On the morning of December 29, 2008, Sheri Sangji began a routine chemical reaction to synthesize an experimental anti-obesity drug. Seventeen days later, she was dead, having succumbed to third degree burns she suffered when the reaction that morning went terribly wrong. The consequences of this tragic accident—including criminal charges against Patrick Harran, in whose lab Sangji was working when she died, and UCLA, where Harran is a professor—have led to a re-examination of safety training and protocols across the country and, in particular, at the ten campuses of the University of California.

Anyone who has worked in a lab at Berkeley has almost certainly experienced some aspect of the UC system’s existing safety program, be it an instructional video on safe laboratory techniques, an online tutorial in the disposal of hazardous materials, or radiation safety training. These measures can sometimes feel overly laborious and time-consuming, provoking questions about whether such training is effective or even necessary. Sangji’s death has brought these questions into sharp relief within the UC system, and efforts to overhaul safety programs in the wake of the accident have illuminated the murky nature of laboratory safety, particularly in the academic environment. When, for example, do lab coats and safety goggles need to be donned? When should a new graduate student be considered sufficiently experienced to perform dangerous procedures independently? Who is responsible for training new researchers, and who is culpable when things go wrong? Recent changes to UC laboratory safety protocols have begun to clarify existing policies, fill in gaps in safety training, and, officials hope, further diminish the likelihood of future accidents.

Click to view data visualization from an international survey on safety attitudes and practices

“Step 1: generate vinyllithium via…”

Sheharbano “Sheri” Sangji was an ambitious young woman with diverse hobbies and an eye towards a career as an attorney. But it was chemistry that captivated her while studying at Pomona College, and she became actively involved in academic research, co-authoring papers in Organic Letters and the Journal of the American Chemical Society as an undergraduate. After graduating in 2008 with a degree in chemistry, Sangji briefly worked as a synthetic chemist at a pharmaceutical company in Los Angeles before being hired as a research associate by Professor Patrick Harran, a talented synthetic organic chemist in the Department of Chemistry and Biochemistry at UCLA.

On December 29, 2008, Sangji was preparing to produce 4-hydroxy-4-vinyldecane from 4-decanone. This reaction first requires the synthesis of a nucleophile, vinyllithium, from vinylbromide and tert-butyllithium (tBuLi). The banality of these chemical names belies the danger of this reaction, as tBuLi is a highly reactive chemical that spontaneously ignites when exposed to air. The reaction was not a new one for Sangji, and she had previously completed the reaction on October 17, four days after her first day in the lab. On the day of the accident, though, Sangji had to scale up the reaction to roughly three times the original volume. This required more than 150 milliliters of tBuLi for the vinyllithium synthesis.

Credit: Notes from Sangji’s lab notebook on the day of the accident that took her life. The second line illustrates the preparation of vinyllithium from vinylbromide and tBuLi, the pyrophoric chemical that ignited during the experiment and led to Sangji’s death.

tBuLi has the chemical formula (CH3)3CLi and contains a highly polarized lithium carbon bond. It reacts violently with water and is flammable in air, requiring that it be stored and handled in an inert atmosphere at all times. Experiments involving tBuLi therefore require special equipment and techniques that are unnecessary when working under ambient atmospheric conditions. Harran’s laboratory did not have a specific standard operating procedure (SOP) for transferring tBuLi and instead followed the general procedure outlined in a technical bulletin written by Sigma-Aldrich, a major chemical and biochemical products company. Among other details, this document recommends that a syringe not be re-used for multiple transfers and that long, one or two-foot needles be used in place of short needles.

A gas bubbler is sometimes used when working with tBuLi to maintain an air-free, inert environment.

In spite of the guidelines, a detailed investigation revealed that Sangji disregarded, or was unaware of, most of the safety measures advised for tBuLi transfers. Instead of the recommended glass syringe, Sangji selected a 60-milliliter plastic syringe to transfer the 150-milliliter volume in three steps, requiring multiple uses of the same syringe and the use of applied force to draw up the liquid. This method also neglected the so-called “two times rule,” which advises using a syringe with a capacity at least twice the volume of the intended transfer to minimize the risk of inadvertently pulling the plunger out of the syringe barrel. Most significantly, Sangji used a syringe with a two-inch long needle instead of a syringe with the recommended, and much longer, one or two-foot needle. The short needle would have prevented Sangji from easily accessing the reagent at the bottom of the bottle, and investigators believe that she either held or tilted the bottle with one hand while pulling the syringe plunger with the other, a precarious method for handling such a dangerous chemical.

Bottles of pyrophoric chemicals such as tBuLi are often sold in “sure seal” bottles that keep them from being contaminated with oxygen or moisture in the air.

The accident occurred when the plunger became dislodged from the syringe while Sangji was drawing tBuLi into the syringe. The chemical spilled onto Sangji’s hands and clothes and immediately caught fire. Sangji was not wearing a protective laboratory coat at the time and may not have known the location of the lab’s safety shower, because, witnesses reported, she did not run towards it in the moments after the accident. Fortunately, another researcher was nearby, and he wrapped his lab coat around Sangji in an attempt to extinguish the flames. After this proved unsuccessful, he was finally able to put out the flames by pouring water on her from a nearby sink. Sangji suffered third degree burns over more than 40% of her body and spent eighteen days in a burn center before she died on January 16, 2009 as a result of her injuries.

A preventable death

Because many graduate students and postdoctoral researchers are not legally employed by their universities, state and federal agencies often don’t have the jurisdiction to investigate accidents that occur in academic labs. Indeed, many lab accidents are never reported to the government. This was not the case with Sangji’s accident, however, because she was an official employee of the university. The California Division of Occupational Safety & Health (Cal/OSHA) began an investigation soon after the incident and released a ten-page report on May 4, 2009, in which they concluded that Sangji had not been properly trained and fined UCLA more than $30,000 for three “serious” violations of workplace safety laws. UCLA accepted the findings in a press release, and outlined the changes it had made on campus to improve laboratory safety.

Around the same time, Harran published a statement in the Los Angeles Times in which he addressed the accident and Cal/OSHA report. While stressing that Sangji was “an experienced chemist and published researcher” who had worked with pyrophoric chemicals before, he acknowledged that “based on mistakes investigators tell us were made that day, I underestimated her understanding of the care necessary when working with such materials.” Notably, Harran stated that Sangji’s death was the result of a “tragic accident,” and that his lab’s safety inspection and training protocols under scrutiny “are largely unrelated to the accident.”

The structure of pyrophoric tert-Butyllithium (tBuLi) is represented by two resonance structures shown above. The strongly polarized lithium carbon bond allows tBuLi to be used for reactions exchanging lithium with halogens, as Sangji did to create vinyllithium.

tBuLi can react violently with moisture in the air (H2O) to produce flammable isobutane gas and heat, which may ignite in the presence of oxygen. tBuLi also reacts violently with oxygen and carbon dioxide to produce large amounts of heat that in some cases can lead to spontaneous combustion.

Did Sangji’s previous experience working with dangerous chemicals absolve Harran and UCLA of the responsibility to provide the proper safety training? Were documented lapses in safety protocols in the Harran lab relevant to Sangji’s death? In the wake of the first Cal/OSHA report, many called for a more serious investigation into these questions and the potential culpability of UCLA and Harran himself. Sheri Sangji’s sister, Naveen Sangji, released a statement expressing the hope that a more thorough investigation would occur and asking for public support in a “quest for justice and reform.”

This plea did not go unheeded. In December of 2009, Cal/OSHA released a second, more exhaustive, 95-page investigative report that provided a highly detailed account of the accident, of the pre-existing safety culture in Harran’s laboratory, and of Sangji’s training prior to the incident. The report’s conclusions were damning, both for Harran and for UCLA. According to Brian Baudendistel, the senior special investigator who authored the second Cal/OSHA report, “laboratory safety policies and practices utilized by UCLA prior to Victim Sangji’s death were so defective as to render the University’s required Chemical Hygiene Plan and Injury and Illness Prevention Program essentially non-existent.” In addition to noting the “systemic breakdown of overall laboratory safety practices at UCLA,” the report highlighted the University’s failure to “correct recurring hazards to employees,” and blamed the Environmental Health & Safety (EH&S) Department for neglecting to “abate a rather clear and appreciable danger.”

The report also directly condemned Harran. Having documented the numerous ways in which Sangji’s techniques on the day of the accident diverged from accepted protocol, Baudendistel called Harran’s contention that Sangji had been adequately trained in his laboratory to work with tBuLi “nothing short of incredulous.” The report also blamed Harran for failing to provide Sangji with a lab coat, and for neglecting to enforce general use of personal protective equipment in his laboratory. In closing, Baudendistel declared that “if Harran had utilized a standard operating procedure as required and would have properly trained Sangji…[her] death would have been prevented.”

The People v. The Regents and Patrick Harran, Defendants

Two years later, on December 27, 2011, the Los Angeles District Attorney’s office acted on the recommendations of the second Cal/OSHA investigative report and filed three felony charges against the Regents of the University of California and against Harran himself. UCLA risked fines of up to $1.5 million for each of the three counts, and Harran, for whom an arrest warrant was issued, faced up to four and a half years in state prison. Harran became the first professor in the United States to be charged with a felony for his role in a worker’s death.

The news came as a shock to the academic research community, and especially to UCLA, since the first Cal/OSHA report found “serious” but not “willful” violations on the part of UCLA. In an official statement, they called the district attorney’s criminal charges “truly baffling,” particularly in light of the dramatic increase in laboratory inspections and rigorous safety standards that were established by UCLA following the incident.

UCLA didn’t need to wait for the jury, nor did they continue to maintain their innocence. In July of 2012, the district attorney struck a plea deal with the Regents of the University of California contingent on the Regents’ continued cooperation with Cal/OSHA, their promise to establish a $500,000 scholarship in Sangji’s name, and their formal acceptance of responsibility. Harran, however, was not party to the Regents’ settlement and remains on trial. The prosecution seeks to demonstrate that he willfully neglected to train Sangji and to enforce proper use of personal protective equipment in his laboratory. Is he indeed accountable? According to UCLA’s Chemical Hygiene Plan, it is the responsibility of the principal investigator to “continually educate all laboratory personnel on the potential hazards associated with a specific task” and to “determine the level of protective apparel and equipment required to adequately protect lab personnel,” among other duties. Since Sangji was not exhibiting safe practices on the day of the accident and not wearing adequate protective equipment, the district attorney is arguing that Harran is at fault.

Yet the facts are far from indisputable. For one, testimony indicates that lab coats were always available and encouraged in Harran’s lab and that he had previously observed Sangji wearing one in the laboratory. Was it her personal decision to not wear one the day of the accident? Harran also stated that he personally observed Sangji appropriately performing a chemical transfer under an inert atmosphere using a technique similar to that for the tBuLi transfer, and that she was trained by a postdoctoral scientist on the tBuLi reaction itself more than two months before the accident. Whether he can be held legally responsible if this training was inadequate is unclear.

Preliminary hearings commenced last fall and ended with a request from the defense for additional time to prepare written motions moving to dismiss or reduce the charges. At the time of press, the court has not yet reconvened to address the motions and decide whether to proceed to trial.

“In my field we pass down knowledge, one generation to another”

Sangji’s death has highlighted the inherent complexities of providing comprehensive safety training in academic research labs, especially those in chemistry departments that frequently use extremely hazardous and dangerous chemicals. Conversations with graduate students and professors here at UC Berkeley revealed the particular challenges associated with large research groups, where individual safety training by professors is not only unfeasible but may also not be the safest strategy. Many professors have not actively conducted experiments in decades and furthermore advise students and post-doctoral scientists who routinely practice novel methods and techniques that professors may have never directly encountered.

Gary Gallego is a fourth-year PhD student at UC Berkeley studying synthetic organic chemistry in Richmond Sarpong’s laboratory, which has more than 25 members. In such a large group, Gallego says, “the professor actively training students individually is completely unrealistic and would be entirely unworkable. With different researchers running multiple reactions a day which employ a variety of different chemicals in each reaction, it would be literally impossible for the professor to train students in safe handling and use of each chemical.”

Design: Liberty Hamilton, more information at http://www.ehs.berkeley.edu/campus\_policies

Of course, professors do have a clear responsibility to ensure that their laboratory personnel receive adequate training before beginning work in the lab, and that certain resources such as standard operating procedures (SOPs) are available and regularly consulted. Gallego stressed that Sarpong “has been proactive in ensuring all of our members have access to safety information and SOPs, and that they are maintained and updated on a regular basis.” But another question remains: who is responsible for training new laboratory personnel?

In Sangji’s case, Harran delegated this training to a post doctoral scientist in his laboratory who had extensive experience working with pyrophoric chemicals. When asked by the Cal/OSHA investigator about the specific procedure that the postdoc would have taught Sangji for the tBuLi transfer, Harran stated that “in my field we pass down knowledge, one generation to another and this is lore. It’s something that you do almost on a daily basis.” Yet Harran was strongly criticized in the second Cal/OSHA report for not explicitly determining how exactly Sangji was trained, what technique she was taught, and whether it conformed to accepted practices.

Again, however, in practice it would be impossible for every professor to personally ensure that every member of his/her laboratory receives proper training for all hazardous chemicals. Relying in part on senior members of the lab for this training is inevitable, and indeed, permeates academic research. As Gallego explained, “knowledge being passed down from postdocs and more senior graduate students is absolutely a mainstay within a research lab. It is one of the best ways to learn from the experience of others, and is essential to the growth and education of younger students.”

The remains of the plastic syringe Sangji used to remove tBuLi from its container. The accident occurred when the plunger slipped from the end of the syringe, spilling tBuLi onto Sangji’s hands and clothing. Note how short the needle of the syringe is; guidelines recommend using a needle at least one foot long.

But so, too, is the regular consultation of official documents that specify unique hazards and handling instructions for certain chemicals and techniques, such as technical bulletins and SOPs. “Learning techniques ‘by the book,’ especially when it comes to the handling of highly reactive chemicals, is also a practice the Sarpong group employs,” said Gallego. “We maintain a large collection of SOPs which contain detailed information in the handling of a wide range of reactive and toxic chemicals. Before the use of these chemicals for the first time, graduate students and post-docs are expected to read over these procedures at a minimum. A combination of training from experienced students, as well as from strict ‘by the book’ guidelines, is the best way to avoid an accident.”

Of course, even perfect safety training will not eliminate laboratory accidents. Ironically, many accidents happen to experienced chemists who are simply repeating familiar experiments. Professor Matt Francis, a synthetic organic chemist in the Department of Chemistry, explains that “all the accidents I have seen throughout my whole career have involved techniques that people completed a bunch of times before. It’s very unusual for it to be a brand new, obscure practice. Rather, it’s always something where you’ve distilled something ten times and are just doing it once more. That’s when people tend to drop their guard a little bit.” Another obstacle is that a properly trained researcher will not necessarily practice the very same techniques they’ve been taught. Francis used the classroom as an analogy: “It’s just like teaching a class. You can tell somebody the right way to do something in class, and then you give an exam. What percentage of your students do it exactly like you said?”

There is a shared feeling among many researchers at UC Berkeley that the criminal charges filed against Harran are excessive. While certain facts are undeniable—Sangji didn’t practice the correct technique in transferring tBuLi, and she wasn’t dressed properly to be working with such a dangerous chemical—less clear is whether the fault should lie exclusively with Harran. According to Spencer Knight, a first-year PhD student in chemistry at UC Berkeley, “people always want a scapegoat. They want one singular entity to blame. The truth is that safety in an academic lab is a complex, multi-person, multi-faceted kind of problem. It requires effort from all personnel, from students all the way up to university safety officers.” Francis voiced a similar sentiment, saying that “while there’s a need to place blame, the reality is that the situation is much more complex. When awful things like this happen, what’s more useful than saying ‘it was that person’s fault, it was UC’s fault,’ is to ask, ‘what can we do to make things better and prevent accidents from happening in the future?’”

Any Department of Chemistry at any Regents’ campus

The 2012 court settlement between the district attorney and the UC Regents contains detailed, prescriptive tasks to accomplish precisely these goals within all Departments of Chemistry at each of the ten UC campuses. Included among the many new obligations are a mandatory Laboratory Safety Training program to be completed by all principal investigators and all laboratory personnel; comprehensive SOP requirements for a wider range of chemicals; specific procedures to be followed when handling any pyrophoric liquid reagents; and enhanced protective equipment requirements specifically at UCLA. All chemistry lab members working on UC Berkeley’s campus have by now completed a new online laboratory safety training course and formally signed off on their lab’s SOPs.

Michael Kumpf, the director of the UC Berkeley College of Chemistry Environment, Health, Safety & Security Program, has been intimately involved in instituting changes on campus over the last six months to comply with the terms of the settlement. “It hasn’t been easy,” said Kumpf. “We spent the first three months just trying to figure out what we’re going to do, and then we spent the next two months trying to build the tools to get it done, and then we had about one month to actually do it. That was almost an impossible task. But we did it.”

According to Kumpf, taking on changes of this magnitude has required a larger allocation of resources from the Office of the President. “If you want to continue in this mode of maintaining compliance, with the ultimate end goal of achieving a fundamental change in safety culture, it’s going to take people and money,” explained Kumpf. “As soon as you stop the people and the money, you’re going to see this drift back to where we were. It’s been proven before, and it’ll happen again.” Kumpf’s office worked closely with campus EH&S, as well as newly hired consultants at UC Berkeley, to help chemistry research groups write the new SOPs and track progress in completion of the new safety training program. So far, Kumpf has already seen concrete improvements in the department’s safety culture.

UC Berkeley can expect more changes on the horizon. There are currently three system-wide policies in draft form that would apply to all departments at all ten UC campuses, not just the Department of Chemistry. The policies cover protective equipment requirements in the laboratory, safety training requirements for laboratory personnel, and new restrictions on minors in the lab workplace. The implementation date for enactment of the new policies has already been postponed, due in part to the “tremendous response in comments,” said Kumpf. “We have several representatives here who are on the steering committee, to just look at the policy and the tools within the policy that allow labs to customize the requirements based on the hazards that people who are working with materials might face.” Part of the problem is that in its current form, the policy mandates that all researchers must wear lab coats and safety goggles at all times while occupying or entering a laboratory area, something that is impractical for the many labs on campus in which desks and office space occupy the same room as laboratory benches. Kumpf agreed that the policy, as proposed, is “not very realistic, not with the configurations we have. So we’re working on getting policies that make sense, and that have some flexibility built in.”

Design: Liberty Hamilton

Clearly, there will be no one solution to address all the safety issues that can arise in academic research. The topic is intrinsically complex and sensitive, and even with all possible safeguards in place, laboratory accidents will inevitably occur. Kumpf remains optimistic. “While there’s still a lot of room for improvement, you just have to continue to be diligent and do what you’re capable of doing, in the allotted time and with the resources you have. You have to continue working hard just to make things better.”

Sangji’s death was tragic, and something that should never have happened. But as Francis put it, “when something tragic like this happens, you hope that something comes out of it to make things better. I do think a lot of the safety awareness is better. I think a lot of the mandatory training is improved. So there are a lot of positive things that have come out of this.”

This article is part of the Spring 2013 issue.