The grid battery fire in Arizona last year ended with an explosion, but disagreement has emerged on how exactly it began.

Owner Arizona Public Service recently concluded that a defect in a single lithium-ion battery cell caused it to heat up, triggering a chain reaction that destroyed a whole rack and released gases that later exploded. But LG Chem, the manufacturer responsible for the battery cells and modules, published its own report with a different conclusion.

According to LG Chem’s "progress report," compiled by scientific investigation firm Exponent, the physical evidence does not support the battery defect theory. The report, filed with Arizona's utility regulators last month, instead theorizes that external causes heated up the battery.

At stake is the reputation of a long-time leader in the lithium-ion manufacturing space. Batteries are known to degrade over time, sometimes forming spindly metal deposits called dendrites, which can eventually reach across a cell and cause a short. When a short circuit heats up, it can trigger thermal runaway. But for LG Chem's cells to exhibit such behavior in just two years — the age of the battery system in Arizona when the incident occurred — would constitute a manufacturing failure.

“The APS event is well known, so when a report gets published as the final report, that does generate a lot of questions,” said Peter Gibson, vice president of energy storage at LG Chem, in a recent interview. “We felt a need to make it clear that the report was not consistent with our perspectives.”

APS declined to comment on LG Chem's methods or findings.

"We stand by the investigation we led into the April 2019 McMicken event and the independent report we commissioned to compile expert analysis into our final findings," spokesperson Jenna Rowell said in an email. "The report we released reflects the lessons we learned and will guide APS’ future actions with battery energy storage."

The utility's new requirements should rule out the explosive outcome, even if another battery cell heats up. In that sense, whether or not a particular cell kicked off the McMicken disaster has little bearing on the safety of future projects. 

The stakes of the conflicting report have more to do with LG Chem's reputation as a responsible battery manufacturer, and the industry's understanding of the risks of lithium-ion battery technology.

Conflicting explanations of the root cause

APS brought in energy consultancy DNV GL to write its report; that firm reviewed the forensic investigation findings to compile its conclusions. Exponent, a firm that specializes in investigating battery failures, conducted its own testing.

The wreckage posed an epistemological problem for investigators because the cell at ground zero for thermal runaway and those around it burned up in the process. Without conclusive evidence from the source, investigators had to look to fragmentary clues to piece together what happened.

Cells taken at random from elsewhere in the battery system, and from its twin system at Festival Ranch, showed “lithium metal deposition and abnormal dendritic growth,” DNV GL notes.

"Because the evidence of Lithium metal deposition and abnormal dendritic growth was sufficiently present in the random samples that were analyzed, it was determined to a reasonable degree of scientific certainty to be the anomaly that caused the initial cell failure and ensuing thermal runaway," battery safety expert Davion Hill wrote in the APS report.

But LG Chem asserts that the investigation needed to consider other explanations. “A cell failure can be caused by either a short circuit or heating from an external source,” Gibson said. “Any serious analysis needs to consider both.”

LG Chem tested one of those cycled cells with deposits similar to those DNV GL examined and found that it did not conduct electricity and thus was unlikely to be pure lithium metal.

“It is impossible for a non-conductive deposit to establish an internal cell short circuit, carry current, resistively heat and cause thermal runaway,” Exponent's report says.

However, Exponent also notes that some of the deposit samples reacted in air in manners consistent with lithium metal. "It is difficult to reconcile the facts that the deposits have been shown to be non-conductive and that they have reactivity in air that is similar to that of conductive lithium metal," the report explains.

Extreme lithium plating, of the sort that could cause a short circuit, necessarily reduces the cell’s ability to store energy, which would lead to a decline in performance. But it's not clear if a reduction in one cell's performance could show up in the data logs. The Exponent report includes a section on "Failure Data Limitations" that shows how McMicken's data logging did not capture a granular picture of what transpired at the cell level.

Exponent also found physical signs of electrical arcing in the battery rack, which could have provided an external heat source to initiate thermal runaway. It found that the damage on the batteries indicates an "attack" by an external heat source.

These observations raise questions about the timeline that may be impossible to resolve. If a metallic deposit appeared non-conductive when LG Chem examined it, were all similar deposits also non-conductive when they first formed or when the thermal runaway began? Did the electrical arcing that left marks on the rack happen right before the thermal runaway or in the chaos that ensued?

Reputational damage control

Even if observers did blame LG Chem’s cell for starting the fire, it’s not clear that this would seriously affect the company’s prospects.

Both reports agree that the real damage came from other factors, namely, the propagation of thermal runaway and the accumulation of unvented explosive gas. Battery developers recognize that individual cells sometimes fail; the industry since has made headway in designing systems to limit the damage of a single cell failure.

LG Chem has dealt with fires at facilities it supplied before. After South Korea paid out a generous subsidy for battery projects, business boomed, and then a spate of battery fires broke out. Around a dozen of them occurred at projects supplied by LG Chem, Gibson said.

“In Korea, there were an awful lot of systems developed by relatively inexperienced system integrators,” he said.

Some projects revealed fundamental design problems, evidenced by things like rust and water leakage. After investigating the problem, the Korean government recommended limiting the maximum state of charge. At that point, LG Chem proactively exchanged battery modules from that era as an additional safety precaution, Gibson said.

“If we see something where there’s really a need to do a retrofit to enhance safety…we are very proactive,” Gibson said. "It happens very rarely."

That episode did not dislodge LG Chem from its Tier 1 cell supplier status, nor did it jeopardize the business of supplying electric vehicle batteries to automakers like GM. LG Chem's automotive battery business is roughly 10 times larger than its stationary grid storage business, Gibson said.

Whatever happened at McMicken, it hasn’t shaken the trust of Fluence, which bought the cells from LG Chem and integrated them into the system for APS.

When asked if Fluence still buys from LG Chem, COO John Zahurancik said, “We work closely with multiple top-tier battery providers, including LG, to ensure the safety and quality of battery cells used in our systems. Under no circumstances would we ever consider using battery technology we did not believe could be deployed and operated safely.” 

Even after the McMicken fire, LG Chem cells found a home in LS Power's Gateway storage facility, the largest lithium-ion battery in the world. LG Chem's safety standards convinced LS Power that it was a good choice to supply 250 megawatts for the Southern California plant, which began operations this summer.

Other chemistries rising

A broader shift in customer preference is underway in the grid battery sector, however.

The stationary storage industry got an early boost from piggybacking on the supply chains for the electric vehicle industry. The energy-dense nickel-manganese-cobalt-oxide (NMC) batteries designed for cars doubled as powerful batteries for the grid.

But the APS fire strikingly illustrated the explosive potential of those NMC batteries. Another chemistry, lithium-ferrous-phosphate or LFP, has gained ground due to a reputation for safety and lower costs, although it packs less energy density.

“The safety question has been one of several reasons why the energy storage market has given the Chinese LFP vendors a harder look,” said Daniel Finn-Foley, energy storage director at Wood Mackenzie. “As they’ve given them a second look, they’re liking what they see, and it’s moving LFP vendors toward a market-leading position.”

Indeed, NMC’s market share for grid storage has already peaked and will continue to decline as LFP gains ground, according to new research from Wood Mackenzie. LFP supplied just 10 percent of the market in 2015, but its share will reach 30 percent by 2030.

That shift elevates manufacturers in China that weren’t considered Tier 1 in the early days of U.S. grid battery development. But the trend does not necessarily mean LG Chem will be left behind.

The South Korean company maintains a sizable roster of technical talent to develop and commercialize new chemistries as the market demands. LG Chem has produced LFP in the past, Gibson noted, but he says he is "still very much of the opinion that NMC is preferred" due to its power density and performance characteristics.

If the market shifts as the analysts predict, it will coincide with meteoric growth in storage deployments. That means NMC would serve a smaller slice of a much larger pie than today's grid storage market.

This is the second in a series of exploring the repercussions of the most prominent battery fire in recent U.S. history. The first installment, on how the storage industry has already improved its safety procedures, is available here.