(560ab) Large-Scale Exploitation of Glucose Reductive Aminolysis: Comparison of Jet-Loop and Trickle-Bed Reactor Performance

text-align:justify;line-height:150%"> 150%">Introduction

The reductive aminolysis of
glucose with dimethylamine (DMA) has been presented as a renewable route
towards dimethylaminoethanol (DMAE) and tetramethylethylenediamine (TMEDA),
which can serve as precursors for surfactants [1]. In our previous work we have
elaborated on the elucidation of the reaction mechanism and construction of a
corresponding kinetic model for glucose reductive aminolysis [2]. In the
current work we aim at optimizing the product yields in two very different
reactor types, i.e., a jet-loop and a trickle-bed reactor [3, 4]. To achieve
this goal 1D reactor simulations were performed, using the previously
constructed kinetic model, and a comparison was made with pilot tests.

text-align:justify;line-height:150%"> 150%">Results and Discussion

A graphical representation of the
reactors, along with the reactor mass balances, has been shown elsewhere for
the jet-loop reactor with a reactor vessel and loop volume of 4.75 m³
and 0.63 m³ respectively and for the trickle-bed reactor with a
length of 1.17 m and a diameter of 0.015 m [5]. Verification experiments
have been performed for the latter at the pilot scale. Removing the generated
heat in the jet-loop reactor is achieved using a heat exchanger in the
recycling loop while a water cooled jacket serves this purpose in the trickle-bed
reactor.

Figure 1: Molar product yields of the reductive
aminolysis of glucose with dimethylamine as a function of time in the jet loop
reactor. a) (n_glucose⁰ = 2250 mol, n_DMA/n_glucose⁰ = 12 mol
mol^-1, n_H2/n_glucose⁰ = 5 mol
mol^-1, p_tot = 7.5 MPa,
T = 398 K, W_cat = 35 kg). b) (n_glucose⁰ = 850 mol,
n_DMA/n_glucose⁰ = 40 mol mol^-1,
n_H2/n_glucose⁰ = 10 mol mol^-1,
p_tot = 7.5 MPa, T = 390 K, W_cat = 35 kg).

The composition of the mixture in
the jet-loop reactor is presented in Figure 1 (a). The selectivity towards the
TMEDA amounts to 37%, while that towards DMAE is limited to 16%. A major
fraction of the feed is converted into degradation products, such as
dimethylformamide. The temperature exhibits a maximum as a function of the
time, simultaneously with the intermediates concentration and, hence, when
hydrogenation reactions, which generate the most pronounced exothermicity,
reach their highest rates. This maximal temperature increase is limited to 13 K
and the heat is efficiently removed as evidenced from the fast temperature
decrease after the maximum. Yet, given the high activation energy of the
degradation reactions, the latter occur to a significant extent during this
(limited) temperature excursion.

When varying the temperature
while keeping all other operating conditions fixed TMEDA exhibits a maximum
yield of 42% at 386 K (not shown). At higher temperatures degradation
significantly increases and, hence, the formation of both DMAE and TMEDA is
reduced. Considering higher amounts of DMA in the feed, while keeping the other
operating conditions fixed, leads to higher TMEDA yields (not shown). The
latter is a direct result of the lower glucose concentration resulting in less
pronounced degradation.

Figure 1 (b) shows the optimized
product spectrum in the jet-loop reactor. Simulation results critically depend
on the catalyst mass, a total catalyst mass of 35 kg seeming to be the most
interesting one. Lower catalyst masses result in product spectra with more
degradation while higher catalyst masses result in more DMAE formation and,
hence, a lower TMEDA yield. The initial amount of glucose is 850 mol and a
molar ratio of DMA to glucose of 40 mol mol^-1 is selected, while the
temperature is 390 K and the total pressure is 7.5 MPa. Ultimately, the TMEDA
yield amounts to 57%, while the DMAE yield amounts to 12%. Thus, the overall
product spectrum is very similar to the one obtained on the lab scale [2].

While efficient heat removal was
achieved in the jet-loop reactor, this was not the case for the trickle-bed
reactor, see also Figure 3 (a), and, hence, isothermal simulations were
performed first to assess the effect of the individual operating conditions,
see Figure 2.

Figure 2: Molar DMAE and TMEDA yields as a function
of varying operating conditions in the isothermally operated trickle bed
reactor. (a) Variation of n_DMA/n_glucose⁰ while
F_glucose⁰ = 2.57·10^‑6 mol
s^-1, n_H2/n_glucose⁰ = 120 mol
mol^-1, T = 390 K, p_tot = 7.5 MPa,
(b) Variation of n_H2/n_glucose⁰ while F_glucose⁰ = 2.57·10^-6 mol
s^-1, n_DMA/n_glucose⁰ = 40 mol
mol^-1, T = 390 K, p_tot = 7.5 MPa,
(c) Variation of temperature while F_glucose⁰ = 2.57·10^‑6 mol
s^-1, n_H2/n_glucose⁰ = 120 mol mol^-1,
T = 390 K, p_tot = 7.5 MPa, (d)
Variation of F_glucose⁰ while n_H2/n_glucose⁰ = 120 mol
mol^-1, n_DMA/n_glucose⁰ = 40 mol
mol^-1, T = 390 K, p_tot = 7.5 MPa.

A higher n_DMA/n_glucose⁰
ratio results in a significant TMEDA yield enhancement, from 36% at a n_DMA/n_glucose⁰
ratio of 40 mol mol^-1 to 57% at a n_DMA/n_glucose⁰
ratio of 120 mol mol^-1, as can be seen from Figure 2 (a).
Several phenomena contribute to the enhanced TMEDA yield, firstly the higher n_DMA/n_glucose⁰
ratio leads to lower glucose concentrations and, hence, leads to suppressing
the degradation reactions. Secondly, more DMA molecules are required during
TMEDA formation as compared to DMAE formation and, hence, a higher n_DMA/n_glucose⁰
ratio yields more TMEDA. Thirdly, the keto‑ enol tautomerism, which is a
key step in tuning the selectivity between DMAE and TMEDA, is base-catalyzed by
DMA. Finally, higher mass transfer coefficients are achieved with a higher DMA
feed rate.

Figure 2 (b) shows that a higher
n_H2/n_glucose⁰ ratio results in higher desired
product yields, mainly TMEDA, and a reduced amount of degradation.

A higher temperature results in
more pronounced degradation and a corresponding decrease in the TMEDA yield as
can be seen from Figure 2 (c). The DMAE yield exhibits a weak optimum but
overall the DMAE yield remains below 1%.

Higher total feed flow rates
again result in enhanced mass transfer, see Figure 2 (d) for the variation of F_glucose⁰
and, hence, result in enhanced hydrogenation rates. This ultimately leads to a
significant increase of the TMEDA yield from 18% to 59% when F_glucose⁰
is increased from 1.29·10^-6 mol s^-1 to 5.14·10^-6
mol s^-1, nevertheless the DMAE yield remains below 5%.

Figure 3: Molar product yields of the reductive
aminolysis of glucose with dimethylamine in the trickle bed reactor. (F_glucose⁰ = 2.57·10^‑6 mol
s^-1, n_DMA/n_glucose⁰ = 120 mol
mol^-1, n_H2/n_glucose⁰ = 120 mol
mol^-1, p_tot = 7.5 MPa,
T = 390 K, ρ_bed = 1000 kg_cat
m_r^-1) (a) Calculated product yields. (b) Experimentally
measured product yields as a function of time.

Figure 3 (a) shows the simulated
glucose aminolysis product spectrum when the reaction exothermicity and heat
exchange are taken into account. It can be seen from this figure that temperature
control is rather poor leading to an unavoidable temperature increase of 18 K.
Correspondingly, a maximum TMEDA yield amounting to 37% is simulated, which is
about 20% lower than the isothermally simulated TMEDA yield as a consequence of
more pronounced degradation at the higher temperatures (not shown). The
experimental validation in the considered reactor is shown in Figure 3 (b).
The steady state TMEDA yield amounts to 40%, while almost no DMAE is formed.
Thus, the model succeeds in predicting the trickle bed reactor yields
accurately when the most important phenomena are accounted for.

text-align:justify;line-height:150%"> 150%">Conclusions

text-align:justify;line-height:150%"> 150%;font-weight:normal">Industrial-scale glucose aminolysis simulations have
shown that TMEDA is the most abundantly obtained product in a jet-loop reactor
with yields amounting to 57%, while a non-negligible amount of DMAE is formed.
The product spectrum, thus, corresponds closely to the one obtained in
(fed-)batch mode on the lab scale, thanks to an acceptable temperature control.
This temperature control is, on the contrary, the main issue in the trickle-bed
reactor where TMEDA yield losses up to 20% are observed as a consequence of
poor temperature control and, hence, more pronounced degradation at higher
temperatures. The model predicted and experimentally observed TMEDA yield of
40%, hence, indicate that mass transfer and temperature control should be
enhanced to exploit this complex chemistry to its full potential.

text-align:justify;line-height:150%"> 150%">Acknowledgement

This work was supported by
Flanders Innovation & Entrepreneurship VLAIO (IWT) via the intermediary of
FISCH/CATALISTI, contract 145020–Carboleum.

text-align:justify;line-height:150%"> 150%">References

-35.45pt;line-height:normal">1.         Pelckmans,
M., et al., ACS Catalysis, 8 (2018): p. 4201-4212.

-35.45pt;line-height:normal">2.         Poissonnier,
J., et al., Applied Catalysis B: Environmental, 227 (2018): p. 161-169.

-35.45pt;line-height:normal">3.         Stitt,
E., Chemical Engineering Journal, 90 (2002): p. 47-60.

-35.45pt;line-height:normal">4.         Ranade,
V.V., R. Chaudhari, and P.R. Gunjal, Trickle bed reactors: Reactor
engineering and applications. 2011: Elsevier.

-35.45pt;line-height:normal">5.         Poissonnier,
J., J.W. Thybaut, and G.B. Marin, Industrial & Engineering Chemistry Research, 
56 (2017): p. 14192-14199.