A molecular spin-photovoltaic device

See allHide authors and affiliations

Science  18 Aug 2017:
Vol. 357, Issue 6352, pp. 677-680
DOI: 10.1126/science.aan5348

A spin-valve solar cell

Electronic spin currents can be measured with a spin valve—a device that injects charge carriers from one ferromagnetic electrode to another through a semiconductor layer. Some organic semiconductors can have long spin-carrier lifetimes and can also generate charge carriers through the photovoltaic effect. Sun et al. fabricated a spin valve based on C60 and showed that the spin current could be modulated by the photocurrent. At certain light intensities, the sign of the photocurrent could be changed using an applied magnetic field, an effect that could potentially be harnessed for sensing applications.

Science, this issue p. 677


We fabricated a C60 fullerene–based molecular spin-photovoltaic device that integrates a photovoltaic response with the spin transport across the molecular layer. The photovoltaic response can be modified under the application of a small magnetic field, with a magnetophotovoltage of up to 5% at room temperature. Device functionalities include a magnetic current inverter and the presence of diverging magnetocurrent at certain illumination levels that could be useful for sensing. Completely spin-polarized currents can be created by balancing the external partially spin-polarized injection with the photogenerated carriers.

Molecular materials such as C60 and Alq3 [tris(8-hydroxyquinolinato)aluminum] can preserve the spin polarization of electrical carriers on time scales of milliseconds (1, 2). This property, along with their ability to alter both the energy and spin polarization of metallic ferromagnetic (FM) surface states (16), makes them attractive for use in spin-transport devices. Spin populations can also play a crucial role in electron-hole pair recombination, which could be exploited in optoelectronic devices such as photovoltaic cells and light-emitting diodes (711). For example, a simple dependence of the photocurrent on a large external magnetic field of several tesla was observed in molecular photovoltaic devices with materials such as donor-acceptor blends, among others (810). However, a full interplay between light and spin-polarized currents in molecular devices has been lacking.

We present the design and characterization of a molecular spin-photovoltaic (MSP) device that has several distinctive characteristics relative to previously reported inorganic spin-photovoltaic devices (1214). It shows a switchable resistance and photovoltaic effect under low magnetic fields at room temperature. It has complex functionalities; for example, we demonstrate a magnetic current inverter. Lastly, the device can generate a completely spin-polarized current under specific light irradiation conditions.

The MSP device has a spin-valve geometry, which is composed of two metallic FM layers (Co and Ni80Fe20, respectively) sandwiching a C60 molecular film, a well-tested material for both photovoltaic (1517) and spintronic applications (1821) [Fig. 1A and fig. S1 (22)]. We obtained reproducible results for more than 10 samples, in part by using a leaky AlOx barrier (semi-oxidized Al) and a low-temperature molecular growth process (23, 24) [a rigid-band energy map of the MSP device is shown in fig. S1; device fabrication details are included in the supplementary materials (22)]. In our device, as in conventional spin valves, one of the FM materials (here, Co) injects spin-polarized carriers into the semiconductor layer, and the other FM layer is the spin detector. If the spin polarization of the electrical carriers is preserved across the C60 layer, the electrical current flow changes depending on the relative orientation of the magnetization of the FM layers (2026). This change in electric current under the application of a magnetic field can be denoted as magnetocurrent (MC) and defined as MC (in percent) = (IPIAP)/IAP × 100, where IP and IAP are the currents for parallel and antiparallel orientations of the magnetization, respectively.

Fig. 1 Schematic of the device, together with its magnetocurrent and photovoltaic characterization.

(A) Schematic representation of the C60-based molecular spin-photovoltaic (MSP) device, composed (from bottom to top) of Si, SiO2, Co, AlOx, C60, and Ni80Fe20. B, magnetic field; hv, photon energy. (B) Magnetocurrent (MC) on the MSP device measured at 295 and 80 K with a bias of 10 mV in dark conditions. The MC is defined as the relative change in electrical current [MC (in percent) = (IPIAP)/IAP × 100, where IP and IAP represent the current for parallel and antiparallel orientations of the magnetization of the Co and Ni80Fe20 electrodes, respectively] as a function of magnetic field. The arrows in the image indicate the orientations of the magnetization. (C) Current-voltage curves measured with and without white-light irradiation (7.5 mW/cm2) at room temperature when the relative orientation of the electrodes was parallel (Iout, output current; Vapp, applied voltage). The open-circuit voltage (VOC) and the short-circuit current (ISC) are indicated on the curve. In the measurement setup, the Co electrode is grounded.

Our optimized MSP device delivered considerable MC values of 6.5% at 295 K and ~15% at 80 K (Fig. 1B) [MC data at other temperatures are shown in fig. S2 (22)]. These values are in consonance with previous reports for C60 layers (1821). Our spin-valve structure resembles simple molecular photovoltaic cells, and we observed a photovoltaic effect under white-light irradiation (7.5 mW/cm2 illuminating an area of 1 cm2) at room temperature (Fig. 1C). The open-circuit voltage (VOC) and short-circuit current (ISC) of the device—defined as photogenerated bias at zero current and photocurrent measured without applied bias, respectively—are relatively small at room temperature, given that the device is a single-layer molecular solar cell (27), we used low-intensity light irradiation (equivalent to 7.5 × 10−2 Sun), and our ferromagnetic electrodes are not fully transparent [fig. S3 (22)].

We illustrate integrated spin and photovoltaic operation in Fig. 2. In a short-circuit solar cell (Fig. 2A), free carriers photogenerated in the C60 layer were driven by the built-in potential between the two magnetic electrodes and generated an output current (Iout) equal to ISC (Fig. 2A, 2). These carriers did not produce any spin-mediated magnetoresistance when the magnetization of the FM electrodes was switched with a magnetic field, because they were intrinsically non–spin-polarized (Fig. 2B).

Fig. 2 Device functioning principle, together with its photovoltaic characterization in a magnetic field.

(A) The MSP device operating as a short-circuited single-layer molecular solar cell without (diagram 1) and with (diagram 2) light irradiation. h+, hole; e, electron. (B) Current versus magnetic field measurement with and without light irradiation at 80 K (short-circuit mode), corresponding to the diagrams in (A). The photogenerated current is not sensitive to the magnetic field. (C) A MSP device operating in open-circuit mode under the application of a magnetic field, in the parallel (diagram 1) and antiparallel (diagram 2) configurations. (D) Applied voltage versus magnetic field measurement at 80 K in the parallel and antiparallel configurations (equal to VOC because we operated in open-circuit mode), corresponding to the diagrams in (C). VOC increased when the magnetic moments of the ferromagnetic (FM) electrodes became antiparallel. (E) I-V curves measured under white-light irradiation (7.5 mW/cm2) for the cases in which the magnetic moments of the FM electrodes were parallel and antiparallel. The changes in VOC and Iout are indicated on the curves. The measurements shown were performed at 80 K.

For open-circuit MSP operation, the applied voltage (Vapp) must compensate the photogenerated voltage (VOC) as it transports carriers from one magnetic electrode to the opposite while driving charge recombination. For parallel magnetization, the spin-polarized carriers created at the Co electrode could reach the NiFe electrode and recombine with the photogenerated holes (28, 29). In this case, Vapp between the two electrodes corresponded to VOC,P (Fig. 2C, 1). For the antiparallel case, the collected holes in the NiFe electrode could only be compensated by the injected spin-polarized electrons in the presence of an enhanced photovoltage (VOC,AP) through the spin-filtering effect at the FM anode (Fig. 2C, 2). The extra generated photovoltage (ΔVOC) is defined as the difference between the photovoltage for antiparallel and parallel configurations of the magnetic electrodes in the MSP device (ΔVOC = VOC,AP – VOC,P). ΔVOC is the spin-photovoltaic response caused by the spin-polarized charge carrier accumulation at the molecule/anode interface (Fig. 2D).

We introduce a magnetophotovoltage (MPV) to quantitatively describe the spin-photovoltaic effect that represents the photovoltage change under the application of a magnetic field [MPV (in percent) = ΔVOC/VOC,P × 100). Our MSP device had MPV values of 10.8% at 80 K and 4.6% at room temperature.

All of the different effects described above can be experimentally summarized in the I-V measurements shown in Fig. 2E. In the case of Vapp = 0, Iout was insensitive to the magnetic field and showed a constant value, ISC (Fig. 2, A and B). However, in the case of Vapp ≠ 0, Iout and VOC values were affected by the external magnetic field change (that controlled the parallel or antiparallel alignment of the magnetization of the electrodes), and their changes are indicated in the figure as ΔI and ΔVOC. Because of the increased VOC, the MSP device increased in power conversion efficiency at 80 K from 5.2 × 10−6 to 5.9 × 10−6 (30).

Other functionalities can also be achieved that arise from the relation between spin transport and light irradiation in our device. In Fig. 3A, we show the current versus magnetic field (I-B) measurements of the MSP device obtained under a constant 10-mV voltage bias for various intensities of irradiated light. The output current could be shifted by the photovoltaic effect as the irradiated light intensity increased and its sign adjusted to be either negative or positive. At the specific range of light illumination where the I-B curve stretched over the zero of current (so IP was negative and IAP was positive), the changes in current flow direction depended on the magnetic field (Fig. 3A). For a light intensity of 4.67 mW/cm2, we obtained IP = –IAP, and the MSP device acted as a perfect magnetic current inverter (Fig. 3B). Given the large changes in the output current obtained for different light illuminations, together with the ΔI current value that remained constant between parallel and antiparallel magnetic configurations (Fig. 3, A and C), the traditional calculation of magnetocurrent does not have a clear meaning in this scenario. For example, and as depicted in Fig. 3C, we can achieve a MC tending to infinite for particular light intensity values, stressing again the link between spin transport and light. This very large figure of merit could be useful in complex magnetic sensors, especially in those requiring a negligible baseline, or as a light sensor. The specific value of the light intensity at which the MC maximizes depends on the voltage applied, and hence an extra tunability could be obtained.

Fig. 3 Operation of the device at a constant applied voltage bias under different light irradiation conditions.

(A) Manipulation of I-B curves in a MSP device by the irradiated light intensity under a constant bias of 10 mV at 80 K. (B) In the particular case of IP = –IAP, the flow direction of the current can be freely inverted by the magnetic field controlling the orientation of the magnetic electrodes, without changes in its absolute value. (C) Output current Iout (IAP and IP) and normalized magnetocurrent response versus the irradiated light intensity; the curves were calculated from discrete data measured as shown in (A).

Because of the intermixing between different degrees of freedom, the manipulation of the I-B curves could also be achieved by modifying the applied bias under a constant light irradiation. In this case, both the magnitude and the sign of the I-B curves were modified (Fig. 4A). The comparison between both modes of operation—light intensity and applied bias—was observed directly, as shown in Fig. 4B.

Fig. 4 Operation of the device at a constant light irradiation condition under different applied voltage biases, together with the more complex electro-optical modulation with varying voltage and light.

(A) Manipulation of I-B curves in a MSP device under different applied biases and constant light irradiation of 7.5 mW/cm2 at 80 K. (B) Change in current (ΔI) under the application of a magnetic field, obtained as a function of both applied light irradiation and applied voltage bias. (C) Electro-optical modulation, varying both the applied voltage bias and the light irradiation. The input values are selected in such a way that Vapp exactly cancels VOC,P. The device output is a spin photogenerated current.

The qualitative difference between the light and bias manipulations of the MC (Fig. 4B) is shown schematically in Fig. 2. The change in the applied bias modifies the amount of accumulated spin charge carriers at the anode/molecule interface, whereas the change in light irradiation modifies the amount of photogenerated carriers. If we adjusted in each case the applied bias to cancel the open-circuit bias created by the light, a spin photogenerated current could be modulated (Fig. 4C and fig. S4). In this case, the output current of the device was set to zero for a parallel orientation of the magnetization of the electrodes, and the change in the magnetization alignment led to the generation of the spin-polarized output current (equal to ΔI). Moreover, the MC recorded is a pure spin-related signal, caused entirely by the spin-valve effect. This effect could lead to a more accurate fitting of the experimental data to some existing theoretical models and, in general, to a better understanding of the spin transport in molecular materials (31, 32). Moreover, and with potential relevance to molecular optoelectronic applications, the magneto-photocurrent signal was also observed at room temperature (fig. S3). In traditional molecular spintronics devices, such as spin valves, a relatively large power consumption is created by the background current. However, because in our case the base current is set to zero owing to the electro-optical modulation, the operating power (equal to I2R, where R is the electrical resistance of the device) for obtaining the same magnetocurrent response is reduced to <1% when compared with standard molecular spin valves (21, 23, 24).

Supplementary Materials

Materials and Methods

Figs. S1 to S4

Reference (33)

References and Notes

  1. Materials and methods are available as supplementary materials.
  2. Acknowledgments: The authors acknowledge financial support from the Ministry of Science and Technology of China (grants 2016YFA0200700 and 2017YFA0206600); the European Union 7th Framework Programme under the Marie Curie Actions (256470-ITAMOSCINOM), the European Research Council (257654-SPINTROS), and the NMP project (263104-HINTS); the Spanish Ministry of Economy (project no. MAT2015-65159-R); the Basque Government (project no. PRE_2016_2_0025); the National Natural Science Foundation of China (grant 21673059); and the CAS Pioneer Hundred Talents Program. All data are reported in the main paper and supplementary materials.
View Abstract

Navigate This Article