Team:FAU Erlangen/Results

iGEM Erlangen


Growing Biofilms

Reliably growing biofilms covering the whole surface of the conducting glass plates we were using for our solar cells was a major challenge during the course of our project. Biofilms grow best in semi-dry environments and while protocols for growing biofilms are abundant in literature, most of them are either not meant for covering greater areas than a well in a 96 well plate or they require rather complicated setups. Both these kinds of protocols were not attractive to us, as we wanted a quick and easy method to cover the entire substrate area with biofilm. Initially, we attempted to grow the films in petri dishes on a rotary shaker. We filled the dishes with medium just enough to barely cover the glass plates and hoped that the circular motion would cause the medium to slosh onto and off the glass, thus create alternating wet and dry conditions, thus enabling the film to grow. Unfortunately, this approach was hindered by the water evaporating overnight and accumulating on the underside of the lid, which resulted in a permanently dry glass surface – not an ideal environment for bacteria to grow.

In the end, we switched to a rocking seesaw shaker. This way, the medium containing the bacteria was sloshed from one side of the dish to the other, on the way flowing across the glass surfaces. This yielded adequate biofilms for our purposes.

Figure 1: Crystal Violet stained biofilm on FTO glass.

Binding of inorganic material

The second part of our project is binding of inorganic nanostructures, such as zinc oxide or cadmium sulfide to the biofilm. We used a modified CsgA, which is a subunit of Curli fibers. Curli are amyloid fibers that Escherichia coli and some other bacteria produce to enhance the structure and stability of biofilms. These proteins were ideal for our purposes, as they are secreted by the cells and require no further purification before use.

To bind inorganic material, we inspected literature for peptides which are known to adhere to zinc oxide and metal sulfides. We chose a zinc oxide binding peptide (part:BBa_K2169137) (related to Tomizaki et al.) and a zinc sulfide binding peptide (part:BBa_K2169138) (related to Mao et al.) that had been shown to bind cadmium sulfide as well. Those peptides were fused C-terminally to CsgA (part:BBa_K2169000) and expressed in E. coli. The resulting biofilms were incubated with the inorganic compound and either sintered (heat inactivation) or UV-sterilized for further characterization.


Binding of Heavy Metals

Due to the metal binding domains on the curli fibers, the biofilms are able to bind not only zinc oxide, but also heavy metals. This might present a safety application since biofilms could scavenge toxic heavy metal ions out of wastewater.

Copper and nickel were tested first. The corresponding metal salt was dissolved in 50 ml of distilled water, yielding a concentration of 0.05 mol/l. Absorbance spectra at t = 0 were recorded. Afterwards, a biofilm grown on a glass substrate was put into a beaker and submerged in the solution so that the biofilm is just covered. The solution was stirred gently. Absorption spectra were recorded after 5, 10, 20 and 30 minutes. During this time, the metal ions were supposed to bind to the curli, leading to a decrease in metal concentration and therefore a decrease in absorbance of the solution.

Figure 1: Absorption spectra of copper (left) and nickel (right) solutions in dependence of time after exposure to biofilm glass slides.

Unfortunately, the curli could only bind very few metal ions, leading to ambiguous results. The change in concentration was too small for the UV/Vis analyzer to detect reliably.
We then tested a different approach utilizing gold ions. In the same setup, the biofilms were covered in a solution of chloroauric acid. Gold ions were reduced with ascorbic acid, leading to formation of gold nanoparticles in solution. When the glass substrates were removed the beaker, a golden-red color was observed, leading to the conclusion that gold was bound to the cysteine subunits of the curli. Unfortunately, optical microscopy did not show whether nanoparticles were bound, so this outcome still needs to be verified using electron microscopy.

Better results have been obtained upon binding cadmium salts. For this setup, a solution of Cd(NO3)2 with c = 50 mmol has been spread on a biofilm grown on an FTO substrate. After a few minutes of incubation, nanoparticles have been formed by addition of a Na2S solution with c = 150 mmol. By doing this, the difficultly soluble compound CdS precipitates in the form of nanoparticles, which can also be observed by an immediate color change from transparent to bright yellow-orange. After the reaction, all residual liquid was removed. The remaining color on the biofilm seen in Figure 1 implies that cadmium was bound by the biofilm.

Figure 2: CdS nanoparticles on a biofilm grown on an FTO substrate. The bright yellow color indicates that CdS is present.


ZnO Mineralization



Figure 1: TEM images of ZnO microflowers related to syntheses by Zhang et al. (2004).


From the TEM images of the ZnO synthesis it could be concluded that no autoclave is needed to produce the microflowers, which was used by Zhang et al. (2004). The purpose using highly branched systems, is an increase of adsorbing molecules on the surface. Further, the branched ZnO clusters can be anchored more simply in a biofilm, than flat or spherical nanoplatelets. The size of the microflowers was a general problem for our proof of concept. As the particles could cover most of the distance from one electrode to the other. This implies that biofilm can be used as template to stabilize the structures, however it is no crucial factor. To circumvent this issue, we changed the procedure to a direct mineralization process, where the biofilm is directly involved in the synthesis on ZnO.



Figure 2: ZnO nanoparticles grown in presence (left) and absence of curli fibers (right) on a p-doped silica wafer

The REM images (Figure 1) visualize the difference between ZnO growth in presence and absence of curli fibers on the glass surface. In presence of curli, the particles tend to grow isotropically, with fibers coated around an assembly of individual particles. The ZnO nanoclusters provide average sizes of 0.9 µm with a standard deviation of 0.2 µm, considering 50 clusters. In absence of curli fibers, needle like structures form, which are highly uniform in shape with an average length of 1.1 µm and an average width of 0.55 µm. The size and shape is likely controlled by other factors such as sugars, less complex protein structures in the biofilm matrix or salts in the medium.



Figure 3: Figure 3: ZnO needles grown in the absence of curli fibers in an autoclave at 121°C for 20 min on a p-doped silica wafer.

This result is confirmed by the autoclave mediated synthesis (Figure 2), where similar needle like structures with equal size appear. Accordingly, the capping agents of these structures have to be included in the growth medium or biofilm.



Figure 4: REM images of mineralized biofilms grown on a FTO glass slide with 500 x magnification (left) and 2000 x magnification (right)

The biofilms grown on a FTO glass slide provide more homogeneous layers in contrast to the biofilms grown on a silica wafer (Figure 3). On the bacteria a network like structure can be found, which is most likely mineralized as the layers were sintered before measuring, to kill the bacteria. Beside the larger network like structures smaller clusters can be identified at higher magnifications. The amount of ZnO for this preparation is low compared to the high amount of organic material and hence we started to increase the ZnO ratio.



Future Prospect

For further enhancement of the solar cells the mineralization process can be improved to obtain higher surface area covered with ZnO. One part of this work must be a reliable prediction of the biofilm quality as every biofilm can differ slightly in its composition and the amount of ZnO binding domains. Also the structure and size of ZnO can be adjusted more accurately. Here a specific surface pattern on bacteria which is called S-layer. The potential surface roughness provides high loading capacities for dyes or other photosensitizers, which enhances the efficiency of the biofilm solar cell drastically.


Solar cell results

During our project, we have shown that fluorescent proteins like GFP can work as photosensitizers on solar cells. However, the quantum yields of those fluorescent-protein-sensitized solar cells were not higher compared to those of normal DSSCs, where a dye in hibiscus tea is used to absorb light. One possible reason is that the absorption maximum of this tea is at about 517 nm. This wavelength belongs to the part of the spectrum most prominent in sunlight. That means that the spectrum of sunlight and the absorption spectrum of hibiscus tea have a high overlap. GFP absorbs at 400 nm, where the intensity of sunlight is not as high anymore, leading to lower quantum yields.

However, GFP is a fluorescent protein, which means that it not only absorbs, but also emits light. The emission maximum of GFP is at 510 nm and therefore very close to the absorption maximum of the tea. We hypothesized that combining both fluorescent protein and tea in one solar cell would allow the emission of the protein to excite the dye again and thus lead to higher quantum yields than either would achieve individually.

By examining this issue further, we have observed that when the tea dye binds zinc oxide on the solar cell, its electronic structure changes slightly and its absorption spectrum shifts to 597 nm, which is yellow light. Since GFP emits green light, the overlap is not as high as expected. The solution to this might be the use of the red fluorescent protein, mCherry, in combination with tea to achieve the highest quantum yields.

Results

For the first solar cell tests, the influence of biofilms was tested for both TiO2 and ZnO. The results are shown below:

Figure 1: Comparison of different solar cells with and without biofilm. Photosensitizer = hibiscus tea. All cells have been sintered after the addition of TiO2/ZnO nanoparticles.

As can be seen, biofilms greatly increased both current and voltage. In the case of TiO2, the biofilms were three times as effective as the TiO2 layer without biofilm. For ZnO, an improvement of 70% was reached. The reason is that the application of biofilms as bottom layer can improve the adhesion of the oxide layers to the surface. Additionally, the restructuring of the surface by the biofilm layer increases the overall area of the solar cell.

It should be noted that the current for solar cells without biofilm could be increased up to the level of those with biofilm by adding an excess of TiO2. In other words, the biofilm solar cells require less metal oxide semiconductor to provide equal or higher currents. The standard cells (see fabrication procedure in the lab journal) using ZnO and TiO2 provide extremely low currents, which is likely caused by the irregular morphology of their surface after the sintering process. Frequently, the semiconductor layers did not attach properly to the FTO glass, thus reducing the potential currents significantly.

Instead of first growing a biofilm and later adding pre-synthsized ZnO nanoparticles to this system, we chose the different approach by adding Zn(OH)2 as a salt to initiate the biomineralization process. Different concentrations of Zn(OH)2 were added and current was measured with and without sintering the ZnO layer. The results are shown below.

Figure 2: Comparison of biofilm-ZnO solar cells using different concentrations of Zn(OH)2. Photosensitizer = hibiscus tea. ZnO layer has been formed by biomineralization.

Without sintering, adequate current values close to those of solar cells without a biofilm were recorded. However, sintering the ZnO layer led to a higher current level. For a Zn(OH)2 concentration of 100 mM, a current of 1498 µA was achieved, which is remarkable for any kind of DSSC. Compared to the unsintered cell, this corresponds to an 83-fold improvement.

Solar cells without a biofilm but with fluorescent proteins as photosensitizers have been tested as well. The results are shown below.

Figure 3: Comparison of solar cells without biofilms using hibiscus tea and GFP as photosensitizers. All metal oxide layers have been sintered after addition of TiO2/ZnO nanoparticles.

TiO2 cells using GFP instead of tea performed quite well. The average current and voltage of those cells has been even higher than the average of cells using tea. However, the improvement lies within the standard deviation of TiO2 + tea cells, meaning that it cannot be generalized that GFP is a better photosensitizer than hibiscus tea. When looking at the same cells using a ZnO layer, GFP cells performed very poorly. Here it becomes noticeable that hibiscus tea is simply a better absorber for sunlight than GFP. In sunlight, wavelengths between λ = 500 – 510 nm possess the highest intensity. While the absorbance maximum of hibiscus tea of λabs = 517 nm matches that quite closely, the absorbance maximum of GFP with λabs = 475 nm is much more off.

To further improve this system, the idea of tandem solar cells using both fluorescent proteins and hibiscus tea emerged. When the emission spectrum of the protein and the absorbance spectrum of the dye have a sufficient overlap, emitted photons from the proteins can excite dye molecules a second time and therefore increase the quantum yield, leading to higher currents. Some tests have been carried out using tea and GFP or tea and mCherry. While results with GFP were rather poor, cells with mCherry and tea provided higher currents than cells using only tea as photosensitizer. However, differences were in the range of standard deviations and therefore not significant enough to lead to any general conclusions. Another drawback is the morphology of the TiO2, since without a stabilizing biofilm underneath the TiO2 layer did not attach to the glass substrate very well. Both those problems still need to be tackled, making the idea of tandem solar cells just a concept for now.

Future Prospect

As an outlook, tandem solar cells using fluorescent protein and hibiscus tea could prove to be very efficient. Taking into consideration that the adhesion of metal oxide layers to the glass substrates have been much better when biofilms are grown underneath, solar cells with biofilm, tea and fluorescent protein should be tested as well.

Another very interesting approach would be to combine our project with the iGEM project from TU Darmstadt 2014. They focused on a biological synthesis for organic dye molecules by E. coli, namely anthocyanins, to use them in Grätzel cells. By combining their dye synthesis with our biomineralization, completely biological solar cells could emerge.


References

  • Zhang, Hui; Yang, Deren; Ji, Yujie; Ma, Xiangyang; Xu, Jin; Que, Duanlin (2004): Low Temperature Synthesis of Flowerlike ZnO Nanostructures by Cetyltrimethylammonium Bromide-Assisted Hydrothermal Process. In J. Phys. Chem. B 108 (13), pp. 3955–3958. DOI: 10.1021/jp036826f.


  • Tomizaki, K. Y., Kubo, S., Ahn, S. A., Satake, M., & Imai, T. (2012). Biomimetic alignment of zinc oxide nanoparticles along a peptide nanofiber. Langmuir, 28(37), 13459-13466.


  • Mao, C., Flynn, C. E., Hayhurst, A., Sweeney, R., Qi, J., Georgiou, G., ... & Belcher, A. M. (2003). Viral assembly of oriented quantum dot nanowires. Proceedings of the National Academy of Sciences, 100(12), 6946-6951.