This article provides an overview of the photovoltaic (PV) devices currently available, and the spectral characterization methods used in the assessment of their efficacy in the goal of harnessing significant amounts of energy from the sun and artificial sources of light.
PV devices (or solar cells) are thus-called because they depend upon the photovoltaic effect to produce a voltage (and current) when exposed to light.
Although Becquerel discovered the photovoltaic effect in 1839, the history of practical PV devices does not start until 1954 with the demonstration by Bell Laboratories of a silicon solar cell with a conversion efficiency of 6% and the following development of PV as the main energy source for satellites in orbit, starting with Vanguard I in 1958.
This extra-terrestrial application certainly continues in parallel with the application of PV devices as a renewable energy source and in consumer devices, albeit with varying terms of reference.
Although high-budget space applications place more of an accent on energy production efficiency, with cost a secondary consideration, the use of PV for renewable energy and consumer products is highly focused on cost reduction.
Other than the (significant) cost in manufacturing and installing PV devices, the energy produced is basically free, and extensively available. It is for this reason that solar energy is progressively being seen as a viable source for universal energy requirements, and for use in consumer devices.
PV Device Materials
PV device types are commonly classified in three generations. The first generation devices, adopting the technological advances of the microelectronics sector, are based on single-junction crystalline silicon which presently remains the most popular material for PV devices.
High-efficiency devices are possible due to the quality material used, yet at the cost of expensive manufacturing methods. Additionally, because of relatively poor light absorption, devices have to be several hundred microns thick, signifying a substantial raw material cost.
As a move towards cheaper PV devices is imminent, recourse is being made to second generation devices, which gain from thin-film technology.
As these materials have good light absorption, up to a factor 100 less material is needed – an instant material saving result. Besides this, process methods are much less complex and thus cheaper, and, considerably in some cases, can be processed at low temperatures paving the way to the possibility of developing devices based on a flexible plastic substrate.
Fact is that nothing comes free. These devices are also characterized by an inferior material quality which results in the loss of light-generated electrons to defects in the structure, and a corresponding drop in efficiency.
Third generation devices include a huge panoply of technologies, aimed at refining the efficiency of current devices, or incorporating different materials and material structures.
These include adding materials responding to different spectral regions, multiple junction devices, mirror/lens-based concentrators to boost the level of light exposure of the device, up to the equivalent of hundreds of suns, and organic polymer and dye-sensitized devices, based on economical technology and available on flexible substrates.
Image Credits: foxbat/shutterstock.com
Light Conversion Processes
The operation of a PV cell requires the absorption of light to produce either an electron-hole pair or exciton and the separation of charge carriers to an external circuit.
In a semiconductor pn junction, an incident photon of the precise wavelength (and therefore energy) excites an electron from the valence band to the conduction band, causing a hole to form. An in-built voltage across the pn junction sweeps away these carriers to an outer circuit before they can re-combine.
A similar process takes place in organic cells, whereby light incident on the organic layer excites an electron to the lowest unoccupied molecular orbit (LUMO) causing a hole to form in the highest occupied molecular orbit (HOMO). The electronic properties of either the electrodes or a heterojunction are customized to separate the charge carriers, to be passed to an external circuit.
The situation in a dye-sensitized solar cell, consisting of an oxide layer (typically TiO2), a dye, and an electrolyte redox system, is rather different.
An electron in the dye is excited by incident light, and conveyed to the conduction band of the TiO2 layer and then to the external circuit. The dye takes an electron from the electrolyte via a process of oxidation. The oxidized electrolyte diffuses to the counter electrode where electrons are re-added having flown via the external circuit.
Spectral responsivity of a selection of current PV devices, to be compared with the AM1.5G reference spectrum
Impact upon Test Methods
These very diverse physical processes have a remarkable consequence in the measurement of spectral response where measurements with a monochromatic probe should be completed under real-time illumination by a solar simulator. To distinguish between the photocurrent produced by the solar simulator from that produced by the monochromatic probe, the latter is controlled with the use of an optical chopper.
As carrier transport in semiconductors and organic PV devices is extremely fast compared with desensitized devices, it follows that in the latter case very low chopping frequencies or CW operation of the monochromatic probe is essential.
Whether the illuminating source is the sun or artificial light, it is obviously desirable to acquire PV devices that respond to as much of the wide source emission range as possible and have the highest responsivity, or conversion efficiency.
In practice, devices react only to a limited range of wavelengths, as shown in the image below, limited at long wavelengths by the material band gap, and at short wavelengths by material absorption. Device spectral response relies on a large number of factors including device design, material system, and electrical contact etc. and for single junction pn, devices are limited by the Schockley–Queisser limit.
The combination of these parameters may be assessed taking into consideration the device energy conversion efficiency.
Together with an I-V measurement, the following quantities offer the most information of material and PV device function.
The spectral response (A W-1) of a PV device provides information on the physics behind the global device, taking into consideration not only the material, but also the reflectance and transmittance of the device.
This measurement is accomplished by shining a monochromatic probe beam onto the sample and noting the photocurrent generated as a function of wavelength.
Care must be taken to guarantee the probe beam is not shaded by electrical connections, or that the shading is considered by modifying the resulting response.
The probe is first characterized, using a detector referred to as responsivity (A.W-1) to define the power in the beam. Ensuing measurement of the photocurrent generated by the device under test as a function of wavelength enables spectral responsivity determination.
The conditions under which this measurement should be performed are discussed further on.
External Quantum Efficiency (IPCE)
The external quantum efficiency (EQE) is defined as the number of electrons provided to the external circuit per photon incident on the device, and is directly acquired from the spectral response measurement by the argument given below.
The number, n, of electrons produced by the device, n = (It/e), where I is the produced current, t time, and e the charge of the electron.
The number, m, of photons, incident on the sample, m = Pt/ EV, where P is the power in the beam, t time, and EV the photon energy. The quantum efficiency, η, is defined as,
η = 100.n/m = 100.(It/e)/(Pt/EV) = 100.(I/P).(EV/e)
=> η = 100.S.(hc/e).(1/ λ) ≈1239.84.S/ λ (%)
Where S is the spectral responsivity in A.W-1 and λ the wavelength in nm. EQE can thus be established directly from a measurement of the spectral response.
Calculation of Jsc
The measured spectral response may be used to predict the anticipated device short circuit current density, Jsc, under regular testing conditions. This is merely calculated by estimating the following integral over the spectral range of response of the device under test.
Where Jsc is in A.m-2, St (λ) is the device spectral response, in A.W-1.nm-1 and E0 (λ) the AM1.5 reference spectrum in W.m-2.nm-1.
I-V measurements of PV devices, used to define amongst other things the device Jsc and Isc, should be done under AM1.5 illumination.
A reference cell is frequently used to establish the irradiance of the solar simulator used. Where the spectral response of the reference cell varies from the test cell, the mismatch factor should be computed from:-
Where Sr (λ) is the reference device spectral response, St (λ) the test device spectral response, E0 (λ) the AM1.5 reference spectral distribution, and E (λ) the solar simulator spectral distribution. As these parameters appear in both numerator and denominator, normalized quantities may be used.
Where essential, the irradiance of the solar simulator should be set to:-
Where Gref is the reference irradiance of 1000 W•m-2
Reflectance and Transmittance
In an idyllic world, all photons reaching a PV device are conveyed only to the active region where the conversion process takes place.
As a result of the refractive index of the materials used, light is reflected from the front surface of the device (to moderate which anti-reflection coatings are applied). In the case of thin-film devices, one should also take into account that light may be conveyed through the sample.
The total reflectance (diffuse and specular), and the total transmittance (diffuse and normal), of the device can be measured with the help of an integrating sphere.
Besides providing a means of determining device IQE (below), these measurements also allow the characterization of anti–reflection coatings and the transmission of thin film layers.
Internal Quantum Efficiency
Based on the above measurements of transmittance and reflectance, the EQE can be altered to consider only the portion of the incident light reaching the active region, to produce the internal quantum efficiency.
This allows a proper understanding of the material properties of the device.
Internal quantum efficiency is just expressed as:
IQE = EQE/ (1-R-T)
Where R is the reflectance, and T the transmittance of the sample.
Luminescent Material Characterization
The physical process of luminescence can be harnessed in two circumstances.
1. The efficiency of PV devices may be improved considerably by down-shifting and up-converting non-absorbable photons, to yield photons having an energy better suiting the device spectral response.
2. Fluorescent solar collectors are made up of a mixture of fluorescent dyes embedded in a transparent medium. Absorbed incident sunlight is re-emitted at longer wavelengths and conveyed to the edge of the collector by total internal reflection, where a solar cell is situated.
Standard Testing Conditions
So as to perform measurements of spectral response/ EQE (IPCE) in a reliable manner, transferable between laboratories, standard testing conditions are important. Indeed, in the case of certain PV technologies, measurement is possible only under highly specific conditions.
International standards are published guide the effective measurement of spectral response/EQE (IPCE) ofphotovoltaic devices, for example:
- IEC 60904-8:2014 - Photovoltaic devices - Part 8: Measurement of spectral responsivity of a photovoltaic (PV) device
- IEC 60904-8-1:2017 - Photovoltaic devices - Part 8-1: Measurement of spectral responsivity of multi-junction
Basically, these standards stipulate the conditions of the measurements of PV devices requiring, for instance, the use of a solar simulator approximating AM1.5G at an irradiance of 1000 Wm-2, and controlled sample temperature.
One must also take into account certain specific technologies or cell architectures which will admit to testing only under very precise conditions.
The use of the solar simulator physically places the device in real operating conditions, and is associated mainly with material quality.
In poorer quality material, where the crystal structure is faulty, there are traps and defects in which generated carriers are lost, causing erroneously reported spectral response.
Light from the solar simulator produces a large number of carriers in the material which can pump defects and traps, guaranteeing that the carriers produced by the probe beam are not thus lost, in contrast to measurements with the probe without solar bias.
Indeed, the application of solar simulators in the case of monolithic multi-junction (MJ) devices is vital as discussed already.
Spectral response/EQE measurements are typically performed under short circuit conditions.
With multi-junction solar cells however, the voltage over non-tested junctions can cause the junction under test to operate in reverse bias, which is not desired, particularly where the subcell has a low reverse breakdown voltage.
With thin-film solar cells, the spectral response/ EQE may be voltage-dependent. Defects in the material cause a reliance on the electric field in the space charge region for minority carrier collection, leading thus to a reliance on the device operating voltage.
Sample temperature control is also a crucial consideration since as sample temperature rises, the properties of the devices are altered (one attains with device heating a red shift in the response), and its efficiency can decline as more energy is lost to phonons in the lattice.
A means of countering the heating effect associated with exposure to the light bias should be used at all times.
Moreover, it may be helpful to explore the temperature dependence of PV technology to assess more accurately device operation in the field.
Whilst standards1 exist for the assessment of the spectral response of monolithic multiple junction (MJ) solar cells, the measurement process is far from standard. Without such guidance, however, it is probable that in the case of subcells having low reverse breakdown voltage or low shunt resistance, substantial measurement errors will result.2
1 ASTM- E2236-05 “Standard Test Methods for Measurement of Electrical Performance and Spectral Response of Nonconcentrator MultijunctionPhotovoltaic Cells and Modules”
2 Meusel et al, “Spectral Response Measurements of Monolithic GaInP/Ga(In)As/Ge Triple-Junction Solar Cells: Measurement Artifacts and their Explanation”, Prog. Photovolt: Res. Appl. 2003; 11:499–514
In aiming to measure the spectral response/EQE of monolithic MJ solar cells, such as the III-V monolithic GaInP/GaInAs/Ge triple junction solar cell depicted here, the measurement of component subcells on a separate basis is not possible since they are epitaxially grown on one substrate and connected by tunnel diodes.
Subcell spectral response must thus be established by making use of the effect of current limitation, accomplished through light biasing.
Besides biasing the subcell under test at a level of one sun to mimic use conditions, it is essential to light bias the non-tested subcells at a higher intensity such that the former produces the smallest photocurrent and is thus current limiting. It follows that good control of the light bias source spectrum is essential.
Since the current of the MJ cell is restricted by the subcell under test, it follows that the non-tested cells with surplus photocurrent work close to their Voc. If the MJ cell is tested in short circuit conditions, then a negative voltage approximately equal to the sum of the Voc of the other cells is placed across the tested subcell.
Though the photocurrent produced by light biasing may be thought to be a constant, the presence of the monochromatic probe gives rise to variations in subcell operating voltage, depending on the probe wavelength and the response range of the current limiting subcell under test.
Where the subcell under test has unsuitable properties such as a low reverse breakdown voltage or a low shunt resistance, repeatedly displayed in low bandgap materials such as germanium, this can cause variations in the measured photocurrent, and the faulty reporting of spectral response/EQE: the outcome depends mainly on the true nature of the I-V curves of both the subcell under test and that of the non-tested subcells.
Beyond the response range of the subcell under test, the monochromatic probe causes an increase in operating voltage of the non-tested cells, compensated for by a decline in the operating voltage of the subcell under test. This may result in an increase in Jsc, exhibiting a response where one is not anticipated. Within the subcell response range, the presence of the monochromatic probe will directly cause an increase in Jsc, transferring the non-tested cells to lower operating voltage, and compensated by an increase in the voltage of the tested subcell. The resulting current may be less than that anticipated in response to the monochromatic probe, causing a lower reported spectral response/EQE. Both effects are usually observed phenomenon in the measuring the bottom cell of the GaInP/GaInAs/Ge triple junction solar cell.
Shifting the external voltage of the cell will likely diminish both of the above effects: in moving to higher gradient regions of the I-V curve of the non-tested cells, lower shifts in subcell operating voltage are faced, giving rise to less difference in the current of the cell under test.
Optimization of Light Biasing
Increasing the photocurrent produced by the non-tested subcells causes increased gradient of the I-V curve in the proximity of their operating voltage, whilst decreasing the photocurrent produced by the subcell under test will result in a reduction of operating voltage closer to Voc where the gradient of the I-V curve is steepest.
Both will result in less variation in the current of the cell under test, and is attained through appropriate filtering.
MJ Cell Example
For the accurate measurement of the GaInP/ GaInAs/Ge triple junction solar cell, the following process is suggested.
Current limiting may be confirmed by spectral response measurement at several levels of non-tested cell light bias intensity.
GaInP Top Junction
The GaInP junction reacts at ~300-700 nm. The device under test should be illuminated concurrently by a solar simulator at one sun bias, and a second simulator is filtered with a red long-pass filter. The spectral response may then be measured straight over the long range 300-800 nm.
GaInAs Middle Junction
The GaInAs junction reacts at ~500-900 nm. The device under test should be illuminated concurrently by a solar simulator at one sun bias, and a second simulator is filtered with a blue band pass filter, conveying also in the infrared. The spectral response may then be measured straight over the extended range 300-1100 nm.
Ge Bottom Junction
The Ge junction responds ~900-1800 nm. The device under test should be illuminated concurrently by a solar simulator at one sun bias, and a second simulator is filtered with an IR rejection filter.
To take this subcell to short circuit condition, an external voltage should be used. Under two-fold solar simulator irradiation, the operating voltage, Vop of the top two junctions is recorded (usually ~2-2.5V). The voltage across the device should then be fixed to –Vop to take the third junction into short circuit.
The spectral response may then be measured straight over the long-drawn-out range 800-1800 nm. Where the accurate bias voltage is used, a maximum spectral response must be recorded. This may be proved by repeating the measurement at other levels of voltage bias.
This information has been sourced, reviewed and adapted from materials provided by Bentham Instruments Limited.
For more information on this source, please visit Bentham Instruments Limited.