Molecular size reforming of undersized and oversized polyoxymethylene dimethyl ethers

Abstract

Polyoxymethylene dimethyl ethers (CH₃–O–(CH₂O)_n–CH₃, PODE_n) are potential environmentally benign coal-based diesel fuel blending compounds. They are synthesized from dimethoxymethane (DMM) and paraformaldehyde (PF). Among the PODE_n homologues, PODE_3–4 have a good match with diesel as a fuel, while the undersized PODE_1–2 and oversized PODE_n>4 have unsuitable properties. This work studied the molecular size reforming of undersized PODE_1–2 and oversized PODE_n>4 by two different methods, namely, self-reforming and reacting with DMM over an acidic ion exchange resin. The molecular size reforming of PODE_1–2 and PODE_n>4 by self-reforming gave a high concentration of formaldehyde, which shifted the distribution to longer chain PODE_n and formed solid PF. In contrast, PODE_1–2 and PODE_n>4 were mainly converted to PODE_3–4 by the reaction with DMM and the high concentration of formaldehyde was also diminished. The equilibrium reorganized molecular size distribution of PODE_1–2 and PODE_n>4 followed the Schulz–Flory distribution. A proposed kinetic model described well the molecular size reforming pathways of PODE_1–2 and PODE_n>4. A methanol-to-PODE_n close-loop process was proposed to enhance atom-economy by recycling the PODE_1–2 and PODE_n>4 streams.

---

1. Introduction

The coal chemical industry is oriented to the development of high efficiency, safety, cleanliness, and optimum utilization. Coal-based compounds as alternative fuels or chemicals are receiving increasing attention for alleviating environmental pollution and reducing dependence on petroleum.^1,2 Methanol is a key intermediate chemical in several important coal-to-chemical processes, such as methanol-to-olefins (MTO),^3,4 methanol-to-propylene (MTP)^5,6 and methanol-to-aromatics (MTA).^7,8 Recently, the methanol-to-oxygenated fuel^9–12 process has become an important route to covert coal to fuel blending compounds. Among these, polyoxymethylene dimethyl ethers (PODE_n) are the most promising green diesel fuel blending compounds because diesel engines can use these without needing to be modified.^11–14 Particulate pollutants in the combustion of a mixture of PODE_n and diesel are significantly reduced due to the oxygen content in PODE_n.^15,16 Considering the properties of PODE_n-diesel blending oil, especially the density, heating value and combustion emissions, the proposed blending ratio of PODE_n in diesel is 10–20%.

PODE_n refers to the compounds with the formula CH₃–O–(CH₂O)_n–CH₃, and PODE₁ is dimethoxymethane (DMM, CH₃OCH₂OCH₃). PODE_n can be synthesized from an end-group provider (DMM or methanol) and a chain-group provider (trioxane, paraformaldehyde or formaldehyde solution). The potential feedstocks are all important derivatives of methanol. Table 1 gives a comparison of the MTO, MTP and MTA processes and the methanol-to-PODE (MTPODE) process in terms of the reaction temperature, pressure, catalyst, reactor, and feedstock consumption. MTO, MTP and MTA processes are typically operated at a higher temperature than the MTPODE process, thus they have higher energy consumption. The methanol consumption (t methanol/t product) is a key parameter used to measure the economics of a process. The methanol consumption of the MTPODE process is 1.3–1.4 t methanol/t product, which is much lower than that in the MTO, MTP and MTA processes. The methanol consumption was calculated based on a MTPODE process developed by our group^11,17–19 and shown in Fig. 1. In the MTPODE process, methanol is oxidized to formaldehyde, which is condensed to paraformaldehyde (PF) or trioxane. Then methanol reacts with formaldehyde to form dimethoxymethane (DMM, CH₃OCH₂OCH₃). Finally, DMM reacts with PF or trioxane to produce the PODE_n compounds.

Table 1 Comparison of methanol-to-olefin (MTO), methanol-to-propylene (MTP), methanol-to-aromatic (MTA) and methanol-to-PODE (MTPODE) processes

Technology	Temperature/°C	Pressure/MPa	Catalyst	Reactor	Methanol consumption/t methanol per (t product)
a Methanol-to-formaldehyde section.b Methanol reacts with formaldehyde to DMM section.c DMM reacts with formaldehyde to PODE section.
MTO (UOP)	350–550	0.1–0.5	SAPO-34	Fluidized-bed	2.85
MTO (OCP)	420–480	0.276	SAPO-34	Fast-fluidized/fixed-bed	2.54
MTP (Lurgi)	460–480	0.16	ZSM-5	Fixed-bed	3.22–3.52
FMTP	380–450	0.2–0.3	SAPO-34	Fluidized-bed	3.36
MTA (Tsinghua)	400–500	0.1	ZSM-5	Fluidized-bed	2.35
MTPODE (Tsinghua)	300–350^a	0.05–0.10^a	Fe–Mo^a	Fixed bed^a^,^b	1.30–1.40
	70–90^b^,^c	0.2–0.3^b^,^c	Solid acid^b^,^c	Fluidized-bed^c

---

Fig. 1 Methanol consumption in the methanol-to-PODE (MTPODE) process (ME: methanol; FA: formaldehyde; DMM: dimethoxymethane; PF: paraformaldehyde).

The fuel properties of the PODE_n homologues (n from 1 to 8) are compared with those of commercial diesel in Table 2.^20–22 Among the PODE_n homologues, PODE_3–4 match the diesel fuel properties well except for a higher density. However, PODE_1–2 do not meet a security criterion and would cause vapour lock during combustion due to their low flash point. Furthermore, their low viscosity will cause wear of the fuel injection pump. PODE_n>4 have poor low temperature fluidity, high melting points and high viscosity, which significantly affect their use in an ambient environment. Our previous work²³ showed that the formation of PODE_n from DMM and paraformaldehyde followed a sequential propagation mechanism. The molecular size distribution of the PODE_n compounds obeyed the Schulz–Flory distribution.²³ Zhao et al.²⁴ also verified the application of Schulz–Flory distribution model for molecular size distribution during the synthesis of PODE_n. Besides the targeted PODE_3–4 compounds, a significant amount of undersized PODE_1–2 and oversized PODE_n>4 compounds were also produced in the reported processes,^22,24–28 as listed in Table 3. Therefore, the molecular size reforming of PODE_1–2 and PODE_n>4 to give the targeted PODE_3–4 compounds by recycling the byproducts is needed in the industrial process, which was also the basis and precondition for the methanol consumption calculation in Fig. 1.

Table 2 Typical properties of PODE_1–7 compared with diesel^20–22

	Diesel (ASTM D 975)	DMM	PODE2	PODE3	PODE4	PODE5	PODE6	PODE7	PODE8
a Temperature equivalent to 90% of distilled volume.
Density at 25 °C/g cm^−3	0.80–0.86	0.86	0.96	1.02	1.07	1.10	1.13	1.16	1.20
Melting point/°C	—	−105	−65	−41	−7	18.5	58	—	—
Boiling point/°C	288^a	42	105	156	201	242	280	—	—
Viscosity at 25 °C/mPa s	1.04–1.92	0.58	0.64	1.05	1.75	2.79	4.14	5.81	7.80
Cetane number	40–65	29	63	78	90	100	104	—	—
Oxygen content/wt%	—	42.1	45.3	47.1	48.2	49.0	49.6	50.0	50.3

---

Table 3 Molecular size distribution of PODE_n compounds in different production processes

Ref.	Temperature/K	Feedstocks	Catalyst	Reactor	Molecular size distribution/wt%
					PODE1–2	PODE3–5	PODEn>5
22	353	Tri + DMM(nDMM : nCH2O = 1.38 : 1)	Amberlyst 36	Batch autoclave	75.75	22.95	1.29
23	373	PF + DMM(nDMM : nCH2O = 2 : 1)	H2SO4	Batch autoclave	67.00	29.39	3.61
24	353	MeOH + CH2O(nMeOH : nCH2O = 1 : 3)	Acid cation exchange resin	Fixed-bed	40.84	50.22	8.94
25	353	Tri + DMM(nDMM : nCH2O = 1.19 : 1)	Amberlyst 46	Batch autoclave	65.05	30.15	4.80
26	403	MeOH + tri(nMeOH : nCH2O = 1 : 2)	S/Fe	Batch autoclave	49.44	41.70	8.87
27	443	Tri + DMM(nDMM : nCH2O = 1 : 1)	Alkanesulfonic acid-functionalized ionic liquids	Batch autoclave	57.49	34.81	7.70

---

To recycle undesired PODE_n compounds, Burger et al.^22,29 proposed to feed the unconverted reactants and the PODE_n of undesired chain lengths (n = 1, 2 or >4) back into the reactor in a PODE_n production process using DMM and trioxane as feedstock. They supposed that PODE_1–2 underwent chain propagation reactions while PODE_n>4 were cleaved when they were fed back into the reactor. However, a significant amount of long chain PODE_n compounds, even PODE_n>10, were formed in the industrial process,²⁹ which indicate that there were different reforming pathways for the PODE_1–2 and PODE_n>4 compounds.

The present work is to develop an effective method for recycling the PODE_1–2 and PODE_n>4 compounds produced in the MTPODE process. The molecular size reforming of the undersized PODE_1–2 and oversized PODE_n>4 compounds by the two different reactions of self-reforming and reaction with DMM over an acidic ion exchange resin was investigated. A previously developed kinetic model was used for the description of the molecular size reforming pathway. The equilibrium reorganized molecular size distribution of PODE_1–2 and PODE_n>4 was found to follow the Schulz–Flory distribution. Based on the experimental results, a MTPODE close-loop process was proposed for recycling the undersized PODE_1–2 and oversized PODE_n>4 compounds.

2. Experimental

2.1 Materials

The PODE_1–2 and PODE_n>4 samples were obtained from a 10kt/a PODE_n demonstration plant using the technology developed by our group in Shandong Yuhuang Chemical (Group) Co., Ltd. in China. Previous works^19,27 showed that water is a compound that causes side reactions when producing PODE_n. Therefore, anhydrous feedstocks of DMM and PF were used in the 10kt/a PODE_n demonstration plant.

DMM (analytic reagent grade, >99 wt%) was purchased from Alfa Aesar-Johnson Matthey. PF (polymer grade, >96 wt%) was purchased from Sinopharm Chemical Reagent Co., Ltd. The NKC-9 ion exchange resin of H⁺ type was provided by Tianjin Bohong Resin Technology Co., Ltd.

2.2 Experimental

The experiments were conducted in a batch stirred autoclave to study the molecular size reforming of PODE_1–2 and PODE_n>4 by self-reforming or reacting with DMM. A detailed description of the experimental setup was reported elsewhere.^11,23 The NKC-9 ion exchange resin was used as the catalyst because it had high activity and good stability for PODE_n production from PF and DMM.^11,23

For each experiment, the feedstocks (PODE_1–2, PODE_n>4 and DMM) were first loaded into the reactor. Then, the reactor was sealed and heated. When the reactor temperature reached the desired value, the catalyst was released from a glass bottle and uniformly dispersed by stirring at 500 rpm. This time was used as the initial reaction time. The products were sampled at different reaction times and quantitatively analysed by gas chromatography-mass spectrometry (GC-MS). A detailed description of the GC-MS analysis method was reported elsewhere.^11,19,23,30 In brief, the method was established based on mass conservation without standard PODE_n samples.³⁰ The relative mass response factor of PODE_n compounds could be described as f(_PODEn) = 1.140ⁿ⁻¹ with f_PODE1 = 1.³⁰ The carbon balance was within ±3%. Detailed carbon balance data were provided in the ESI.† The concentration of formaldehyde was determined by the sodium sulfite titration method described in ASTM D2194-02 (2012). More detailed description of calibration and good reproducibility of the sodium sulphite titration method in this system was reported elsewhere.^19,31,32

3. Results and discussion

3.1 Reactions and molecular size distribution

In the reaction scheme for the formation of PODE_n from DMM and PF catalysed by NKC-9 ion exchange resin, PF depolymerizes to monomeric formaldehyde and dissolves in the liquid phase due to the acidic catalyst (reaction (1)). PODE_n reacts with monomeric formaldehyde to form longer chain molecules PODE_n+1 (reaction (2)). This scheme was supported by the experimental results that PODE₂, PODE₃, PODE₄, etc. appeared one by one in the system.^11,15 Burger^22,27 proposed a similar scheme for the formation of PODE_n from DMM and trioxane.where k_p and k_d are the rate constants of the forward (propagation) and reverse (depolymerization) reactions, respectively.

The sequential propagation mechanism leads to a Schulz–Flory equilibrium molecular size distribution described by eqn (3).²³

where a is

The molecular size distribution of PODE_n is determined by the reaction temperature and effective molar ratio of DMM/CH₂O. An equivalent DMM/CH₂O molar ratio M_eq. is defined as:

where C_PODEn is the mole concentration of PODE_n, and C_FA is the mole concentration of formaldehyde.

In principle, the propagation of PODE_n by insertion of formaldehyde is similar after formaldehyde has been produced in the same concentration from decomposition of trioxane or paraformaldehyde. However, the decomposition rate and equilibrium of trioxane and paraformaldehyde are different, which result in different formaldehyde concentration in solution. The decomposition of formaldehyde may depend on its polymerization degree. Xia et al.³³ set the degree of paraformaldehyde polymerization as 3 to reduce the computational cost. However, the paraformaldehyde used in this work had a polymerization degree of 8–100. In addition, Xia et al.³³ used sulfonic acid-functionalized ionic liquids as catalyst while this work used cation exchange resin. Due to these factors, the sequential formation of PODE_n was observed in this work, different from the mechanism suggested by Xia et al.³³

According to the reaction mechanism, the production of PODE_1–2 and PODE_n>4 is unavoidable in a single pass PODE_n synthesis system. Fig. 2 shows a block flow diagram of the first stage PODE_n demonstration process using DMM and PF as feedstock. In this process, DMM and PF were fed into a fluidized bed reactor to produce PODE_n using the strong acid ion exchange resin as catalyst. After reaction, the reaction mixture was fed into a rectifying sequence and separated into different fractions. The DMM stream was separated as the overhead fraction in the pre-distillation tower and was directly fed back into the reactor. The side withdrawal mainly contained PODE₂ and a portion of the DMM (PODE₁), thus it was named as the undersized PODE_1–2 stream. A portion of unreacted formaldehyde was dissolved in the DMM stream and PODE_1–2 stream. The remaining formaldehyde from the reaction mixture was separated by extractive distillation using water as the extractant. Formaldehyde and water formed azeotrope at top of the extractive distillation tower. The extracted formaldehyde solution was used as feedstock to produce DMM. The desired PODE_3–4 stream and oversized PODE_n>4 stream were separated in a vacuum distillation tower.

Fig. 2 Block flow diagram of the first stage PODE process.

In our first stage PODE_n demonstration process, the yield of PODE_1–2 and PODE_n>4 stream were 33.15 wt% and 1.68 wt% under the typical conditions of feeding ratio n_DMM : n_CH2O = 2 : 1 and reaction temperature T = 80 °C. As listed in Table 4, the PODE_1–2 stream contained 13.68 wt% PODE₁, 73.14 wt% PODE₂, 2.30 wt% PODE₃ and 10.88 wt% formaldehyde, and the PODE_n>4 stream contained 73.48 wt% PODE₅ and 26.52 wt% PODE₆.

Table 4 Compositions of undersized PODE_1–2 stream and oversized PODE_n>4 stream in the first stage PODE process

Stream	Formaldehyde/%	PODE1/%	PODE2/%	PODE3/%	PODE4/%	PODE5/%	PODE6/%
PODE1–2	10.88	13.68	73.14	2.30	0.00	0.00	0.00
PODEn>4	0.00	0.00	0.00	0.00	0.00	73.48	26.52

---

3.2 Molecular size self-reforming of PODE_1–2 and PODE_n>4

To recycle the PODE_1–2 and PODE_n>4 streams, the chain propagation of PODE_1–2 and depolymerization of PODE_n>4 towards PODE_3–4 are the desired reaction pathways. The self-reforming experiments were conducted in an autoclave reactor at 80 °C with 5 wt% NKC-9 catalyst for both the PODE_1–2 and PODE_n>4 streams. The equivalent DMM/CH₂O molar ratio M_eq. was 0.82 and 0.24 for the PODE_1–2 and PODE_n>4 streams, respectively.

The molecular size distribution of the products from PODE_1–2 self-reforming is shown in Fig. 3. Fig. 3(A) shows the mass fractions of PODE_1–6 and formaldehyde as a function of reaction time. The main species, PODE₂, was quickly consumed and its fraction decreased from 72.85 wt% to 24.39 wt% in the first 10 min. On the one hand, PODE₂ decomposed into DMM and formaldehyde; on the other hand, PODE₂ reacted with formaldehyde forming PODE₃. The concentration of formaldehyde increased from 10.88 wt% to 18.26 wt% in 10 min, indicating that the depolymerization reaction was faster than the chain propagation reaction during this period. The concentration of PODE₃ showed a peak value of 9.84 wt% at 5 min. After 5 min, the amount of PODE₃ decreased and correspondingly, there was a sequential increase of PODE_4–6. Note that the amount of PODE_n>3 was 2.51 wt% at 10 min due to a low initial amount. Overall, the self-reforming of the PODE_1–2 stream mainly proceeded in the depolymerization direction, which formed large amounts of DMM and formaldehyde. Due to the formation of formaldehyde, the molecular size distribution shifted to longer chain PODE_n compounds.²⁷

Fig. 3 Molecular size reforming of the PODE_1–2 stream by self-reforming (reaction temperature: 80 °C, NKC-9 catalyst loading: 5 wt%). (A) Mass fractions of PODE_1–6 and formaldehyde (FA) at different reaction times; (B) plots of ln(w_n/(n + 1.533)) as a function of n at different reaction times, where w_n refers to mass fractions of PODE_n; (C) comparison between experimental (symbols) and predicted (lines) mass fractions of PODE_n compounds, C_F = 1.40 mol L⁻¹ in the kinetic calculation; (D) parity plots of the experimental and predicted values.

Fig. 3(B) shows the relationship between ln(w_n/(n + 1.533)) and n at different reaction times, where w_n is the mass fraction of PODE_n. The measured equilibrium molecular size distribution followed the Schulz–Flory distribution with a = 0.537 in eqn (3). In this system, the factor a was determined by the reaction temperature and equivalent DMM/CH₂O molar ratio. During the molecular size self-reforming process, the data points of ln(w_n/(n + 1.533)) gradually shifted to the equilibrium line with a = 0.537.

A kinetic model was proposed in our previous work¹⁹ based on the sequential propagation mechanism of PODE_n, and assuming second order kinetics for propagation and first order kinetics for depolymerization. The rate constants for propagation (k_p) and depolymerization (k_d) and the reaction equilibrium constants (K_n) were found to be independent of molecular size. The pre-exponential factors A_p for propagation and A_d for depolymerization were 1.84 × 10⁴ L mol⁻¹ min⁻¹ and 5.36 × 10⁶ min⁻¹, respectively. The activation energy E_p for propagation and E_d for depolymerization were 39.52 kJ mol⁻¹ and 52.01 kJ mol⁻¹, respectively. The concentration of PODE_1–6 at different reaction times were calculated from the initial reactant concentrations using the kinetic model. During model calculation, the concentration of formaldehyde in homogenous solution (C_F) was used as an input parameter, as illustrated in our previous work.¹⁹ In most cases the C_F become constant within 10 min in this work, and the calculated molecular size distribution was found insensitive to the C_F values, therefore the C_F was regarded as constant during the kinetic calculations. The C_F data used in the calculations were annotated below the corresponding figures. Good agreement was obtained between the experimental and calculated weight fractions of PODE_1–6, as shown in Fig. 3(C) and (D). While a similar pseudo-homogeneous model was not capable to describe the formation of PODE_n from DMM and trioxane, indicating the differences between trioxane and paraformaldehyde when reacting with DMM.¹⁹

It is difficult to analyse the molecular size distribution during the PODE_n>4 self-reforming process, because the reaction mixture became a slurry in several minutes due to the formation of a large amount of CH₂O. In this case, the kinetic model was used to predict the molecular size distribution, as shown in Fig. 4. Fig. 4(A) shows the mass fractions of PODE_1–8 and formaldehyde as a function of reaction time. PODE₅ and PODE₆ were quickly consumed and their concentrations decreased from 74.48 wt% and 25.52 wt% to 31.39 wt% and 19.41 wt% in 10 min, respectively. Meanwhile the mass fraction of formaldehyde increased from zero to 9.10 wt%. The maximum concentrations of PODE₄ (21.23 wt%) and PODE₇ (6.61 wt%) appeared at 10 min. The polymerization of formaldehyde produced some solid PF powder, which were suspended in the reaction solution and increased the viscosity of the suspension. PODE₃ and PODE₈ showed their maximum concentrations at 30 min. At equilibrium, the mass fractions of DMM and PODE₂ increased to 21.32 wt% and 12.73 wt%, respectively, and the total mass fraction of formaldehyde, including both the dissolved formaldehyde and solid PF, increased to 51.33 wt%. As a result, the molecular size self-reforming of the PODE_n>4 stream formed a large amount of PODE_1–2 and formaldehyde and a small amount of PODE_7–8 (0.76 wt%). Fig. 4(B) shows the variation of ln(w_n/(n + 1.533)) as a function of n at different reaction times. The equilibrium molecular size distribution agreed well with the Schulz–Flory model with a = 0.801. The calculated mass fractions of PODE_1–8 during the self-reforming of PODE_n>4 is shown in Fig. 4(C).

Fig. 4 Molecular size reforming of the PODE_n>4 stream by self-reforming (reaction temperature: 80 °C, NKC-9 catalyst loading: 5 wt%). (A) Predicted mass fractions of PODE_1–8 and formaldehyde (FA) at different reaction times; (B) plots of predicted ln(w_n/(n + 1.533)) as a function of n at different reaction times, where w_n refers to mass fractions of PODE_n; (C) predicted mass fractions of PODE_n. C_F = 1.75 mol L⁻¹ in the kinetic calculation.

The direct recycling of PODE_1–2 and PODE_n>4 stream to the reactor decreased the value of M_eq., thus effecting a change from the equilibrium molecular size distribution. A high concentration of formaldehyde shifted the distribution to longer chain PODE_n with n even larger than 10,^27,29 and also prevented the depolymerization of fresh PF feed. This accounts for the results reported by Burger et al.²⁹ that a significant amount of long chain PODE_n compounds, even PODE_n>10, were formed in a process that recycled the PODE_1–2 and PODE_n>4 streams to the reactor. The resulting high concentration of solid PF would deteriorate the fluidization quality when a fluidized bed reactor is used.

3.3 Molecular size reforming by reacting with DMM

Based on the self-reforming results discussed in Section 3.2, we proposed a new approach to change the molecular size distribution by reacting PODE_1–2 or PODE_n>4 with DMM. In this approach, the amount of added fresh DMM was the most important parameter. In this work, the PODE_1–2 and PODE_n>4 streams were respectively mixed with DMM to form new streams with M_eq. = 2 : 1. The reaction temperature was 80 °C and the NKC-9 catalyst loading was 5 wt%.

The molecular size distributions of the products of PODE_1–2 reforming with DMM are shown in Fig. 5. Fig. 5(A) shows the mass fractions of PODE_1–6 and formaldehyde as a function of reaction time. The concentration of PODE₂ quickly decreased from 36.80 wt% to 26.11 wt% in 10 min, and the desired conversion, namely, to PODE₃, was dominant. The concentration of DMM slightly increased first and showed its peak value of 60.19 wt% at 10 min. Meanwhile the concentration of formaldehyde slightly increased from 5.56 wt% to 6.06 wt%, indicating that the chain propagation reactions were slower than the reversed depolymerization reactions during this period. After 10 min, the concentration of DMM gradually decreased to its initial level. At equilibrium, almost all converted PODE₂ compound underwent chain propagation to PODE_3–6, among which PODE_3–4 had a selectivity of 87.05 wt%. Finally, the concentration of formaldehyde decreased to 1.91 wt%. Fig. 5(B) shows the plots of ln(w_n/(n + 1.533)) as a function of n at different reaction times. The equilibrium molecular size distribution data at M_eq. = 2 : 1 agreed well with the Schulz–Flory model with a_e = 0.438. This was very close to the experimental value a_e = 0.444 with DMM and PF at the DMM/CH₂O molar ratio 2 : 1 reported in our previous work.¹⁹ This indicated that the factor a_e is determined by M_eq. of a stream and is independent of the stream composition at a specific temperature. The comparison of the experimental and calculated molecular size distribution of PODE_1–6 and the corresponding parity plot were shown in Fig. 5(C) and (D), respectively. Good agreement was obtained. Compared to the molecular size distribution of the products of PODE_1–2 self-reforming (Fig. 3), the reaction of the PODE_1–2 stream with DMM at M_eq. = 2 : 1 mainly converted PODE₂ to the desired products PODE_3–4, and the high concentration of formaldehyde was also diminished.

Fig. 5 Molecular size reforming of the PODE_1–2 stream by reaction with DMM (reaction temperature: 80 °C, NKC-9 catalyst loading: 5 wt%). (A) Mass fractions of PODE_1–6 at different reaction times; (B) plots of ln(w_n/(n + 1.533)) as a function of n at different reaction times; (C) molecular size reforming pathways of PODE_1–2 stream by reacting with DMM, C_F = 1.22 mol L⁻¹ in the kinetic calculation.; (D) parity plot of the experimental and calculated values.

Fig. 6 illustrates the experimental results of the reforming of the PODE_n>4 stream when reacted with DMM. The mass fractions of PODE_1–8 and formaldehyde at different reaction times are shown in Fig. 6(A). PODE₅ and PODE₆ were quickly consumed and were mainly converted to the desired PODE_2–4. In the first 10 min, the mass fraction of DMM decreased from 73.72 wt% to 58.44 wt%, and the concentration of formaldehyde slightly increased from zero to 1.90 wt%. The concentration of PODE_7–8 showed a small peak value at 20–30 min and decreased to the equilibrium value of 0.37 wt%, indicating an effective control of chain propagation to long chain PODE_n. The final concentration of formaldehyde was 2.48 wt%. For a further analysis of the molecular size distribution, ln(w_n/(n + 1.533)) was plotted as a function of n at different reaction times in Fig. 6(B). The equilibrium molecular size distribution data at M_eq. = 2 : 1 were well described by eqn (3) with a_e = 0.441, which was very close to the experimental value a_e = 0.444 at the same conditions.¹⁹ Good agreement between the experimental data and the proposed kinetic model¹⁹ are shown in Fig. 6(C) and (D). Different from the self-reforming of PODE_n>4 stream (Fig. 4), the reforming of the PODE_n>4 stream by reaction with DMM at M_eq. 2 : 1 mainly converted PODE_5–6 to the desired product PODE_2–4.

Fig. 6 Molecular size reforming of the PODE_n>4 stream by reaction with DMM (reaction temperature: 80 °C, NKC-9 catalyst loading: 5 wt%). (A) Mass fractions of PODE_1–8 and formaldehyde (FA) at different reaction times; (B) plots of ln(w_n/(n + 1.533)) as a function of n at different reaction time; (C) molecular size reforming pathways of PODE_n>4 stream by reacting with DMM, C_F = 1.22 mol L⁻¹ in the kinetic calculation.; (D) parity plot of the experimental and calculated values.

3.4 Conceptual close-loop MTPODE process

Based on the experimental results discussed in Sections 3.2 and 3.3, different PODE_n streams with the same M_eq. gave the same equilibrium molecular size distribution after reaction. Therefore, after the reforming of PODE_1–2 and PODE_n>4 by reaction with DMM at M_eq. = 2, these byproducts can be recycled by feeding the equilibrium products directly into the rectifying sequence without disturbing the operation stability of the system. Compared with the process in Fig. 2, an improved process of close-loop MTPODE is proposed, as shown in Fig. 7. In this process, methanol is the sole raw material. First, methanol is oxidized to formaldehyde (in the solution state) which is then condensed to paraformaldehyde powder (PF). Then methanol reacts with formaldehyde in an acetalation reactor (for example a catalytic distillation tower) to produce DMM. The DMM and PF are fed into the PODE_n synthesis reactor I. The PODE_n products and unreacted feedstocks are fed into a rectifying sequence to separate the DMM, PODE_1–2, FA, PODE_3–4 and PODE_n>4 streams, similar to that described in Fig. 2. The FA solution from the FA separation tower is fed back to the acetalation reactor and the PODE_1–2 and PODE_n>4 streams are fed into the PODE_n synthesis reactors II and III, respectively. The products from the PODE_n synthesis reactors II and III are fed into the rectifying sequence for separation. This process is a closed cycle because the desired PODE_3–4 is the only product. Compared to the first stage PODE_n process in Fig. 2, the close-loop MTPODE process in Fig. 7 would achieve the recycling of PODE_1–2 and PODE_n>4 and give PODE_3–4 as the only product. This would give an advantage to the MTPODE process in terms of methanol consumption as shown in Fig. 1. The close-loop MTPODE process in Fig. 7 had been designed for a 300kt/a PODE_n plant in China.

Fig. 7 Schematic of the proposed methanol-to-PODE recycling process. (ME: methanol, FA: formaldehyde, PF: paraformaldehyde, DMM: dimethoxymethane, PODE: polyoxymethylene dimethyl ethers).

4. Conclusions

The molecular size reforming of undersized PODE_1–2 and oversized PODE_n>4 compounds by the two reactions of self-reforming and reaction with DMM over an acidic ion exchange resin was studied. Molecular size self-reforming of the PODE_1–2 and PODE_n>4 streams resulted in a high concentration of formaldehyde, which shifted the distribution to longer chain PODE_n and formed solid PF. Molecular size reforming of the PODE_1–2 and PODE_n>4 streams by reaction with DMM with M_eq. at 2 : 1 gave the desired reactions, i.e., chain propagation of PODE_1–2 and depolymerization of PODE_n>4 to give PODE_3–4. The use of the byproducts of a MTPODE process by recycling the PODE_1–2 and PODE_n>4 streams to react with DMM in independent reactors and then feeding the equilibrium products directly into the rectifying sequence was proposed. The proposed kinetic model gave good descriptions of the molecular size distribution of the production of PODE_1–2 and PODE_n>4 reforming.

Conflict of interest

The authors declare no competing financial interest.

Acknowledgements

We acknowledge the help of professor Dezheng Wang and the financial support by the China Postdoctoral Science Foundation funded project (2015M570113) and the National Natural Science Foundation of China (No. 21476122).

References

1. G. Liu and E. D. Larson, Energy Procedia, 2014, 63, 7315–7329.
2. L. M. Balster, E. Corporan, M. J. DeWitt, J. T. Edwards, J. S. Ervin, J. L. Graham, S.-Y. Lee, S. Pal, D. K. Phelps and L. R. Rudnick, Fuel Process. Technol., 2008, 89, 364–378.
3. X. Aihua, W. Li and S. Yulin, Energy Fuels, 2014, 28, 3339–3344.
4. J. Lefevere, S. Mullens, V. Meynen and J. Van Noyen, Chem. Pap., 2014, 68, 1143–1153.
5. B. Jiang, X. Feng, L. Yan, Y. Jiang, Z. Liao, J. Wang and Y. Yang, Ind. Eng. Chem. Res., 2014, 53, 4623–4632.
6. A. K. Jamil, O. Muraza, M. Yoshioka, A. M. Al-Amer, Z. Yamani and T. Yokoi, Ind. Eng. Chem. Res., 2014, 53(50), 19498–19505.
7. H. Ji, Q. Zhang, B. Wang, C. Li and H. Shan, Catal. Lett., 2014, 144, 1860–1867.
8. G. Q. Zhang, T. Bai, T. F. Chen, W. T. Fan and X. Zhang, Ind. Eng. Chem. Res., 2014, 53, 14932–14940.
9. Q. Tang, H. Xu, Y. Zheng, J. Wang, H. Li and J. Zhang, Appl. Catal., A, 2012, 413, 36–42.
10. K. D. Vertin, J. M. Ohi, D. W. Naegeli, K. H. Childress, G. P. Hagen, C. I. McCarthy, A. S. Cheng and R. W. Dibble, Methylal and methylal-diesel blended fuels for use in compression-ignition engines, SAE Technical Paper, 1999.
11. Y. Zheng, Q. Tang, T. Wang, Y. Liao and J. Wang, Chem. Eng. Technol., 2013, 36, 1951–1956.
12. M. Marchionna, R. Patrini, D. Sanfilippo, A. Paggini, F. Giavazzi and L. Pellegrini, Stud. Surf. Sci. Catal., 2001, 136, 489–494.
13. L. Pellegrini, M. Marchionna, R. Patrini, C. Beatrice, N. Del Giacomo and C. Guido, Combustion Behaviour and Emission Performance of Neat and Blended Polyoxymethylene Dimethyl Ethers in a Light-Duty Diesel Engine, SAE Technical Paper, 2012.
14. G. J. An, X. D. Wang, C. B. Lu, C. H. Xiong, Y. J. Zhou, Y. W. Liu and H. Y. Shang, Appl. Mech. Mater., 2014, 672–674, 676–679.
15. Y. Ren, Z. Huang, H. Miao, Y. Di, D. Jiang, K. Zeng, B. Liu and X. Wang, Fuel, 2008, 87, 2691–2697.
16. M. Natarajan, E. A. Frame, D. W. Naegeli, T. Asmus, W. Clark, J. Garbak, A. Manuel, D. Gonzalez, E. Liney and W. Piel, Oxygenates for advanced petroleum-based diesel fuels: Part 1. Screening and selection methodology for the oxygenates, SAE Technical Paper, 2001.
17. J. Wang, Q. Tang, S. Wang, T. Wang, S. Chen and Y. Wang, Fluidized Bed Reactor and Method for Preparing Polyoxymethylene Dimethyl Ethers from Dimethoxymethane and Paraformaldehyde, US Pat., No. 20,150,273,426, 1, October 2015.
18. J. Wang, Y. Zheng, S. Wang, T. Wang, S. Chen and C. Zhu, Method for Producing Polyoxymethylene Dimethyl Ethers, US Pat., 20,150,291,722[P], October 15 2015.
19. Y. Zheng, Q. Tang, T. Wang and J. Wang, Chem. Eng. Sci., 2015, 134, 758–766.
20. R. H. Boyd, J. Polym. Sci., 1961, 50, 133–141.
21. J. Wu, Z. Xu, Z. Liu and B. Wang, J. Chem. Eng. Data, 2005, 50, 966–968.
22. J. Burger, M. Siegert, E. Ströfer and H. Hasse, Fuel, 2010, 89, 3315–3319.
23. Y. Zheng, Q. Tang, T. Wang and J. Wang, Chem. Eng. J., 2015, 278, 183–189.
24. Y. Zhao, Z. Xu, H. Chen, Y. Fu and J. Shen, J. Energy Chem., 2013, 22, 833–836.
25. J. Zhang, M. Shi, D. Fang and D. Liu, React. Kinet., Mech. Catal., 2014, 113, 459–470.
26. Q. Wu, M. Wang, Y. Hao, H. Li, Y. Zhao and Q. Jiao, Ind. Eng. Chem. Res., 2014, 53, 16254–16260.
27. J. Burger, E. Ströfer and H. Hasse, Ind. Eng. Chem. Res., 2012, 51, 12751–12761.
28. H. Li, H. Song, L. Chen and C. Xia, Appl. Catal., B, 2015, 165, 466–476.
29. J. Burger, E. Ströfer and H. Hasse, Chem. Eng. Res. Des., 2013, 91, 2648–2662.
30. Y. Zheng, Q. Tang, T. Wang and J. Wang, J. Chem. Eng. Chin. Univ., 2015, 29, 505–509.
31. G. Maurer, AIChE J., 1986, 32, 932–948.
32. J. Kiernan, Microsc. Today, 2000, 1, 8–12.
33. F. Wang, G. Zhu, Z. Li, F. Zhao, C. Xia and J. Chen, J. Mol. Catal. A: Chem., 2015, 408, 228–236.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra08255f

---