A recent article in the Journal of Agricultural and Food Chemistry (HOUESSOU, MALOUG, LEVEQUE, DELTEIL, HEYD, & CAMEL, 2007; 55, p.9719-9726) suggests that we are right to stop roasting shortly after second crack. The article points out that roasting coffee starts to produce some pretty ugly compounds (polycyclic aromatic hydrocarbons) such as pyrene and chrysene at 250-260 degrees celsius (480-500f). These compounds are considered carcinogenic, are found in tar and creosote, and are used in pesticides, among other stuff.

Since second crack occurs typically between 445-455f, it would make sense from a taste perspective that most espresso lovers stop the roast shortly after 2nd crack begins. Ignoring the health aspects of the compounds being produced at higher temperatures, the presence of these compounds in tar and creosote suggests that they might be responsible for the flavours that match the source descriptions in over-roasted coffee. That would suggest that we not only stop the roast well below the 480 mark, but also keep the maximum roaster temperature to something below 480 - 500 as well.

I believe that the article is available through subscription only (or university library access), so I post only a portion of the abstract below. The authors' mention of infusion relates to brewing coffee using what they describe as "an electric coffee maker":

"Formation of phenanthrene, anthracene, and benzo[a]anthracene in coffee beans was observed at temperatures
above 220 °C, whereas formation of pyrene and chrysene required 260 °C. Low levels of benzo[g,h,i]perylene were also noted for dark roasting under 260 °C, with simultaneous partial degradation of three-cycle PAHs, suggesting that transformation of low molecular PAHs to high molecular PAHs occurs as the roasting degree is increased. The PAH transfer to the infusion was quite moderate (<35%), with a slightly lower extractability for dark-roasted coffee as compared to light-roasted coffee. "


Acrylamides, which are the nasties that we are supposed to avoid in fried or grilled starchy foods, have also recently been found in coffee. Some estimates (and I got this statistic from Wikipedia) suggest that our coffee is responsible for as much as 20%-40% of our intake of acrylamides. The article found that acrylamides are produced rapidly at the beginning of the roast and decrease exponentially as roast continues from 220-240C (425-465f). So drinking light roasts apparently does not decrease our exposure to nasty compounds.

It sounds like the target of second crack is not only the ideal for brewing coffee as espresso, it is also the sweet spot for minimizing our exposure to the nasties in roasted coffee.