Photochemical microfluidic synthesis of vitamin D3 by improved light sources with photoluminescent substrates

Wenhao Niu,Yuanzhi Zheng,Ying Li,Le Du,Wei Liu

The State Key Laboratory of Chemical Resource Engineering,Beijing Key Laboratory of Membrane Science and Technology,Beijing University of Chemical Technology,Beijing 100029,China

Keywords:Microfluidics Photochemistry Composites UVB emission Vitamin D3

ABSTRACT This study presents a novel technique for the controllable preparation of photoluminescent substrates to enhance the photochemical microfluidic synthesis of vitamin D3 .The dip-coating method to prepare the substrates was experimentally optimized,and the corresponding emission behaviors were systematically investigated.The substrates were successfully used to enhance the ultraviolet B(UVB)emission of a low-power light source(e.g.,an 8 W lamp),whose UVB emission intensity was increased by approximately 11 times.By virtue of the novel light source,the productivity of a single set of photochemical microreactor with a 12-meter-long channel(0.6 mm i.d.) was increased to 1.83 kg·a?1,which was 42%higher than that of a 100 W lamp,and no cooling devices were used.The method is simple and has great potential to replace traditional medium-pressure mercury lamps for UVB-irradiated photochemical reactions.

1.Introduction

Photochemical production has drawn increasing attention owing to its advantages such as high selectivity,environmental friendliness,and improved process safety[1–4].In traditional batch reactors,light attenuation(according to Bouguer–Lambert–Beer law)normally exists due to the large equipment scale and the large depth of liquid [5,6].With the development of microfluidics,photochemical synthesis has attracted wide attention[7,8].The characteristic size of microreactors is on the order of millimeters or even micrometers,significantly reducing the effect of light attenuation [9,10].In addition,the scale-up of microreactors could be achieved by numbering-up without scale-up effects[11–16].Nevertheless,there still exit two major challenges for photochemical microfluidic systems:one is the light source,and the other is the fabrication material of the reactors.

The feasibility of a light source used in photochemical processes mainly depends on whether its emission wavelength is consistent with the absorption characteristics of reactants.Additionally,the energy costs of the light source(including both the light energy and the cooling energy)need to be considered,since it could consume more than a half of the energy for some photochemical processes [17,18].For example,light-emitting diodes(LEDs)such as vis-LEDs(LEDs emit visible light)and UV-LEDs(LEDs emit ultraviolet light)in the UVA range(315–400 nm) have replaced a large number of tradition mercury lamps [19].However,in the UVB range(280–315 nm),the power of LEDs is relatively low.Taking the 315 nm LEDs as an example,the irradiation power(Pel)is 5 mW·cm?2and irradiation efficiency(ηel)is 1.07×10?2,which are approximately 95%and 90%lower than those of the 365 nm LEDs,respectively[20,21].Normally,low-pressure mercury lamps(1–10 W)could not meet the requirements of UVB emission intensity,and mediumpressure mercury lamps(100–1000 W)with higher heat release have to be used for the photochemical synthesis in the UVB range.

On the other hand,the fabrication materials of microreactors also limit practical applications of mercury lamps.The channels or tubes are normally made of polymers such as fluorinated ethylene propylene(FEP),perfluoroalkoxy(PFA),and polydimethylsiloxane(PDMS),since soft materials are feasible to be integrated with light sources[22,23].However,both the low-and medium-pressure mercury lamps have a strong emission at 254 nm,and aging of polymers is accelerated under irradiation at this wavelength[24,25].Mercury lamps have to be used with filters or cold traps(between the light source and the reactor)made of silicate or borosilicate glass[6,19].The transmittances of borosilicate and silicate are only 58%and 13%at 303 nm,respectively.In this case,the 303 nm emission(mercury lamps emit 254,303,and 365 nm in the UV range)is significantly reduced.To complement the emission in the UVB range,higher electrical power is required,which further increases the energy cost.

With the development of photoluminescent materials,these problems might be solved by appropriate transformation.For example,Gd3+-doped ZnB2O4and LaB3O6:Bi,Gd particles could absorb the radiation at 254 nm and emit at 312 nm[26,27].If these photoluminescent particles were coated onto the substrates between light sources and reactors,desired transformation could be achieved.Even if the silicate or borosilicate filters were used,it could still substantially enhance the UVB irradiation,and the energy costs could be correspondingly reduced.In this regard,the improvement of UVB irradiation might be achieved by appropriately integrating photoluminescent materials into light sources.

In our previous studies,uniform coating of nanodispersions onto film substrates has been realized and the coating method has been demonstrated[28,29].On this basis,if the LaB3O6:Bi,Gd photoluminescent particles were coated onto the surface of glass substrates or vessels,it could not only enhance the irradiation in the UVB range,but also reduce the 254 nm irradiation to avoid the aging of the polymer channels or tubes.

In this study,the photochemical production was conducted under enhanced UVB irradiation by integrating photoluminescent substrates into low-power lamps.The photoluminescent substrates were prepared by dip-coating with photoluminescent particles.The UVB emission intensity of the light through the particle-coated substrates was experimentally assessed.The effect of UVB irradiation on a practical photochemical process—the synthesis of vitamin D3(VD3)was systematically investigated.

2.Materials and Methods

2.1.Materials

LaB3O6:Bi,Gd photoluminescent particles were purchased from Xinlian Materials,Co.Tetraethylorthosilicate (TEOS,98.0%) and methyltrimethoxysilane (MTMS,99.0%),hydrochloric acid (HCl,36.5%),isopropanol (99.5%),and guanidine hydrochloride (GuHCl,99.0%)were purchased from Tianjin Yongda Chemical Works,Co.The silane-coupling agents KH560 (97.0%) and KH570 (97.0%) were purchased from in Shuguang Organic Chemical Works,Co.7-Dehydrocholesterol(7-DHC,95.0%),vitamin D3(VD3,99.0%),and tertbutyl methyl ether (t-BME,99.0%) were purchased from Sigma-Aldrich Co.

2.2.Preparation of photoluminescent substrates

Certain amounts of TEOS,MTMS,KH560,and KH570 were mixed in a beaker,and isopropanol was then added to obtain the solution A.By mixing HCl and GuHCl at a certain ratio,their aqueous solution was prepared as the solution B.The solutions A and B were then evenly mixed and placed at room temperature for 24 h to obtain a silica sol.Afterwards,different amounts of LaB3O6:Bi,Gd particles were then dispersed into the silica sol to obtain a suspension for dip-coating.

The clean substrates including glass chips (for spectra measurement)and vessels(for photochemical production)made of quartz or borosilicate,were washed by a HCl aqueous solution(3 mol·L?1)and then put into an oven for drying.The dip-coating method is illustrated in Fig.1a.The substrates were immersed into the prepared LaB3O6:Bi,Gd suspension and were then pulled up by the gearing of an injection pump at a certain rate from 1 to 12 cm·min?1.After dip-coating,the particle-coated substrates were pre-dried at 50 °C for 20 min,and then dried at 80°C for 4 h.

2.3.Photochemical synthesis of vitamin D3 by enhanced UVB irradiation

To verify the effect of enhanced UVB emission intensity by the photoluminescent substrates on practical photochemical production,vitamin D3(VD3)was chosen as the target since the UVB light was the most favorable condition for one of the key steps [30].The reaction pathway is illustrated in Fig.S1 in the Supplementary material.In the experiments,two 8 W low-pressure mercury lamps with a strong emission at 254 nm were used for the two-step synthesis—one lamp for ring opening and the other integrated with the particle-coated vessel(made of quartz or borosilicate) for ring closure.The vessel was placed between the 8 W low-pressure mercury lamp and reactor tubes,as shown in Fig.1b.Transparent FEP tubes(0.6 mm i.d.)were coiled on the outside of all the vessels and served as the microchannels.As a matter of fact,when the inner diameter was larger than 2 mm,the photochemical reaction efficiency was significantly decreased in the exploratory experiment.According to the Bouguer–Lambert–Beer law,light attenuation occurs when the thickness or depth of liquid increases.

In the photochemical experiments,7-DHC was dissolved in t-BME at a concentration of 10 g·L?1,and this solution was then pumped using a metering pump.The reactant solution and N2gas (0.3 MPa) was injected into a PEEK T-mixer to generate segmented flow (Video S1 showing the formation of segmented flow is available in the Supplementary material).The 7-DHC solution successively passed through the photochemical reaction zones and the thermochemical reaction zone.Recycle operation was adopted at higher flow velocities,to ensure that the efficiencies were compared at the same reaction time.Specifically,the N2gas and solution were separated in a glass vessel when the outflow was collected.Then the solution was recycled into the T-mixer for further reaction.The circulation was stopped when the photochemical equilibrium was achieved.After collection,all the samples were stored at ?20°C before the analysis by high-performance liquid chromatography(HPLC).To prevent undesired oxidation during the photochemical reaction,the solution was pumped after 5 min of N2blowing in the tubes,while the sampling bottle was connected with a N2balloon.

For comparison,the ring-closure step was also conducted using a 100 W medium-pressure mercury lamp instead of the 8 W lamp with particle-coated vessel.In this case,a cold trap made of borosilicate or silicate was employed for heat exchange and filtering off the radiation below 260 nm at the same time.

2.4.Characterization and analysis

Morphologies of the particle-coated substrates were observed by scanning electron microscopy (SEM;TM3000,Hitachi).A fibercoupled spectrometer(Ocean Optics USB4000 UV–vis-ES)was used to assess the emission characteristics of light sources with or without the photoluminescent substrates.The composition of photochemical products was determined by HPLC(Shimadzu UFLC-XR)with a C18 column and a UV detector(the detection wavelength was 282 nm).Details of HPLC analysis and compositions of the reaction are provided in Section S2 of the Supplementary Material.

3.Results and Discussion

3.1.Optimization of dip-coating conditions for preparing photoluminescent substrates

Quartz and borosilicate vessels were used as the particle-loaded photoluminescent substrates,which were then integrated with 8 W low-pressure mercury lamps as improved light sources.The emission characteristics were compared with those of the traditional light sources,which employed 100 W medium-pressure lamps as well as silicate and borosilicate cold traps as the filters,as shown in Fig.2.According to the original emission spectrum of the 8 W lamp(Curve I),the intensity at 254 nm is much greater than those in the UVB region.After dip-coating,UVB emission at 312 nm was promoted when the 254 nm light passing through the coating layers(Curves II and III),which resulted in the transition emission from6P7/2→8S7/2by the doped Gd3+[31].However,not all the 254 nm light was absorbed by the coating layers,unabsorbed light could penetrate through quartz and a weak emission was maintained at 254 nm(Curve II).Borosilicate can filter out the light below 260 nm(as previously shown in Fig.3),and thus only the 312 nm light transmitted,while its emission intensity(2.7×104)at 312 nm is about 25%lower than that of the quartz(3.6×104).Considering the aging problem of polymer tubes due to the 254 nm irradiation,borosilicate was adopted for subsequent experiments.Quartz was chosen as the control group to assess whether the weak emission at 254 nm would have adverse effects on the photochemical synthesis.

Fig.1.(a)Experimental setup for dip-coating of photoluminescent particles onto glass substrates;(b)schematic diagram of photochemical synthesis of VD3 with different light sources.

Fig.2.Emission spectra of different light sources:(I)8 W lamp;(II)8 W lamp with quartz vessel;(III)8 W lamp with borosilicate vessel;(IV)100 W lamp with borosilicate cold trap;(V) 100 W lamp with silicate cold trap.Mass fractions of photoluminescent particles in the coated layers of vessels are all 50 wt%.

By contrast,the UVB emission of a traditional 100 W mediumpressure lamp integrated with a silicate or borosilicate cold trap is not sufficient considering its electrical power.This is because the UVB emission wavelength of medium-pressure lamps is 303 nm,at which the transmittance of borosilicate is about 58% and that of silicate is only 13%.The intensities of UVB emission are only 2.6×104and 0.5×104,respectively.In this case,the power of cooling devices had to be increased due to the high heat release of medium-pressure lamps.

Fig.3.Transmittances of quartz,borosilicate,and silicate filters.

Experiments were also conducted to determine the optimal dipcoating conditions and the corresponding emission intensities(the average value of 10 different locations for each sample)were measured,as shown in Fig.4.With the increase of the particle mass fraction,the emission at 312 nm was enhanced (Fig.4(a),while the emission at 254 nm was prevented since borosilicate filtered out the light below 260 nm.After the fraction was increased to higher than 50 wt%,the intensity became stable.In this case,the sites on the substrate are saturated,and it was difficult to adhere more particles for further enhancement.Accordingly,the mass fraction of the suspension was chosen to be 50 wt%,at which the number of dip-coating operation could be minimized.The effect of pulling rate during dip-coating was investigated,as shown in Fig.4(b).The emission intensity decreased with the increase of the pulling rate until it reached 3 cm·min?1.This is because enough time was required for cross-linking and for adhering enough particles onto the substrate surface.As such,a pulling rate of 3 cm·min?1was adopted.The changes of emission intensity with immersion time were shown in Fig.4(c),exhibiting that the intensity first increased and then remained stable.After immersion for more than 3 min,the substrate surface could be completely covered and particles could no longer be adhered.In addition,the effect of number of dipcoating operation was investigated,as shown in Fig.4(d).After five times of dip-coating,the intensity tended to be stable,since the coating agent was not enough to provide sufficient adhesion for more layers of particles.

Morphologies of the photoluminescent substrates were observed by SEM,as shown in Fig.5.With multiple times of dip-coating,the number of particles covered on the substrate surface simultaneously increased.The blank areas were gradually covered with particles during the multiple operations.A uniform dispersion of particles was also achieved,as shown in Fig.5(c) and (d).After 5 times,the coverage was visibly 100%.It suggests that the coating agent was not enough to accommodate more layers of particles,which was consistent with the spectra results.

3.2.Practical use of photoluminescent substrates for photochemical synthesis of VD3

By virtue of the prepared photoluminescent substrates,their practicality in photochemical synthesis of VD3was further assessed.In the first step,a large amount of tachysterol(T3)was obtained by ring opening of 7-DHC under 254-nm irradiation.In the second step,most of T3was converted into previtamin(P3)during ring closure under UVB irradiation.As a matter of fact,P3can be readily transformed into VD3(as explained in Section S3 in the Supplementary Material),and the yields of P3and VD3were taken as a total yield(YP)in the following experiments.In addition,recycle operation was adopted at higher flow velocities,to ensure that the efficiencies were compared at the same reaction time.

Fig.4.Changes of average emission intensity at 312 nm under different dip-coating conditions:(a)mass fraction of photoluminescent particles in coating suspensions;(b)pulling rate of substrates;(c)immersion time before pulling-up;(d)number of dip-coating operation.

Fig.5.SEM images of photoluminescent substrates with different numbers of dip-coating operation:(a)1;(b)3;(c)5;(d)7.

The ring-opening step was conducted at 254 nm,as shown in Fig.6a.Since the energy of short waves is relatively high,the 8 W low-pressure mercury lamp was used as the light source.The highest YPwas achieved when the conversion of 7-DHC was 79%.With further increased residence time,the yield of side-products(YSP)began to increase and YPdecreased gradually,which might be caused by over-irradiation.Accordingly,the first step was stopped at the conversion of 79%(with a resistance time of approximately 80 s),providing reactants for the ring-closure step.

Fig.6.(a)Changes of conversion and yields of ring-opening reaction;(b)effect of flow velocity on yields of products and side-products.Reaction conditions:gas–liquid phase ratio of 1∶1,temperature of 20°C.

To ensure that the turbulence in the liquid segments was intensive enough for minimizing over-irradiation,the effect of flow velocity was assessed,as shown in Fig.6b.At the same reaction time of 80 s,the conversion increased with the increase of flow velocity,and YSPdecreased simultaneously.This is because with the increase of flow velocity,the turbulence intensity was enhanced owing to secondary flow in the liquid segments.Molecules near the tube wall could be dynamically transferred and over-irradiation was effectively avoided.When the flow velocity was higher than 0.1 m·s?1,the inhibition of over-irradiation reached the maximum level.

To promote the transformation from T3into P3and VD3,ring closure requiring UVB irradiation was conducted with different light sources.Four types of light sources were employed,including the traditional 100 W medium-pressure mercury lamp with silicate and borosilicate cold traps,the 8 W low-pressure mercury lamp with particle-coated borosilicate and quartz vessels.The yields(i.e.,YPand YSP)were experimentally investigated and compared to assess the practical use of the different irradiation strategies,as shown in Fig.7.

Compared with the 100 W lamp with borosilicate cold trap(Fig.7b),the maximum YSof the 8 W lamp with particle-coated borosilicate vessel(Fig.7c)was also around 39%,as well as YST(approximately 1.1%).It is also worth noting that although the light of 260–280 nm can penetrate through borosilicate,less side-products were produced compared with the 100 W lamp with the silicate(filtering out light below 280 nm)cold trap(Fig.7a),which should be attributed to the faster reaction rate owing to the stronger UVB irradiation.The traditional strategy of the 100 W lamp with silicate glass seriously limited UVB irradiation,which took three times of reaction period to obtain the same yield as the 8 W lamp.In addition,the reaction was even faster for the 8 W lamp with the particle-coated quartz vessel(Fig.7d),indicating that accelerating reaction was more important than filtering out all the light with short wavelengths.

3.3.Modification of a microreactor with novel light source

Although the particle-coated quartz vessel could not completely filter out all the short waves,the reaction efficiency was relatively high.However,the FEP tubes were aged due to high the energy of short waves during long-term operation,as shown in Fig.8.After a total irradiation time of 300 h,the FEP tube visibly turned yellow for the 8 W lamp with the particle-coated quartz vessel (Fig.8b).As a matter of fact,polymers such as FEP tend to degrade under UV irradiation.This is mainly due to the photoaging of polymer matrix,since some organic groups absorb UV radiation of specific wavelengths (especially the short-wave UV radiation).In this case,the bonds are cut off and new groups that cause polymer to turn yellow are formed.Although the degree of damage was not as high as that of the 100 W lamp with the borosilicate cold trap(Fig.8c),it still significantly reduced the effective utilization of photons,which was unfavorable for photochemical reaction.In addition,no changes in the appearance and the emission efficiency of the photoluminescent substrate were observed after it was used for 300 h.

Fig.7.Changes of yields in the ring-closure step for photochemical synthesis of VD3 with different light sources:(a)100 W lamp with silicate cold trap;(b)100 W lamp with borosilicate cold trap;(c)8 W lamp with particle-coated borosilicate vessel;(d)8 W lamp with particle-coated quartz vessel.Reaction conditions:gas–liquid phase ratio of 1:1,temperature of 20°C.

Fig.8.Photographs of FEP tubes during the photochemical process:(a)before irradiation;(b)irradiation of 300 h by the 8 W low-pressure mercury lamp with particle-coated quartz vessel;(c)irradiation of 300 h by the 100 W medium-pressure mercury lamp with borosilicate cold trap.

In this regard,a modified microreactor with quartz tubes(0.6 mm i.d.)was designed and manufactured,as shown in Fig.9a.Combined with the 8 W lamp and particle-coated quartz vessel,the microreactor could not only provide higher transmittance,but also withstand short waves and long-term irradiation.The reaction results with the novel reactor are shown in Fig.9b.At the same product yield of 39%,the required reaction time was only 120 s.Side-products were not massively generated even in the case of incompletely filtering out short waves.The high transmittance of quartz can make more effective use of photons to accelerate reaction before numerous side-products are produced.In other words,changing the reactor material from FEP to quartz not only solves the aging problem of reactor tubes,but also achieves the same yield in a shorter time.The strategy of combining quartz tubes and particlecoated vessels takes full advantage of the penetrated radiation.

The productivities and energy consumptions of different light sources have also been compared according to their productivities in the photochemical synthesis of VD3,as shown in Table 1.The productivity of a single set of microreactor(a lamp with a tube of 12 m)is 1.16–1.83 kg·a?1.Owing to the particle-coated substrates,the productivities of the strategies with 8 W lamps were even higher than those of the 100 W mercury lamps.In particular,with the 8 W lamp equipped with quartz tubes and particle-coated quartz vessel,it produces the most VD3up to 1.83 kg·a?1,i.e.,122 million tablets(15 μg per tablet).The energy consumption is reduced by 86%–95%,and the consumption by heat exchange is also avoided,which is quite high with traditional mediumpressure lamps.Furthermore,the coating method is similar to that of fluorescent UV lamps,which could run continuously for 1200–1600 htheoretically.It should also be noted that if a medium-pressure mercury lamp was used instead of the photoluminescent substrate and the lowpressure lamp,FEP tubes could not be used for more than 200–300 h.The use of photoluminescent substrates and low-pressure lamps would cause less damage to the reaction equipment.Therefore,it can be concluded that the enhance of UVB irradiation based on the prepared photoluminescent substrates is feasible,showing great potential to replace the traditional medium-pressure mercury lamps for photochemical production of VD3.

Fig.9.(a)Photograph of modified microreactor with quartz tubes;(b)Changes of yields in the ring-closure step for photochemical synthesis of VD3 with 8 W low-pressure mercury lamp,particle-coated quartz vessel,and quartz tubes.Reaction conditions:gas–liquid phase ratio of 1:1,temperature of 20°C.

Table 1 Comparison of photochemical microfluidic production of VD3 with different emission strategies①

4.Conclusions

In this study,the photoluminescent particle-coated substrates with high emission efficiency were successfully prepared by dip-coating.The UVB emission intensity of the 8 W lamp was increased by approximately 11 times with this novel substrate,and the undesired UVC radiation was simultaneously decreased.By virtue of the substrates,modified light sources were successfully used in the photochemical microfluidic synthesis of VD3under UVB irradiation.The corresponding productivity of a 12 m channel with a diameter of 0.6 mm was increased to 1.83 kg·a?1,which was 42%higher than that of a 100 W lamp and no cooling devices were used.The experimental results also confirm that the enhanced UVB irradiation plays a key role in minimizing side reactions.This method exhibits considerable potential to replace traditional medium-pressure mercury lamps for photochemical production of VD3and other steroids.

Future studies should focus on enhancing the feasibility of particlecoating onto other substrates,such as polymer tubes or microchips.Moreover,integrating various types of photoluminescent particles into a single light source to conduct multistep reactions is also worth investigating.

Acknowledgements

We gratefully acknowledge the support of the National Natural Science Foundation of China (21978008,21606008),the State Key Laboratory of Chemical Engineering(SKL-ChE-17A02),the Fundamental Research Funds for the Central Universities(JD2017).

Supplementary Material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.cjche.2020.07.016.