Characterization of Multi-Junction Solar Cells Using Combined Atomic Force Microscopy and Confocal Optical Spectroscopy

Table of Contents

Formulation of the Problem
Experimental Results and Discussion
Topoplogy of the PV Semiconductor Surface
Observations Under Photoexcitation
About NT-MDT


The sun is an abundant, easily accessible power source which is presently underutilized and will probably be the only choice for electric power in the near future. It is believed that the best method of solar power generation is using the photoelectric method used in solar cells (SCs).It is envisioned by the EU that at least 3% of the electric power will be provided from solar installations by the year 2020.

Formulation of the Problem

Currently, the highest efficiencies are exhibited by MJ SCs based on semiconductor nanoheterostructures. MJ SCs comprise a number of sub-cells with p-n junctions and barrier layers of a number of semiconductor materials. The arrangement of these subcells is in the order of decreasing energy bandgap from the photosensitive surface to the substrate connected by oppositely connected tunnel diodes. Hence the whole solar spectrum energy is segmented and collected causing high efficiencies. It is essential to understand that the most inefficient subcell determines the total efficiency of an MJ SC. Determining the constituent layers of this kind of composite device is possible with indirect, integral measurement methods and mathematical simulations. Information obtained thus is not always unambiguous as it requires the solution of multivariate inverse problems.

An unambiguous determination will require separate monitoring of the operation of all the constituent subcells.

Experimental Results and Discussion

An example is detailed below about how the NTEGRA Spectra AFM confocal Raman fluorescence Probe NanoLaboratory (PNL) is used in the study of MJ SCs based on a GaInP2/GaAs/Ge heterostructure having three p-n junctions. The total number of layers is more than 20 and individual layers are less than 20 nm thick as shown in Figure 1.

Figure 1. Schematic of an MJ SC with three subcells. Designations: various tints of pink, p-type layers of the heterostructure; light blue tints, n-type layers; and yellow, highly conducting layers of tunnel diodes and contact layers. The digits show the p-n junctions in the subcells based on (1) Ge, (2) GaAs, and (3) GaInP2.

The Kelvin probe force microscopy (KPFM) technique was used to determine the surface potential profile variations of a cross-section-cleaved SC with respect to the intensity, wavelength, and beam position of a laser excitation source. Based on the schematic of the layers in Figure 1, the distance between the p-n junctions of neighboring subcells based on GaAs and GaInP2 is less than a micrometer.

Response monitoring of the response, surface potential variation, of a separate subcell was enabled by focusing the excitation laser into a submicrometer spot. An objective having a numerical aperture of 0.7 and resolving power of 400 nm was used in the confocal laser microscope integrated in the NTEGRA Spectra PNL. An AFM cantilever is arranged below the objective allowing simultaneous optical excitation and AFM measurements as shown in Figure 2b. It is important to note that the instrument enables independent and synchronized laser spot scanning and sample scanning using a piezo driven mirror and sample piezo-scanner respectively.

Figure 2. (a) Schematic of layers in an MJ SC with the same color designations as those in Fig. 1. The three p-n junctions are shown by arrows. (b) Schematic of experiment. Optical micrographs of the edge of the cleaved surface of a SC during a KPFM experiment under focused photoexcitation of (c) the p-n junction in Ge with a blue laser (473 nm) and (d) p-n junction in GaAs with a red laser (785 nm). Latin numerals designate: (I) Ge substrate, (II) III-V layers (GaAs and GaInP2 ), (III) free space,and (IV) KPFM cantilever. The optical microscope is focused on the Ge substrate in Fig. 2c,and on the III-V layers in Fig. 2d.

Topology of the PV Semiconductor Surface

Some of the results obtained in the first Case Study are listed below:

  • It is observed as shown in Figures 2c and 2d, near the cleaved surface edges light spots from red and blue lasers focused on the p-n junctions in the subcells of Ge and GeAs respectively.
  • There is a sharp change in the surface topology in the left half of the topographic image shown in Figure 3a. In this particular region, the Ge substrate’s smooth relief changes to a striated topology of the III-IV layers.
  • Figures 2b and 2c show that III-V crystal materials are cleaved easily to form atomically smooth and perfectly planar surface only along 110 basal planes.
  • Ge and Si crystals cleave along a different crystal planes.
  • Ge substrate is two times thicker than all other MJ SC layers and hence cleavage propagation directions are predominantly based on the substrate
  • The contact potential difference (CPD) map in Figure 3b shows features in alignment with expected integrated potential differences in the bulk heterostructure under total darkness conditions.
  • The CPD map proves that near the p-n junction on the GaAs subcell surface there is a decrease in the CPD signal instead of the peak as shown in Figure 3c
  • The light band region corresponds to well doped transition layers between Ge and GaAs subcell
  • These discrepancies are seen because the semiconductor structure surface material varies from the bulk potential by near-surface band bending not known for an arbitrarily cleaved sample.

Figure 3. KPFM study of the cleaved surface of an MJ SC in the dark. During measurements both contacts to the MJ SC were grounded. (a) Topographic image of the cleaved surface profile, measured in semi-contactmode (the color-scale contrast spans the height variations of 0.85 μm). (b) Map of the CPD signal measured in the second pass in the absence of an external photoexcitation (the color-scale contrast spans the CPD variations of 1.05 V). (c) Smoothed equilibrium profile of the built-in potential (from model). Schematic of the layers: arrows with digits show the p-n junction positions in the subcells (see also color designations in Fig. 1). Measurement parameters: AFM laser with a wavelength of 650 nm used in the system for cantilever deflection detection, noncontact VIT_P probe, resonance at 257 kHz, surface potential signal was measured at 100 nm lift height and Uac=2 V.

Observations Under Photoexcitation

When the semiconductor surface is exposed to light with a photon energy more than the band gap of the material, the photocarrier separation by the near-surface field results in minority carriers emerging at the surface, which makes the band bending smaller. This mechanism is applied for semiconductors with surfaces depleted of majority carriers, in which the surface photovoltage has the opposite sign of that of majority carriers. In a complicated structure, photocarriers can be separated not only in the near-surface field, but also in the bulk due to the field of built-in barriers. For instance, it is possible to predict changes in the surface potential on illumination of a single p-n junction. Due to the photocarrier separation in the near-surface field, the p side is charged negatively, and the n side positively. In contrast, the separation of photocarriers in bulk material from the field of the p-n junction charges the p side positively, and the n side negatively.

Some of the observations are listed below:

  • If the number of photocarriers separated in the field of the p-n junction exceed those separated in the near-surface field, then the surface photovoltage will decrease, passing from the p side to the n side.
  • If the contacts to the p and n sides are shorted, then the contribution from the bulk separation is eliminated, and the surface photovoltage will increase on such a transition.
  • Figure 4 shows two sets of simulated and measured photovoltage profiles from a cleaved surface in alternate photoexcitation of p-n junctions in three subcells of MJ SCs.
  • The first set of profiles, Figs. 4a-4c, was obtained with blue laser excitation (wavelength λ= 473 nm), and the second, Figs. 4d-4f, with red laser excitation (λ = 785 nm). The photoexcitation densities were approximately the same in both cases, 2-3 mW/ m . The focal spot diameter D was calculated using the Rayleigh criterion D = 1.22 /NA, where λ is the laser wavelength, and NA = 0.7 is the numerical aperture of the objective.
  • Determination of the surface photovoltage profile was by the difference of PD values determined under photoexcitation and in the dark.

The simulation process was done with the following conditions:

  • The contacts to the MJ SC are shorted
  • The photovoltage appearing in the bulk of a p-n junction exposed to light is distributed among the barriers of two nonilluminated p-n junctions.
  • Also, the capacitances of these two junctions are considered to be equivalent.
  • The light from the blue laser is absorbed by all layers in the MJ SC and light from the red laser is not absorbed by the wide band gap GaInP2 layers.
  • In the experimental photovoltage profiles, the arrows show a dip in Fig. 4b and a peak in Fig. 4c. The simulation predicts these specific features.
  • The GaAs subcell is insulated from the contacts to the MJ SC by the potential barriers at the p-n junctions of the neighboring subcells. In case it is exposed to blue light, separation of photocarriers in the field of the p-n junction causes electrons to get ejected into the n layers of this subcell. Hence, a negative potential appears in the bulk of these n layers and in the p layers of the GaInP2 subcell. Due to the photocarrier separation in the near-surface field, the surface of the p layers is also negatively charged in relation to the bulk.
  • The joint effect of both processes forms a deep dip in the surface photovoltage profile when it passes across the p layers of the GaInP2 subcell, as seen in Fig. 4b. If red light is used, no photocarriers are generated in the wide-bandgap GaInP2 layers. Consequently, the dip should be less pronounced, which is indeed observed in Figure. 4e. When the GaInP2 subcell is exposed to blue light, a positive potential appears in the bulk of its p layers and is transferred to the n layers of the GaAs subcell.
  • The photoeffect at the surface of the n layers is also positive, and a peak corresponding to these layers appears in the photovoltage profile

Figure 4. Comparison of experimental and simulated data. (a-c) Photoexcitation with laser light (λ = 473 nm) focused on the p-n junctions in (a) Ge, (b) GaAs, and (c) GaInP2 . (d-f) Photoexcitation with laser light (λ = 785 nm) focused on the p-n junctions in (d) Ge, (e) GaAs, and (f ) GaInP2 . Designations: SPV, experimental surface photovoltage profile. A simulated profile is also given above each plot. Below, under all the plots are shown schematics of layers in MJ SCs (with the same color designations as those in Figs. 1-3).omparison of the experimental and simulated data. (a-c) Photoexcitation with laser light (λ = 473 nm) focused.


The conclusions from the study are listed below:

  • The study of a solar cell with three subcells based on Ge, GaAs, and GaInP2 in a NTEGRA Spectra PNL showed that it is possible to monitor the operation of each subcell separately.
  • The experimental surface photovoltage profiles obtained comply with the results of the qualitative simulation.
  • This agreement between the experimental data and the simulation results show that there are no parasitic barriers in the multijunction solar cell under study for the chosen photoexitation densities.
  • It should be noted that the NTEGRA Spectra PNL, integrating AFM with optical spectroscopy techniques offers a significantly broader set of capabilities for solar cell diagnostics than that considered in the present communication.
  • The following measurement techniques with submicrometer and nanometer spatial resolution that include the following are possible:
    • surface topography
    • local conductivity
    • variations of potentials and charges
    • built-in or induced by external bias or photoexcitation
    • evaluation of compositional homogeneity and material defects
    • spatial and spectral variations of transmittance, reflectance, and other optical properties
    • localization of nonradiative recombination regions
    • monitoring of p-n junction positions
    • monitoring of heterointerface transitions
    • mapping of mechanical stresses

All of these measurement scan be used to optimize the solar cell technology. For example, the internal design of solar cells can be optimized through correlation of regions having the maximum photovoltaic conversion efficiency with data on variation of the chemical composition, layer thickness, profile, defects and optical parameters.

About NT-MDT

NT-MDT has 550 employees, including Ph.D. scientists, many of whom are leaders in their field. The company has more than 600 installations in 39 countries, and has been operating in the APM market for more than 15 years, achieving worldwide distribution of their devices. NT-MDT's clients include Universities and colleges, laboratories, governments, research centers and scientific companies of all sizes in the nanotechnology field.

This information has been sourced, reviewed and adapted from materials provided by NT-MDT Co.

For more information on this source, please visit NT-MDT Co.

Ask A Question

Do you have a question you'd like to ask regarding this article?

Leave your feedback