In the study of odor coding, we correlate characteristics of the odorant stimulus with measures of response. To do this properly, we need to have a good understanding of the odorant stimulus. This task is simplified by using individual odorant chemicals having properties that are either known or easily estimated. Typically, researchers in olfaction purchase an odorant chemical from a vendor, who either has produced it directly or has acquired it from a manufacturer. The production of these chemicals entails either chemical synthesis or purification from a natural source. Regardless of the method of preparation, there is invariably a small amount of material other than the desired compound that also is present in the product as contamination.  

In the case of chemical synthesis, the starting reagents and solvents can linger, and minor side-reactions producing different products usually occur. Distinct products also can be generated from minor contaminants in the starting materials. The manufacturer attempts to isolate the desired products to increase their purity. In the case of purification from a natural source, a variety of steps are used to enrich the desired product. Generally, these purification schemes are based on known properties of the desired compound such as charge, size, boiling point, and relative solubility in different solvents. Nevertheless, other compounds, often related to the compound of interest, typically ride along during the isolation, and the solvents used during the purification steps also can linger and produce side reactions. The degree of contamination, and the nature of the contaminants, can vary depending on the original source for the purification as well as on the isolation methods employed.  

Vendors usually offer an estimate of the purity of the label compound, and most researchers select the highest purity of odorant chemical that is compatible with their research budget. (When purification is difficult, the buyer pays for the additional effort.) Nevertheless, an apparently high estimate of purity can mislead researchers for a variety of reasons. First, the estimate of purity applies to the state of the product immediately following preparation, and does not foretell the possible spontaneous degradation of the desired compound into additional chemicals during storage and use. Important degradation reactions include oxidation by air, photo-oxidation, and hydrolysis by trace amounts of water. Storing material cold, in the dark, under nitrogen atmosphere, and in tightly capped containers can reduce the extent of spontaneous degradation. Second, the estimate of purity applies to the compound in the solid or liquid phase, rather than in the vapor phase used as the olfactory stimulus. A small amount of a very volatile contaminant in the liquid or solid phase could lead to that contaminant being predominant in the vapor phase. Third, the accuracy of an estimate of purity depends on the methods used to separate the compounds during analysis, as well as the methods used to detect the contaminants. If separation during analysis uses a method resembling an isolation step, then the same contaminants surviving purification might be cloaked during analysis. The best strategies are to use at least two different methods of separation during analysis, and the gold standard is gas chromatography followed by mass spectroscopy (GC/MS), which also is very sensitive. Fourth, even very minor vapor phase contaminants can be the strongest olfactory stimuli if they have unusually high affinity interactions with odorant receptors. 

Compounds responsible for the overall odor of a preparation can be determined by using GC/olfactometry, wherein components of a mixture are separated by gas chromatography and detected by using a sniff port and a trained human observer (Acree, 1997). Occasionally, the compound that is principally responsible for the odor of the mixture can elute from the column in a region removed from the principal components as detected by using flame ionization. Using GC/olfactometry, minor amine and sulfur-containing compounds have been shown to be responsible for many off-odors in food and beverage preparations (Dreher et al., 2003; Lunden et al., 2002). In his archive of perfume and flavor chemicals, Arctander frequently mentioned differences in perceived odor between different sources of individual chemicals, and he generally attributed these differences to variations in chemical purity (Arctander, 1994). 

Given the known impact of odorant contaminants on perceived odors, it is not surprising that contaminants can have a large effect on odorant-evoked neural activity patterns. In a recent study (Ho et al., 2006), we mapped activity in response to a homologous series of straight-chained alkanes, each of which was reported to be at least 99% pure by the manufacturer. From our research involving other homologous series, we predicted increasingly ventral responses with increasing carbon number. We observed activity shifting in the predicted direction, but in addition to activating ventral glomeruli, the 15-carbon alkane pentadecane also stimulated more dorsal glomeruli in a region that had been stimulated by smaller members of the series. Interestingly, we had observed that increasing carbon number had been associated with a lesser odor up to the 14-carbon alkane tetradecane, but that odor intensity increased again for pentadecane. We tested the possibility that the preparation of pentadecane we had used was contaminated with some other odorous chemical by purchasing material of even higher purity (99.8%). Indeed, the more pure pentadecane was odorless to us, and it evoked 2-DG uptake only in the ventral part of the olfactory bulb, suggesting that the more dorsal glomeruli had been activated by a contaminant. We showed that rats could distinguish the two odorant samples spontaneously in an odor habituation assay (Ho et al., 2006). It may seem unexpected that a 0.8% difference in purity could make such a large difference in odor responses, but it helps to think in terms of the contaminant, which could be 5 times more concentrated in the 99% material (1%) than in the 99.8% material (0.2%). If the contaminant were to have characteristics similar to a smaller alkane such as octane, as suggested by the evoked pattern, then the contaminant could be hundreds of times more volatile than the label material, which would explain further the impact of the contaminant. 

Even a small amount of impurity can have a large effect on evoked activity patterns, as demonstrated here for pentadecane. The major regions of 2-DG uptake activated by the 99% pure odorants were not present when a 99.8% pure odorant was used.

In a subsequent study on double and triple bonds, we found that different lots of the same material from the same vendor can lead to different evoked activity patterns, and that different vendors reporting the same high level of purity for a given compound can lead to different evoked activity patterns (Johnson et al., 2006b). In these scenarios, the simplest pattern obtained for a given high odorant concentration (and the parts of the pattern that overlap for all preparations) are likely to indicate the actual response to the label compound.

We also suspect that contaminants contribute to the patterns that we and others have observed when using straight-chained aldehydes as odorants. At high concentrations, these compounds evoke activity in rostral and dorsal glomeruli that overlap with those stimulated by carboxylic acids of the same carbon number (Rubin and Katz, 1999; Johnson and Leon, 2000a; Wachowiak and Cohen, 2001; Johnson et al., 2004). This overlap was suspicious, given that aldehydes oxidize spontaneously in air to produce the corresponding acid. Indeed, we analyzed samples of our aldehyde odorants immediately after opening the reagent bottles and found up to 1% contamination by acid (Johnson et al., 2004). After exposures in which nitrogen was the gas used to volatilize the odorant in an attempt to reduce oxidation, the amount of acid increased appreciably, suggesting an ongoing oxidation process despite the precaution (Johnson et al., 2004). In addition, we know that the olfactory system is at least two orders of magnitude more sensitive to acids than to aldehydes (Johnson and Leon, 2000a), making it possible for even a small acid contaminant of an aldehyde to have a significant glomerular response. We also have found evidence of carboxylic acid in preparations of aliphatic methyl esters, which suggests that hydrolysis of the ester bond may produce significant levels of contaminant, as well. Indeed, activity patterns evoked by methyl and ethyl esters overlap with those of acids in the rostral and dorsal parts of the olfactory bulb (Johnson et al., 2002, 2004).  

Because it is generally difficult to rule out the influence of odorant contaminants, each pattern on our site (and all other responses reported in the literature) should be viewed with a modicum of caution, especially for compounds predicted to have low volatility, where the presence of a more volatile contaminant could dictate the overall pattern. It also is helpful to evaluate odorant patterns evoked by a systematic series of small changes in odorant molecules, where unexpected responses can be noted and their origin determined. We provide a number of such systematic series on this website.