Level: Alchemist

Reading time: 16 minutes

Do you think that there is any danger of the metals in the husk of brewing cocoa?  I found this study that had been conducted by As You Sow
- do you have any thoughts on this?  Do you think that depending on the region, levels may be higher? 

I’m a little surprised at myself for not having written about this sooner.  I’m going to go ahead and give away the punch line.  In short, I’m not concerned about metal contamination in cocoa beans and brewing cocoa.  Let’s get into why.

My thoughts are that I am very suspect about the data....and specifically the lack of data.  That's as a chemist with 20 years in the lab.  Chocolate is a difficult matrix and honestly I'd want to see the data with resulting MRLs (minimum reporting limits) before I actually believed all those companies had lead and cadmium in their chocolate above regulatory levels.

This kind of response from a company makes me skeptical:

“However, the non-profit refused to share the precise lead and cadmium levels found. “Since our notices are the initiation of litigation, we cannot share data without it losing its confidentiality privileges,” said van Vliet.”


After that, to my understanding both Pb and Cd are both primarily mechanical/process contaminants and not region or growth contaminants.   There is some data that Cd can occur naturally in cocoa.


But the levels are quite low.

I particularly dislike As You Sow's reference to a study 'proving' their point here:


In contrast, lead concentrations of manufactured cocoa and chocolate products were as high as 230 and 70 ng/g, respectively, which are consistent with market-basket surveys that have repeatedly listed lead concentrations in chocolate products among the highest reported for all foods.

On the surface, it looks terrible until you look at the units.  230 ng/g. That is 0.23 ug/kg and or 0.00023 mg/kg or ppm.  Here is an FDA quote based on levels and its recommendations:

The agency said that although the new recommendations of 0.1 parts per million (ppm) are not legally required, it is prepared to take enforcement action against any candy product containing lead levels that may pose a health risk.   This overrides previous guidance provided by the FDA , which stated that regulatory action would be considered against candy with lead levels that exceed 0.5 ppm (mg/kg)

That is kind of an old reference, but even looking at the hardcore Prop 65, I don't see an issue.

If we go to the lead listing in Prop 65 and look at the NSRL and MADL( No Significant Risk Levels (NSRLs) and Maximum Allowable Dose Levels (MADLs)) you will find this:

NSRL- oral (μg/day) 15
MADL - oral (ug/day) 0.5

Doing a little basic maths on that 0.23 ug/kg, that means you would have to eat over 2 kg/day of chocolate to reach Prop 65 levels.  Really?  No.

My thoughts?  It's about the same conclusion as pretty much all the other conclusions I found in rigorous studies.  There is some in soil but it's low and has a low correlation to what is in the beans.   And that a lot more data and studies are needed but it really seems first indications are that it's not a major issue.  Likewise the new refrain that metals are 'concentrated in the husk' is way overblown.  From what I could see, the levels were slightly higher in the husk, but to me, concentrated implies something like an order of magnitude which I've never seen data for.

 I agree it can be in chocolate, but as of yet, every reference I can find where a company responds says they are in compliance.  I'm no lover of big chocolate, but have trouble believing they would knowingly ignore or falsify data. 

My gut feeling is As You Sow is being alarmist.  In principle I can agree with the philosophy of “there is no healthy level of Lead” but it conflicts with my knowledge that if you look low enough you will find contaminants everywhere and somehow we all keep surviving and thriving. 

It is also worth reiterating those studies are aimed at processed cocoa beans, not cocoa beans.  And that is a HUGE difference.

To that end, let’s move onto talking about brewing cocoa, husk, total metal test results and what that tells you and what it does not.

When I worked in the lab doing metals analysis where the concern was safety and compliance the first step was to do a scan to get a baseline of the metal content and then, based on the level, decide (by well established protocol) whether we needed to analyze for total metals, dissolved metals or extractable metals, all dependent upon what the results were being used for and what risks were being assessed.

For instance it doesn’t matter if an item contains toxic levels of something if the concern is whether it will leach into ground water and be hazardous.  We (and the government/EPA) were only concerned about the what could be extracted under very particular conditions.

If we were given a sample of cocoa husk for instance, the first step would be to digest (treat it with a concentrated acid so the metals are liberated) so we could scan for the approximate total metal content.  If the level was high and above a regulatory level would we then process the sample again and see what the extractable metal content was.  There is a well-established procedure known as a TCLP or Toxic Characteristic Leaching Procedure that we used often.. 

This was put into place because metals it turns out are not readily soluble in toxic quantities when bound in a great many matrices.  It was a very rare day in the lab where we found any detectable metals in TCLP extracts.  They just didn't extract.

Considering the TCLP extraction took 24 hours of constant agitation, under semi-aggressive conditions (pH < 7  or < 4 depending on the substance in question), I just can’t see metals being liberated from cocoa husk brewed 5 minutes in basically static neutral water.

Do I know this for sure?  Of course not. It should be tested, but no one seems to be doing that.  All the data I can find are from ICP-MS scans.

Let’s talk a bit about ICP-MS Scans.  First off, that stands for Inductively Coupled Plasma – Mass Spectrometer.  Without going too deep (we are really just going to scratch the surface) samples (those digested or TCLP extracts above) are sprayed into a chamber containing a Plasma to ionize the metal, the metal ions are introduced Spectrometer and the Masses are Scanned.  The intensity of the mass scan correlates to how much is in the sample.  The sweet thing about an ICP-MS is that the user can decide whether to scan all the masses at once (a “Scan”) or give each mass individual attention.  For scans, a little bit of sensitivity (you can’t see as low) is given up for the sake of time (it is fast, hence the term scan).  Accuracy is usually also given up to some degree (explained below).

Sigh, I can tell already I am diving deeper than I want without giving you enough context.

I recently saw results for Lead in cocoa beans from an ICP-MS Scan.  My big issue is that the company made a safety decision based on that result.  Not to chastise the lab too badly, but they really should have had a disclaimer on the result that basically said “not for compliance purposes” recognizing that because the analysis was a scan, the result had a moderate chance of being inaccurate.  Maybe high, maybe low.  Really you have no way of knowing.  Recall I said you give up  something for the quick scan.  Standard laboratory policy in every lab I have ever worked in is that scans can never be used for compliance and that we made sure the customer knew that clearly.  If they needed defendable data they needed an analysis focused on the metal in question, extracted under the appropriate conditions to give the needed results.

And still I ramble.  My apologies.  Semi-hypothetical example time.  Joe comes into the lab and says he wants to know how much lead and cadmium is in his cocoa beans.  He’s given the choices, sees that he can have each analysis (compliance defendable but not stated) for $150 or a whole metals scan (wow, 20 metals!) for $200.  Ha! What a deal, he takes the scan.

A week later he gets a report back that says there is 0.4 ug/kg lead and 0.2 ug/kg cadmium in the cocoa.  He panics a little but does his own maths and works out that based on how much he is using in his brewing cocoa drink, that the levels are fully safe.  He even recognizes that those numbers could be elevated, and that the extracted amount could be much lower.  It all sounds perfectly logical and for someone that is not used to laboratory analysis, hasn’t don’t anything wrong.

But here is the problem.  

Those results could just as easily be 1.4 ug/kg each and now the brewing cocoa could have too much cadmium and lead.  Or there might not be any metal there at all (frankly, more likely).  The issue is you simply don’t know.  Just because you see a number, doesn’t make it real.  You need verification and fast scans just don't provide that..

The main issue with a scan is that a couple assumptions are made.

  • The system is calibrated with a single point and it is assumed all the samples are completely free of interferences.  This is rarely true except for very clean water samples, but is good enough for a scan (and the price).
  • The instrument responds in a perfectly proportional way, i.e. twice the amount of metal gives twice the amount of signal.  It’s mostly true at medium levels, but at very low levels the error can get very large.

Ok, I give up.  I have to start diving deeper because I seem incapable of keeping it simple today.

ICP-MS are often touted as having a huge dynamic range. Sometimes 12 orders of magnitude.  That is truly awesome. That means they can see (under certain circumstances) from 1 ppq (part per quintillion) to 100 ppm (parts per million).  One of those circumstances is a clean matrix.  Cocoa and most solids are not clean (like drinking water).  Another condition is not being in scanning mode.  And what may not by apparent by that dynamic range statement is the relationship may not be perfectly linear.

Look at these three graphs. 

icp curves.jpg


In a Scan (top left), one middle level standard is chosen (20 ug/ml) and it is assumed that there is no signal if there is no metal (so the line goes through 0,0, the origin).  Looking at the graph to the right, you see an interference free calibration curve made with multiple points and it does indeed go through 0,0.  The trouble comes up with certain complex matrices where even if there is no metal present, the system gives a signal and so the whole line is elevated and doesn’t go through the origin.  So, when you put in your digested cocoa bean sample and get a signal of 0.2 (the left hand axis) a Scan will tell you there is about 2.5 ug/ml but a complete, and defendable analysis (the $150 one) will tell you there was nothing there since the calibration solution that had no lead also had a signal of 0.2.  See what I mean?

How can this happen?  Is that laboratory not doing their job?  Not at all.  These can be incredibly small amounts of material and laboratories use reagents (acid and such) that may contain again, incredibly tiny amounts of these substances that can add up in glassware, concentrate in samples as more is added to break down the sample (fats like in cocoa butter usually take additional acid) and the matrix itself can cause extra signal.  Hence the disclaimer that most labs put on scan results.

The wall of text aside, the main point is that upon seeing a result from a scan, for a heavy metal, in a food product, I would have hoped the laboratory would have encouraged the customer to both consider a defendable, multiple-point calibration curve analysis AND suggested they consider an extraction that better emulated (and would probably have reduced interferences greatly) the use the cocoa beans were being put to, i.e. brewing so the customer’s needs were met.  The best situation is that the customer could have just submitted to the lab a sample of the exact extract they wanted to know about since that at the end of the day is the important number.  Not the total number, but how much might be consumed.

Until I see results of significant levels of lead or cadmium in brewed cocoa, analyzed by something other than a quick scan, I’m going to stay in the unalarmed category. 

Oh, and I won’t eat 2 kg chocolate a day.