenesko 2012). The above mentioned equations (Equation (4) and Equation (5)) are used to quantify the leak rate of a sealed package. However, the leak rate changes with the time and pressure (total and partial) and is different for different molecules. In a package encapsulated at 1 atm of N2 placed in ambient air (1 atm), no viscous flow is present because the total pressure is equal inside and outside of the package. However, the partial pressure of the N2 in the package is 1 atm while in the outside is 0.78 atm. Thus, N2 will leak out until the partial pressure reaches equilibrium. O2 will leak into the package because the partial pressure of O2 is 0.21 atm in the air and zero atm inside the package. Over time, the leak rate will decrease due to a decrease in the partial pressure gradient. At equilibrium, the composition of the gas in the package will be the same as in air. The time at which equilibrium is reached will depend on the size of the leak channel of the encapsulation (Greenhouse, Lowry, and Romenesko 2012 (Kähler, Lofink, and Reinert 2020).

Perfluorocarbon gross leak test
The perfluorocarbon gross leak test consists in immersing the device in type I detector fluid and pressurizing the chamber for a defined time. The type I detector fluid is a perfluorocarbon liquid with a boiling temperature between 50 °C and 95 °C. The bombing (pressurization) conditions are presented in Table 1 (Department of Defense 2019).

Pressure / Pa Minimum pressurization time / hour Test condition C1
Test condition C3 2.07 × 10 5 23.5 12 3.10 × 10 5 8 4 4.14 × 10 5 4 2 5.17 × 10 5 2 1 6.21 × 10 5 1 0.5 7.24 × 10 5 0.5 N/A  (2019)) In test condition C1, after the bombing phase, the device is removed from the chamber with the type I detector fluid and placed on a second chamber with a type II indicator fluid at 125 ± 5 °C. The type II indicator fluid is a perfluorocarbon liquid with a boiling temperature from 140 °C to 200 °C. The device must be observed from the moment of the immersion for a minimum period of 30 second, unless rejected earlier. If a stream of bubbles or two or more large bubbles originate from the same point the device is rejected (Department of Defense 2019).
Due to the difficulty in detecting the bubbles, the observation step must use a magnifier and a nonreflective black background while the device is illuminated (Department of Defense 2019).
In test condition C3, after the bombing phase, the device is placed in a chamber at 125 ± 5 °C connected to a perfluorocarbon vapor detector that measure the amount of the type I detector fluid that evaporates. Devices where the equipment detects more than 0.167 µL of type I detector fluid have gross leaks (Department of Defense 2019).
The perfluorocarbon gross leak test involves temperatures that are not compatible with the DSSCs and PSCs. Some of the components of these solar cells are temperature sensitive and degrade at temperatures higher than 85 °C (Mesquita, Andrade, and Mendes 2019).

The weight gain test
The weight gain test identifies if a package has a gross leak by measuring the weight difference of the package before and after being immersed in a fluid under pressure. The weight of the device is measured after a pre-heating step at 125 °C for a minimum of 1 hour. The device is then placed in a chamber and immersed in a detector fluid, e.g. a perfluorocarbon fluid with boiling point between 50 °C and 110 °C. The chamber is pressurized to 5.17 × 10 5 Pa for 2 hours. Devices that cannot withstand this condition can be submitted to 3.10 × 10 5 Pa for 10 hours. After the bombing phase, the weight of the devices is measured within 4 minutes following the removal from the fluid. The weight gain of the device is then calculated (Department of Defense 2019). Devices with an internal volume of ≤ 0.01 cm 3 and > 0.01 cm 3 are rejected if they gain more than 1 mg and 2 mg, respectively (Department of Defense 2019). This test method includes a heating step, similar to the perfluorocarbon gross leak test, that is not compatible with the PSCs and DSSCs.

Penetrant dye test
The penetrant dye test is a destructive test that is used to determine the location of gross leaks. The devices are place in a chamber filled with dye and the chamber is pressurized to ca. 7.24 × 10 5 Pa for a minimum of 3 hours. The bombing condition of 4.14 × 10 5 Pa for 10 hours can also be used. After bombing step, the devices are washed using a suitable solvent for the particular dye used in the test. Afterwards, the devices are examined under an ultraviolet light source using a magnifier for any evidence of dye penetration. The detection of dye inside a package is considered a failure (Department of Defense 2019). The MIL-STD-883 define that the dyes Zyglow, Fluorescein and Rhodamine B with maximum reflection at 365 nm, 493.5 nm and 556 nm, respectively, shall be used (Department of Defense 2019). This test cannot be used to determine the gross leaks of encapsulated PV devices since it is a destructive test. Any device, hermetic or non-hermetic, will become non-functional.

Helium leak rate test
The helium leak rate test is the most used test to characterize the hermeticity of several types of electronic devices. This test uses helium as a tracer gas to measure the fine leaks of packages. Helium is a suitable tracer gas because is one of the smallest molecules, is inert, non-toxic and has a low concentration in air. This test method is comprised of two phases, a bombing phase and a measurement phase. In the bombing phase, the devices are placed in a chamber and pressurized to a specific pressure of helium for a certain time. Only helium must be present in the chamber, therefore, the chamber must be evacuated before filling with helium. Otherwise, a series of vents and helium refills can be performed. During this step, helium will enter the package cavity due to partial pressure difference (Department of Defense 2019). After the bombing phase, the chamber is depressurized, and the devices are transferred to a second chamber connected to a mass spectrophotometer. The second chamber is evacuated, forcing the helium that entered the cavity of the package in the bombing phase to leak out. The amount of helium that leaks out is measured by the mass spectrophotometer (Department of Defense 2019). This technique has two test method: a fixed method and a flexible method. The fixed test method defines the bombing pressure and period (t1), the maximum dwell time to measure the sample after bombing phase (t2), the internal volume of the device (V) and the reject leak rate limit (R1) in Pa·m 3 ·s -1 . These conditions are presented in Method 1014 of MIL-STD-883 standard. Devices with leak rates lower than the reject leak rate limit are considered hermetic (Department of Defense 2019).
In the flexible test method, it is possible to choose the bombing period (t1), the dwell time for the measurement of helium leak rate (t2) after depressurization and the bombing pressure (PE), provided that it is ≥ 2.03 × 10 5 Pa of helium. The reject helium leak rate limit (R1), in Pa•m 3 •s -1 He, is calculated by the Howl-Mann Equation (6) with the previous chosen values, the internal cavity volume of the device and the equivalent leak rate limit (L) ( Table 2). The devices are considered hermetic if the measured helium leak rate (R) by the mass spectrometer is lower than the calculated reject helium leak rate limit (R1) (Department of Defense 2019).
where R1 is the reject helium leak rate limit in Pa•m 3 •s -1 He, L is the equivalent leak rate limit in Pa•m 3 •s -1 air, PE is the bombing pressure in Pa, P0 is the atmospheric pressure in Pa, MAir is the molecular weight of air in grams, MHe is the molecular weight of tracer gas He in grams, t1 is the bombing period in seconds, t2 is the dwell time for the measurement of helium leak rate after depressurization in seconds and V is the internal cavity volume.

L reject limit / Pa•m 3 •s -1 air (for Hybrid Class H, and Monolithic Classes B, S, Q and V*)
L reject limit / Pa•m 3 •s -1 air (for Hybrid Class K*) ≤ 0.05 5 × 10 -9 1 × 10 -10 0.05 < V ≤ 0.4 1 × 10 -8 5 × 10 -10 > 0.4 1 × 10 -7 1 × 10 -9 * Equivalent leak rate limits (L) for the various classes of microcircuits  (2019)) The helium leak rate test has several limitations. This method does not differentiate between helium leaking from a package and desorption of helium from the test chamber surface. The mass spectrophotometer reads the total amount of helium removed from the test chamber. This problem is especially important for packages composed by polymers substrates or sealants, due to high adsorption of helium in these materials. In these cases, it is possible that the leak rate measurement is reading only the desorption of helium, thus providing an inaccurate value (Costello, Desmulliez, and McCracken 2012).

Cumulative helium leak test (CHLT)
The cumulative helium leak test is a variation of the helium leak rate test. The test procedure is the same as in the helium leak rate test. The main difference is the use of an additional cryogenic pump in the detection system. A cryogenic pump is capable of extracting all of the gases in the test chamber except noble gases which include helium (Greenhouse, Lowry, and Romenesko 2012). If a device has a gross leak, the volume of helium in the test chamber increases rapidly at the start of the test. If a device has a fine leak, the volume of helium increases slowly over time. The rate of change in the volume of helium is used to determine the leak rate. This determination method allows to measure gross and fine leaks. This test is capable of measuring leak rates from 1 Pa•m 3 •s -1 to 4 × 10 -15 Pa•m 3 •s -1 (Department of Defense 2019). The CHLT has a higher sensitivity compare to the traditional helium leak rate test (1× 10 -13 Pa•m 3 •s -1 ) (Kähler, Lofink, and Reinert 2020). Moreover, virtual leaks due to helium desorption of the surface of a package can be detected through signal reading. Real leaks have a linear rate while virtual leaks exhibit an exponential rate. However, CHLT, like the helium leak rate test, is dependent on the volume of the packages (Kähler, Lofink, and Reinert 2020). The fixed bombing conditions (pressure and time) for different internal volume cavities and the related reject leak rate limits can be found on method 1014.17 of MIL-STD-883 standard.
The flexible method can also be used; this method is similar to the flexible method in the helium leak rate test (Department of Defense 2019).

Krypton-85 radioactive tracer gas test
The radioisotope element krypton-85 is used as a tracer gas in the Krypton-85 radioactive tracer gas test. Krypton-85 is a radioisotope with a half-life of ca. 11 years and when decays it emits beta and gamma rays (Costello, Desmulliez, and McCracken 2012;Greenhouse, Lowry, and Romenesko 2012). The gamma rays can penetrate package walls, thus the krypton-85 that leaks inside the package can be read by measuring the gamma radiation. The measuring step does not require vacuum to force the tracer gas to leak out as in the helium leak rate test.
Kr-85 radioactive tracer gas test is independent of cavity volume and is more suitable industry application where is necessary to test high number devices. The main disadvantage of this method is the use of radioactive material that requires specialized handling. Moreover, over time the testing apparatus becomes radioactive. This test must be complemented with a gross leak test (Costello, Desmulliez, and McCracken 2012). Activity of the radioisotope is measured in microcuries per atmosphere cubic centimeter. The concentration of Kr-85 in the tracer gas mixture of Kr-85/air needs to be higher than 1000 µCi·Pa -1 •m -3 . Curie, Ci, is a unit of radioactivity of a material. The devices are placed in a chamber and the chamber is filled with the tracer gas mixture up to a certain pressure for a period determined by Department of Defense (2019): where QS is the maximum allowed leak rate in atm•cm 3 •s -1 Kr-85 (based on L values from Table 2. 1.71 Pa•m 3 •s -1 Kr-85 = 1 P•m 3 •s -1 air), is the counts per minute above the ambient background after pressurization, s is the specific activity in μCi•Pa -1 •m -3 of the Kr-85 tracer gas, k is the counting efficiency of the scintillation crystal used to measure Kr-85 (This k-factor must be determined for the combination of scintillation crystal used for the measurement with the device geometry), Pe is the bombing pressure in Pa (must be higher than 2.03 × 10 5 Pa ), Pi is the original internal pressure of the device in Pa and t is the pressurization period in hours.
After the bombing step, the devices are removed from the chamber and, within 1 hour, the counts per minute must be measured using a scintillation crystal system. The actual Kr-85 leak rate (Q) of a tested device is given by Department of Defense (2019): = Actual read of counts per minutes × A device is considered hermetic if Q is lower than QS.

Optical leak test
The optical leak test is based on the measurement of the deflection of a package when submitted to pressure difference. To use this test the package must be able to deflect at least 7.25 × 10 -7 µm·Pa -1 (Department of Defense 2019). The package is placed in a chamber where a pressure difference (by vacuum or pressurization) is applied and an interferometer measures the deflection of the package over time. When a device with a leak is pressurized, it will bend inwards. Then, as the gas leak into the package and the pressure starts to equalize, the curve begins to decrease. If a device with a leak is placed under vacuum, first will bend outwards because the pressure is higher inside the package. Then starts to bend inwards as the gas leaks out. By measuring the deflection over time, it is possible to determine the leak rate. The leak rate of the device is given by Equation (9). The deflection stiffness (µm·Pa -1 ) of a non-leaking sample must be measured to obtain accurate leak rates (Department of Defense 2019).
where, OL is the implied leak rate of the test in Pa•m 3 •s -1 , V is the internal free volume of the package cavity in m 3 , t is the test duration time in seconds, Pa is the chamber test pressure in Pa as a function of altitude, ΔPi is the chamber test pressure in Pa at the of the test, ΔPf is the chamber test pressure -leakage in Pa (Leakage is the change in pressure inside the package during the test. Leakage = deflection movement (µm) / deflection stiffness (µm·Pa -1 ). This test can measure gross and fine leaks. A package fails the gross leak test if no deflection is detected or if the package deflects in the beginning of the test but quickly returns to the unpressurized position. A package passes the fine leak test if the OL is lower than the values presented in Table 2 (Department of Defense 2019).
The lid of the package must be able to deflect otherwise the optical leak test cannot be applied. Furthermore, this technique has sensitivity issues since it depends on the stiffness and thickness of the lid, the geometry of the sealed cavity, the test duration and the sensitivity of the measuring equipment (optical interferometer). The optical leak test has lower sensibility than other hermeticity methods being able to read leak rates > 1 × 10 -10 Pa•m 3 •s -1 (Costello, Desmulliez, and McCracken 2012).

Fourier transform infrared spectroscopy (FTIR)
This method uses Fourier transform infrared spectroscopy (FTIR) to determine the concentration of a gas inside a device. The concentration is measured through the absorption of infrared radiation (IR) by the molecules of the gas. A reference FTIR measurement is performed to identify the IR spectrum of the cavity of a package before the test. The test begins by pressurizing the devices using a tracer gas, similar to helium leak rate test. The bombing conditions are arbitrary and should be chosen so that enough tracer gas enters the device. Typically, N2O is used as tracer gas due to similar molecule size to N2 and high absorbance in the IR spectrum. Others tracer gases such as SF6, OCS, HCl, H2S can be used instead of N2O, however most of the mentioned gases are toxic (Veyrié et al. 2005). After bombing, FTIR is used to measure the concertation of N2O in the package and monitored its variation over time. The package walls must be transparent to the IR radiation to use this method (Veyrié et al. 2005).
To determine the leak rate, the transmission spectrum of the device after bombing is divided by the transmission spectrum of the reference (before bombing). Thus, transmission rate of the tracer gas in the package is obtained. The partial pressure (p) of the tracer gas is calculated from the transmission rate by Lellouchi et al. (2010): where, T' is the transmission rate for the absorption wavelength of the tracer gas, l is the cavity thickness, ω is the molar coefficient of the gas for the specified wavelength, ℛ is the universal gas constant and T is the temperature. Lastly, the leak rate (L) is determined using the modified Howl-Mann Equation (11). The rejected leak rate limits for this test are shown in Table 2 where p is the partial pressure of the tracer gas, L is the equivalent leak rate limit in Pa•m 3 •s -1 air, PE is the bombing pressure in Pa, P0 is the atmospheric pressure in Pa, MAir is the molecular weight of air in grams, M is the molecular weight of tracer gas in grams, t1 is the bombing period in seconds, t2 is the dwell time for the measurement after depressurization in seconds and V is the internal cavity volume. The FTIR method can measure leak rates up to 1 × 10 -13 Pa•m 3 •s -1 . FTIR measurements can suffer from noise and interference peaks due to reflection of the IR signal in the cavity of the package. Moreover, this test is not standardized and a gross leak test must be performed to rule out gross leaks (Costello, Desmulliez, and McCracken 2012).
The applicability of FTIR technique to determine the hermeticity of glass frit encapsulated solar cells is limited. Solar cells are frequently fabricated on glass substrates. Glass is a material with high transmittance in the visible and near-infrared spectrum up to 3 000 nm and low transmittance from 3 000 nm up to 4 600 nm (Rubin 1985). Most tracer gases used in the FTIR method have absorption peaks in the infrared region between 2 500 nm and 20 000 nm (Lellouchi et al. 2010). The small wavelength region overlapped limits the tracer gases that can be used and make the test more susceptive to noise and interference.

Residual gas analysis (RGA)
The residual gas analysis is used to measure and quantify the gas elements present inside sealed devices. This test allows to measure the encapsulation atmosphere, the moisture and oxygen content inside the device and the presence of bombing gases of previous tests. Other gases, such as CO2, CH4 and NH3, and other solvents can be identified. A mass spectrophotometer is used to analyzed all the gases (Department of Defense 2019).
Prior to the test, the devices must be preconditioned at 100 °C for 16 -24 h. The device must be transferred to the testing apparatus within 5 minutes after the prebaking step (Department of Defense 2019). The testing apparatus is composed of vacuum chamber with an opening connected directly to the mass spectrophotometer. The chamber must be heated to the preconditioned temperature of 100 °C. Moreover, the chamber must contain a piercing mechanism that can pierce the testing sample. This test method is destructive (Department of Defense 2019). The sample is placed on the vacuum chamber at 100 °C and the chamber is evacuated to a vacuum level which ensures no background interference. The package is then pierced, allowing the gas in the cavity to escape and flow into the mass spectrophotometer for analysis. A device is considered a failure if the pressure difference in the chamber during testing is greater than ± 15 %, which can indicate a non-hermetic device or the piercing was not accomplished (Department of Defense 2019). The RGA test is expensive, time demanding and requires expert analysis of the results. Moreover, the volume of the devices is limited by the testing apparatus (Costello, Desmulliez, and McCracken 2012). RGA is not appropriate for characterizing the hermeticity of packages due to its destructive nature. However, it can be used to analyze the long-term stability of the encapsulated devices. As long as the encapsulation is hermetic, the degradation byproducts of solar cells are trapped inside of the sealed cavity and can be identify by RGA. Therefore, this test method is appropriate for solar cells research laboratories (Kähler, Lofink, and Reinert 2020).

Conclusion
Hermeticity tests are used to measure and characterize the hermeticity of glass frit encapsulated perovskite solar cells and dye-sensitized solar cells devices. Devices with leak rates lower than 1 × 10 -7 Pa•m 3 •s -1 are considered hermetic. The hermeticity test are divided in gross and fine leak tests. The gross leak tests include the perfluorocarbon gross leak test, the weight gain test and penetrant dye test. These tests cannot be use to qualify the gross leakage of encapsulated solar cells due to high temperatures requirements and its destructive nature of the tests. The fine leak tests include the helium leak rate test, cumulative helium leak test (CHLT), Krypton-85 radioactive tracer gas test, optical leak test, Fourier transform infrared spectroscopy (FTIR) test and residual gas analysis (RGA). RGA is a destructive test; therefore, it is not suitable to measure the hermeticity of devices. A common limitation of the other tests is the use of a bombing phase that can produce 'oneway' leaks that normally do not exist in normal conditions. Moreover, none of the tests considers the mechanical stress applied to the devices under the bombing phase or vacuum measurement phase in the case of the helium leak rate test and CHLT. The mechanical stress limits the application of the tests to relatively small area devices (< 7 × 7 cm 2