September 11, 2009: Researchers at the National Institute of Standards and Technology (NIST) say they have a new method to improve temperature calibration for microfluidic systems.
Typically, reactions in microfluidic systems require some form of heating, and monitoring temperature changes in fluid volumes ranging from microliters to subnanoliters — DNA analysis relies on precise temperature cycling, for example. An alternative to thermometers and temperature probes, which are generally not effective at these dimensions, are noninvasive temperature-sensitive fluorescent dyes (e.g., rhodamine B), whose intensity is inversely proportional to temperature (i.e., as temperatures go up, their intensity goes down). The technique, though, requires basing all reading on the fluorescence at a single reference temperature. Others (NIST researchers, in 2001) have devised "calibration curves" to relate temperature to rhodamine B fluorescent intensity, using a reference temperature of about 23°C, but those are only good for that one temperature.
The new work, described in a paper for Analytical Chemistry, shows that changing the reference point in a microfluidics system (e.g., higher temperatures when first heated) introduces significant errors with such dye intensity calculations using current method — up to an 11°C range of error (-3°C to 8°C) for calibration equations derived at ~23°C, using a simple linear correction for a 40°C reference temperature, according to lead researcher Jayna J. Shah.
To address this, the team devised mathematical models to correct for the shift during reference temperature changes; with this they created generalized calibration equations that can be applied to any reference temperature. Among the applications is amplifying microfluidic DNA (producing numerous copies) by the polymerase chain reaction (PCR), which requires cycling a microfluidic device to be cycled through temperatures at three different zones, starting around 65°C — a dye intensity-to-temperature ration would have to be based on that temperature, not the aforementioned 23°C, Shah notes.