PV durability and reliability issues

>While no abbreviated test program can predict with 100% certainty that a module will properly perform in an environment for 25+ years …

ALLEN ZIELNIK, Atlas Material Testing Technology LLC, Chicago, IL USA

While no abbreviated test program can predict with 100% certainty that a module will properly perform in an environment for 25+ years, an improved methodology that more closely simulates the real environment, while still maintaining reasonable acceleration, can certainly be a valuable tool to assess the question of survivability.

The photovoltaic industry needs PV modules that will perform for ~25 years or longer in the field, as well as a reliable means to determine that viability. These are necessary not only for product development and warranty considerations, but also in terms of risk to financial stakeholders and overall economic viability—not to mention key safety concerns.

While there are initial PV qualification tests, such as the IEC and UL requirements, among others, they are neither intended to, nor capable of, predicting long-term performance. As a result, there has been an evolution in the application of accelerated life testing (ALT) and accelerated environmental testing (AET) to the service life prediction (SLP) of PV modules and systems.

25 years and beyond

To understand reliability, we must first define what we mean. For simplicity, we will say a PV module fails to provide service if its power output decreases by more than 20% after 30 years in its use environment. Also, “a high probability” means that 95% of the modules in the field will achieve this success [1],” or the like.

For most PV products, there are many additional concerns regarding what constitutes failure or unacceptable performance. Notably, safety after aging is a critical concern to protect life and property. There may also be critical aesthetic concerns, such as with BIPV products, which affect the viability (but not necessarily the power generation) of the product. If a module discolors in the Mojave Desert, no one may care, but if it discolors on a building façade, everyone may care (at least one BIPV producer is out of business due to cosmetic failure).

At the Department of Energy “Accelerated Aging Testing and Reliability in Photovoltaics Workshop II” held in 2008, Akira Terao [2] (Sunpower Corp.) pointed out that, even in the mature c-Si module arena, there are still many remaining reliability challenges, including:

  • 25-year warranty. This is still a barrier; how does a company prove 25-year life?
  • Ill-defined field conditions. The same warranty must apply for all conditions for the same modules.
  • Harsh and varied outdoor conditions.
  • Materials used near their limits; how to accelerate the effects on material already being used near its limit.
  • Limited acceleration factors. There are few available; industry must rely on long tests instead. Long test time can be a hindrance to market introduction.
  • Cumulative effects, positive feedback loops. Challenging to determine and test for all interactions in the field.

The best approaches to reliability engineering include using the standard Weibull bathtub curve to determine the physics of failure for each mode.

According to John Wohlgemuth (BP Solar), “Today, BP Solar offers a 25-year warranty on most of its crystalline silicon PV modules…while the modules have to last for 25 years of outdoor exposure, we cannot wait 25 years to see how they perform… no BP/Solarex module has been in the field longer than ten years. Even the oldest 20-year warranty modules have only been in the field 15 years.”

Wohlgemuth adds, “Examples of accelerated stress tests of use for PV include:

  • Thermal cycling;
  • Humidity-freeze;
  • Damp heat;
  • Mechanical load both static and dynamic, and
  • Ultraviolet exposure [3].”

Figure 1. Failure rate vs. lifetime.

Reliability concerns associated with PV technologies

Quoting directly from a monograph [4] by Dr. Sarah Kurtz of NREL:

“General reliability issues across all PV technologies are:

  1. Corrosion leading to a loss of grounding
  2. Quick connector reliability
  3. Improper insulation leading to loss of grounding
  4. Delamination
  5. Glass fracture
  6. Bypass diode failure
  7. Inverter reliability
  8. Moisture ingress

Continuing, Dr. Kurtz said: “In addition, there are issues specific to the individual technologies, to name a few:

i. Wafer silicon: Light-induced cell degradation, front surface soiling, effect of glass on encapsulation performance, reduced adhesion leading to corrosion and/or delamination, busbar adhesion degradation, junction box failure;

ii. Thin film silicon: electrochemical corrosion of SnO2, initial light degradation;

iii. CdTe: interlayer adhesion and delamination, electrochemical corrosion of SnO2:F, shunt hot spots at scribe lines before and after stress;

iv. CIS: interlayer adhesion, busbar mechanical adhesion and electrical, notable sensitivity of TCO to moisture, moisture ingress failure of package; and

v. OPV: photolytic instability, moisture induced degradation, moisture ingress failure of package.”

We often hear that people interpret passing the IEC 61215 or 61646 qualification tests is proof that a product has been tested and shown to be durable and reliable. This is simply not true; like many ALT tests, the IEC environmental stress test protocols are designed primarily to test the infant mortality period of the above-referenced bathtub curve (Fig. 1) and do not adequately or realistically stress a module in the way that nature does.

During life, a product loses performance attributes according to the accumulated damage model. Continued damage (thermal, photolytic, mechanical, hydrolytic, etc.) inflicted over a long time—and influenced by the daily and seasonal diurnal cycles—takes a toll.

According to Wohlgemuth, “While qualification tests are important, they have limitations because the stress levels are, by design, limited… so, passing the qualification test means that the product has met a specific set of requirements, but doesn’t say anything about which product is better for long-term performance, nor does it provide a prediction of product lifetime [3].”

Reliability vs. durability

Classic reliability testing is primarily concerned with measuring outright failure, such as time to failure, mean time between failures, number of failures per “n” units or operations, etc. A variety of accelerated life test methodologies (ALT, highly accelerated life testing [HALT], highly accelerated stress screening [HASS], etc.) are used in this pursuit.

The general methodology is to apply higher (sometimes considerably so) levels of stress than actual use conditions—but over a shorter period of time—to try to predict longer term performance. This methodology carries the caveat that it may induce failures that would not naturally occur, but can be very useful for studying failure modes and product robustness.

Another ALT approach is to use near-normal stresses but applied over a much shorter time period (such as cycling of a door hinge). The IEC tests predominantly use the first approach; for example, the 85°C/85%RH damp heat test condition is not very realistic in terms of actual service. These ALT approaches work best for mechanically-related single-stress failures, and response is often non-linear for chemically induced changes or where degradation is dependent on sequential or overlapping mechanisms.

Durability testing, however, is primarily concerned with realistically stressing products to predict long-term performance and is concerned with routes to failure (mechanisms), rates of performance, or property loss, etc. A loss of material or product durability may lead to catastrophic failure (i.e, loss of reliability). Another critical aspect of “durability to the environment” testing is that weather and climate have multiple inter-related stressors varying continuously in both short- and long-term patterns—something extremely difficult to reproduce in basic test equipment.

What is usually referred to in photovoltaics as “reliability,” such as no more than 1% power loss per year to a maximum of 20%, is actually a durability issue.

A sequenced approach

Many other industries, such as automotive and building products, have long learned and developed a general testing approach to weather durability testing. While lifetime expectations or product complexity may be less than that for PV, the needs are the same.

The progession is:

  • Material-level tests to select suitable products;
  • Component-level tests to include processing variables and some material-material interactions; and
  • Product-level tests to test final design and manufacturing, including transportation and installation.

To assess PV module durability, we usually start with the design failure modes and effects analysis (FMEA; or FMECA—failure modes, effects, and criticality analysis), and add a materials-level analysis. This helps to understand unique potential failure modes as well as potential test bias, both for optimizing the test methodology and for interpreting test results.

Next, materials-level tests are undertaken, especially for any polymeric materials (such as connectors, encapsulants, or topsheets) to determine their durabilities. We progress to component-level or pre-production unit testing, if warranted. Such testing can be helpful in determining processing variables, such as laminating conditions, or detecting basic problems, such as edge sealing or adhesion issues. The next step is full-module (or product, such as BIPV) testing.

An improved testing regimen

The primary module durability testing to date has simply relied on extending the existing IEC qualification tests to longer duration (e.g., more hours or more cycles). From our perspective, involving extensive work with PV materials and module manufacturers (as well as other industries) over the past years, and with millions of various products and materials on exposure in labs and test sites around the world, this approach has several flaws.

Although to study or force failure modes can have value for ALT testing, it isn’t a realistic approach for estimating long-term weather durability. It limits the number of simultaneous stresses (such as temperature cycling without humidity or solar radiation), uses stress levels not representative of specific end-use climates, and fails to deliver the stresses in the complex short- and long-term cycles of the natural environment. In weathering testing, the rule is: do a different test (than nature), get a different result.

To improve weather test modules to predict the likelihood of 25+ years durability, Atlas has been developing an improved test methodology. The fundamental characteristics are:

  • Parameters are selected based on three major PV use climatic zones: arid desert, tropical/subtropical, and northern/temperate, plus one additional global composite of all three sets of boundary conditions.
  • Additional test modifiers of urban/industrial (e.g., hydrocarbons, soot); windblown dust/dirt; acid rain; mechanical loading; and coastal/marine.
  • Combining multiple stresses into single test cycle modules, such as combined temperature and humidity cycling and humidity freeze with solar radiation to better simulate the natural environment.
  • A series of test sequences, combining corrosion, condensing humidity, thermal/humidity/freeze/solar, outdoor solar tracking, UV preconditioning, etc.—all performed on the same module.
  • Periodic visual inspections, I-V curve measurements, and thermal imaging.

Utilizing a variety of accelerated environmental testing (AET) devices and techniques, at realistic climate-specific stress levels and delivered in cycles that mimic the natural environment, this methodology can be run prior to, concurrent with, or following the IEC qualification tests, and is available now.


Service life prediction of complex products is an evolving discipline that often requires data and techniques that are not available. But while this methodology does not purport to address true SLP, it does have a 95-year foundation in empiricism shown to provide practical results. While no test program can predict with 100% certainty that a module will properly perform in an environment for 25+ years (except for real-time 25 year testing, of course), an improved methodology that more closely simulates the real environment while still maintaining reasonable acceleration can certainly be a valuable tool to answer the question, “Will my module last 25+ years?”


1. T. McMahon, G. Jorgensen, R. Hulstrom, “Module 30 Year Life: What Does it Mean and Is It Predictable/Achievable?,” National Renewable Energy Laboratory, Reliability Physics Symposium, 2008 (IRPS 2008); IEEE Inter., April 27, 2008-May 1, 2008 pp.: 172–177.

2. U.S. Department of Energy, Accelerated Aging Testing and Reliability in Photovoltaics Workshop II, Summary Report, April 1 & 2, 2008.

3. J. Wohlgemuth, “Reliability of PV Systems, Reliability of Photovoltaic Cells, Modules, Components and Systems,” edited by Neelkanth G. Dhere, Proc. of SPIE ,Vol. 7048, 704802-1, (2008).

4. S.Kurtz, “Reliability concerns associated with PV technologies,” www.nrel.gov/pv/performance_reliability/pdfs/failure_references.pdf

Allen Zielnik received his associates degree in electronics engineering from DeVry U. and BS in chemistry from Michigan State U., and is a senior consultant within the Solar Energy Competence Center of Atlas Material Testing Technology, 4114 North Ravenswood Ave., Chicago, IL 60613 USA; 773-327-4520; info@atlas-mts.com


More Photovoltaics World Issue Articles
Previous articleModule pricing: Rational, or just plain nuts?
Next articleLeading Edge

No posts to display