Tayfun Gokmen

Band tailing and efficiency limitation in kesterite solar cells

We demonstrate that a fundamental performance bottleneck for hydrazine processed kesterite Cu2ZnSn(S,Se)4 (CZTSSe) solar cells with efficiencies reaching above 11% can be the formation of band-edge tail states, which quantum efficiency and photoluminescence data indicate is roughly twice as severe as in higher-performing Cu(In,Ga)(S,Se)2 devices. Low temperature time-resolved photoluminescence data suggest that the enhanced tailing arises primarily from electrostatic potential fluctuations induced by strong compensation and facilitated by a lower CZTSSe dielectric constant. We discuss the implications of the band tails for the voltage deficit in these devices.

High Efficiency Cu2ZnSn(S,Se)4 Solar Cells by Applying a Double In2S3/CdS Emitter

High-efficiency Cu2ZnSn(S,Se)4 solar cells are reported by applying In2S3/CdS double emitters. This new structure offers a high doping concentration within the Cu2ZnSn(S,Se)4 solar cells, resulting in a substantial enhancement in open-circuit voltage. The 12.4% device is obtained with a record open-circuit voltage deficit of 593 mV.

Low band gap liquid-processed CZTSe solar cell with 10.1% efficiency

A low band gap liquid-processed Cu2ZnSn(Se1−xSx)4 (CZTSSe) kesterite solar cell with x ≈ 0.03 is prepared from earth abundant metals, yielding 10.1% power conversion efficiency. This champion cell shows a band gap of 1.04 eV, higher minority-carrier lifetime, lower series resistance and lower Voc deficit compared to our previously reported higher band gap (Eg = 1.15 eV; x ≈ 0.4) cell with similar record efficiency. The ability to vary the CZTSSe band gap using sulfur content (i.e., varying x) facilitates the examination of factors limiting performance in the current generation of CZTSSe devices, as part of the thrust to achieve operational parity with CdTe and Cu(In,Ga)(S,Se)2 (CIGSSe) analogs.

Electronic properties of the Cu2ZnSn(Se,S)4 absorber layer in solar cells as revealed by admittance spectroscopy and related methods

Admittance spectra and drive-level-capacitance profiles of several high performance Cu2ZnSn(Se,S)4 (CZTSSe) solar cells with bandgap ∼1.0–1.5u2009eV are reported. In contrast to the case for Cu(In,Ga)(S,Se)2, the CZTSSe capacitance spectra exhibit a dielectric freeze out to the geometric capacitance plateau at moderately low frequencies and intermediate temperatures (120–200u2009K). These spectra reveal important information regarding the bulk properties of the CZTSSe films, such as the dielectric constant and a dominant acceptor with energy level of 0.13–0.2u2009eV depending on the bandgap. This deep acceptor leads to a carrier freeze out effect that quenches the CZTSSe fill factor and efficiency at low temperatures.

Valley susceptibility of an interacting two-dimensional electron system.

We report direct measurements of the valley susceptibility, the change of valley population in response to an applied symmetry-breaking strain, in an AlAs two-dimensional electron system. As the two-dimensional density is reduced, the valley susceptibility dramatically increases relative to its band value, reflecting the systems strong electron-electron interaction. The increase has a remarkable resemblance to the enhancement of the spin susceptibility and establishes the analogy between the spin and valley degrees of freedom.

Photoluminescence characterization of a high-efficiency Cu2ZnSnS4 device

We report on low-temperature (4u2009K) photoluminescence of an 8.3% efficient Cu2ZnSnS4 photovoltaic device. Measurements were recorded as a function of excitation intensity, and the evolution of the resulting spectra is discussed. The spectra indicate that the radiative recombination is characteristic of heavily compensated material with a high quasi donor-acceptor pair density, as determined by the relationship between peak height, peak position, and excitation intensity, as well as the carrier lifetimes at different wavelengths. The blue-shift of the defect-derived peak position is used to estimate the quasi donor-acceptor pair spacing and density. The data indicate an average pair spacing of roughly 3.3u2009nm, yielding an overall total radiative-defect density of ∼1.3u2009×u20091019u2009cm−3.

High-efficiency thin-film InGaP/InGaAs/Ge tandem solar cells enabled by controlled spalling technology

In this letter, we demonstrate the effectiveness of the controlled spalling technology for producing high-efficiency (28.7%) thin-film InGaP/(In)GaAs/Ge tandem solar cells. The controlled spalling technique was employed to separate the as-grown solar cell structure from the host Ge wafer followed by its transfer to an arbitrary Si support substrate. The structural and electrical properties of the thin-film tandem cells were examined and compared against those on the original bulk Ge substrate. The comparison of the electrical data suggests the equivalency in cell parameters for both the thin-film (spalled) and bulk (non-spalled) cells, confirming that the controlled spalling technology does maintain the integrity of all layers in such an elaborate solar cell structure.

Minority carrier diffusion length extraction in Cu2ZnSn(Se,S)4 solar cells

We report measurement of minority carrier diffusion length (Ld) for high performance Cu2ZnSn(S,Se)4 (CZTSSe) solar cells in comparison with analogous Cu(In,Ga)(S,Se)2 (CIGSSe) devices. Our Ld extraction method involves performing systematic measurements of the internal quantum efficiency combined with separate capacitance-voltage measurement. This method also enables the measurement of the absorption coefficient of the absorber material as a function of wavelength in a finished device. The extracted values of Ld for CZTSSe samples are at least factor of 2 smaller than those for CIGSSe samples. Combined with minority carrier lifetime (τ) data measured by time-resolved photoluminescence, we deduce the minority carrier mobility (μe), which is also relatively low for the CZTSSe samples.

Semi-empirical device model for Cu2ZnSn(S,Se)4 solar cells

We present a device model for the hydrazine processed kesterite Cu2ZnSn(S,Se)4 (CZTSSe) solar cell with a world record efficiency of ∼12.6%. Detailed comparison of the simulation results, performed using wxAMPS software, to the measured device parameters shows that our model captures the vast majority of experimental observations, including VOC, JSC, FF, and efficiency under normal operating conditions, and temperature vs. VOC, sun intensity vs. VOC, and quantum efficiency. Moreover, our model is consistent with material properties derived from various techniques. Interestingly, this model does not have any interface defects/states, suggesting that all the experimentally observed features can be accounted for by the bulk properties of CZTSSe. An electrical (mobility) gap that is smaller than the optical gap is critical to fit the VOC data. These findings point to the importance of tail states in CZTSSe solar cells.

Acceleration of Deep Neural Network Training with Resistive Cross-Point Devices: Design Considerations

In recent years, deep neural networks (DNN) have demonstrated significant business impact in large scale analysis and classification tasks such as speech recognition, visual object detection, pattern extraction, etc. Training of large DNNs, however, is universally considered as time consuming and computationally intensive task that demands datacenter-scale computational resources recruited for many days. Here we propose a concept of resistive processing unit (RPU) devices that can potentially accelerate DNN training by orders of magnitude while using much less power. The proposed RPU device can store and update the weight values locally thus minimizing data movement during training and allowing to fully exploit the locality and the parallelism of the training algorithm. We evaluate the effect of various RPU device features/non-idealities and system parameters on performance in order to derive the device and system level specifications for implementation of an accelerator chip for DNN training in a realistic CMOS-compatible technology. For large DNNs with about 1 billion weights this massively parallel RPU architecture can achieve acceleration factors of 30, 000 × compared to state-of-the-art microprocessors while providing power efficiency of 84, 000 GigaOps∕s∕W. Problems that currently require days of training on a datacenter-size cluster with thousands of machines can be addressed within hours on a single RPU accelerator. A system consisting of a cluster of RPU accelerators will be able to tackle Big Data problems with trillions of parameters that is impossible to address today like, for example, natural speech recognition and translation between all world languages, real-time analytics on large streams of business and scientific data, integration, and analysis of multimodal sensory data flows from a massive number of IoT (Internet of Things) sensors.