Spin-Controlled Photoluminescence in Hybrid Nanoparticles Purple Membrane System.

Spin-dependent photoluminescence (PL) quenching of CdSe nanoparticles (NPs) has been explored in the hybrid system of CdSe NP purple membrane, wild-type bacteriorhodopsin (bR) thin film on a ferromagnetic (Ni-alloy) substrate. A significant change in the PL intensity from the CdSe NPs has been observed when spin-specific charge transfer occurs between the retinal and the magnetic substrate. This feature completely disappears in a bR apo membrane (wild-type bacteriorhodopsin in which the retinal protein covalent bond was cleaved), a bacteriorhodopsin mutant (D96N), and a bacteriorhodopsin bearing a locked retinal chromophore (isomerization of the crucial C13═C14 retinal double bond was prevented by inserting a ring spanning this bond). The extent of spin-dependent PL quenching of the CdSe NPs depends on the absorption of the retinal, embedded in wild-type bacteriorhodopsin. Our result suggests that spin-dependent charge transfer between the retinal and the substrate controls the PL intensity from the NPs.

In the last few decades, the combination of biosystems and nanostructures has been applied to produce new photonic and electronic devices.[1−3] Owing to the special properties of the biosystems, like structure recognition, self-assembly abilities, and complex responses to light, new hybrid biodevices may be used to express novel functions, those that are not available in the common solid-state photonic/electronic devices. Purple membrane, which includes bacteriorhodopsin (bR), is perhaps one of the most promising biosystems that has been explored for photo/electronic applications.[4−6] The present work focuses on the optical and spintronic properties of bR, an integral membrane protein in the purple membrane (PM) of Halobacterium salinarum, which is usually found as two-dimensional crystalline patches. The retinal, imbedded inside bR, absorbs light (at about 570 nm), which triggers a photocycle including several intermediates, which leads to pumping of protons across the membrane from the internal cytoplasm to the external medium. As a result, a proton gradient is generated that is used for ATP synthesis in the cell.[7,8] The proton transfer within PM is accomplished by charge separation followed by de- and reprotonation of the retinal.[9] Note that the retinal is located at the center of the PM, i.e., 2.5 nm from both sides of the PM surfaces.[1] Relevant to this work is the realization that this distance is smaller than the typical Förster radius (5 nm) in energy-transfer processes.[10]

Excitonic interactions in nanobiohybrid structures based on semiconductors, nanoparticles (NPs), and photochromic biomolecules including bR and green fluorescent protein (GFp) were observed before.[1,2] This phenomenon is associated either with fluorescence quenching, fluorescence enhancement, or nonradiative energy transfer from NPs (donor) to the retinal (acceptor) and is commonly related to the Förster resonance energy-transfer (FRET) process.[3] This process is sensitive to the distance between donor and acceptor and can be modulated by the size of the NPs,[11] irradiation intensity, structure, and morphology of the biosystem.[12] Besides these extrinsic parameters, properties inherent to the NP-Bio system such as dipole–dipole interactions[13] and chirality[14,15] play a significant role in the overall photoluminescent (PL) behavior of such a hybrid system. It is also known that pure sources of spin-polarized electrons contribute to the spin-dependent PL in semiconductor[16] and spin-dependent exciton formation in a π-conjugated system.

Recently, spin-dependent electron transport through wild-type (WT) bacteriorhodopsin (WT bR) was reported.[17] This phenomenon has been attributed to the chirality of the protein and is another manifestation of the chirality-induced spin selectivity effect.[18,19] In a sequential work it was also demonstrated that the degree of spin polarization can be controlled by light.[20]

In the present work, we investigated the PL from CdSe NPs that are adsorbed on a PM that contains bR.[1] The hybrid system, NPs, and the PM are deposited on ferromagnetic Ni-alloy (see the Supporting Information (SI)). We demonstrate that the PL intensity strongly depends on the direction of magnetization of the ferromagnetic substrate. By investigating the system with modified bR, we are able to conclude that the alternation in the PL intensity is controlled by spin-specific electron transfer from the substrate to the retinal, which consequently affects the efficiency of energy transfer between the NPs and the retinal.

Results and Discussion

Figure schematically presents the experimental system in which a PM that includes bR is adsorbed on a ferromagnetic substrate (Ni- alloy) and CdSe NPs are adsorbed to bR on top of the membrane (see SI). The CdSe NPs’ absorption peak is at 630 nm. Circular dichroism (CD) spectroscopy performed on the adsorbed membrane confirms that the helical structure of the proteins in bR is retained when the membrane is adsorbed on the Ni-alloy surface and that the CD spectra are identical to those observed in our former studies.[17] A well-defined absorption band centered at ∼570 nm confirms that retinal is covalently attached to the WT bR and its mutant (D96N) (see the SI, Figure S1). This observation is consistent with the proteins not being significantly altered. Emission spectra of the adsorbed bR were also measured and confirm the intact structure. In order to adsorb the NPs to bR, CdSe NPs were dissolved in toluene, and the samples containing bR were immersed in this solution for 3.5 h. Afterward, the samples were sonicated for 20 s to remove the excess and weakly bound NPs. To confirm that the structure was not disturbed, we compared the emission spectra of bR on Ni-alloy before and after it was suspended in toluene for 3.5 h (see SI Figure S5). In addition, the absorption band at 570 nm indicates that immersing the bR multilayer in toluene does not disturb the bR structure. The disappearance of the absorption band at 570 nm for apo membrane bR indicates that the retinal–protein covalent bond was cleaved, and therefore it can be used for the control experiment (see SI Figure S1). In order to confirm the surface coverage, atomic force microscopic measurements were carried on drop-casted WT-bR and APO-bR. The results show that the surface coverage is almost 100%, and the data confirm that these drop-casted bR on the surface always form multilayer (see SI, Figure S2). The orientation of bR membrane on the substrate was studied in our previous work, and it was found that the membrane is deposited such that the more negative side is facing the substrate.[20]

Figure 1

Schematic diagram of an experimental setup for measuring PL. A hybrid structure containing semiconductor nanoparticles (CdSe) of ∼6–7 nm diameter, which are adsorbed to the bacteriorhodopsin (either WT, APO, mutant, locked retinal, or reconstitute), imbedded in a PM which is deposited on a Ni-alloy surface. Magnetic field (field strength ∼0.5T) is applied normal to the surface of the sample while measuring the PL spectra.

To elucidate the effect of spin-dependent fluorescent, we recorded the PL spectra with the magnetic field applied normal to the Ni-alloy surface in two opposite directions. Figure shows the PL spectra from hybrid NPs-bR (WT) recorded with the magnetic field applied up (↑H) and down (↓H). A clear change was observed with an average peak-to-peak ratio of the PL spectra for two directions of the magnetic field of 1.4(±0.1):1 (down:up). The results represent the average over five different sets of samples, when the maximum ratio observed was 2.5:1 (inset in Figure ).

Figure 2

PL spectra of CdSe NPs on WT-bR with two different magnetic field directions (H ∼ 0.5T). Inset shows the maximum effect obtained for WT-bR.

In principle, the magnetic field-induced change in the PL intensity may result either from variation in the rates of the energy transfer or the charge transfer. In order to verify the contribution of the charge transfer, we performed contact potential difference (CPD) measurements that are sensitive to the work function (Φ) of the sample. When the sample is illuminated and charge transfer occurs, a surface photovoltage (SPV) is developed with a shift in the work function of the material. Details of the measurements and instrumentation are given in ref (21).

Figure presents the SPV signal observed for a PM containing WT bR, with and without NPs. The two columns on the left show the SPV signal (the change in the CPD signal upon illumination) when the sample was illuminated at 532 nm (green light). The amount of charge transferred as a result of the illumination is almost constant with and without CdSe NPs, i.e., the amount of charge transfer from NPs is negligible, and most of the charge transfer occurs between the WT-bR and the surface. The positive sign of the signal indicates hole transfer from the retinal to the substrate. In order to confirm this conclusion, SPV measurements were performed with illumination at 630 nm (red light). The purpose of using red light was to monitor the charge transfer with negligible retinal excitation (630 nm should mostly excite the CdSe NPs). Indeed, when the experiment is performed with red light, there is almost no response to the light (i.e., when the SPV signal is low, charge transfer between the NPs and the surface is negligible, as shown in Figure ). Therefore, this result suggests that the PL intensity from the NPs is mainly affected by the energy-transfer rate, and variation in the energy-transfer efficiency should account for the observed magnetic field. Since the energy-transfer rate itself should not be spin dependent and therefore cannot depend on the sign of the magnetic field, we assume that there is a cooperative effect leading to the magnetic field-dependent PL. Namely, there is a charge-transfer process between the retinal and the magnetic substrate, which is magnetic field dependent. The charge transfer affects the efficiency of energy transfer between the NPs and the retinal; hence it affects the PL. Therefore, the magnetization of the substrate “gates” the energy transfer from NPs to the retinal by controlling the charge transfer from the retinal to the substrate. The validity of the proposed model will further be strengthened by the results described below.

Figure 3

SPV (ΔCPD) of WT bR deposited on Ni-alloy surfaces with and without CdSe nanoparticles illuminated in two different wavelengths and the control for the bare Ni-alloy substrate.

To pinpoint the role of the retinal chromophore as the intermediate in the process, we have modified the WT bR structure by site-directed mutagenesis to form the D96N mutant. Details on the modifications were published elsewhere.[15] Experiments similar to those that led to the results in Figure were repeated for a CdSe NPs-bR (D96N) hybrid structure and no effect of the magnetic field could be observed (Figure A). Moreover the PL signal is stronger than in the case of WT bR, indicating that there is no efficient energy transfer.

Figure 4

PL from CdSe NPs for substrates that are magnetized normal or antinormal to the surface for various mutations of the retinal. (A) Mutant (D96N), (B) APO bR, (C) locked retinal, (D) reconstituted bR.

The lack of a magnetic field effect, in the case of the mutant, is accompanied by an inefficient charge-transfer process, as observed in the SPV experiment (Figure ). The SPV signal is about −50 mV vs about −30 mV obtained with the bare substrate. Here we observed negative values which indicate electron transfer from the NPs and not positive values (hole transfer), as was observed in the WT bR. Since in the mutant the M photochemically induced intermediate is accumulated, the retinal in the bR mutant dark state cannot be excited, and the lack of a magnetic field effect suggests that retinal excitation is essential for quenching the fluorescence of the NPs.

Figure 5

SPV (ΔCPD) of mutant and APO bR deposited on Ni-alloy surfaces illuminated at 533 nm.

As another control, magnetic field-dependent studies were performed on apo bR, which lacks the retinal–protein covalent bond, leading to absorption at 360 nm instead of 570 nm in the WT bR. Hence, the retinal should not be able to accept any energy transferred from the NPs. Indeed, Figure B shows no change in PL intensity upon changing the direction of the magnetic field. This observation proves that the excited retinal plays an important role in the observed effect. In this context, SPV data for an apo bR-modified surface with and without CdSe NPs (Figures ) show an inefficient charge transfer both between the CdSe NPs and the surface and between the retinal and the surface (the low response that was observed may come from a small portion of bR in which the retinal is still bonded to the protein).

For additional verification of the role of the retinal in the magnetic field-controlled PL, the external magnetic field-dependent PL process was studied with two other modified bR forms. The first is an artificial pigment derived from a synthetic “locked retinal” in which the retinal has been modified such that isomerization around the crucial C13=C14 double bond is prevented (see Methods). The second is a reconstituted bR to evaluate the effect of mere bleaching and retinal reconstitution processes. In the first case, no change in the PL intensity was observed, as a function of the direction of the magnetization of the substrate (Figure C), whereas for the reconstituted bR, a clear difference in PL intensity was found (Figure D). However, the intensity ratio (up:down = 1.2:1) is smaller than for WT bR, and the sign of the magnetic field effect is reversed.

The existence of the effect in the reconstituted bR indicates that the absence of the effect in the locked artificial pigment is a property of the pigment itself and does not originate from the reconstitution process. The switch in the sign of the signal may result from some structural changes in the reconstituted bR versus the WT. In recent studies it was found that structural changes may affect the sign of the spin preferred in the electron transfer.[22] Finally, the effect of the direction of the magnetic field on the PL was monitored on a sample containing only WT bR without NPs (see SI, Figure S3). No effect of the magnetic field was found.

To probe the importance of simultaneously exciting the NPs and the retinal, we conducted the experiments on WT bR with NPs using 488 nm light. At this wavelength, the absorption of the NPs is higher than at 514 nm; however, the retinal absorption is much weaker. A significant change in the PL intensity was found (see SI, Figure S4); however, the intensity ratio (down:up = 1.15:1) is smaller than in the case of illumination at 514 nm. This observation can be explained only if the effect involves a cooperative mechanism and it cannot be explained by simple spin-dependent charge transfer between the NPs and the surface.

Based on extensive experimental work with several modified forms of bR, we propose the following mechanism (see Scheme ) for the observed magnetic field-controlled PL. It assumes that after irradiation at 514 nm, both CdSe NPs and the retinal are excited, and subsequently energy is transferred from the excited CdSe NPs to the excited retinal. Furthermore, it is assumed that energy transferred does not occur from the excited CdSe NPs to the retinal ground state. If the excitation of the retinal is followed by the photocycle, no energy transfer can occur from the NP to the retinal all through the photocycle time. A competitive process involves electron transfer from the magnetic substrate to the excited retinal. This process prevents the retinal excited state to isomerize and to initiate the regular bR photocycle. In our previous work, we found that electron transmission through WT bR is spin specific.[20]

Scheme 1

Process Leading to Magnetically Controlled PL from the CdSe NPs

(A) The system consists of a ferromagnetic substrate (gray on the right) on top of which a PM that includes bacteriorhodopsin (bR) is deposited. CdSe NPs are adsorbed to the bR. Commonly upon excitation of bR with green light, a photocycle is initiated by isomerization of the retinal chromophore. (B) In the system studied, the same light also excites the NPs. (C) If the ferromagnetic substrate is magnetized so that spins can be injected from it into the chiral bR, following photoexcitation of the bR, an electron will be transferred to the hole in the excited bR, the retinal will be quenched to its ground state so that it can absorb a photon again and form the excited state to which energy transfer can occur, and the PL signal will be reduced. (D) In the case that the magnetic substrate is magnetized so that the spin transfer to the excited retinal is slower, the retinal excited state will complete its lifetime, and the regular photocycle will take place, reducing the probability for multiple retinal excitation and therefore reducing the probability for energy transfer from the NPs to the retinal.

Since the electron transfer between the substrate and the retinal is spin selective, the direction of magnetization of the substrate defines its rate. If the substrate is magnetized so that it has a substantial density of populated states of the correct spin, namely, the spin that can be transferred from the substrate to the excited retinal through the chiral protein, then the electron transfer from the surface to the retinal is efficient, and the retinal is quenched to its ground state so that it can absorb a photon again and form the excited state to which energy transfer can occur. On the other hand, if the substrate is magnetized in the opposite direction, the electron transfer from the substrate to the retinal is blocked, due to the very low density of the populated states having the correct spin, and the retinal excited state continues to its regular photocycle. Consequently the time it takes for the retinal to return to its ground state is longer, reducing the probability for multiple retinal excitation and therefore reducing the probability for energy transfer from the NPs to the retinal.

In short, the mechanism we propose assumes that for observing efficient energy transfer from the NP to the excited retinal, the excited retinal has to be quenched fast so as to be ready for reabsorption of photon. The return of the retinal to its ground state can be hindered by the photocycle process or by the slow electron transfer from the substrate, when it is magnetized in the “wrong” direction.

The proposed model is based on two photons that must be absorbed by the system simultaneously. Figure indeed indicates that this effect depends on the intensity square.

Figure 6

Difference in the fluorescence intensity for a magnetic field of the substrate pointing up versus down, as a function of the laser intensity. The results are fitted to the intensity square with R = 0.95.

The strong effect of the magnetic field observed can be rationalized by the short lifetime of the excited retinal (about 0.5 ps) versus that of the excited NPs (ns). Hence, the retinal can be excited many times within the lifetime of the NP, and therefore the probability for energy transfer from the NPs to the retinal is enhanced by up to a factor of 2000.

The model presented above is confirmed by studying the systems that contain bR with various modifications. In the D96N mutant, the M intermediate is accumulated, following photoexcitation, and therefore there is no retinal in the excited state that can accept energy from the NPs, and the retinal does not return to its ground state. The same holds for the apo system. The lack of an effect in the “locked” system also indicates that the effect is associated with the lifetime of the retinal excited state. Since retinal isomerization is prevented in the locked pigment, its excited lifetime is prolonged from 0.5 ps (WT) to 20 ps,[23,24] and therefore the probability of the retinal to return to its ground state increases, and it can be excited multiple times so that efficient energy transfer from the NPs to the retinal can occur at both directions of the magnetic field.

Conclusion

Our results indicate a system in which spintronic properties control PL. In the hybrid NP-bR structure, the chirality of the protein induces spin-dependent electron transfer from a magnetic substrate to the retinal moiety, inside the bR. This electron transfer depends on the direction of the substrate’s magnetization. The electron-transfer process “gates” the energy transfer from the excited NPs to the excited retinal.

Methods

Figure presents the structure of the retinal and of the “locked retinal” used in the current study.

Figure 7

Structure of the retinal (on the left) and the “locked retinal” (on the right). In the latter no isomerization can occur upon photoexcitation.

Absorption Spectra

The absorbance spectra were measured on a HP 8453 UV–vis spectrometer, and the CD spectra of bR films were measured on a Chirascan spectrometer, Applied Photo Physics, England.

Surface Photovoltage (SPV) Measurements

SPV measurements were used to quantify the amount of photo-induced charge transfer in these assemblies. CPD measurements provide the work function (Φ) of the sample by measuring the electric potential between the sample and a reference plate. When the sample is illuminated and charge transfer occurs, a SPV is created; i.e., a work function shift occurs. Based on the SPV measurements, it is possible to determine the direction and the extent of the electron transfer (for a given illumination intensity) between the NPs and the substrate. If an electron is transferred from the NP layer to the substrate, the surface’s work function decreases, whereas the work function increases if electrons are transferred in the opposite direction. Thus, the sign of the SPV signal provides the direction of electron transfer, and its magnitude is proportional to the change in the work function resulting from the light-induced dipole moment that arises from charge transfer between the NPs and the substrate.

Photoluminescence (PL) Measurement

The PL measurements were carried out by using a LabRam HR800-PL spectrofluorimeter microscope (Horiba Jobin-Yivon). Typically, 514 nm laser light (argon-ion CW laser at ∼15 mW/cm2) has been used for excitation of CdSe NPs. The incident light was impinged on the surface at an angle 90° to the sample surface. Prior to collecting the PL spectra, an area (typically, 20 au × 20 au) and number points (25 points) within this area were defined in order to map the surface. Afterward, the PL spectra were collected using a microscope (with a 5× high-working distance lens). During the measurement, a confocal aperture (1100 μm) was fully opened, and the integration time was maintained at 15 s. Finally, the spectra were presented after averaging out the PL of individual points within the defined area at two different magnetic orientations. The external magnetic field dependence on the PL of the CdSe NP assembly, connected to the magnetic substrate via bR, was investigated at room temperature.