Misconceptions in the Exploding Flask Demonstration Resolved through Students' Critical Thinking.

As it connects to a large set of important fundamental ideas in chemistry and analytical techniques discussed in high school chemistry curricula, we review the exploding flask demonstration. In this demonstration, methanol vapor is catalytically oxidized by a Pt wire catalyst in an open container. The exothermicity of reactions occurring at the catalytic surface heats the metal to the extent that it glows. When restricting reactant and product gas flow, conditions may favor repetitive occurrence of a small explosion. We show how mass spectrometry and infrared spectroscopy allow for unravelling the chemical background of this demonstration and discuss various ideas on how to use it in a classroom setting to engage students' critical thinking about chemical research. Along the way, we show that two commonly published ideas about the chemical background of this demonstration are incorrect, and we suggest simple tests that may be performed in a high school setting either as an addition to the demonstration or as a student research project.

Introduction

There are, at most, a few good demonstrations that illustrate principles of heterogeneous catalysis. The “exploding flask” demonstration is probably the best known example.[1−4] A preheated catalyst wire is suspended over a thin layer of liquid methanol and catalytically oxidizes its vapor. The oxidation reaction is exothermic enough to make the catalyst wire glowing hot. Using an Erlenmeyer flask instead of a beaker restricts replenishment of consumed O2 from the air. Repetitive explosions may occur. The intensity of the explosion can be increased using a divider in the neck of the Erlenmeyer flask.[1,3] When a beaker is used, the catalyst may remain visibly hot, but no explosions are observed.[2] YouTube provides various video examples of this demonstration, generally based on Pt as the catalyst material.[5−8] Other metals may be used though. De Gruijter used copper.[3] Battino et al. showed that Pd, Ni, Ni/Cu, and Ag also work,[1] whereas Weimer claimed that Cu, Al, Fe, and Ni show no reactivity at all,[4] a systematic comparison for a wide range of metal catalysts and various C1–C4 alcohols showed that Fe, Rh, Ir, Ni, Pd, Pt, Cu, and Ag all catalyze the oxidation of simple primary and secondary alcohols in this demonstration. Only for tert-butanol, no reactivity was observed.[9]

The demonstration is visually attractive and has a number of easily tested variables, that students may propose when asked for minor alterations that could help in learning to investigate general aspects of the chemical reaction. Catalyst material, the fuel, gas flow restriction, distance between catalyst and the liquid level, and the temperature of the liquid all affect the demonstration’s observables. All of these directly or indirectly affect the kinetics of the reaction either through changes in the (relative) concentrations of reactants or the catalytic activity of the material toward combustion of the specific fuel. There is, however, another major benefit to implementing this demonstration in a high school curriculum. When continuing the classroom discussion following the demonstration, questions that relate to and connect many aspects of the curriculum will arise. How do gas flow conditions, thus stoichiometry, result in absence or presence of repeated explosions? What is exploding? Is it the methanol vapor? What is that pungent smell? Formaldehyde? What is causing the glow of the solid catalyst wire, whereas the chemistry, as written down in a chemical equation, contains only gas phase reactants and products? While a survey of the literature suggests that experimental variables have been explored rather well for this demonstration, the provided explanations, among others in this Journal, are highly doubtful. A guided classroom discussion will bring this to light. We discuss two obvious questions within the framework of such a classroom discussion, assuming a knowledge level characteristic for students nearing graduation.

What Are the (Elementary) Chemical Reactions Underlying This Demonstration?

Multiple authors claim that the key reaction in this demonstration, as performed with methanol, is its decomposition to formaldehyde:[1,2]This claim arises from the pungent smell that may be observed when being (too) close to the reaction vessel. However, one can point out that formaldehyde can be smelled at very low concentrations.[10] In combination with the observed hot catalyst wire, it is actually impossible that it is the major product as the decomposition of methanol to formaldehyde in reaction is endothermic. Students can check the reaction enthalpy calculation within seconds as CH3OH and CH2O are generally very close in standard thermodynamic tables and the only other product is an element. Formaldehyde may therefore be no more than a (minor) side product of the reaction. This poses the question what (elementary) chemical reactions occur at the catalyst surface that explain both the formation of CH2O and the obvious heating of the Pt wire.

Starting with the known product H2CO, the reactions that must occur are schematically illustrated in Figure .

Figure 1

Catalytic reaction of methanol on a Pt surface. Reaction rate constants are labeled in accordance with reactions appearing in the text.

Upon impinging onto the Pt surface from the gas phase, CH3OH dehydrogenates. If it only loses 2 H atoms that individually bind to the surface, the remaining adsorbed H2CO moiety can desorb intact, explaining the pungent smell. The adsorbed H atoms form H2 (g) by recombinative desorption. By neglecting the reverse reactions, the overall reaction may thus be written as a simple summation ofThese surface reactions, especially the adsorption on and desorption of hydrogen at Pt, most commonly on the Pt(111) and vicinal surfaces, are extremely well studied in the catalytic and surface science communities, among others by ourselves.[11−14] This reaction has served as the model system to the surface science and reaction dynamics communities in testing basic theoretical assumptions, for example, the Born–Oppenheimer approximation, for breaking a single bond.[15] Adsorption to and the subsequent decomposition of methanol on Pt (see, e.g., refs (16−18)) as well as H2CO (see, e.g., ref (19)) have been studied. Reactions –4 explain the appearance of the pungent smell in the demonstration, but they do not explain that the Pt wire remains hot throughout the demonstration as the overall reaction is endothermic.

An exothermic surface reaction is required for the catalyst to remain hot or even increase in temperature in between explosions when oscillatory behavior is observed. When asked, students are likely to recognize that the H2 (g) from reaction can be combusted, yielding heat. As it is obviously the wire that becomes very hot, the combustion must, however, occur at the surface of the metal. This catalyzed chemical reaction is thus the oxidation of surface-bound atomic hydrogen produced from reaction :Catalytic oxidation of hydrogen by Pt is also considered a model system and has been studied extensively using the surface science approach (see, e.g., refs (20−23)). The required surface-bound atomic oxygen arises from dissociative adsorption of O2 (g) on the Pt catalyst, which in itself is also well-studied (see, e.g., refs (24−28)):Having introduced the left part of the surface reactions shown in Figure , students may be expected to come up with reactions and 7 when asked. Now, students may be asked to calculate using their thermodynamic table whether the additional oxidation by O2, or in other words the partial oxidation of methanol to formaldehyde in air, may explain the continuous glowing of the catalyst wire.

At this point, one may diverge into different directions to discuss various related chemical principles. For example, one could ask students for the expected effects when changing the relative concentrations of O2 (g) and CH3OH (g). Experimentally, this is done, for example, by restricting the air flow, additional heating of the liquid, or changing the position of the catalyst wire relative to the air-conducting orifices and the liquid level. As the concentrations affect the rates of dehydrogenation versus partial oxidation of methanol, this opens up the opportunity to introduce the difference between two of the three most important concepts in heterogeneous catalysis: reactivity toward a reactant and selectivity toward a desired product. In an oxygen-deficient environment, H2 will build up. In an oxygen-rich environment, the partial oxidation is likely to dominate, and H2O will be formed at the expense of H2. One could also take the discussion in the direction of the different catalytically active materials that can be used in this demonstration. The required dissociative adsorption of O2 varies with the type of metal. Copper is known to oxidize fairly easily and extensively (think of the Statue of Liberty), whereas Ag and Pt do not and are therefore used as jewelry.

Whatever the course of the discussion, it is noteworthy that selective dehydrogenation of methanol to formaldehyde turns out to be difficult. In industry, it is generally performed using a Ag or Fe–Mo oxide catalyst (see ref (29) and references therein). A large fraction of methanol is unfortunately lost to byproducts. What byproducts would students expect? It is likely that, in parallel to desorbing, formaldehyde reacts further on the Pt catalyst surface:The surface-bound CO may either desorb intact or, depending on the availability of surface-bound O atoms, also be oxidized to CO2, thus giving complete oxidation of methanol to CO2:These processes are illustrated in Figure .

Figure 2

Schematic illustration of catalytic dehydrogenation of H2CO on a Pt surface in combination with potential oxidation of surface-bound products. Reaction rate constants are labeled in accordance with reactions appearing in the text.

The oxidation of CO in reaction provides additional heat to the catalyst surface and is, in part due to its extremely high turnover frequency, probably the most studied reaction in the surface science community (see, e.g., ref (30) and references therein). Students could again use their thermodynamic table to compare the heats of reaction for the various partial and complete combustions.

A recent density functional theory study of methanol oxidation on Pt outlines the energy profile of the reaction up to the formation of CO.[16] Combined with other studies,[31,32] we estimate the energy profile for the complete oxidation on the most abundant surface plane of Pt, that is, Pt(111), in Figure . We provide it here, as in a classroom setting, it may be used to illustrate the continuing downward energetic path for this reaction beyond the initial dehydrogenation of CH3OH. Hence, the path should favor formation of CO or CO2 via reactions , 9, and 10 at the expense of H2CO desorption, that is, reaction . As CO and CO2 have no smell, they go by undetected to a demonstrator relying only on olfactory detection of gaseous products. This allows a teacher to pose the question how one could experimentally investigate which products are made and how much of each of them. If previously discussed, students may give answers, for example, mass spectrometry and infrared spectroscopy for identification or quantitative analysis of gaseous products. In the Experimental Section, we provide data that may be used and also serve to show that the use of a single experimental technique may often not allow for answering scientific questions in full.

Figure 3

Energy profile for complete catalytic combustion of CH3OH on Pt(111) based on refs (16, 31, and 32).

What is the Cause of the Explosions?

Besides the incomplete consideration of surface chemical reactions, the literature is only speculative on the origin of the gas phase explosions that may be induced in this demonstration. Without providing evidence, it is generally ascribed to methanol vapor.[5,6,8] However, considering the reaction mechanism provided above, H2 is likely present in the gas phase. The relative rates of H2 production from reaction and the consumption of atomically bound H in reaction will vary with surface temperature. Reactions and 8 provide the atomic, surface-bound H atoms. Hence, the H2 (g) concentration will vary with surface temperature and may just as well lead to a situation where the gas phase explosion limit of an H2/O2 mixture is reached:The mechanism of this reaction in the gas phase is complex, involves multiple radical species (see, e.g., ref (33)), and may be initiated by H or O atoms emitted from the Pt catalyst at high temperature. Pt is used in scientific laboratories as a source of gas phase O atoms from dissociative adsorption of O2 followed by emission of O (g). The same mechanism probably produces H atoms, although generally one would use hot W as the source of H atoms as tungsten is considerably cheaper and has a higher melting point. Notably, the concentration range where methanol forms an explosive mixture (between the lower and higher explosion limit), is much narrower (6.7–36.0 vol %) than for hydrogen (4.0–75.0 vol %).[34] The expected vapor pressure of methanol inside the flask with a liquid temperature of 50 °C is 0.55 bar. Hence, the gas mixture is too rich in methanol to be explosive, which supports the suggestion that hydrogen is responsible for the explosion. Although this may exceed the appropriate level for a discussion in a high school situation, the essential idea that the composition of the gas mixture inside the vessel depends on the details of gas flow, catalyst temperature, etc. are surely within reach. One could ask students to formulate a hypothesis. If standard analytical techniques have been discussed previously, students may also be asked how they would investigate the chemical composition of the gas mixture in the vessel, for example, prior to, during or after the explosions, and what results would prove their hypothesis.

In the following, we describe how we have studied this demonstration with second-year undergraduate students at Leiden University in The Netherlands using parts of regular scientific (ultrahigh vacuum, UHV) equipment (see, e.g., ref (35)) for mass spectrometry (MS) and infrared (IR) spectroscopy. Such equipment is generally not available at high schools. For this reason, we present our data such that they may be used as part of a classroom discussion or in written examples or exercises. Also, we have experimented with much simpler equipment and indicator reagents that are available at high schools and easily provide qualitative proof and allow students to arrive at the same conclusions regarding the chemical background of the exploding flask demonstration.

Experimental Section

Figure schematically represents our system used to probe gaseous products during Pt-catalyzed methanol oxidation by MS. It consists of a glass reaction tube with two necks positioned on opposite sides and a glass-to-metal connection at the top. Through the top flange, a stainless steel tube protrudes into the vessel. It samples the gas mixture just above the catalyst. The sampled gas stream is split. One part is led toward a differentially pumped quadrupole mass spectrometer (QMS, Pfeiffer QME200). The other part enters a bubbler, preventing the system from pressurizing and capturing potentially toxic, water-soluble vapors. The liquid methanol is kept at 50 °C using a water bath and thermostat (Heidolph MR Hei-Standard, not shown in Figure ). Just above the liquid surface, a polycrystalline Pt gauze (0.04 mm diameter wire) is positioned through one of the two necks. It is heated in a blue butane flame prior to insertion. A type-K thermocouple, spot-welded to the catalyst, measures the temperature. It is read out using a PID controller (Eurotherm 2416). To control the O2 concentration, a mixture of O2 and Ar is provided through the other neck into the reaction vessel using flow controllers (Brooks F-201C and Brooks 5850TR). The Ar carrier gas replaces the unreactive N2 from air and serves as an internal standard in case the QMS shows intensity drift. It also reduces the chance of formation of NO during explosions. The volume of the system is approximately 800 mL.

Figure 4

Setup used to sample the gas mixture for MS. The temperature of the Pt gauze catalyst is read out using a thermocouple. The overpressure is released via a bubbler, while the composition of the product stream is determined by a differentially pumped mass spectrometer.

An FTIR spectrometer (Bruker VERTEX 70) is used for IR identification of gaseous products. A rotary vane pump evacuated an IR cell with NaCl windows placed inside the FTIR spectrometer. The vacuum was used to draw the gas mixture from the standard demonstration equipment (as depicted in ref (1)), where we replaced the Erlenmeyer flask with a Büchner flask. A short tube with a valve connected the IR cell and the Büchner flask, ensuring little loss and a short transfer time. Gas was drawn into the IR cell when the catalyst was glowing hot or immediately after explosions when the Pt catalyst was rapidly cooling. IR experiments were performed with methanol kept at a constant temperature of 50 °C and a preheated Pt catalyst wire.

The presence of CO2 as a product may be verified qualitatively with limewater. A transparent, saturated solution of Ca(OH)2 becomes turbid in the presence of CO2. Figure schematically illustrates how we adapted the demonstration equipment to this purpose. We used a three-necked flask, positioned with a stand, and containing a layer of methanol kept at 50 °C by a water bath. One neck of the flask was connected to a bubbler filled with limewater. Through another neck, compressed dry air was introduced. A preheated Pt catalyst was hung from a septum, which closed the system.

Figure 5

Schematic drawing of the setup used for indicating CO2. The compressed air drives the reaction mixture through the limewater.

Similar to limewater indicating CO2, Tollens’ reagent can be used to show the presence of aldehydes.[36,37] This reagent was prepared by adding excess 2 M KOH (BOOM) solution to 0.05 M AgNO3 (ACROS ORGANICS) solution forming insoluble Ag2O, and then adding excess 2 M NH4OH (Riedel-de Haën) forming soluble [Ag(NH3)2]NO3. This complex is colorless in solution. Reduction to metallic silver by aldehydes yields a dark, metallic coloring of the solution. We used the equipment shown in Figure with Ca(OH)2 being replaced by Tollens’ reagent.

Hazards

The risks related to the exploding flask demonstration in general have been described extensively in the literature.[1,2,4,9] In short, the used chemicals and the reaction products pose risks due to their toxicity and flammability. The demonstration should be performed in a well-ventilated area, preferably a fumehood. Extra care should be taken while heating the catalyst or alcohol with a flame. Some additional remarks are necessary for the specific extensions of the experiment described here. When using a closed system, care should be taken not to overpressurize the system and to prevent the occurrence of explosions of the gas mixture. It is recommended to put appropriate shielding in place, especially when tuning the gas flow. Tollens’ reagent should not be stored longer than strictly necessary as the highly explosive silver nitride is formed during its degradation.[36]

Results and Discussion

Results Using Scientific Laboratory Equipment

Figure shows background-corrected mass spectra obtained without (blue) and with (red) the hot Pt catalyst present in the reaction vessel under otherwise identical conditions while no explosions occurred. The two spectra are normalized to the Ar partial pressure using the integral of the peak at m/z = 40 (Ar+). The appropriateness of this correction is demonstrated by the resulting perfect overlap of the peak appearing at m/z = 20 (Ar2+). Figure shows the tenth spectrum after insertion of the Pt catalyst in comparison to the spectrum without the catalyst from a continuous series of spectra, each requiring 60 s to collect. After this amount of time, initial changes in the gas composition resulting from letting in air when inserting the catalyst have mostly disappeared, the spectrum is stable and may be interpreted quantitatively. The two spectra differ clearly with respect to m/z = 2, 28–32, and 44. Minor differences are also observed for m/z = 18, 17, 16, 15, and 12. The peaks in the regime m/z = 28–32 require a detailed analysis as they may result from multiple species.

Figure 6

Mass spectra of the gas composition in the reaction vessel for methanol heated to 50 °C and a 45 vol % O2 and 55 vol % Ar stream at 83 mL/min flow rate without (blue) and with a Pt catalyst at 660–680 K (light red) present.

Figure , without the chemical labels, may be used to test students’ ability to interpret mass spectra and discuss the implications toward the chemical background of the exploding flask demonstration. For example, the spectrum without the catalyst present may be used to assign chemicals introduced into the vessel; m/z = 40 for Ar+, m/z= 32 for CH3OH+, and the same for O2+. Subsequently, the changes in the spectrum can be assigned. The large peak at m/z = 2 can only be assigned to H2+ and the peak at m/z = 44 is assigned to CO2+. In relation to the chemical mechanism underlying the demonstration, the appearance of peaks at m/z = 2 and 44 when the Pt catalyst is inserted proves that H2 and CO2 are formed catalytically by Pt in significant quantities. The minor increase for m/z = 18 may be asked to be ascribed to prove the formation of H2O+. In case MS fragmentation has been discussed (as, e.g., in Dutch VWO and the International Baccalaureate Chemistry syllabi), the increase in m/e = 12 (C+), 16 (O+) may be asked to be related to fragmentation of CO2+ and 17 (OH+) and 16 (O+) to H2O+. Other minor changes in the spectrum that may be related to catalytic formation of CO and H2CO require more scrutiny of the spectrum as described below.

Figure shows three mass spectra over the range of m/z = 27 to 33. Let us consider the possible contributors to these peaks in the red spectrum, that is, when there is no chemical reaction for a lack of the Pt catalyst. First, CH3OH is present abundantly due to heating the liquid to a temperature only 15 K below its boiling point. The parent ion of methanol, H3COH+, (m/z = 32) is rather unstable and fragments in the QMS ionizer to H3CO+ (m/z = 31), H2CO+ (m/z = 30), HCO+ (m/z = 29), and CO+ (m/z = 28). Hence, each peak in the red spectrum contains a contribution of methanol. The peak at m/z = 32, however, also contains a contribution of O2+, which is added as a reactant via the O2/Ar stream.

Figure 7

Mass spectra of methanol at 50 °C with the Pt catalyst in for 300 s (dark red), 600 s (light red), and without catalyst (blue).

When the Pt catalyst is added to the system, consumption of CH3OH would lead to a drop to the intensity of all peaks in this part of the spectrum. However, production of H2CO (g) from consecutive reactions and 4 would (approximately quantitatively) counterbalance the drop for m/z = 30, 29, and 28 from creation of H2CO+, HCO+, and CO+. Catalytic production of CO via reactions and 9 would also compensate at m/z = 28 (CO+). The same peak however also reflects CO2 from fragmented CO2+. Hence, catalytic dehydrogenation and oxidation would probably not affect peaks at m/z = 28, 29, and 30 much. However, this would not be the case for m/z = 31, which is unique to CH3OH. One can easily image a question for students regarding this exception.

As can be judged from the spectrum in Figure , the experiment suffered from another complication that makes it difficult to judge by MS whether CO and H2CO are produced. All peaks actually increase. This is the result of local heating of the liquid methanol when the hot Pt catalyst (its temperature being on the order or 500–1000 K) is introduced ∼1 cm above the liquid–gas interface. The increased vapor pressure affects all peaks in a nearly identical way. There is only one exception when judging the differences after 300 (dark red) and 600 s (light red). Although for all peaks between m/z = 29 and 32, the methanol contribution has dropped a minor amount (dark being lower than light red), the change is inversed for m/z = 28. It is tempting to consider this proof that CO is being formed from CH3OH. However, only the very bright student will recall that changes in m/z = 28 may also reflect variations in production of CO2. Hence, MS only provides clear evidence that the exploding flask demonstration produces CO2 and H2 and hints at the production of H2O.

As vibrational spectroscopy is generally also discussed in high school syllabi, we investigated the product gases using an FTIR. Integrated intensity bands for CH3OH and H2CO are rather similar and also CO2 and CO are easily detected with characteristic frequencies uniquely identifying each species. Figure a shows IR absorbance spectra of gas drawn from the Büchner flask before introducing a Pt wire into the system and when the catalyst is present and glowing. Without the catalyst, strong absorbances near 1000, 1600, 3000, and 3700 cm–1 appear. The peaks around 1000 and 3000 cm–1, as well as the smaller peak around 1300 cm–1, result from methanol vapor. The bands around 1600 and 3700 cm–1 are related to the presence of gas phase water. For comparison, the spectra were normalized to the peak around 1000 cm–1, which led to a good overlap of the other methanol peaks.

Figure 8

(a) IR spectra of methanol at 50 °C without (blue) and with (red) a glowing Pt catalyst; (b) enhanced IR spectra of methanol at 50 °C without (blue) and with (red) a glowing Pt catalyst.

After introducing the glowing Pt wire, the gas phase water peaks increase in intensity and several additional peaks are observed. The peaks around 2380 and 800 cm–1 can be ascribed to CO2, confirming our conclusion from the mass spectrometry measurements. In addition, when zooming in at the area around 2140 cm–1 (shown in Figure b), the characteristic absorbance of CO is found. The P- and R-branches are even rovibrationally resolved. The broad feature around 3500 cm–1 results from liquid water, which condensed on the windows of our IR cell, its intensity also varying in each experiment. Hence, the combination of IR and MS studies prove that the exploding flask demonstration produces CO2, CO, H2, and H2O.

If formaldehyde was present in the gas phase at concentrations comparable to CO, its characteristic peaks near 2850, 1750, and 1165 cm–1 should be observable. From these, only the latter has no overlap with the other peaks in the spectra. When zooming in at the area around 1165 cm–1, no peak is observed. Even the integrated spectra, which corrected for the absorption cross-sections contain quantitative information on the detected molecules, show no indication of a peak in this area. Thus, the only possible conclusion is that the amount of formaldehyde produced is extremely small.

When allowed to use a reference book, provided with a table of typical absorption frequencies, or with already assigned peaks in these spectra (as in Figure ), students may be asked to discuss the absorption spectra prior to and after insertion of the catalyst. Their assignment of the peaks and consideration of the changes allows them to arrive at the same conclusion that, beyond olfactory identification of formaldehyde, there is no evidence of it being an important product or intermediate in the chemical background of the exploding flask demonstration.

Finally, Figure shows IR spectra taken when the catalyst was glowing (red) and immediately after an explosion occurred (green). Much like the spectra in Figure , the characteristic peaks of products CO and CO2 appear as well as remaining reactant CH3OH. As the amount of gas sampled from the reaction vessel varied slightly from experiment to experiment, our spectra should not be interpreted quantitatively here. However, relative changes allow the student to get one step further in understanding the demonstration. To do so, we normalized the spectra in Figure to the CO2 peak at 2380 cm–1. From these data, it can be determined that after the explosion, the CO/CO2 ratio has increased with a factor 3.5. This suggests either a lack of O2 required for CO oxidation (presumably because of its consumption in the gas phase explosion) or a changed product selectivity resulting from different temperature-dependencies of k9 and k10 as the temperature of the catalyst drops on the order of several hundreds of degrees after the explosion. On a similar note, one might argue that the relative amount of H2O (g) decreases, but this difference is much more subtle. As in the case with the glowing catalyst, the explosion does not produce a detectable amount of formaldehyde.

Figure 9

Infrared spectra of the gas mixture resulting from oxidation of methanol with Pt catalyst while the catalyst is glowing (red) and immediately after an explosion.

Result Using High School Accessible Equipment

The experiment performed with the apparatus shown in Figure with methanol kept at 50 °C showed turbidity of the limewater solution within minutes. Although qualitative by nature, the experiment reinforces that CO2 is a major product of the chemical reactions occurring in this demonstration. When repeating the experiment with Tollens’ reagent, it also turned cloudy. However, we did not find any sufficient reference to the selectivity of this reagent to formaldehyde, other aldehydes, and other species possibly present in our gas mixture. Thus, we verified whether the same reagent would also react with CO, CO2, methanol, and the used pressurized air. Unfortunately, Tollens’ reagent also reacts with CO. Hence, it cannot be used to discriminate between CO and formaldehyde. However, with a working catalyst the reagent turned cloudy within a few minutes, while this took a few seconds for the CO experiment. This suggests that the concentration of molecules reacting with Tollens’ reagent in our reaction mixture is on the order of a few per mille. As formaldehyde becomes irritating at concentrations above 1.0 ppm, it seems highly likely that CO is formed at significant concentrations.

Conclusion

Experimental results clearly indicate that the surface chemistry underlying the heterogeneous catalysis demonstration is significantly more complex than claimed previously. From all possible paths in the network of parallel reactions of surface-bound intermediates (desorption versus decomposition of CH2Oads, desorption versus oxidation of Hads, and desorption versus oxidation of COads), two reaction paths seem dominant: recombinative desorption of H2 produced from sequential dehydrogenation of CH3OH to COads and the complete oxidation of the methanol’s carbon atom to CO2. IR spectroscopy clearly identifies CO as a product. The selectivity toward partial oxidation seems higher following an explosion. This is rather logical as O2 must be consumed in the explosive gas phase oxidation processes, which compete with dissociative adsorption of O2 on the catalyst surface. The abundant presence of H2 in the gas phase and the very low H2/O2 explosion limit make it likely that the catalytic dehydrogenation of methanol is the cause of formation of an explosive mixture. The cause is not the presence of methanol and oxygen in the gas phase, which do not change much throughout the cycle as evidenced by only minor changes in mass spectra. Finally, formation of CH2O is confirmed by its typical smell when opening the apparatus, but its absence in IR spectra and very low olfactory detection limit suggest that it is a minor side product.

We conclude with the notion that this demonstration has beautiful complexity and wide applicability veiled by an apparent simplicity. It requires very few materials, available at nearly every high school, and provides clear, straightforward observations. Addition of the limewater bubbler can be used to show that CO2 is formed in addition to the CH2O which can be smelled. This can be used as an illustratory introduction of topics, for example, parallel reactions in kinetics and selectivity, in addition to it visualizing important aspects of heterogeneous catalysis. Other topics, for example, stoichiometry and gas phase reactions (explosions), mass and heat transfer, diffusion, and convection are also directly related and can be incorporated in discussions with students. Finally, rather simple variations of the experiment allow for extensive student research, either as an individual project or for entire classes. For example, one could research the effect on the intensity or frequency of the explosion by changing, for example, temperature of the liquid fuel, the position of the catalyst relative to the liquid surface and the inlet of the O2, and the shape and size of a divider.[1,3] In our parallel tested demonstration in this issue, we show that a large number of different fuels can be used and which transition metals catalyze their combustion using the same equipment (see ref (9)). Readers interested in kinetic studies of CH3OH reactions over well-defined Pt surfaces are referred to a recent study performed under somewhat lower pressure conditions.[38]