Direct Conversion of Glycerol into Formic Acid via Water Stable Pd(II) Catalyzed Oxidative Carbon-Carbon Bond Cleavage

Abstract

Using our tridentate NHC-amidate-alkoxide Pd(II) complex, we developed a catalytic method for oxidative C-C bond cleavage of glycerol. The glycerol was degraded exclusively to formic acid and CO₂. Two possible degradation pathways were proposed through ¹³C labeled studies.

Glycerol, also called glycerin, is a main by-product of biodiesel production and traditional soap manufacturing processes.¹ A rapid increase in biomass conversion has produced a massive stockpile of glycerol, and its transformation to value-added chemicals has been in great demand. A number of methods have been reported to provide C₂ ~ C₃ chemical products such as glyceraldehyde, glyceric acid, hydroxypyruvic acid, tartronic acid, glycolic acid, and oxalic acid.² Because these C₂ ~ C₃ products have been short of practical use, the formation of C₁ products such as formic acid has attracted much attention for future energy applications.³ There are few examples have been introduced through hydrothermal oxidation, heterogeneous catalysts, and electrocatalytic oxidation.⁴

One noteworthy application of formic acid is the DFAFC (direct formic acid fuel cell), which has been of increasing popularity compared with hydrogen and methanol based fuel cells because of their ease of refuelling, efficiency, and safety. As an emerging technology, DFAFC is currently being tested by major producers of portable electronics in phones, laptops, and computers.⁵ In an effort to find a potential source of formic acid, we embarked on the development of new oxidative carbon-carbon bond cleavage methods of glycerol mainly because the previously reported conditions failed to degrade glycerol as shown in scheme 1.

Scheme 1.

Limitations of known conditions on glycerol degradation.

Representative examples included Isbell’s alkaline hydrogen peroxide and our hydrogen peroxide/ammonia water conditions.⁶ Under these conditions, various aldoses were oxidatively transformed to formic acid whereas glycerol (1) barely reacted. In addition, these procedures converted ketoses to both formic and glycolic acids (2), while glycolic acid was resistant to further degradation. These shortcomings prompted us to undertake studies on oxidative degradation using organometallic catalysts.

Recently, we reported that NHC-palladium complexes including 4 were highly stable and still reactive enough to facilitate C-H activation of relatively unreactive hydrocarbons.⁷ Their high stability in nucleophilic solvents such as water and alcohol could allow for conditions amenable to oxidative carbon-carbon bond cleavage of glycerol. These processes could provide C₁ ~ C₂ products encompassing carbon dioxide, formic acid (3), and glycolic acid (2). Using known palladium catalysts and our Pd(II) catalyst 4, we evaluated the feasibility of oxidative degradation of glycerol (table 1).

Table 1.

Catalytic oxidative carbon-carbon bond cleavage of glycerol with various oxidizing agents in the presence of Pd catalysts at room temperature.^a

entry	catalyst	oxidant	yield (%)d	2/3 (× 10−5 mol)e
1	PdCl2	H2O2	trace	trace/trace
2	Pd(OAc)2	H2O2	6	0.16/0.20
3	4	H2O2	41	1.50/8.05
4	4	t-BuO2H	trace	-/-
5b	4	Oxone	10	0.39/1.52
6	4	K2S2O8	-	-/-
7c	4	O2	-	-/-

^aAll reactions were performed with glycerol (10 mg, 10.8 × 10⁻⁵ mol), Pd catalyst (5 mol%). Entry 1–3: 30% H₂O₂ (0.4 mL) in H₂O (0.1 mL) was added. In the entry 4, tBuOOH (70% in H₂O, 0.4 mL) was used. Entries 5~6: 10.8 × 10⁻⁵ mol of oxidant (entry 5: oxone, entry 6: K₂S₂O₈) in 0.3 mL of H₂O was used. Entry 7: O₂ was bubbled in 0.3 mL of H₂O solution. All reactions were performed at room temperature for 6 hours..

^bRun in H₂O (0.4 mL).

^cContinuous flow with O₂.

^dConversion yield of glycerol.

^eThe amount of each product was determined by ¹H NMR spectral analysis using MeOH as the internal standard.⁸

Regarding oxidative degradation by hydrogen peroxide, commercially available Pd complexes including PdCl₂ and Pd(OAc)₂ didn’t offer meaningful improvement over KOH or NH₄OH (entries 1 and 2).⁶ However, NHC-Pd complex 4 exhibited significant consumption of glycerol at room temperature to furnish formic acid as the major product (entry 3). We also noticed that the ratio of formic acid to glycolic acid produced was much higher than that of the reaction with Pd(OAc)₂ despite its low yield. These results might indicate our catalyst degraded both glycerol and glycolic acid unlike other Pd salts or basic conditions. We screened other oxidants such as tert-butyl peroxide, oxone, K₂S₂O₈, and molecular oxygen, most of which were ineffective (entries 4 ~ 7). Similarly to hydrogen peroxide, oxone provided a higher ratio of formic acid to glycolic acid compared to the Pd(OAc)₂ case (entries 2 and 5).

In the catalytic processes using 4, one equivalent of glycerol can produce either three equivalents of formic acid or one equivalent of formic acid and glycolic acid each. To understand how many equivalents of formic acid can be produced from glycerol, it was necessary to understand degradation pathways. In this context, we carried out the oxidative cleavage reaction using 1,3-¹³C-labeled glycerol and 2-¹³C-labeled glycerol in the presence of NHC-Pd complex 4 and hydrogen peroxide at 60 °C for 6 hours, under which conditions we tried to consume most of glycerol (cf. table 2, entry B).

Table 2.

Yields of reactions in the figure 3

entry	glycerol (%)a	glycolic acid (%)b	formic acid (%)c
B	trace	11	39
C	trace	20	50
D	19	28	57
E	0	0	61

^aRemaining glycerol/added glycerol × 100.

^bMoles of glycolic acid/moles of added glycerol × 100.

^c(Moles of formic acid/moles of glycerol × 100)/3, assuming one mole of glycerol produced 3 mole of formic acid. Moles of each product were calculated by using methanol as an internal NMR reference.

As shown in figure 1, the reaction of 1,3-¹³C-labeled glycerol (5) afforded both ¹³C-labeled formic acid (6) and unlabeled formic acid (3) in a 2 to 1 ratio, as well as a small amount of ¹³C-labeled glycolic acid on the β-carbon (7). In the case of 2-¹³C-labeled glycerol (8), a 1 to 2.5 ratio of ¹³C-labeled formic acid (6) to unlabeled formic acid (3), as well as 1-¹³C-labeled glycolic acid (9) were observed. In addition, ¹³C-NMR analysis further confirmed the assignment by ¹³C-¹²C coupling for ¹³C-labeled glycolic acid (7) (δ= 60.6 Hz at 176.0 ppm). Therefore, these results indicated that the formic acid produced contained both the secondary and primary carbons of glycerol. On the other hand, the carbonyl carbon of glycolic acid would stem only from the secondary carbon of glycerol while the carbinol carbon would originate from the primary carbons.

Figure 1.

¹H-NMR spectra for the oxidative degradation reactions of 1,3-¹³C-glycerol (A) and 2-¹³C-glycerol (B).

Since the observed products didn’t satisfy the mass balance, we supected that we lost some carbons in the form of carbon dioxide, and examined such possibility (scheme 3). In the presence of NHC-Pd complex 4, both acids gradually disappeared over time to form carbon dioxide. For example, glycolic acid led to formic acid in 10% yield after 3 hours while formic acid gave no detectable products except carbon dioxide. In addition, unlabeled formic acid (3), but no ¹³C labeled formic acid (6) was detected when (1-¹³C)glycolic acid (9) was reacted with hydrogen peroxide. It was evident that 2-¹²C and 1-¹³C in glycolic acid were incorporated into formic acid and ¹³CO₂, respectively. These results were consistent with the aforementioned glycol oxidation patterns.

Scheme 3.

Potential degradation of glycolic acid and formic acid in the presence of NHC-Pd(II) catalyst 4.

These labeled experiments suggested two possible degradation pathways. If formic acid was derived equally from all three carbons in gylcerol, the ratio of ¹³C-labeled formic acid (6) and unlabeled formic acid (3) should be 2 to 1 in figure 1 (A), and 1 to 2 in figure 1 (B). Even though the observed ratios were close to the theoretical ones, these ratios were still different from the expected ones by 10–25%. Thus, we assumed a major pathway would oxidize all three carbons to formic acid. Since small amounts of glycolic acid was detected, we considered that another pathway through glycolic acid was active concomitantly. In fact, both C-1 and C-3 in glycerol could be converted to formic acid whereas the C-2 would fail to give formic acid and instead furnish CO₂ as aforementioned. As a consequence, one could expect more ¹³C-labeled formic acid than ¹²C-formic acid in figure 1 (A) and more ¹²C-formic acid than ¹³C-labeled one in figure 1 (B), respectively.

Based on these results and previously reported studies, two possible degradation pathways can be proposed as in scheme 4.^3,9,10,11 One mechanistic pathway would form three equivalents of formic acid from each glycerol molecule while another mechanism would lead to two equivalents of formic acid and one part of CO₂ through the glycolic acid intermediate. As the incipient product, Pd-glycerol adducts could be generated and oxidized to aldehyde 10, which would undergo rapid C-C bond cleavage to release formic acid and another aldedyde 11. Further oxidative cleavage of 11 could afford the second equivalent of formic acid and formaldehyde, which eventually would generate the third equivalent of formic acid. Meanwhile, the first oxidative C-C bond cleavage product 11 can be oxidized to yield glycolic acid 2, which subsequently can be degraded to formaldehyde and CO₂, ultimately furnishing one equivalent of formic acid. In summary, dual mechanistic pathways would contribute to our catalytic processes.

Scheme 4.

Potential oxidative degradation pathways of glycerol.

In the reaction mixture, it was unable to detect any aldehyde (10, 11, or formaldehyde). This could indicate that aldehyde compound was not completely released from Pd complex but underwent C-C bond cleavage to form formic acid and another aldehyde. Even though aldehyde was released from Pd, it could be oxidized quickly by the oxidant to form acid (2 or formic acid). Therefore the following catalytic cycle could be proposed (scheme 5). Glycerol adduct 12 could be oxidized to form 13. Subsequently, one equivalent of H₂O₂ could generate one equivalent of formic acid and 14. Following third oxidation, formaldehyde adduct 15 could be formed while releasing another equivalent of formic acid. The last formic acid could be produced from this formaldehyde.

Scheme 5.

Catalyic pathway for the formation of formic acid

To seek optimal conditions, we investigated various factors including amounts of hydrogen peroxide, reaction temperatures, and reaction times. As depicted in figure 2 (A), the formation of formic acid increased upon higher concentration of hydrogen peroxide. When we raised reaction temperatures gradually (25 to 60 °C), the amount of formic acid was increased until 40 °C, then decreased at higher temperatures (figure 2 (B)) presumably due to the overoxidation to carbon dioxide. Additionally, longer reaction times were not sufficient to enhance the formation of formic acid (figure 2 (C)). In addition, when the amount of Pd (4) was increased to 10 or 15% from 5%, overall yield of formic acid as well as consumption of glycerol dropped due to the degradation of formic acid and hydrogen peroxide by the catalyst.

Figure 2.

Concentration changes vs. volume of H₂O₂ (A), reaction temperature (B), and time (C) for the oxidative degradation of glycerol: glycerol (red square), formic acid (blue diamond), and glycolic acid (green triangle).

Because excess H₂O₂ and high reaction temperatures caused the degradation of formic acid to CO₂, we decided to employ slow addition of hydrogen peroxide at mild temperatures, and evaluated the time and temperature dependence on the yields of both formic and glycolic acids using ¹H-NMR techniques (figure 3 and table 2). As shown in figure 3 and table 2 (entries B and C), the formation of formic and glycolic acids was increased slightly by the slow addition of hydrogen peroxide. Although glycerol was completely consumed under these conditions, glycolic acid still remained. In efforts to avert glycolic acid and maximize the amount of formic acid, hydrogen peroxide was slowly added at 0 °C for 3 hours and stirred at room temperature over 8 hours (table 2, entry D), glycerol was not completely consumed despite higher yields of formic and glycolic acids. Finally, when we added hydrogen peroxide slowly at 0 °C for 3 hours and additional H₂O₂ at 60 °C for 3 hours, we observed exclusively formic acid in 61% yield (table 2, entry E).

Figure 3.

¹H-NMR study for the degradation pathway of glycerol: (A) starting glycerol, (B) 0.4 mL H₂O₂ for 6 hours at 60 °C, (C) 6 hour slow addition of 0.4 mL H₂O₂ at 60 °C, (D) 3 hour slow addition of 0.4 mL H₂O₂ at 0 °C and 8 hour stirring at room temperature, (E) 3 hour slow addition of 0.3 mL H₂O₂ at 0 °C and 3 hour slow addition of 0.2 mL H₂O₂ at 60 °C.

In conclusion, we successfully demonstrated a method to produce formic acid as the sole product via the oxidative NHC-Pd catalyzed carbon-carbon bond cleavage of glycerol with hydrogen peroxide as an oxidizing agent. The direct conversion of glycerol into formic acid was facilitated under mild conditions. As we proposed, dual cleavage pathways were likely to be active. Based on these mechanisms, we sought optimal conditions to avoid the glycolic acid intermediate and over-oxidation of formic acid to carbon dioxide. Our catalytic conditions can be useful for the degradation of carbohydrates and biomass including starch and grass, which will be reported in due course.

Scheme 2.

Glycerol degradation in the presence of Pd(II) catalysts.