Pursuing high-efficiency photovoltaics with novel alloys
By forming an intermediate band, dilute nitrides promise to propel single-junction solar cell efficiency to new highs
BY KIN MAN YU FROM THE CITY UNIVERSITY OF HONG KONG AND WLADEK WALUKIEWICZ FROM LAWRENCE BERKELY NATIONAL LABORATORY
Throughout the world, governments are encouraging greater use of renewable energy in a bid to cut carbon dioxide emissions. This is helping to drive up the deployment of solar cells, the majority of which are made from silicon.
Although silicon currently offers the best bang-per-buck, it has major downsides, including an efficiency that is not that high − and has little room left for improvement. These weaknesses are a consequence of the incredibly wide range of solar photon energies. The sun's radiation extends from 0.4 eV in the mid infrared to 4 eV in the deep ultraviolet, and it is impossible for a single-junction cell to convert all of this energy efficiently. Photons with energy above the band gap must lose their excess energy to heat, while those below it will not be absorbed.
To increase power conversion efficiency, the solar spectrum has to be divided into parts that are absorbed by separate cells with different energy gaps. This is the basic premise of the multi-junction solar cell, which features junctions from different materials connected in series.
The most established class of multi-junction cell is based on group III-V compounds − and it may have a bottom cell made from germanium. This particular multi-junction cell is an established commercial technology, dominating power generation on satellites and generating megawatts of electricity on the ground, where it lies at the heart of solar concentrating systems. However, the chips are very expensive, due to the high cost of the materials and complexities associated with epitaxial growth.
Figure 1. A simplified band diagram of a (left) multi-junction (MJ) and (right) an intermediate-band solar cell (IBSC). In a MJ cell, semiconductor materials with different bandgaps are stacked on top of each other with the material, with the largest gap on top. This allows absorption of a large fraction of the solar spectrum. In a IBSC, solar radiation can be absorbed by the valence band (VB) to either the intermediate (IB) or conduction band (CB), and at the same time by the IB to the CB.
A promising alternative − proposed more than 50 years ago by Martin Wolf, when he was working at the Hoffman Electronics Corporation in El Monte, California − is to use a single-junction device with multiple bands. This approach − that realises the same objective as using semiconductors with different gaps in multi-junctions cells − involves the introduction of a narrow band of partially occupied states in a semiconductor band gap. This modification to the single-junction cell facilitates transitions from the valence band to the conduction band through a sequential absorption of two lower-energy photons (see Figure 1 for a simplified band diagram). In a sense, the intermediate band plays the role of a stepping-stone, allowing sub-bandgap photons to excite electrons across the band gap.
These intermediate-band solar cells have the potential to deliver a massive hike in efficiency. Detailed balance calculations in 1997 by Antonio Luque and Antonio Marti from the Technical University of Madrid show that a power-conversion efficiency of 63 percent is possible when the intermediate band is ideally located. This value is far higher than the limit of what is possible for a two-junction tandem cell. What's more, if a junction can be made from a material with two bands of intermediate states, efficiency can reach almost 72 percent.
Fulfilling this promise of very high efficiency has not been easy, due to difficulties in identifying a semiconductor with an appropriate band structure. One option is to turn to quantum confinement effects in semiconductor superlattices and quantum dots. This has produced some success. However, it is difficult to produce a device that incorporates enough low-dimensional semiconductor material to have an intermediate band with sufficient absorption, while managing the strain in the structure so that it does not impair material quality. If this degrades, optical quality also falls. What's more, there is limited scope to adjust the position of the intermediate band.
Highly mismatched alloysOur partnership between the City University of Hong Kong and Lawrence Berkeley National Laboratory is pursuing a different approach, based on highly mismatch alloys. The new class of materials, discovered by our research group at Berkeley, are formed by alloying distinctly different semiconductors.
Figure 2. Energy band structure of GaN0.024As0.976,calculated using the band anti-crossing (BAC) model.
Figure 3. (a) The structure and band diagram of a blocked intermediate-band (BIB) device with the intermediate band (IB) disconnected from the contacts. Transitions generating electron-hole pairs utilizing the IB are denoted as hν1 and hν2. The transitions from the valence band (VB) to the conduction band (CB) are represented by hν3. (b) The structure and the band diagram of a UIB device with the IB connected to the backside contact.
The two most prominent, extensively studied groups of highly mismatched alloys are dilute nitrides and dilute oxides. In the former, column V atoms in III-V compounds are partially replaced with nitrogen; and in the later, oxygen partially replaces column VI atoms in II-VI compounds.
We choose to focus on dilute nitrides. They command attention because the partial substitution of arsenic with nitrogen produces a drastically different band structure in the resulting GaNxAs1-x.
The band structure of this class of highly mismatched alloy is well described by the band anti-crossing model. For example, the bandstructure of GaN0.024As0.976 calculated using this model shows that replacement of arsenic with 2.4 percent of nitrogen results in a splitting of the conduction band into two subbands: a lower, relatively narrow band that is located at 1.1 eV above the valence band and forms an intermediate band; and an upper band, shifted 0.5 eV higher than the original conduction band of GaAs (see Figure 2 for the band diagram). This dilute nitride is ideal for making intermediate solar cells, because it has optical transitions for photon energies of 0.9 eV, 1.1 eV and 1.9 eV.
One of the key requirements for an intermediate band solar cell is electrical isolation of the intermediate band from the charge-collecting contacts. This condition ensures that the operational voltage is determined by the largest band gap.
To demonstrate the importance of this pre-requisite, we have produced a pair of devices by MOCVD. The first has a blocked intermediate band, created by incorporating AlGaAs blocking layers on the surface and substrate sides (see Figure 3 (a)). The control, which highlights the pitfall of failing to isolate the intermediate band from the charge collecting contacts, has the intermediate band connected to the substrate side of the device (see Figure 3 (b)).
Another essential ingredient for the intermediate band solar cell is that its intermediate band has to be partially filled with electrons. Taking this step increases the optical absorption of the low energy photons (0.9 eV -1.1 eV) that promote electrons from the intermediate band into the conduction band. We satisfy this requirement by doping the GaNAs absorber with tellurium. This produces an electron concentration in the intermediate band of 1016 to mid-1017/cm3.
We have measured the external quantum efficiency of our devices (see Figure 4). As we predicted, in the unblocked intermediate band control, the intermediate band takes on the role of the conduction band. This creates a device that acts as a single gap photovoltaic cell, with characteristics determined by the band gap of about 1.1 eV. At higher photon energies, photocurrent falls off rapidly. In stark contrast, in the blocked intermediate band device, the photocurrent clearly exhibits two thresholds. The first of these occurs at about 1.1 eV and corresponds to the transition from the valence band to the intermediate band; and the second, at 2 eV, is associated with optical excitation directly from the valence band to the conduction band. The origin of both these thresholds has been confirmed by photomodulated reflection spectroscopy.
With this particular device, the backside blocking layer prevents electron transport from the intermediate band to the back contact. So, when the cell is illuminated, electrons that accumulate in the intermediate band absorb another low-energy photon, and are then promoted to the conduction band, where they can be collected at the back contact. It is also possible to promote electrons directly from the valence band to the conduction band "“ this is the origin of the higher energy threshold at 2 eV.
In short, the results provided by the external quantum efficiency measurements are very encouraging. The blocked intermediate band device has all the essential features of the multi-band solar cell: the open-circuit voltage is determined by the largest band gap; and sub-band light can contribute to the photocurrent through sequential absorption of two photons.
One issue that can plague intermediate band solar cells based on highly mismatched alloys is a weak coupling of the three optical transitions between the valence, intermediate and conduction bands. To determine whether this is the case in our devices, we have measured the electroluminescence from our intermediate band solar cells (results are shown in Figure 5). We find that the blocked intermediate-band device produces two-colour electroluminescence under reverse and forward bias. This is great news, as it provides a clear demonstration of the presence of the two optical transitions necessary for properly operating an intermediate-band cell. As expected, forward biasing of the unblocked intermediate band structure produced a solitary electroluminescence peak, resulting from transitions between the intermediate band and the valence band.
Figure 5. Electroluminescence spectra of the blocked intermediate band (BIB) structure as shown in Figure 3 (a) under forward (green) and reverse (yellow) bias. The electroluminescence of the reference unblocked intermediate band (UIB) structure is also shown (blue). The energy band diagram of the BIB structure under forward and reverse bias is also shown.
Although our efforts have provided clear evidence of an operational, intermediate-band solar cell based on highly mismatched alloys, the devices need to get better. Improvements must include an increase in the optical coupling strength between the intermediate band and the conduction band, an increase in the lifetime and the diffusion length of electrons in the conduction band, and an optimisation of the doping in the absorber layers.
Gains in performance might require the introduction of new, highly mismatched alloys. This could be a new alloy composition of dilute nitrides that provides a better location of the intermediate band relative to the conduction band. Recently there has been progress in synthesis of new highly mismatched alloys, which may be more suitable for intermediate band solar cell applications. These include dilute II-VI oxides as well as GaAsP-based dilute III-V quaternary nitrides.
