How do we use oxygen isotopes as temperature proxy?

As far as I know scientists use oxygen isotope 16 to 18 ratio in air trapped in glaciers (or in old foraminiftera shells) as proxy for temperature in the past. I know that $\ce{^18O}$ is heavier, and it precipitates faster, and $\ce{^16O}$ is lighter, and it evaporates faster.

I understand that when it's warmer, there is more evaporation and more heavy isotope left in water. Vapor moves toward poles, where it falls as rain or snow and goes back to water (this doesn't count) or stays in glacier for hundreds of thousands of years.

So when it's warm, we should have more $\ce{^18O}$ in ocean sediments, and less in arctic glaciers, right? So more $\ce{^16O}$ (less $\ce{^18O}$) in glaciers -> warm period.

On the other hand, when it's cold, precipitation occurs earlier while air moves to the poles, and heavy isotope falls down earlier, on bigger latitudes. In the air that goes to the poles there is only little heavy isotope left. So more $\ce{^16O}$ (and less $\ce{^18O}$) in glaciers -> cold period. This is contrary to the above. So how to figure out temperature from ice and sediments cores?

• When you say "more 18O" do you really mean to say "higher d18O values"? As temperature increases, doesn't mass dependent fractionation become less and less important? At a certain point kinetic energy overwhelms the system. For example, in a water-rock system, high dO suggests lower temperatures. Also, when you say "ocean sediments" this is not the same as saying "in the ocean" right? I'm not clear on what the oxygen atom is in the sediments, water? SiO? – equant Jan 23 '15 at 21:35
• From en.wikipedia.org/wiki/Proxy_%28climate%29 "The calibration was initially done on the basis of spatial variations in temperature and it was assumed that this corresponded to temporal variations. More recently, borehole thermometry has shown that for glacial-interglacial variations ... temperature changes were twice as large as previously believed." and "To produce the most precise results, systematic cross-verification between proxy indicators (e.g. tree rings, pollen, etc.) is necessary for accuracy in readings and record-keeping." – farrenthorpe Jan 23 '15 at 23:21
• @equant When temperature increases, so does evaporation, so I think fractionation too. By ocean sediments I understand forams shells. It's built from CaCO3 and this shells are used as temperatures proxy. – amorfis Jan 24 '15 at 20:23
• @farrenthorpe Thank you for this information, but this doesn't answer my question. I want to understand the mechanism that makes isotopes proxy for temperatures in the past, and this doesn't help. – amorfis Jan 24 '15 at 20:27
• @amorfis you explained the mechanism in your question... I was offering a comment not an answer. – farrenthorpe Jan 24 '15 at 22:33

This is somewhat a general answer. According to the Royal Chemical Society webpage Chemical climate proxies (Evans, 2013), the rarer and heavier $\ce{^18O}$ isotope in water generally evaporates after and condenses (and precipitates) out earlier than the lighter and more common $\ce{^16O}$ isotope, shown in the diagram below:

Source: University of Michigan's page Past Climates on Earth

The colder the environment, the quicker the atmospheric water vapour condenses, hence the quicker the $\ce{^18O}$ isotope is depleted, giving a lower ratio of $\ce{^18O}$ to $\ce{^16O}$ (referred to by the RSC page as $\delta{\ce{^18O}}$) by the time it reaches the poles. In contrast, a warmer environment would mean that the water vapour condenses later and slower, meaning that less $\ce{^18O}$ laden water vapour condenses, resulting in a higher $\delta{\ce{^18O}}$ value.

In short, it is not so much of the amount of $\ce{^16O}$, but rather the ratio of $\ce{^18O}$ to $\ce{^16O}$ isotopes present, otherwise known as $\delta{\ce{^18O}}$.

The following Earth Observatory diagram demonstrates the present-time process, showing the differences between warm and cold environments:

Source: Earth Observatory's page: Paleoclimatology: The Oxygen Balance

• Thank you for such a detailed answer. Yes, I know it's O16 to O18 ratio that matters. So as far as I understand, more O18 in ice (in ice core) means that climate was warmer at the time ice formed, right? – amorfis Jan 28 '15 at 8:20
• @amorfis yes, that is pretty much correct. – user889 Jan 28 '15 at 8:22