Free formaldehyde - a practical guide15 December 2003
Formaldehyde or methanal is a highly volatile, noxious molecule which is omnipresent in our environment due to biological and synthetic production. For synthetic purposes, it is normally handled as a 30% solution in water called formalin or formol, but more diluted solutions are employed as bactericides or preservatives for eg cosmetics and concentrations like 100ppm can be found in cigarette smoke or 10-20ppm in green apples. Even the human blood usually contains 2-3ppm of formaldehyde as a metabolism intermediate. While the acute toxicity has been proven without doubt, the latent effects of formaldehyde are still under discussion and only in 2002 has it been reclassified from the list of suspected mutagenes into group 5 saying 'no mutagenic effects observed if the MAK threshold value of 0.3ppm in air is not surpassed.' Regarding carcinogenicity, this decision is still pending. For the production of leather, the use of formaldehyde can be classified into three groups: Highly concentrated formalin for (pre-)tannage and crosslinking of certain finishes, in diluted versions as preservatives in certain products for wet-end and finishing and as a starting material for the production of synthetic aromatic and resin tanning agents. All these uses can lead to a leather article which is contaminated with trace amounts of formaldehyde. To determine the contamination of the leather, a number of analytical procedures have been developed which are described in Table 1. Apart from the dynamic diffusion method currently under development at the LGR at Reutlingen, all these methods have been set up to test leather and leather articles only, and not the products to be used for its production. It is of essential importance, especially for all the liquid extraction methods, that the tested sample is not water soluble! Unfortunately, most retanning materials are water soluble. To minimise the risk to users and the environment, threshold values have been set up. Since the different detection methods have a different sensibility, such threshold values vary from 0.05ppm up to 300ppm. Thus, it is obvious that the detection method has to be specified whenever a certain limit is requested. But even within the parameters of one detection method, the actual value can vary greatly, since different institutions are interpreting the scientific facts about the hazard level in a different way and are, furthermore, applying different safety margins to eliminate any kind of risk. A list of typically requested threshold limits and their providers are given in Table 2. It should be noted that so far, there is no legal threshold limit for consumer articles like leather goods but only the request to specify the content if certain levels are achieved. The MAK-value of 0.3ppm (maximum concentration in the air at the workplace) on the other hand is greatly dependent on the local environment. The participation of a leather article to this value can best be estimated by the test chamber method. But even this will practically always result in much too high numbers, since the ratio of leather surface to air volume from within the test chamber is hardly ever reached in a normal setup. And from experience, we know that there are generally no problems to comply even with the threshold limit of 0.05ppm for the test chamber. Nevertheless, what can be done if a certain leather article has to be produced complying with one of the above mentioned regulations? The easiest thing would be to completely avoid any formaldehyde containing products during the production process. But since even the air contains a certain amount of formaldehyde, it seems more reasonable to request any product to comply with a certain limit of free and/or bound formaldehyde. Unfortunately, it has been shown in earlier works that there is no linear correlation between the amount of free formaldehyde in a leather processing product and the value detected for the leather article. Apart from the systematic errors during the detection, this is mainly due to the fact that formaldehyde can be bound to collagen and this reaction is greatly dependent on the reaction conditions, ie the retanning process and other products used therein. This only leaves us with the possibility of testing the final leather article with whatever method is requested and to demand from the provider that a given process chemical will not participate in the contamination of the article with formaldehyde. To give a guideline, Table 3 shows the different types of leather process chemicals that can be classified regarding their potential to contaminate the leather. If only high quality products from the three green groups are used, the risk of finding a problematic level of free formaldehyde in the leather is minimised. Using formaldehyde for crosslinking, tanning or as a preservative will give an increased amount of free formaldehyde in the leather. However, this case seems relatively easy to handle. The maximum possible contamination can easily be calculated but is generally much lower than this value due to the consumption of formaldehyde. And, once reacted, the formaldehyde is bound fairly well to the matrix. Resin tanning agents are the last group and most difficult to evaluate, which is why I will concentrate on them. To synthesise a classical resin tanning agent, different amino and amido compounds are reacted with formaldehyde to form a water soluble polymer of large molecular size. Due to the size, resin tanning agents have good filling properties and bind selectively to loose areas of the skin. Those properties makes them valuable and indispensable for certain leather articles. Most resins are based on either dicyandiamide or melamin as nitrogen compound and unfortunately their reaction with formaldehyde is a reversible equilibrium process. Depending on the requested retanning properties of the resin, different reaction conditions, reagent ratios and starting materials are used which, by themselves, lead to products with different amounts of free formaldehyde after equilibration. To optimise the tanning properties, those resins are then typically blended with other synthetic auxiliaries for penetration, buffering etc. The resulting commercial product again has a different level of free formaldehyde and different behaviour during the testing conditions, in which the reaction equilibrium is practically always shifted to liberate more formaldehyde. Especially, the test methods based on an aqueous extraction and detection in aqueous solution, such as the DIN 53315, create such conditions and are therefore extremely sensitive to variations in the process. To illustrate this, we have run a series of determinations conducted with the samples from the same piece of leather which are shown in Table 4. It can clearly be seen that the size of the sample particles (powdered or only cut) and time and temperature of extraction all have a large influence on the stated value. But even if all this is done according to the protocol, one can still get false positive results if the sample is not immediately measured after one hour. And this can easily happen if the laboratory is working with a high degree of automation. One single chromatograph taken with HPLC takes about 30 to 60 minutes to be processed and, therefore, a sample might even have to wait overnight in the automatic feeder of the HPLC at elevated temperatures within the device. What is the reason for this deviation? In theory, the experimental setup should eliminate such a problem. After the extraction, the 'formaldehyde source' leather is removed by filtration and the reaction of dissolved free formaldehyde with the detection reagent should be completed before measurement. As the IR-analysis of the test extract shows (Table 5), not only formaldehyde is extracted from the leather samples but a certain amount of resin tanning agent, too. In case of a dicyandiamide based resin, it can be identified by its absorption peak at 2170cm-1. Due to the high dilution of this extract, the shift of equilibrium away from the resin towards the reaction product of formaldehyde with the detection reagent is greatly favoured. And in the presence of enough detection reagent, practically all the formaldehyde used for the synthesis of the resin can be found in the end. So there are several factors to be considered in dealing with a question of analysis of free formaldehyde, which I would like to elaborate now. As mentioned, resin tanning agents are used to fill especially the emptier parts of the skin, eg the belly region. Therefore, a Danish bovine wet-blue was sampled from different regions of the skin and one set of samples retanned together in one drum (effect of different affinity) and one set separately each (effects of different exhaustion/fixation). As a standard, one sample was retanned without the addition of the resin. The results of the subsequent DIN analysis is displayed in Table 6. As can be seen, the addition of resin tanning agent increases the detected value from 3ppm to about 70ppm with relatively small deviation between the single samples. It should be mentioned now, that the standard used during most of the following experiments was always this basic process done by the same technician in the same equipment with identical products and starting materials and analysed again by only one technician. However, the values found for the standard varied from about 55ppm to 105ppm, which has to be considered a system imminent error. Since the reproducibility of the analysis of one single sample was reasonably better, the experimental programme was completed nevertheless. Comparisons, however, were restricted to be within one single experimental set (6 or 12 samples) or to cases of very similar results for the standard. It is to be expected that a higher offer of resin product will result in a higher contamination of the leather, which could be demonstrated in the experiment in Table 7. To distinguish between dicyandiamide and melamin based resins, a number of commercially available resin tanning agents have been tested at the same level of product offer. Since the exact composition of the products regarding resin and auxiliary content is not known, the experiment has a limited value regarding the quality of the resin itself and displays only the facts for the product. In addition, the filling and softening capacity of the different products may vary and an industrial tanning process might use different amounts for the same article (see Table 8). It can be concluded that there is no general preference to be made between dicyandiamide or melamin based products, and that all products tested create a relevant amount of formaldehyde. Since an offer of 5% seems reasonable for a technical process and remembering the typical threshold values, a product should be considered 'less problematic' up to a detected number of 100-150ppm of formaldehyde in this experiment. What can be done, if a requested threshold value is surpassed and the resin tanning agent employed seems to be indispensable for the quality of the article? To start the work, a commercial formaldehyde reducing synthetic auxiliary was employed that had been shown to be effective in prior work (see Table 9). As can be seen from the measurement with a diffusion method which is likely to detect only free formaldehyde in the dry crust, the addition of the auxiliary syntan after the offer of resin largely reduces the amount of free formaldehyde at the end of the leather production process. Unfortunately, the effect on the measurement with the DIN 53315 is much smaller, since this auxiliary is not capable of fixing the resin agent to the leather matrix and the extraction of resin leads to the creation of detectable formaldehyde as discussed above. To optimise the behaviour towards the DIN test, attempts have been made to increase the fixation of the resin to the matrix. In a first experiment, tanning agents and ionic retanning agents have been used, but only with limited success as can be seen in Table 10. Vegetable extracts, which can be employed with great efficiency for the prevention of chrome VI formation, also do not have any positive effect on the detected level of free formaldehyde. Better results where achieved when polymeric fatliquors and softening polymers where added to the retanning process. With the addition of 3% (solids) of product, a decrease of free formaldehyde of up to 40% was observed. One plausible explanation is an 'indirect fixation' by the fact that the leather becomes less hydrophilic and, therefore, is less extractable. This would correspond with the experience, that waterproof leathers generally have less problems with free formaldehyde (Table 11). Some more experiments that have been run will not be discussed in detail here but can be integrated in the summary of this experimental programme about free formaldehyde in leather and its prevention (Table 12). Acknowledgments: I like to thank Ahmet Kaplan and Georg Pesch for conducting the experiments and measurements and Dr Nils Brinkmann for valuable input and discussions.