MASTER'S THESIS. Synthesis of NaX coated honeycombs in a continuous flow reactor

Transcription

1 MASTER'S THESIS 2010:076 Synthesis of NaX coated honeycombs in a continuous flow reactor Antoine Kinder Luleå University of Technology Master Thesis, Continuation Courses Chemical and Biochemical Engineering Department of Chemical Engineering and Geosciences Division of Biochemical and Chemical Engineering 2010:076 - ISSN: ISRN: LTU-PB-EX--10/076--SE

2 Abstract The aim of this work was to grow zeolite NaX on ceramic cordierite monolith wall in a continuous flow reactor. Zeolite NaX is interesting for CO2 adsorption and has already been synthesized in a batch reactor. Growing it in a continuous flow reactor should allow faster production. This work has been done to determine whether it is feasible or not and to optimize the process in case of success. In this report, we show that it is clearly feasible, especially as demonstrated by two samples. One film was grown on a 1200 channels per square inch (cpsi) support that was coated with aluminum oxide. The other was grown on a 400 cpsi monolith. Both were synthesized with a flow rate of 6 ml/min which produced a better film than what was obtained in a batch reactor. However, more synthesis solution was required. Considering the optimization of the process, it is shown that one of the main issues is the formation of a bubble under the monolith that plugs the entrance of some of the channels and causes the formation of a film of very bad quality. This can be avoided by using a different setup. The temperature tests showed that it is possible to grow the film already at 80 C, and also that the film synthesis at 100 C was difficult to control, with a lot of parasite reactions. The synthesis solution also needs to be optimized to obtain better results. The best solution used herein contained 10% more aluminum than the standard recipe, but further increase of the aluminum content might also improve the synthesis. The last parameter that has been studied is the flow rate. Increasing the flow rate results in better films but a part of the reactants is wasted because it does not react. This parameter is directly linked to the reactants concentration and temperature. It needs to be optimized accordingly.

3 Table of contents Introduction! 3 I. Scope of the work! Scope of the previous work! Scope of the present work! 4 II. Experiments! Preparation of the monoliths! Zeolite film growth! Characterization of the samples! 6 III. Results and discussion! Calibration of the reactor temperature! 7 Figure 1: Evolution of the temperature in the batch quartz reactor! Synthesis results! Different reaction temperature! Batch reactor! Continuous flow reactor with 1200 cpsi supports! Continuous flow reactor with 400 cpsi supports! Weight gain during the synthesis! X-ray diffraction! 38 Conclusion! 39 Appendix! 40 Appendix 1: Recipe of the (80NaO2:Al2O3:9SiO2:5000H2O) synthesis solution! 40 Appendix 2: Recipe of the synthesis solution with 10% additional aluminum! 41 References! 42 2

4 Introduction The aim of this work was to grow NaX zeolite on ceramic cordierite monoliths in a continuous flow reactor. Zeolites are microporous crystalline aluminosilicates with a three-dimensional tetrahedra framework that makes it an adsorbent. Different microstructures can be found, but the one used in this work is the FAU framework which has pores running perpendicular to each other, with a diameter of 0.7 nm and a cavity of 1.2 nm where the channels intersect. Scheme 1: FAU three-dimensional structure [1] Two different zeolites have the same FAU framework whose names are X and Y. The difference between these two is the Si/Al ratio which varies between 1 and 1.5 for zeolite X and between 1.5 and 5.6 for zeolite Y. These structures are interesting for their selective adsorption capacities. Zeolite NaX, which was chosen for this work, has a strong potential to be used for removing CO2 from flue gases. A support is required to grow the zeolite as a structured absorbent. In this regard, the most commonly used absorbents are pellets set as a packed bed. The mass transfer in this kind of system is then controlled by the diffusion path, which is determined by the radius of the adsorbent particles and/or the size of zeolite crystals. When decreasing the size of the pellets, the pressure loss is increased which leads to energy loss on the system. This is the reason why so much attention is given to monolith supports. The adsorption capacity per column volume is of the same magnitude as for packed bed, while the pressure drop is about 3-5 times lower. [2] It was chosen to grow the zeolite film in a continuous flow reactor because previous work had been done in a batch reactor and the method required to repeat the synthesis five times. To do it in one step in a continuous flow reactor would lead to a gain of time. 3

5 I. Scope of the work! 1.1 Scope of the previous work Previous work was focused on growing very thin films on the walls of ceramic cordierite monoliths to offer a competitive alternative to traditionally used adsorbents in packed bed. To reach this goal, the objectives of the previous work were: To study the effect of different synthesis routes on the growth of zeolite films on cordierite monoliths To study the effect of different ceramic cordierite monoliths on zeolite film growth To investigate the CO2 and NOx adsorption properties of the structured adsorbents To compare the adsorption performance of the structured adsorbents with traditionally used bead-type adsorbent in a VSA process [3]! 1.2 Scope of the present work This work was focused on optimizing the zeolite film growth in order to gain some time in the process. To reach this goal, the aim of the present work was: To obtain the same results in a continuous flow reactor as the one obtained in a batch reactor To optimize the synthesis regarding different parameters like temperature, aluminum concentration in the synthesis solution and flow rate in the reactor The structured adsorbent were characterized by scanning electron microscopy (SEM) and X-ray diffraction (XRD). 4

6 II. Experiments! 2.1 Preparation of the monoliths NaX films were grown on two different types of ceramic cordierite monolith. The difference between these two types was the number of channel per square inch. The two sizes were 400 and 1200 cpsi. The supports were cut to a cylindrical shape that was 2 cm diameter and 4 or 8 cm high. However, the quality of the 1200 cpsi monoliths had to be improved before use because the supports themselves were too porous to be seeded. It was done with an aluminum oxide slurry which was prepared by mixing 222g of aluminum powder, 150g of water and 11.1g of dispersant. The slurry was then stirred for one hour in order to be homogenized. Meanwhile, the supports were rinsed six times with an 0.1M ammonia solution, then treated for 10 minutes in a 0.4 wt% cationic polymer solution (prepared by diluting a commercial solution of a cationic polymer in distilled water, the ph was adjusted to 8 with an ammonia solution). Finally, they were rinsed six times again with the ammonia solution. When the slurry was ready, part of it was poured in a plastic tube in order to dip the supports in it. The slurry being viscous, it doesnʼt enter the channels easily. Because of that, the monolith has to be dipped carefully in the tube in order to be sure that all the channels are filled with the slurry. The tube was then stirred for one minute. After that, the tube was pulled out of the tube and the excess of slurry in the channels was removed with compressed air. The dipping and drying were repeated three times. The supports were then dried overnight at ambient temperature, then overnight again in a 80 C oven and finally sintered at 1200 C for four hours with a 5 C per minute heating and cooling rate.! 2.2 Zeolite film growth First, the supports (400 and 1200 cpsi) were rinsed six times in an 0.1M ammonia solution. They were then treated in the 0.4 wt% cationic polymer solution for 10 minutes to render the surface positively charged. After charge reversal and an 0.1M ammonia rinsing to remove the excess of polymer solution, the monoliths were treated in a 1 wt% FAU seed solution. Thereafter, the supports were rinsed with ammonia again to remove the seeds excess. Two solutions were tested in order to check the impact of the aluminum concentration on the synthesis. One solution (solution 1) was the previously used synthesis solution and the other one (solution 2) had an additional 10% aluminum The composition of the two different solutions that were mainly used for synthesis can be found in appendix 1 and 2. The synthesis was then launched with the following setup: Monolith Scheme 2: zeolite film growth setup 5

7 In this setup, the synthesis solution is pumped from the beaker to the pre-heater, which is needed when the flow rate is relatively large in order to elevate the synthesis mixture temperature near to the synthesis temperature. Then, the solution is pumped in the reactor where the support is in order to grow the film. The films were grown for 6 hours and 40 minutes with different flow rates (0.6, 4 and 6 ml/min).! 2.3 Characterization of the samples The samples were mainly analyzed with a scanning electron microscope (SEM) on top and cross section views. For large samples, the photos were taken along certain channels and also at various heights across the diameter of the support. X-ray diffraction (XRD) was also utilized to determine the phases present in the samples. 6

8 III. Results and discussion! 3.1 Calibration of the reactor temperature The first step of the work was to calibrate the temperature in the reactor. The first checks were done in the plastic batch reactor. It took approximatively 45 to 50 minutes to reach steady state. With the oil bath at C, the temperature in the reactor was 91.0 C. Then, the time it took for the temperature to rise in the quartz reactor, set as a batch, was determined. Data are given in the following table and figure: Time (min) Synthesis solution temperature ( C) Heat transfer fluid temperature ( C) 1 24,9 2 30,0 3 37,5 53,7 4 46,6 62,5 5 56,1 72,0 6 65,3 78,2 7 73,9 86,0 8 81,9 93,2 9 89,6 99, ,0 100, ,2 100, ,0 100, ,5 100, ,6 100, ,8 100, ,9 100, ,8 100, ,9 100, ,0 100, ,0 100, ,0 100, ,0 100, ,0 100,0 Table 1: Temperature rise as a function of time 7

9 100,0000 Temperature versus time Temperature ( C) 75, , , Time (minutes) Synthesis solution temperature ( C) Heat transfer fluid temperature ( C) Figure 1: Evolution of the temperature in the batch quartz reactor Figure 1 shows that it takes approximatively 10 minutes for the temperature to reach steady state. This is 35 minutes less than for the plastic batch reactor. It is an important data for the analysis of the difference between the synthesis in the plastic and quartz reactor since the reaction temperature is not reached at the same time in both cases. In other words, the actual synthesis time differs in both types of reactor since the protocol was the same for both batch reaction. The temperature of the heat transfer fluid was set to 91.9 C to reach 91.0 C in the reactor.! 3.2 Synthesis results!! Different reaction temperature Size of the support Characteristics of the support Type of reactor : 1cm; height: 2cm 1200 cpsi, aluminum coated Batch, plastic Temperature of the reactor 80, 85, 90, 95 and 100 C Reaction time Synthesis solution 1 hour and 20 minutes Solution with 10% extra aluminum Table 2: Characteristics of the reactions 1 1 Two different solutions were used for synthesis. The first is given in part 2.2. The other is a solution with 10% extra aluminum. A table containing the exact recipes can be found in appendix 1 and 2. 8

10 The film was grown at 5 different temperatures (80, 85, 90, 95 and 100 C) in a batch reactor during 1 hour and 20 minutes. The plastic reactor was put in an oil bath which had its temperature set to one of the five temperatures in the synthesis solution. The monolith was then rinsed six times with a 0.1M ammonia solution. The results were the following: Figure 2: Cross section view of the film grown at 80 C Figure 3: Cross section view of the film grown at 85 C 9

11 Figure 4: Cross section view of the film grown at 90 C Figure 5: Cross section view of the film grown at 95 C Figure 6: Cross section view of the film grown at 100 C 10

12 These experiments were mainly done to know at which temperature the reaction starts. The pictures show that at the five different temperatures, the film is quite dense and continuous for a single and short step synthesis, which means that the reaction has started at 80 C. The samples made at 80 and 100 C were investigated more carefully than the others to determine whether the film had any gradient. The supports were then observed on both sides and in the middle as shown in Fig. 7: X X X +15 mm +7.5 mm 0 mm Figure 7: Top view of the 80 C sample, from left to right: three different zooms (10,000x, 25,000x and 50,000x); from top to bottom: one side, middle and other side (+0, 7.5 and 15mm) The micrographs in Fig. 7 show that in this sample, there was no major difference from one side to the other but the grains seems to be slightly smaller on the +15mm side. Next table shows the average thickness at the three positions: +0mm +7.5mm +15mm nm nm nm Table 3: Average thickness (in nm) of the zeolite film grown at 80 C 11

13 The values in this table have to be taken with precaution because the number of micrographs at each location was not the same and thickness was measured randomly and not integrated over the whole surface. The same procedure was repeated with the 100 C sample: X X X +15 mm +7.5 mm 0 mm Figure 8: Top view of the 100 C sample, from left to right: three different zooms (10,000x, 25,000x and 50,000x); from top to bottom: one side, middle and other side (+0, 7.5 and 15mm) In Fig. 8, we can see a difference between one side and the other. The grains are much larger on the +15mm side. The film has, as for the 80 C, a gradient in film quality which is also reflected by the data in next table, even though these values must be taken with precaution for the same reason as before: +0mm +7.5mm +15mm 511,6 nm 747,4 nm 810,0 nm Table 4: Average thickness (in nm) of the zeolite film grown at 100 C Note: the sides were not marked on these samples which explains why the film is thicker on one side for the 80 C sample and the other for the 100 C. In fact, it probably is the same side with regard to the bottom of the reactor. 12

14 A comparison can then be done between the two previous samples: 80 C 100 C +15 mm +7.5 mm 0 mm Figure 9: Top view of: from left to right: 80 and 100 C samples; top to bottom: one side to the other 13

15 The sides with the same characteristics were made to correspond in the micrographs of Figure 9.. It can be seen that the 100 C sample exhibits a much larger grain size, a film that is denser and more continuous than that grown at 80 C. Comparison between Table 3 and 4 indicates the same trend for thickness. Indeed, the largest grains in the 80 C sample are approximatively of the size of the smallest grains in the other sample. The reaction was faster at 100 C but a hydroxysodalite (HS) crystal (or zeolite P) can be observed in Fig 8, 0mm and 10000x. Regarding this issue, it couldnʼt be determined which one of those two phases was present, so this kind of crystals are referred to as HSP crystals (hydroxysodalite or zeolite P) in the following. Moreover, these crystals have to be avoided for this work. To conclude, the film begins to grow at 80 C already. But there is a gradient in the films grown in batch in terms of thickness when comparing one side of a channel to the other. Finally, the film grown on the 100 C sample is much better than that grown at 80 C in terms of grain size and film density. The 100 C sample is further addressed for testing later in this report.!! Batch reactor The aim of the following experiment was to investigate the influence of the nature of the batch reactor by repeating the standard synthesis developed previously [5]. The main parameters of this experiment are gathered in Table 5. Size of the support Characteristics of the support Type of reactor : 2cm; height: 4cm 1200 cpsi, aluminum coated Batch, plastic and quartz Temperature of the reactor 91 C Reaction time Synthesis solution 5 times 1 hour and 20 minutes Solution with 10% extra aluminum Table 5: Characteristics of the reactions The two first syntheses were made in two different batch reactors, one in plastic, the other in quartz. For these experiments, the setup previously shown was not used. The experiments were prepared in a similar way but instead of operating as a continuous reactor during 6 hours and 40 minutes, they were made in five steps of 1 hour and 20 minutes with the reactor as a batch. Between each step, the synthesis solution was renewed and the monolith rinsed six times with a 0.1M ammonia solution. The following pictures in Fig. 10 and 11 were obtained by SEM and show how the zeolite film grew on the supports: 14

16 Figure 10: Top view of the zeolite film grown in the plastic batch reactor Figure 11: Top view of the zeolite film grown in the quartz batch reactor In Fig. 10 and 11, more HSP crystals can be seen in the quartz reactor. The explanation could be that, as explained before, the quartz reactor reaches steady state faster, which means that the reaction is longer. The reactants are then used during a longer time which leads sooner to a lower concentration in aluminum. This low concentration in aluminum could be the reason why the HSP crystals form. Cross section pictures were also taken in order to evaluate the films quality and thickness. Typical results are shown in Fig. 12 and

17 Figure 12: Cross sectional view of the zeolite film grown in the plastic batch reactor Figure 13: Cross sectional view of the zeolite film grown in the quartz batch reactor In both cases, the zeolite films were dense with well inter grown crystals. The average thickness of the two films was also measured and the values are given in Table 6: Plastic reactor Quartz reactor Table 6: Average thickness (in μm) of the zeolite film grown in two different reactors The thickness of the film grown in the quartz reactor was larger due to the longer reaction time. The problem is that more HSP can be found in the sample that was in the quartz reactor, which means as explained before that the reactant concentration was depleted earlier. To optimize the synthesis in the quartz reactor and avoid the formation of HSP, the reaction time should be decreased, or the reactant concentration increased. In summary, no major difference was found for both types of reactor, except the faster time 16

18 to reach steady state in the quartz reactor. Since the nature of the quartz reactor was found not to influence synthesis, experiments using this reactor in a continuous mode were further carried out.!! Continuous flow reactor with 1200 cpsi supports The aim in the rest of this study was to synthesize NaX films in a continuous reactor for the first time. The parameters used for the first set of experiments are given in Table 7. Size of the support Characteristics of the support Type of reactor : 2cm; height: 4cm 1200 cpsi, aluminum coated continuous flow reactor Temperature of the reactor 91 C Reaction time Synthesis solution 6 hours and 40 minutes Solution with 10% extra aluminum Table 7: Characteristics of the reactions For this reaction, the setup described in the experimental part was used. The goal of this part was to check different flow rates: 0, 0.6 and 6 ml/min. Since 0 ml/min corresponds to a batch reactor, it was tested to know what would happen without changing the solution. Figure 14: Top view of the 0mL/min sample As shown in Fig. 14, a lot of HSP crystals grew in the 0mL/min sample, which typically happens when the concentration of the reactants in the solution is too low. Zeolite crystals transform into HSP crystals. This is illustrated later in this work with SEM pictures of the transformation. In addition, the film was not dense and its thickness was around 500 nm which is rather thin considering the reaction duration. As for the real continuous flow reactor, a flow rate of 0.6 ml/min was chosen to have the same residence time as in the batch reactor with five synthesis steps, while 6 ml/min was 17

19 chosen to test a higher flow rate. Figure 15: Top view of the 0.6 ml/min sample The left picture in Fig. 15 shows that the support is barely covered by a film. This is certainly due to a seeding problem. The picture on the right hand side shows a film that is not too dense and with some HSP crystals. Figure 16: Cross section view of the 0.6 ml/min sample As shown in Fig. 16, the average thickness is slightly below that of the batch sample with five steps despite the same residence time. The overall quality is worse than the standard batch experiment with 5 synthesis steps, although it shows that it is feasible in one step. 18

20 Figure 17: Top and cross sectional view of the 6mL/min sample As can be seen in Fig. 17, the film synthesized with a flow rate of 6 ml/min had large grains, and was dense and continuous. Also, the thickness was larger than for the 0.6 ml/ min experiment. Next table compares the average thickness in both samples: 0.6 ml/min 6 ml/min Table 8: Average thickness (in μm) of the zeolite film The film was carefully investigated in the 6 ml/min sample. Fig. 18 shows the difference in growing behavior of the film between channels, from the edge to the center of the sample at the same height. Ch. 1 Ch. 2 Figure 18: Top view of channel 1 to 15 19

21 Ch. 3 Ch. 4 Ch. 5 Ch. 6 Ch. 7 Ch. 8 Figure 18: Top view of channel 1 to 15 (cont.) 20

22 Ch. 9 Ch. 10 Ch. 11 Ch. 12 Ch. 13 Ch. 14 Figure 18: Top view of channel 1 to 15 (cont.) 21

23 Ch. 15 Figure 18: Top view of channel 1 to 15 (cont.) Clear discrepancies can be seen from channel 1 to 15. From channel 2 to 8, the grains are larger and the film is dense and continuous. Channel 9 to 15 have smaller grains, the film is not as dense. Moreover, more HSP crystals can be seen in these channels. Channel 1 is an exception because the channel was blocked by some aluminum from the coating. The 15 th channel correspond approximatively to the middle of the sample but it can be expected that the other side of the sample is identical as shown later. The film gradient, showing the quality of the film, looks like that: Film with bad quality Film with good quality Trapped bubble Figure 19: drawing showing schematically the location and influence of the entrapped bubble. A possible explanation for these discrepancies is that bubbles formed in the synthesis solution and accumulated under the monolith which blocked the access of fresh synthesis solution to certain channels. This is illustrated in Fig 19. Fig. 20 shows a series of micrographs which were taken in channels 2 and 15 at different heights to check whether there were also a film gradient in the longitudinal direction in the 6 ml/min sample. 22

24 Ch. 2 Ch mm 30 mm 25 mm Figure 20: Top view of: left: 2 nd channel; right: 15 th channel; from top to bottom: from 0 to 35mm in the channels with a 5mm step 23

25 Ch. 2 Ch mm 15 mm 10 mm Figure 20: Top view of: left: 2 nd channel; right: 15 th channel; from top to bottom: from 0 to 35mm in the channels with a 5mm step (cont.) 24

26 Ch. 2 Ch mm 0 mm Figure 20: Top view of: left: 2 nd channel; right: 15 th channel; from top to bottom: from 0 to 35mm in the channels with a 5mm step (cont.) Fig. 20 shows that differences between a channel on the outside and one in the middle of the sample can be seen very easily on these series. The grains in the 15 th channel are far smaller than in the 2 nd. There is the presence of HSP crystals and the film is barely continuous. Besides, these pictures show that in the continuous flow reactor, at least in this sample, there is no gradient along the length of a channel. Finally, an edge effect can be seen. At 0 and 35mm, the film looks the same in both channels, the quality being better for the 15 th channel. After these experiments, it was decided to move back to the 400 cpsi support to avoid preparation of the aluminum oxide coating, which is time consuming.!! Continuous flow reactor with 400 cpsi supports The first experiments were performed at a temperature of 91 C, with solutions 1 & 2 and for flow rates of 4 and 6 ml/min. 25

27 Size of the support Characteristics of the support Type of reactor : 2cm; height: 8cm 400 cpsi continuous flow reactor Temperature of the reactor 91 C Reaction time Synthesis solution 6 hours and 40 minutes Both Table 9: Characteristics of the reactions The goal of these experiments was to find out which of the two solutions was the most suitable for the 400 cpsi supports, and also to test two different flow rates: 4 and 6 ml/min. Fig. 21 shows a series of pictures for the 4 ml/min sample using the solution with no extra aluminum. Ch. 1 Ch. 2 Ch. 3 Ch. 4 Ch. 5 Ch. 6 Ch. 7 Ch. 8 Ch. 9 Figure 21: from left to right and top to bottom: top view of channels 1 to 16 (sample with no aluminum added and 4mL/min flow rate) 26

28 Ch. 10 Ch. 11 Ch. 12 Ch. 13 Ch. 14 Ch. 15 Ch. 16 Figure 21: from left to right and top to bottom: top view of channels 1 to 16 (sample with no aluminum added and 4mL/min flow rate) (cont.) The entire diameter of the sample was carefully investigated. Fig. 21 shows two different size of grain depending on the channels. In channel 1 and 2, the film did not grow properly even though it is located on the outside. The channels could have been blocked by a bubble, which is not really probable because trapped bubbles located under the monolith but close to the edge have a tendency to escape easily. Another possibility is that the bubbles could not escape if the monolith was too close to the wall of the reactor. Then, the film in channel 3 to 9 and 14 to 16 presents an acceptable grain size compared to the rest of the sample. In channel 10 to 13 grain size is small and the film does not seem continuous. This is probably due to the bubble blocking the entrance of the channels as mentioned before. A closer look to channel 3, 6 and 10 led to the following results compiled in Fig

29 Ch. 3 Ch. 6 Ch mm 10 mm 20 mm 30 mm Figure 22: Top view of different channels (left to right: n 3, 6 and 10) at different heights, from top to bottom (0 to 30mm in the channel, with a 10mm step) Fig. 22 shows that there is no gradient in each channel. The grains have the same size all along one channel. Next pictures concern the 6mL/min flow rate with the solution without additional aluminum. 28

30 Ch. 1 Ch. 2 Ch. 3 Ch. 4 Ch. 5 Ch. 6 Ch. 7 Ch. 8 Ch. 9 Ch. 10 Ch. 11 Figure 23: from left to right and top to bottom: top view of channels 1 to 11 (sample with no aluminum added and 6mL/min flow rate) The conclusion is the same as for the previous sample. Channel 1 to 4 and 11 to 12 are better than the others. Interestingly, the grain size is about twice the size of the previous sample However, it does not appear more continuous and HSP crystals are present. This raises a question. With a flow rate of 6 ml/min, the film is thicker than with 4 ml/min, and the grains are larger, which is the expected evolution. But the HSP crystals should not have formed with the highest flow rate if they really form when the reactant concentration becomes too low. This may be due to the heating treatment in the pre-heater. Indeed, if 29

31 temperature is too high in the pre-heater, the reaction already begins, which in turn lowers the reactant concentration and worsen the quality of the film. The next two samples were done with the same experiment, but the synthesis solution contained 10% additional aluminum. Ch. 1 Ch. 2 Ch. 3 Ch. 4 Ch. 5 Ch. 6 Ch. 7 Ch. 8 Ch. 9 Ch. 10 Figure 24: from left to right and top to bottom: top view of channels 1 to 10 (sample: 4mL/ min flow rate, 10% more Al in solution) 30

32 The same problem as before can be seen in Fig. 24. The change in grain size is between channel 7 and 10. But as written above, the distribution of the affected channels is not symmetric so it is not possible to extrapolate the results of channel 11 to 16. No HSP can be reported for this sample, and the film is more continuous than with the other synthesis solution, even the 6 ml/min sample. The grains have approximatively the same size as with the previous sample. The next series of pictures in Fig. 25 were obtained for the sample synthesized with a flow rate of 6 ml/min. The photos are taken at three different heights in the sample (0, 37 and 75mm, from left to right) and from channel 1 to 16 (top to bottom): 0 mm 37 mm 75 mm Ch. 1 Ch. 2 Ch. 3 Figure 25: Top view of the sample: 6mL/min flow rate, 10% extra Al in solution; from left to right: 0, 37 and 75 mm height; top to bottom: channels 1 to 16 31

33 0 mm 37 mm 75 mm Ch. 4 Ch. 5 Ch. 6 Ch. 7 Figure 25: Top view of the sample: 6mL/min flow rate, 10% extra Al in solution; from left to right: 0, 37 and 75 mm height; top to bottom: channels 1 to 16 (cont.) 32

34 0 mm 37 mm 75 mm Ch. 8 Ch. 9 Ch. 10 Ch. 11 Figure 25: Top view of the sample: 6mL/min flow rate, 10% extra Al in solution; from left to right: 0, 37 and 75 mm height; top to bottom: channels 1 to 16 (cont.) 33

35 0 mm 37 mm 75 mm Ch. 12 Ch. 13 Ch. 14 Ch. 15 Figure 25: Top view of the sample: 6mL/min flow rate, 10% extra Al in solution; from left to right: 0, 37 and 75 mm height; top to bottom: channels 1 to 16 (cont.) 34

36 75 mm Ch. 16 # # # # # # # # # Figure 25: Top view of the sample: 6mL/min flow rate, 10% extra Al in solution; from left to right: 0, 37 and 75 mm height; top to bottom: channels 1 to 16 (cont.) The bubble problem also appears in this sample shown in Fig. 25 but only in channel 6 to 9. But there is no difference between the top and the bottom of each channel. It seems that there were a seeding problem at the top of certain channels (e.g. 8, 9 and 10), in which the crystals are comparatively larger there, but not continuous. Nevertheless, the grains are larger than in the previous sample, about twice the size and the film is continuous and dense. This sample was the best obtained on 400 cpsi supports and is close in quality to 6mL/min sample grown on a 1200 cpsi,support. The thickness of three samples is compared in Table F6-T91-S2 400-F4-T91-S2 400-F6-T91-S Table 10: Average film thickness (μm) for some experiments 2 These three samples are those with the most data recorded. Either with 1200 cpsi or 400 cpsi supports, the samples grown with a flow rate of 6mL/min show comparable properties, i.e. a continuous and dense film with good thickness. These two samples were the best of all experiments. The lack of data is not the only problem associated with the thickness measurement. There is also the fact that the film is very thick where HSP crystals are formed as shown later with the 100 C samples. The next two experiments were made at 100 C, the solution having the most aluminum, and a flow rate of 4 and 6 ml/min. They have been done in order to see if an increase of the temperature could improve the process by increasing the reaction rate. Size of the support Characteristics of the support Type of reactor : 2cm; height: 8cm 400 cpsi continuous flow reactor 2 In order to simplify the notation, a code has been used, the first number is the number of cpsi, the second is the flow rate (ml/min), the third is the temperature ( C) and the fourth is the solution. 35

37 Temperature of the reactor 100 C Reaction time Synthesis solution 6 hours and 40 minutes Solution with 10% extra aluminum Table 11: Characteristics of the reactions For these two samples, pictures were recorded at the bottom and at the top of the monolith but without ensuring any correspondence between channels at the two different heights. Therefore, it would be wrong to try to analyze the samples along a particular channel. Nevertheless, interesting observations can be made. The obtained micrographs illustrate how hard it is to grow the zeolite film at 100 C, and also the thickness difference between NaX films with or without HSP crystals. Figure 26: Top and cross section view of the 4 ml/min sample with the 10% more aluminum solution, 100 C, bottom part of the sample Figure 27: Top and cross section view of the 4 ml/min sample with the 10% extra aluminum solution, 100 C, top part of the sample 36

38 Fig. 26 and 27 show that the film cracks due to the formation of HSP crystals (bottom part, top view). The top part of the sample (see Fig. 27) exhibited a HSP film. It was the only time this type of film was encountered during all experiments. Fig. 28 and 29 show the 6 ml/min flow rate sample grown at 100 C. Figure 28: Top and cross section view of the 6 ml/min sample with the 10% more aluminum solution, 100 C, bottom part of the sample Figure 29: Top and cross section view of the 6 ml/min sample with the 10% more aluminum solution, 100 C, top part of the sample Again, the film cracked when HSP crystals formed (top view of the top part). Comparing both samples, it seems that the film grows faster when HSP start forming. In the end, the resulting film is usually 1μm thicker.!! Weight gain during the synthesis In order the quantitate the film growth on the samples, the weight gain was measured on the 400 cpsi supports. Of course, it only gives an indication of the crystal growth since the 37

39 NaX and HSP content is not known. The results are given in Table 12. Monolith weight gain 400-F4- T91-S1 400-F6-T91- S1 400-F4-T91- S2 400-F6-T91- S2 400-F4-T100- S2 400-F6-T100- S2 weight before (g) 7,0622 6,9797 7,2368 6,9922 6,7791 6,8969 weight after (g) 7,8128 7,5809 7,7206 7,7555 7,9256 8,0230 weight gain (%) 10,63 8,61 6,69 10,92 16,91 16,33 Table 12: monolith weight gain The 100 C samples gained more weight than the others but accompanied with the formation HSP crystals which was responsible for weight gains from 7-10% to 16%.! 3.3 X-ray diffraction The curve obtained by XRD is shown in Fig. 30. Fig. 30: XRD analysis of the 400 cpsi, 91 C and 6mL/min flow rate sample According to literature [4], all these peaks belong either to cordierite or to zeolite NaX which confirms that the film obtained has the right crystal structure. 38

40 Conclusion The main goal of this work was to show that the synthesis of NaX films is doable in a continuous flow reactor. Interesting results were obtained like for the 400 & 1200 cpsi, 91 C with a flow rate of 6 ml/min samples. The resulting film were continuous and dense, with a large grain size and a thickness of 2 to 3 µm. The problem of this method is that it requires consumption of substantial quantities of synthesis solution compared to the batch process. Therefore, is the higher need of the process in synthesis solution worth the gain in time and in quality? The second goal was to optimize the process, which can be done in many ways due to the number of parameters. First, including a pre-heater with better temperature control might help synthesis. With regard to temperature, 91 C was the optimum value for the two samples mentioned above. Synthesis at 100 C would need real optimization; the parameters used in this work produced too much HSP. Preliminary results in this work suggest that a temperature of 80 C may be enough for synthesis to form a good film. A synthesis solution with more aluminum may also improve the reaction and the quality of the film. Moreover, there is also the problem of the pressure drop in the sample. It causes part of the solution to bypass the sample. It would certainly be a useful improvement to have proper sealing between the sample and the reactor to avoid the solution to flow around the sample. It would force all the solution to go through the channels, and maybe the bubbles that are formed under the support. Another way to tackle this problem is probably to have the inlet above the support and the outlet under. It would cause all bubbles to escape from the reactor without even getting in contact with the monolith. Finally, the flow rate must be optimized by tailoring the other parameters in order to obtain a film with sufficient quality and without wasting too much reactant. 39

41 Appendix! Appendix 1: Recipe of the (80NaO2:Al2O3:9SiO2:5000H2O) synthesis solution Synteslösning EFM 2 (Samma sammansättning som kumakiri, artikel 63 i Fredriks databas) GrundmängTotal mängd 1 Kemikalie TPAOH TEAOH (TMA)2O Na2O Al2O3 SiO2 H2O EtOH NH3 0 TEOS M TPAO %W TPA Bindzil30/360, sol Bindzil30/360, torr Bindzil30/220, , sol ,120 33,12 H2O (Tillsatt totalt) 1, TMAOH5H2O M NaOH lösning 0 0 2,13 2,13 NaOH pastiller 0, ,25 0,25 Al2(SO4)3*18H2O 0, , Al-isopropoxid TEAOH (20%) Natriumaluminat Etanol M ammoniak 0 0 0,96 0,96 Na2SiO3*9H2O (ren) 0, , , Na2SiO3*9H2O (63 %SiO2, 18% Na2O) ,46 36,46 Summa , , , , Molbr , , , Molbr , , , Molbr , , , Dissolve Al2(SO4)3*18H2O in 10g H2O (ca. 1h) and Na2SiO3*9H2O and NaOH in 23,12g H2O (ca. 1.5h), Mix the solutions and stir vigorously for 1-2 minutes 40

42 ! Appendix 2: Recipe of the synthesis solution with 10% additional aluminum 10% MORE Al Synteslösning EFM 2 (Samma sammansättning som kumakiri, artikel 63 i Fredriks databas) GrundmängTotal mängd 1 Kemikalie TPAOH TEAOH (TMA)2O Na2O Al2O3 SiO2 H2O EtOH NH3 0 TEOS M TPAO %W TPAO Bindzil30/360, sol Bindzil30/360, torr Bindzil30/220, , sol ,120 33,12 H2O (Tillsatt totalt) 1, TMAOH5H2O M NaOH lösning 0 0 2,13 2,13 NaOH pastiller 0, ,275 0,275 Al2(SO4)3*18H2O 0, , Al-isopropoxid TEAOH (20%) Natriumaluminat Etanol M ammoniak 0 0 0,96 0,96 Na2SiO3*9H2O (ren) 0, , , Na2SiO3*9H2O (63 %SiO2, 18% Na2O) ,485 36,485 Summa , , , , Molbr , , , Molbr , , , Molbr , , , Dissolve Al2(SO4)3*18H2O in 10g H2O (ca. 1h) and Na2SiO3*9H2O and NaOH in 23,12g H2O (ca. 1.5h), Mix the solutions and stir vigorously for 1-2 minutes 41

43 References [1] International Zeolite Association website: [2] A. Mosca, Structured Zeolite Adsorbents for PSA Applications, Doctoral thesis made in the Department of Chemical Engineering and Geosciences, Division of Chemical Engineering, Luleå University of Technology, 2009, pp [3] A. Mosca, Structured Zeolite Adsorbents for PSA Applications, Doctoral thesis made in the Department of Chemical Engineering and Geosciences, Division of Chemical Engineering, Luleå University of Technology, 2009, pp [4] A. Mosca, Structured Zeolite Adsorbents for PSA Applications, Doctoral thesis made in the Department of Chemical Engineering and Geosciences, Division of Chemical Engineering, Luleå University of Technology, 2009, p. 21 [5] A. Mosca, Structured Zeolite Adsorbents for PSA Applications, Doctoral thesis made in the Department of Chemical Engineering and Geosciences, Division of Chemical Engineering, Luleå University of Technology,