On the chemistry of 1-pyrroline in solution and in the gas phase

1-Pyrroline has a highly characteristic odor, which is employed by living organisms for chemical signaling and other purposes, but the mechanism whereby this odor is formed remains poorly understood. Here we used a combination of ambient mass spectrometry (AMS) and nuclear magnetic resonance (NMR) spectroscopy to experimentally address the mechanistic aspects of 1-pyrroline volatility and other controversies regarding the chemistry of this compound. Our results indicate that in solution the volatility of the monomer species is significantly higher than that of the trimer species, and 1-pyrroline is evaporated mainly in its monomer state. Neat 1-pyrroline is essentially the pure trimer and displays ca. 100-fold lower evaporation rate than the monomer state in solution. In the gas-phase the trimer species is irreversibly decomposed into monomer species. Under equilibrium conditions the vapor of 1-pyrroline entirely consists of monomer species. The evaporation rate of 1-pyrroline in water has a step-wise dependence on the solution pH, the abrupt increase in volatility (>1,000-fold) occurring around the pKa value of 1-pyrroline (6.8). The pronounced step-wise dependence of 1-pyrroline volatility around neutral pH may also be an important evolutionary factor allowing living systems to regulate the odor strength from very weak to very strong with minimal efforts.

The NMR spectral data for both the monomer and trimer obtained from 1-pyrroline sample with different concentrations and solvents indicate equilibration of the monomer with its trimer form on the timescale of dissolution, which is in agreement with earlier studies 13 . This equilibrium is concentration dependent 16 . The spectral data for the both monomer and trimer of prepared 1-pyrroline in 100-fold diluted solution (   The high linearity indicates that the equilibrium between T s and M s obeys third-order kinetics in a wide concentration range. The equilibrium kinetics has a very high speed and is readily re-established upon dilution, as has been shown by earlier studies 13 . The results indicate that at 300 K T s is more thermodynamically stable than M s . From our results we calculated that the free energy of conversion from T s to M s in DMSO-d 6 is equal to ΔG = −RT ln(K eq (s) ) ≈ 3.20 kcal/mol, which is higher than the earlier estimate for ΔG ≈ 2 kcal/mol obtained for 1-pyrroline in DMSO-d 6 by Baker et al. 13 . Our results therefore suggest that the relative stability of T s in solution is higher than that based on earlier estimates. The free energy difference between M s and T s in D 2 O and CDCl 3 was ΔG ≈ 3.2 kcal/mol and ΔG ≈ 1.5 kcal/mol accordingly. Ambient MS analysis. The purity of synthesized 1-pyrroline sample was additionally confirmed by the ambient atmospheric pressure chemical ionization mass spectrometry (APCI-MS) analysis of vapor phase (Fig. S4). Only the presence of 1-pyrroline monomer could be detected in the form of a protonated cation (m/z 70). No trace of the trimer species could be detected even from the pure undiluted sample. If present in the vapor phase, 1-pyrroline trimer would have been observed as a protonated cation at m/z 208. The lack of T g signal in APCI-MS of 1-pyrroline vapor is consistent with the results of earlier microwave spectroscopy experiments that also indicated the lack of T g in 1-pyrroline vapor 15 . Consistent with the results of NMR analysis (Figs. 2, S1, S2), our APCI-MS analysis also indicate the high purity of the synthesized 1-pyrroline sample. No trace of possible volatile contaminants, such as protonated 4-aminobutanal (m/z 88) or protonated pyrrolidine (m/z 72), was revealed in the mass spectrum (Fig. S4). Fig. S5 shows the response range of 1-pyrroline APCI-MS signal (m/z 70) as a function of 1-pyrroline concentration in water solution in the range from 10 to 400 ppb and in working DMSO solution in the range from 1 to 100 ppb. Both plots reveal high linearity between signal intensity and 1-pyrroline concentration in the low concentration range. In order to measure the concentration of 1-pyrroline in concentrated solutions, the corresponding samples were first diluted down to the linear range (Fig. S5), and the original concentration in the concentrated solution was calculated from the signal intensity in the linear range and the known dilution rate.
Evaporative loss of 1-pyrroline in water. Evaporative loss of 1-pyrroline from 4 mL solution in water with a concentration of 10 ppm (pH 6.8) as a function of vacuum (−0.1 MPa) evacuation time was measured by APCI-MS analysis (Fig. 4). The sample was accordingly diluted to 200 ppb before the analysis to the linear response range of the instrument (Fig. S5a). The derived loss of 1-pyrroline was ca. 60% per 30 min. Water loss was ca. 12% weight per 30 min and was replenished every 30 min to the original level. Based on the exponential fit of evaporation kinetics ([C s ] = [C s ] 0 exp(−αt)), the evaporation rate (α) of 1-pyrroline in water is estimated to be ca. 7 times higher than the evaporation rate of water under the same conditions. No evaporative loss of 1-pyrroline was observed over the same time period in the reference experiment when the bottle was open under air without vacuum evacuation.
The effect of pH value. Fig. 5a shows the pH value of 1-pyrroline solution in distilled water as a function of 1-pyrroline concentration. Significant elevation of pH from ∼7 starts at ca. 100 ppm concentration and reaches the level of ∼10 at the concentration 100,000 ppm. The onset of pH elevation roughly correlates with the onset of [T s ] (Fig. 3a). Therefore, we propose that the elevation of pH at high 1-pyrroline concentration is related to the increase of [T s ]/[M s ] ratio and the increased contribution of T s species. Fig. 5b shows the APCI-MS signal intensity of 1-pyrroline in 200 ppb aqueous solution as a function of solution pH value. The pH value was adjusted using PBS buffer, sodium hydroxide solution and hydrochloric acid solution. At 200 ppb only M s is present in solution (Fig. 3a). The pH dependence displays a pronounced step-like shape: more than three-order increase in signal intensity is observed between pH = 6 and pH = 8. This change in intensity corresponds to the transition through the pKa of 1-pyrroline (6.8). 1-Pyrroline, being a weak base, easily binds a proton in water. The equilibrium ratio between neutral and protonated 1-pyrroline species in water is determined by the pH value of the solution. The higher the pH the lower is the degree of 1-pyrroline protonation. Since 1-pyrroline evaporates only in its neutral form, the volatility of 1-pyrroline increases with the pH value of the solution. The growth in 1-pyrroline volatility with pH (Figs. 5b and S6) is most pronounced around pH 6.8, which corresponds to the pKa of 1-pyrroline. At this pH value, the ratio between neutral and protonated 1-pyrroline species is equal to 1:1. Boddeker et al. reported similar effect of pH on the evaporative enrichment of phenol 18 . The authors found that phenolic enrichment depends on the pH value of solution, which determines the equilibrium ratio between undissociated phenol and phenolate ion. Only undissociated phenol evaporates, whereas phenolate ion does not evaporate.
The comparison of relative M s and T s volatility. The assessment of 1-pyrroline volatility based on the amount of evaporative loss is complicated by the poor long-term chemical stability of 1-pyrroline solutions 13 under air and high pH sensitivity 14 . For the unbiased comparison of M s and T s volatility we used aprotic solvent (DMSO) and vacuum conditions in order to eliminate the pH factor, accelerate evaporation and eliminate the possibility of reaction with air. Fig. 4b displays the relative loss of 1-pyrroline in DMSO as a function of concentration over 30 min evaporation assisted by vacuum pumping (−0.1 MPa) at room temperature. The observed relative loss of 1-pyrrolnie was ca. 80% at concentrations below 10,000 ppm. At higher concentrations the rate of evaporative loss showed decrease with 1-pyrroline concentration. The evaporative loss of 1-pyrroline at 200,000 ppm concentration over 30 min is equal to ca. 20%, which corresponds to ca. 7 times lower evaporation rate of 1-pyrroline than in diluted solution. This estimate is consistent with the 7-times lower relative proportion of M s in 200,000 ppm solution than in diluted solutions measured by NMR (Fig. 3a). Apparently, the evaporative loss of 200,000 ppm solution is mainly caused by M s species, and the decrease of evaporative loss with 1-pyrroline concentration occurs mainly due to the decreased relative concentration of M s . Our experiments thereby indicate  that M s is much more volatile than T s in DMSO solution. The pure 1-pyrroline, which is 100% trimer based on NMR data, showed only ca. 1% evaporative loss under the same experimental conditions. Based on these results, we estimate that the evaporation rate of neat 1-pyrroline (100% T s ) is ca. 160 times lower than the evaporation rate of diluted (<10,000 ppm) 1-pyrroline in DMSO (100% M s ) under the same conditions. Stability of T g . Density functional theory calculations at 6-31 + G(d, p) level show that the free energy of M g is 18.8 kcal/mol lower than that of T g , indicating that T g is less stable than M g in the gas phase from the view of thermodynamics. The results of our calculations are consistent with the earlier estimate of 20 kcal/mol by Wiberg et al. 12 . The equilibrium [M g ] and [T g ] are interconnected via the equilibrium constant K eq (g) = [M g ] 3 / [T g ] = e −ΔG(g)/RT = 4.8 × 10 13 (at 300 K). The estimate vapor pressure of neat 1-pyrroline material at 300 K (liquid) is 0.005 atm, as derived by extrapolating the earlier measured data on the vapor pressure of neat 1-pyrroline material at 318-339 K 12 . The partial pressure estimates of M g and T g for neat 1-pyrroline are: P M ≈ 0.005 atm; P T ≈ 10 −27 atm, i.e., one liter of air contains on average ∼10 −5 molecules of T g . In other words, T g is not present in the 1-pyrroline vapor under equilibrium with solution at any physically significant amount. According to earlier studies, the olfactory threshold concentration of 1-pyrroline in water is ca. 20 ppb 1 . This concentration roughly corresponds to the partial 1-pyrroline vapor pressure P threshold ∼ 10 −8 atm, which is ca. 20 orders of magnitude higher than the estimated equilibrium vapor pressure of T g . Overall, our results indicate that T g is extremely unlikely to have any contribution to the odor of 1-pyrroline under equilibrium conditions. The odor of 1-pyrroline is most probably entirely contributed by M g .
Edwards et al. reported that no trimer could be detected in the gas phase by IR and microwave spectroscopy of 1-pyrroline vapor 15 . Interestingly, our GC-MS analysis of 1-pyrroline in CH 2 Cl 2 , apart from the major M g signal (m/z 69) at 4 min, also yielded the signal m/z 207 eluted at 8.2 min, corresponding to the radical cation of T g (Fig. 6). This appears to be the first observation of 1-pyrroline trimer in the gas phase. The major characteristic ions at m/z 41, 42, 68, 69 were observed in the mass spectrum of 1-pyrroline monomer at 4 min, which are consistent with the earlier literature reports 19,20 . As shown in Fig. 6, the T/M ratio in GC-MS experiments is greatly different from that in NMR experiments discussed above at the same solution concentration. A possible reason may be the different ionization efficiency and fragmentation behavior following ionization for the monomer and trimer. Another possible explanation is the partial decomposition of the trimer species, e.g., due to the high inlet temperature (150 °C). Therefore, the ratio of m/z 207 to m/z 69 in GC-MS cannot reflect the true T/M ratio in solution at the same solution concentration. However, the mere observation of m/z 207 (T g ) in GC-MS is very curious as it indicates that under GC-MS conditions T g appears to be rather stable. In fact, the trimer should be in gas state after GC injection and is somehow able to withstand very high temperature of the inlet (150 °C). Detailed studies may be required to evaluate the surprising stability of T g revealed in GC-MS experiments and understand the factors responsible for such stability.

Discussion
The article published in 1992 by Baker et al. presents by far the most detailed description of 1-pyrroline chemistry 13 . However, the current knowledge is yet far from complete. In fact, even the third-order equilibrium kinetics between M s and T s was postulated but has not been rigorously proven in the work by Baker et al. due to the lack of titration measurements. Our NMR data obtained at different dilution rates of 1-pyrroline in different solvents provides a clear-cut evidence for the third-order equilibrium kinetics (Fig. 3). The analysis of obtained concentration curves suggests that the relative stability of T s in solution (ΔG ≈ −3.2 kcal/mol) is considerably higher than the earlier estimate by Baker et al. (ΔG ≈ −2 kcal/mol) obtained using NMR data from a single concentration point 13 . Another poorly understood aspect is the low chemical stability of 1-pyrroline in solution, e.g., in water or DMSO. The interpretation of experimental observations is greatly complicated by the equilibrium between the monomer and trimer 1-pyrroline species in the solution. Our experiments carried at different 1-pyrroline concentrations allowed independent consideration of the monomer and trimer chemistry and indicated that 1-pyrroline displays three distinct volatility behaviors: when present in neat form, in concentrated solution (>10,000 ppm), or in diluted solution (<1,000 ppm). Below we discuss these three cases separately.
1-Pyrroline in neat form is stable and exists purely in the trimeric form 13 . According to our experimental results, the volatility of neat 1-pyrroline is ca. 13 times lower than that of pure water. Our estimate is consistent with the earlier publications that reported the vapor pressure of 1-pyrroline at 300 K equal to 0.005 atm 12 , whereas the vapor pressure of water at 300 K is equal to 0.03 atm. The low volatility of neat 1-pyrroline is most probably related to the strong intermolecular bonding in neat material. Neat 1-pyrroline is a viscous liquid. The odor of neat 1-pyrroline is contributed by M g produced by the spontaneous decomposition of T g .
1-Pyrroline in concentrated solution (ca. 10,000 ppm -200,000 ppm) exists in equilibrium between the monomer and trimer states. The equilibrium obeys third-order kinetics. At room temperature T s is more stable than M s by ΔG ≈ 3.2 kcal/mol (in DMSO-d 6 ). M s in solution is ca. 100 times more volatile than T s . Accordingly, volatility of 1-pyrroline increases with the degree of dilution due to the increase of M s /T s ratio. At 10,000 ppm the evaporation rate of 1-pyrroline from water exceeds the evaporation rate of water by ca. 7 times. Note that the uneven evaporation process does not alter the equilibrium M s /T s ratio due to the rapid equilibrium kinetics: the evaporative loss of M s from solution is continuously replenished to the equilibrium level by the rapid decomposition of T s solutes 13 . The poor chemical stability of concentrated 1-pyrroline solutions discovered in earlier studies 13 is probably related to the high chemical reactivity of T s with the molecules of solvent and air. Baker et al. reported that 1-pyrroline in water solution (~200,000 ppm) stored at ambient conditions in 5-mm NMR sample tubes with loosely covered caps revealed approximately 20 additional 13 C absorptions after three weeks of storage 13 . A rigid layer was formed at the surface of the degraded samples, suggesting that some type of polymerization occurred. We also observed the formation of a rigid upper-layer for 1-pyrroline in concentrated water solution. These observations indicate that the chemical instability of 1-pyrroline in concentrated solutions may be caused by the reactivity of 1-pyrroline with solvent and air. More research of these reactions is still needed to thoroughly understand the mechanism of 1-pyrroline chemical degradation.
1-Pyrroline in diluted solution (<1,000 ppm) exists mainly in the monomer state, and its volatility is independent on the degree of dilution (Fig. 4b). Earlier study by H. Poisel demonstrated that monomeric 1-pyrroline is relatively stable in dilute solution 21 . The higher chemical stability of diluted 1-pyrroline solutions (i.e. in the absence of trimer) suggests that the reactivity of concentrated 1-pyrroline solutions is mainly accounted by the trimer species. The study of 1-pyrroline reactivity, in particular ambient condensation in water, is a subject for further research.
Of particular interest is the step-wise dependence of 1-pyrroline volatility on pH (Fig. 5b). The abrupt increase of volatility corresponds to the transition through the pKa of 1-pyrroline (6.8). At pH < 6.8, 1-pyrroline is mostly protonated and displays lower volatility. At pH > 6.8, 1-pyrroline is mostly neutral and displays more than three-order of magnitude stronger volatility. While pKa for the majority of known nitrogenous volatile bases pKa is around 10 or 11 1 , the neutral pKa value of 1-pyrroline is rather unusual. Therefore, even minor pH alterations around neutral pH will be the cause of pronounced variation in odor strength. This may be an important evolutionary factor allowing living organisms to regulate the odor strength from very weak to very strong with minimal changes in the physiological pH.

Methods
Synthesis of 1-pyrroline. Several methods for the synthesis of 1-pyrroline have been introduced in earlier reports, such as silver (I)-catalyzed oxidation of pyrrolidine [22][23][24] , acid hydrolysis of 4-aminobutanal diethyl acetal 25,26 . In our study, 1-pyrroline was synthesized according to the method described by Ogawa et al. 24 based on easily available raw materials (silver and pyrrolidine) and was chosen for its rapidity and simplicity. Briefly, in a round-bottom flask covered with aluminum foil, a catalytic amount of AgNO 3 (Shanghai Aladdin biochemical Polytron Technologies Inc, purity >99%; 0.37 mmol, 64 mg) was first added to a solution of pyrrolidine (Shanghai Aladdin biochemical Polytron Technologies Inc, purity >99%, 75 mmol, 5.33 g, 6.2 mL) in water (75 mL), after which NaOH pellets were added (0.15 mol, 6 g). To this mixture, a 25% aqueous solution of Na 2 S 2 O 8 (75 mmol, 17.8 g, 70 mL) was added dropwise at ice bath. The reaction mixture was stirred for 3 hours at ambient temperature and was then extracted with CH 2 Cl 2 (100 mL) and saturated NaCl water solution (50 mL) three times. After drying over Na 2 SO 4 and removal of the solvent, the mixture was dissolved in ethyl ether and filtered through the plug of neutral alumina and was evaporated to dryness in vacuum. The resulting transparent viscous oil was used for NMR and MS analysis. NMR analysis. NMR spectra were measured on a Bruker DRX 500 Avance (1H, 500.13 MHz and 13C, 125.77 MHz) spectrometer at 298 K with tetramethylsilane (TMS) as the internal standard. Standard Bruker automated acquisition programs were used for all experiments. All one-dimensional spectra were composed by 64 K data points. The spectral width for 1 H was set at 6024 Hz and for 13 C at 18519 Hz with acquisition times of 5.439 and 1.769 sec, respectively. The resolution of the spectra was 0.565 Hz per point for 13 C and 0.184 Hz per point for 1 H. The numbers of scans for 1 H and 13 C were 16 and 1024, respectively. The following abbreviations were used to explain the multiplicities in Fig. 2: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet. Chemical shifts were assigned relative to TMS. Coupling constants (J) are reported in Hertz (Hz). The spectra and assignments were obtained with neat sample dissolved in different solutions (CDCl 3 /DMSO-d 6 /D 2 O) with different concentrations.
Ambient MS analysis of 1-pyrroline vapor. Ambient MS 27, 28 analysis of 1-pyrroline vapor was done on commercial ion trap mass spectrometer (LTQ-XL, Thermo Scientific, San Jose, CA, USA) using home-made corona discharge ionization 29 and extractive electrospray ionization (EESI) sources as detailed in our earlier studies [30][31][32] . The headspace volatiles of 1-pyrroline solution (8 mL) in a 10 mL centrifuge tube (Solarbio, Beijing, China) were continuously transferred into ionization region via plastic tubing (ID 0.5 mm) assisted by nitrogen gas flow (0.1 MPa, 1 L/min). Ionizing voltage in the both ionization approaches was + 4.5 kV. Pure water and DMSO were used as solvent to prepare working solutions. The angle between the discharge needle and the outer tubing was 30 °C. The distance from the tip of the ion probe to the inlet of the MS capillary was 6mm. Each sample was analyzed in three replicates and collected for 30 s in the m/z range 15-300. The reference pure water or DMSO samples without 1-pyrroline were analyzed for background correction. The reference solutions with known 1-pyrroline concentration were analyzed under identical experimental conditions to construct the calibration curve for quantification purposes.
GC-MS analysis. GC-MS analysis was performed on an Agilent triple quadrupole system (7890 B/7000C) equipped with an Agilent 7693 auto-sampler. The chromatographic separation was done on an HP-5 MS capillary column (30 m × 0.25 mm, film thickness 0.25 μm, Agilent Technologies, USA). The front inlet was kept at 150 °C in the split mode. The GC oven temperature program was as follows: initial column temperature 30 °C, held for 5 min, then programmed to 70 °C at a rate of 10 °C per minute and held for 2 min and finally programmed to 250 °C at a rate of 30 °C per minute and held at 250 °C for 3 min. Helium was used as the carrier gas at a constant flow rate of 1.0 mL/min. The injection volume was 1 μL. The MS detector was used in the electron ionization (EI) mode with an ionization voltage of 70 eV. The ion source temperature was at 230 °C. The transfer line was at 280 °C. The spectra were collected over the m/z range 30-300.
Structure calculations. Structure calculations were done using Gaussian 03 software 33 . The geometries of the target species were optimized using the density functional theory (DFT) method at the B3LYP/6-31 + G(d,p) level. The optimized structures were identified as a true minimum in energy by the absence of imaginary frequencies. Vibrational frequencies of all the key species were calculated at the same level of theory. The optimized structures were displayed by the Gauss View (05) software. The energies discussed here are the sum of electronic and thermal free energy. Calculation parameters and results are shown in output files in Supplementary Material.