The Powerhouse: Inside the Invention of a Battery to Save the World (27 page)

T
he day after the deal closed, Kumar received an e-mail from Damon Frisch, his GM program officer. Frisch would arrive in Newark two days later—on Monday. He would have his entire team with him. They would start work immediately.

The idea was to begin with the launch date—2016—and work backward. Over the following two days, Frisch, Kumar, and their teams would go through every major step of the laboratory process until they had a meticulous schedule of milestones. In just twenty months, this time line would deliver the central nervous system for the electric that Dan Akerson had promised his employees.

Such painstaking planning was vital from GM’s perspective because most cars did not earn large margins for their manufacturers. If GM was to profitably make electrics, there was very little room for error. Kumar said that to deliver the required performance, he would have to start with a better understanding of his materials, using equipment he did not currently possess. He would require access to either synchrotron radiation or neutron diffraction capability, either of which would permit him to peer deeply into his battery. Only the national labs had this capacity—the beam lines at Argonne, Lawrence Berkeley, and Oak Ridge. Kumar would need to obtain beam time and to do so absent the usual three- and six-month-long waiting times.

It would help if GM finally announced the deal. That would supply Kumar the credibility to enter into a conversation with the labs about beam time—to request and be granted favors from them. He thought the labs should waive the usual fees given the industrial and corporate stakes. “I feel this should be a national mission,” Kumar said. The national labs should treat the NMC 2.0 and the silicon-carbon anode as priorities. “Why shouldn’t they work on something that has an American company and is to America’s advantage?” he said. “Why shouldn’t they analyze my material as opposed to publishing many, many papers that are fantastic scientifically but don’t help American industry?”

The GM and Envia teams endorsed the schedule. Now Kumar embarked on the dizzying sequence of milestones. He assumed that disclosure of the deal would follow more or less immediately in order to provide him some negotiating traction with the labs.

Kapadia figured the deal would be made public by early January at the latest. He reckoned on a resurrection of the previously discussed public ceremony; seeing as how the two-hundred-mile car would validate Obama’s goals set out three years before, the White House itself might participate in the news conference. That would help his own efforts. As Kumar advanced on the GM contract, he moved forward on the plan to allow Envia investors to cash out. There no longer seemed to be time to launch an IPO. It again seemed much more sensible to find a buyer. Potential American purchasers remained cautious, so Goldman Sachs took the acquisition proposal to Asia, sending queries to, among other suitors, Japan’s Asahi Kasei and South Korea’s LG and Samsung. These companies had to feel that time was of the essence—that Envia could be swept up any moment by a rival. The GM deal would help create that impression.

But GM again said the announcement had to wait. There was an electronics show the first week of December, which was not a good time for the disclosure. Then the Detroit auto show—also not ideal. After that, Envia, LG, and GM were to be occupied in a brainstorming session in Detroit. That would push back the announcement to the third week of January. Then there was the second Obama inauguration. Which made the latest plan to wait until the next ARPA-E Summit, in February. There would be something poetic about that date given the announcement at the last Summit.

GM was not announcing the deal but neither was it moving back its time line. The deadlines to deliver the working battery remained etched in stone. “I feel pressured,” Kumar said. He had to find and hire a structural cathode scientist, someone like Jason Croy who was familiar with the beam lines. Since A123 had declared bankruptcy, Kumar figured that he could raid it for at least some of the ten or fifteen scientists he required. He would place an ad and start making calls.

The ARPA-E Summit passed without an announcement. But toward the end of February, Hari Iyer received an urgent e-mail from General Motors: it had finished initial tests on Envia’s 400-watt-hour-per-kilogram battery. So far, they had failed to replicate the results announced at ARPA-E. No remedy seemed to work. Could Envia explain why?

Increasingly concerned queries piled up during routine conference calls and meetings with the GM team over the next week. The cell was not reproducing the ARPA-E results. It was supposed to be delivering 400 watt-hours per kilogram, and it wasn’t—not by a long shot. Iyer asked Kumar, who shrugged off his questions. Iyer approached members of the scientific team. Finally he went back to Kapadia. Iyer said, “Perhaps what we told the world and GM isn’t what it is.”

Kumar’s technical team in fact had an inkling of what GM was finding. Two weeks after the ARPA-E Summit the previous year, they had sent another cell to Crane for evaluation. When the result was returned to Envia, it verified the 400-watt-hour-per-kilogram claim—but only on the second cycle. After that, the energy density plummeted. By the fifth cycle, energy density was down to 302. On the hundredth, it was at 267. By the two hundredth, it had dipped to 249, and on the 342nd cycle—the last listed—energy density was at 232. The cell had lost 42 percent of its energy and showed no sign of stabilizing.
1
But the GM team seemed not to achieve even two cycles of the 400-watt-hour-per-kilogram performance.

As part of his explanation for the battery’s problems, a staff scientist had told Iyer something unnerving. Envia, he said, was
not
the inventor of the 400-watt-hour-per-kilogram battery touted at ARPA-E. Not of the entire technology, anyway. The NMC cathode was legit—Envia, as it had told everyone, had transformed the Argonne electrode into an optimized, top-of-the-line product. But not the silicon-carbon anode—the flaunted component that lifted the performance of the battery as a whole was not Envia’s invention.

A year before, cofounder Mike Sinkula had told Kapadia that the anode contained material from a Japanese supplier. Kapadia had dropped the subject after it was waved away as unimportant by Purnesh Seegopaul, the materials scientist on the board of directors. But this fresh claim, heard as GM was raising a fuss, assumed a different meaning. It could not be ignored.

Iyer again confronted Kumar, who said that what he heard was wrong—the anode
was
Envia’s. He was shipping out the anode for treatments such as chemical vapor deposition, the first of five discrete production processes to which the electrode was subjected. The anode was going to Japan for that purpose. It was all routine stuff.

The explanation made sense—much went into the creation of an advanced electrode. If an outside supplier painted a GM car, for instance, no one would challenge the assertion that it was in fact still a GM car. So it was with the anode. Kumar said Envia did the work that mattered.

Kumar said that the most pressing problem was not the anode, but how much voltage GM insisted on applying to the battery. The greater the voltage, the more lithium began to shuttle between the electrodes and the farther a car could go on a single charge. That was the idea behind activating the NMC at 4.5 volts and thus transforming it into NMC 2.0. But such higher voltages also stressed the material and made the Argonne battery fail—it was what caused voltage fade. Kumar wanted to stay well away from the stress limit. For ARPA-E, he had pushed the voltage as high as he dared—to 4.3 volts—and at that level managed to cycle the battery 450 times; to meet GM’s specs, he would have to more than double that to 1,000 cycles, but Kumar was confident he would succeed.

Only, after the deal was signed, when the GM and Envia engineers sat down to map out the work to come, the carmaker surprised him by insisting on the application of
4.4 volts
. While that tenth of a volt may have seemed only a tweak, its impact was magnified on the atomic scale. At that state of charge, atoms began to move around at an accelerated pace, the cathode expanded and contracted with the shuttling of the lithium, and ultimately the material could crack. You began to lose electrical contact between the particles that make up the cathode. Kumar began to achieve at best three hundred cycles, much further from the thousand that he needed. Now his team had not to double the battery’s longevity, but to triple it.

Kapadia remained suspicious. On March 7, 2013, he briefed the Envia board about their primary customer’s unhappiness with the signature product and the doubts about the anode. In an e-mail, he said, “GM has observed a significant and very large disparity between the data obtained from Arpa-E cells and proof of concept cells based on [the] 400 wh/kg technology we sent them for testing.” Until the performance questions were resolved, there should be no further discussion of any news releases or contract ceremonies, Kapadia said.

A week later, Kumar held a heated meeting with Iyer. Kumar disparaged Kapadia for putting his concerns in writing, which made them legally “discoverable”—any potential buyer of Envia would have to be provided such internal correspondence during the due diligence process. It was a bizarre outburst. Iyer said, “If there were no misrepresentations, you should not worry about Atul putting these issues in writing.”

Kumar broke down and told the truth: the anode—the one used in the ARPA-E battery and shipped a year later to GM—hadn’t been serviced in Japan. It had been bought there. What Iyer heard was true. The anode was made by a company called Shin-Etsu.

But creating an anode was only the first step, Kumar continued to say. This being silicon, you had to carry out many additional procedures in the lab in order to make the anode usable. It required treatments, coatings, and nano-processing. That was why he regarded the anode as Envia’s. It was analogous to the Argonne cathode. Because Envia had optimized it, Kumar could legitimately call it proprietary.

Kumar asked Iyer to prevail upon Kapadia to withdraw his allegations about the anode. If Kapadia would do so, Kumar would apologize and make peace with the CEO.

It was a bewildering admission, and to Kapadia an altogether baffling justification when Iyer reported it to him. For two years, Envia had trumpeted its work on silicon. It had done so to potential customers, in particular GM, to peers at conferences, and to the world at large. The ARPA-E announcement was only the most-noticed version. Now one had to wonder what was true. Kumar had it wrong, Kapadia said. If Envia said that the anode material was proprietary, as it had done in its license with GM, then it had to have created the anode. If it was someone else’s, Envia was duty-bound to explicitly disclose that fact. The anode, plain and simple, was not
Envia’s IP. Absent the anode, Envia was no better than the crowd. Panasonic, Samsung, and Argonne itself all were delivering energy density of 215 or 220 watt-hours per kilogram using a standard graphite anode. Kumar’s NMC cathode combined with a graphite anode—unactivated so as not to trigger the voltage fade—was
not
superior, not by more than a few percentage points anyway. Kapadia himself was at fault for championing the battery for so long despite the countervailing data and the complaints. But now he was going the other way: Envia, he said, should come clean in order to try to salvage the relationship with GM.

The Envia executives and board skirmished between and among themselves on what to do. Kapadia and Iyer insisted that the board allow them to go to GM and fess up. But the directors ordered a brave face and continued work. To Kapadia, Seegopaul seemed less concerned about the anode’s origins than that its discovery could upset the chances for his venture fund to cash out.

At the end of March, the two companies—GM and Envia—assembled for their first quarterly review. GM’s contingent was led by Matthus Joshua, head of purchasing for the company’s hybrid program, and Larry Nitz, its head of hybrid engineering. The GM men were blunt. In addition to the shortcomings of the ARPA-E material, Kumar had missed his very first milestone—the provision of one hundred kilograms of NMC cathode powder for the 2015 Volt. They asked Kumar what was going on.

Kumar responded that the problem with the cathode powder was temporary. There had been “a cross-contamination issue” and the Volt commitment would be met. The GM men were “dismissive [and] mocked Kumar,” Kapadia said later. They demanded that he reproduce the ARPA-E results.
2
Kumar had lost credibility. This was not the friendly repartee of prior company-to-company talks.

As for Kumar’s story about chemical vapor deposition, the GM men said that if that was so, they wanted to see a full accounting of materials used in the anode.

Kumar conceded the true story as he had earlier with Iyer. With that, Kumar’s painstakingly presented picture fell apart. The vaunted silicon-anode performance was at best the flawed, undisclosed work of a Japanese American composite.

What was the answer for the GM men? One could be angry at what seemed a possible case of flimflammery. But from their perspective, how would that enable the vehicles on which a high-profile part of GM’s future—not to mention perhaps their own jobs—depended? If Kumar could step up and deliver the ARPA-E performance, GM could sort out the licensing of the Shin-Etsu anode, even if they’d come to it in roundabout fashion.

Joshua and Nitz gave Kumar another quarter. It was no longer important who had invented the silicon-carbon anode. The numbers were the crucial issue. If Kumar could produce the metrics boasted of at the Summit, all could be forgiven.

One thing was certain—none of this, nor the deal itself, was to go public. Nitz said, “The next cell—build or bust” for Envia.

Kapadia’s aggressive outing of the anode seriously strained his relationships within the company. He and Kumar now barely spoke. The board instructed him to restrict his activities to selling the company and resolving a lawsuit by Kumar’s former employer, NanoeXa. The previous month, NanoeXa CEO Michael Pak accused Kumar in a court claim of stealing intellectual property in order to create Envia’s cathode. Kapadia said it was a nuisance suit. But even a nuisance suit could complicate any sale of the company.