#SheriSangji case: The limits of oral transfer of knowledge

As others have also reported, a second Cal/OSHA report has surfaced on the circumstances surrounding the death of University of California, Los Angeles, researcher Sheharbano (Sheri) Sangji from a laboratory fire. The first report, which I covered extensively back in 2009, was the one that led to civil sanctions against UCLA. As is standard in a California workplace fatality, once the civil case was completed, it was turned over to Cal/OSHA’s Bureau of Investigations (BOI) to determine whether it might warrant criminal prosecution. The newly-available report is the one that the BOI sent to the Los Angeles County District Attorney, who filed criminal charges against UCLA and chemistry professor Patrick Harran.

The report paints a pretty damning picture of overall UCLA safety culture at the time of the incident. To quote from the report’s conclusions (report page 90):

Based upon the investigation, it is apparent that the 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. The lack of adequate lab safety training and documentation, lack of effective hazard communication practices, and repeated failure to correct persistent and repeated safety violations within University labs, were all causal deficiencies that led to a systemic breakdown of overall laboratory safety practices at UCLA.

Much of the report just gives added detail to what was already known about the case or documents conventional wisdom about poor lab safety culture in academia. In one interview, postdoctoral researcher Paul Hurley noted that lab coats were seen as optional rather than required in every academic lab he’d worked in. “That’s how my experience has been pretty much everywhere I’ve been, apart from my current job actually,” the report quotes Hurley. Where does he work now? In industry.

Credit: UCLA

The report does reveal, however, a bit more of how Sangji was supervised and trained. To recap the incident, Sangji was using a syringe to transfer about 53 mL of tert-butyllithium (tBuLi), a pyrophoric substance that ignites spontaneously in air, when the barrel came out of the syringe. The chemical splashed on Sangji, who was wearing neither a regular nor a flame-resistant lab coat, and set her clothes on fire. She was burned on more than 40% of her body and died of her injuries.

The report confirms that Sangji did not handle pyrophoric reagents as an undergraduate or during a few months of work at Norac Pharma. At UCLA, Sangji worked in Harran’s lab. Harran told the Cal/OSHA investigator that he observed Sangji do an initial experiment to gauge her experience and competency (the summary of Harran’s interview starts on page 67). Assuming that the experiment Harran referred to is the first entry in Sangji’s lab notebook–the second is her first tBuLi reaction–the procedure involved working in a glove bag and syringing no more than a few milliliters of air-sensitive, non-pyrophoric material. Although some of the fundamental techniques might have been the same, I think many people would argue that in scale and hazard that first reaction was a far cry from Sangji’s tBuLi experiments (synthetic chemists: feel free to weigh in on this point in the comments).

As for handling tBuLi, Harran reportedly told Sangji to ask Hurley for help. Hurley told the Cal/OSHA investigator that he didn’t follow standard operating procedures or other written protocols; rather, he was trained and in turn trained others by word of mouth (the summary of Hurley’s interview starts on page 59). Hurley also couldn’t recall the specifics of his interactions with Sangji. The way Hurley described handling pyrophoric reagents, however, echos some of what contributed to the lab fire that ultimately killed Sangji:

  • Hurley said in the interview that “most people…more often than not did not clamp the bottle” and instead held it with one hand while syringing with the other. Aldrich recommends (pdf) clamping the bottle, which leaves both hands free to manipulate the syringe. (Sangji compounded this by using a 60-mL syringe with a short needle, so would have had to tip the bottle up to get the needle into the liquid while trying to handle a large syringe with one hand.)
  • Hurley said that he always used plastic syringes. Aldrich does not specifically prohibit using plastic syringes, but it does recommend drying syringes in an oven, which effectively means that glass must be used. (Sangji used a plastic syringe. Another postdoc in the lab, Hui Ding, commented that plastic syringes can swell, making them difficult to operate.)
  • Hurley said that he would use a syringe size as close to the reagent volume as possible. Aldrich recommends using a syringe at least twice the volume of the reagent volume, precisely to guard against taking the plunger out to the end of the barrel. (Sangji used a 60-mL syringe for 50 mL or more of material. Ding knew about the twice-the-volume rule.)
  • Hurley said that he would physically pull on the plunger to pull up the reagent. Aldrich recommends using low-pressure, inert gas to push the material into the plunger rather than risk pulling too hard on the plunger and drawing in air. (It’s not clear which approach Sangji used. When Harran was interviewed by Cal/OSHA investigators in the civil investigation, he also said that he’d pull on the syringe. Aldrich scientist Mark Potyen has previously told me that plastic syringe plungers can’t be pushed up by low-pressure gas.)

(What’s not addressed in the report is the issue of transferring material by syringe versus a double-tipped needle, or cannula. On the day of the incident, Sangji was supposed to transfer a total of about 160 mL. For that amount, a cannula is the accepted practice rather than multiple smaller transfers. It’s not clear that anyone ever said that to Sangji. Certainly no one told her to read the Aldrich protocol.)

The Hurley interview clearly illustrates the limit of oral transfer of knowledge. As in a game of telephone, what gets passed on can change over time. Written protocols help guard against that. And making lab workers (re-)write protocols that faculty then review helps to ensure that everyone does, in fact, know what they’re supposed to do.

Other points of note: Sangji started in the lab in October, 2008. Hurley left the lab in November. According to the report, on the day of the incident in December, Harran knew that Sangji was planning to scale up the tBuLi reaction but told the investigator that he didn’t know by how much. Harran seems never to have discussed the hazards or use of tBuLi with Sangji.

Around the web today, I’ve seen a couple of people ask why the DA didn’t file charges against Hurley. My guess is that it comes down to the legal statute, California Labor Code Section 6425, which specifies “Any employer and any employee having direction, management, control, or custody of … any other employee.” Harran clearly fits that description–he hired Sangji, directed her work, and was ultimately in charge of his lab. I don’t see that it applies to Hurley, who was not explicitly assigned to supervise Sangji’s work and who was gone from UCLA by the time of the incident.

Last but not least, one caveat about the report: It is basically a summary of what the investigator found. The full package appears to involve five binders of information, which I do not have (nor, as far as I know, does anyone else outside of Cal/OSHA or the DA’s office). When reading the report, I think it’s important to remember that you’re not getting the full interview transcripts and we don’t know what the investigator may have left out. (I tried to get everything back in 2010, but was told that because the report was part of an ongoing criminal investigation, it was exempt from disclosure under California’s Public Records Act [pdf summary].)

Author: Jyllian Kemsley

Share This Post On


  1. The image in the earlier C&EN article here http://pubs.acs.org/cen/science/87/8731sci1.html does not seem to be of a “glass syringe”, meaning an all-glass syringe of the type depicted in the Aldrich protocol. It looks like a glass barrel but a PTFE/metal plunger. This plunger would not be pushed out by overpressure from the bottle as recommended by Aldrich. We use such glass/PTFE syringes. We do not use all-glass (interchangeable) syringes for this job. In other words, we do not follow the Aldrich protocol.

  2. Word-of-mouth training is usually poor training. The trainee often knows little about the procedure they are being trained to perform and are often told things one time. Often the trainer is doing the training because he/she is about to leave and will not be available as a resource to the trainee. This is why written SOPs are so important. Verbal instructions are not worth the paper they are written on.

  3. I think you’re right about the postdoc. It wouldn’t make sense to charge him since he was no longer working for the lab at the time of the incident. He’s in the report to show what kind of training that Ms. Sangji received.

  4. I think the real problem is why a chemist pulled the plunger out of a syringe. competence problems perhaps? this chemist transfers n-BuLi and worse with plastic syringes for years with no problems.

  5. But wouldn’t it still be relevant if he was supposed to instruct the victim before he left? I am not saying he should be charged, but wouldn’t that make him a part of the whole incident? It seems strange to me to have him excluded from all consideration just because he was not around at the time of the actual accident. I of course agree that word-of-mouth training, while an extensive part of most laboratory protocol, is highly subject to error and attrition. Reminds me of Chinese whispers.

  6. @Peter Scott – That photo came from UCSD and I just watched their video again and saw that they pull up on the plunger, too. I know this is a point of contention in the community–Aldrich’s Mark Potyen even acknowledged to the Cal/OSHA investigator that he’ll use a plastic syringe for volumes smaller than 10 mL.

    Is this partially an issue of the larger syringe and volume that Sangji was using? With a larger syringe that is harder to manipulate, wouldn’t the risk of drawing in air go up? Potyen said that plastic luer-locks don’t work as well as metal, too.

    @qvxb and Unstable Isotope – Thanks for weighing in!

    @captain BS – Well, she was young and inexperienced and most likely trying to completely fill a 60-mL syringe with one hand. Is that a fundamental competency problem or a training problem? How does one *become* competent?

    @Ash – I’m not a lawyer and I haven’t asked the DA, so I don’t have a definitive answer. But I still don’t see that the postdoc had “direction, management, control, or custody” over Sangji. Harran did, and I’m guessing that the DA’s argument here is that it was therefore Harran’s responsibility to ensure that Sangji was trained properly, even if someone else did it.

  7. Probably also relevant in these situations to mention that there are different types of plastic syringes. I assume they were using NORM-JECT brand, which unlike B-D and many others, do not have rubber parts. If you are going to use plastic with organic solvents, NORM-JECT is what you want.

  8. I prefer to transfer all materials, pyrophoric or not, with plastic syringes rather than glass. This is because the barrel of plastic syringes are typically narrowed towards the end to prevent the plunger from being pulled out, whereas glass syringe plungers can slip out quite easily.

    I also take issue with pressurizing, even slightly, containers of pyrophoric materials. This would seem to increase the likelihood of accidental escape of the pyrophoric material from the container.

  9. I like using all-plastic syringes because they are disposable, the plunger is more firmly “held” by the body of the syringe, and they seem to give a better seal than glass. Chemjobber showed that the plunger is still easily removed from these syringes, but still not as easily as for glass syringers (IMO).

    I like putting the Sure-Seal bottle under light positive pressure for two main reasons. First, there is less “fight” to pulling up the syringe. Second, you are protecting your stock of reagent. If you remove a volume of imcompressable liquid from the bottle (under 1 atm), then you are going to create a vacuum in the bottle. The Sure-Seal septa are not perfect once they’ve been pierced, so (wet) air from the atmosphere is going to push into the bottle and foul your reagent. So long as the bottle is under positive pressure, that’s not going to happen. A balloon filled with Ar is usually good enough for syringe work. Of course, the crappy thing is that you end up piercing the seal a second time to get the balloon in, and like I said above, these septa are not very good. Basically, if you’re going to use these Sure-Seal bottles, order them in small volumes because they don’t store well once used.

  10. I’m with Paul and Lee on this one – an inveterate plastic-syringe user. Also pressurize SureSeal bottles, but not too much, because sometimes the plunger will move faster than you anticipate, which leads to spills or leakage.

    I will point out that when I use >10 mL of any major pyrophoric (AlMe3, tBuLi, sec-BuLi, Me2Zn), I like to be in full regalia: labcoat, neoprene gloves, goggles, behind a blast shield. I always have an “Oh crap!” beaker off to the side, where I can aim any problematic syringes.

    When I’m forced to use large quantities, I’ll often buy a small bottle (100 mL) which I titrate (2x) first. I then draw a line on the side at the halfway point, and cannula transfer under nitrogen (~48-52 mL, or +-4% error, but I remain safer!)

  11. The video that I made (linked in my handle) used the same brand of syringe that Sangji was using, I believe. I pulled the plunger past the ridge that Lee was talking about.

  12. – Jyllian (and others – if you are still reading any comments):

    I believe the Aldrich protocol was written for all-glass syringes. These are less tight than all-plastic syringes and as such they are best filled by using a slight pressure from the bottle, exactly as recommended in the Aldrich technical bulletin.

    With plastic syringes, the extra tightness does allow the user to withdraw the plunger by hand, and in fact I think the pressure required for filling them by the Aldrich method would be dangerous: if the needle is not tightly attached or locked in by a Luer lock – and even Luer locks can leak! – the syringe could be separated from the needle, with disastrous consequences as the pressure in the bottle would spill the reagent around through the open needle. As such the correct way to fill a plastic syringe is not the technique described by Aldrich but by gently withdrawing the plunger and using a small positive pressure in the reagent bottle.

    Now, this gentle withdrawal becomes harder to do with larger syringe sizes. In my opinion, 50 or 60 mL syringe sizes should never be used for anything except maybe water. These syringes are so tight that an inexperienced user is likely to use excessive force – causing the plunger to pop out, with consequences that we are discussing here.