TECHNOLOGY:

Suppose a doctor wants to test a blood sample for cancer, and the best way of doing it is to determine the concentration of a particular protein in the blood sample. There are many ways of doing this.

One example is the approach of Haakan Joensson (Royal Institute of Technology, Sweden) and coworkers. They can detect as few as 22 protein molecules per cell with conventional cell sorting instrumentation.

The challenge of characterizing highly complex mixtures.

Now consider a far more imposing challenge. Suppose one wants to diagnose cancer by tracking the change in concentration of a large number of proteins in a cell, an approach which may be less likely to give false positives or false negatives.

It's infeasible to isolate and characterize all of the proteins individually. Cells have many, many proteins in them, all at different concentrations, which fluctuate over time.

Nevertheless, Mikkel West-Nørager (Statens Serum Institut, Denmark) and coworkers have made significant strides in this effort via mass spectrometry. They have identified ovarian cancer with a 68% success rate (note than this is less effective than conventional diagnostic approaches).

However, the instrumentation is expensive, and their research goal is a long-term effort. Furthermore, it might not be necessary to determine absolute protein concentrations; comparison to a standard (i.e. relative concentrations) may be all that is needed, depending on the application.

Cheap and fast, yet effective, instrumentation is needed.

Many non-medical applications have similar needs. Kenji Yamamoto (International Medical Center of Japan, Tokyo) and coworkers have fabricated an electronic nose that can discriminate between yeast and bacteria within 90 minutes, based on the volatile chemicals they emit.

Such sensors can be effective at analyzing a complex mixture of dissimilar chemicals. However, a complex mixture of similar chemicals often poses a much greater challenge; since the molecules are chemically similar, they give similar device readouts unless the device itself possesses a high degree of chemical specificity.

Kenneth Suslick (University of Illinois, Urbana-Champaign, United States) and coworkers have introduced a disposable, color-based sensor that can fulfill this unmet need. As an introductory yet challenging case study, they have rapidly discriminated between the aromas emitted by 10 different brands of roasting coffee beans, as well as different temperatures and durations of roasting.

Why coffee?

It may seem silly to characterize coffee aroma. This research won't help "find the cure for cancer," or something along those lines.

The reason for the scientists' choice of coffee is that roasted coffee bean aroma is a complex mixture of over one thousand volatile chemicals. If the scientists can discriminate between coffee aromas, then surely they will have similar luck in other more socially-useful endeavors.

If it doesn't work, coffee is cheap. This enables the scientists to perform extensive experimental trial and error if necessary.

Principles of device design and operation.

The scientists' device gives a color readout at 36 different spots. Each spot is designed to be chemically responsive (modify its light emission) to four broad classes of chemical properties expected of a molecule: Lewis basicity, Brønsted acidity and basicity, polarity, and redox properties.

For example, if a molecule is slightly polar and a strong Lewis acid, a specific color pattern on the device will be given. A different color pattern will be observed if the molecule is redox active and a weak Brønsted base.

A complex mixture will have a sum total of these four broad properties, and will give a unique color readout. Another complex mixture will have a different sum total of these four broad properties, and will give its own unique color readout.

In contrast to most "electronic noses," these scientists' device readout is strongly dependent on numerous properties of the analyte, not simply the hydrophobic or van der Waals interactions heavily utilized in those devices. This facilitates chemical discrimination between highly similar molecular species.

Device performance.

The scientists tested 10 different brands of coffee fifty-five times total in their device. Due to different growing and processing conditions, the aroma of each one should be different, e.g. due to differences in carbohydrate composition, which fragment into volatile acids under roasting.

The device was able to discriminate between all ten brands within two minutes. It made no errors in identification in any of these experiments.

The scientists then roasted one particular type of coffee bean for a given duration (15 minutes) at six different temperatures (180°C to 240°C). The device was able to discriminate differences in roasting temperature of less than 10°C, and made no errors in identification within at least the 20°C intervals.

The scientists next roasted one particular type of coffee bean at a given temperature (220°C) for different durations (one minute to three hours). Roasting for a longer time will decompose more of the volatile chemicals to carboxylic acids, which should be detectable with the scientists' device.

The device was able to discriminate differences of less than 5 minutes roasting time. Interestingly, light roasting was readily discriminated from medium roasting, which was readily discriminated from heavy roasting, implying further applications in quantitative coffee bean quality control.

The device made no errors among 45 different experiments. However, longer roasting times (greater than three hours) gave less satisfactory discrimination results, probably because most of the volatile chemicals had decomposed (chemically changed) by that point in time.

Overall evaluation.

The scientists' device has several features that enhance its practical utility in real-world settings for a broad range of applications (not just coffee analysis). It is based on a sol-gel, suggesting that it is disposable and is likely to have a long shelf life (at least several months in this example).

It is transparent in the ultraviolet and visible light range, meaning that colors are readily read out from it. Furthermore, it is easily modified through changing the responsive dyes in the device.

This device provides unique information on the sum total chemical composition of a complex mixture; it provides no information on the chemical composition of individual components of a complex mixture. To obtain such information, complimentary instrumentation is required.

However, for applications that only require one to accurately discriminate one complex chemical mixture from another, this technology will be very useful. Expanding it to include more than 36 sensors will enhance its discriminatory capability even further.

for more information:
Suslick, B. A., Feng, L., & Suslick, K. S. (2010). Discrimination of Complex Mixtures by a Colorimetric Sensor Array: Coffee Aromas Analytical Chemistry, 82 (5), 2067-2073 DOI: 10.1021/ac902823w