AFMs shed light on emerging photovoltaics

Donna Hurley2 and Ben Ohler1
1 Oxford Instruments Asylum Research Inc., Santa Barbara, California USA
2 Lark Scientific LLC, Boulder, Colorado USA
Donna Hurley is a consultant on AFM measurement methods and their application to materials science. Previously, she was a senior scientist at the National Institute of Standards & Technology (NIST) and led a team to develop and apply new AFM techniques for nanomechanics. She has a Ph.D. in condensed-matter physics from the University of Illinois–Urbana-Champaign.
The atomic force microscope (AFM) plays an indispensable role in characterizing photovoltaic (PV) materials and devices. It maps surface morphology ranging from crystal grain structure to polymer network dispersion with unmatched spatial resolution while offering numerous ways to image local electrical and functional response with high sensitivity. Furthermore, experiments can be performed under dark or light conditions and in a variety of controlled environments. In this article we focus on two emerging systems, hybrid organic-inorganic perovskites and organic semiconductors, to illustrate the power and versatility of AFMs for PV research.
We thank R. Giridharagopal, B. Huey, and H. Phan for valuable discussions and R. Giridharagopal, D. Ginger, A. Gruverman, I. Hermes, B. Huey, J. Huang, F. Jaramillo, D. Moerman, Y. Shao, K. Sivula, S. Weber, and J. Yuan for assistance with figure preparation. The contributions of everyone at Oxford Instruments Asylum Research are also greatly appreciated.
Corresponding author
Dr. Ben Ohler
Oxford Instruments Asylum Research Inc.
6310 Hollister Ave.
Santa Barbara, CA 93117  USA 
Tel: +1 805 696-6466
As global energy needs keep growing, sustainable technologies based on photovoltaic (PV) materials—those that directly convert light to electricity—offer a promising solution. Widespread solar-cell commercialization currently hinges on reducing cost, which requires increasing conversion efficiency, lowering manufacturing costs, and lengthening device life1.
Success in each area depends on improved characterization techniques with higher spatial resolution. This need is driven by increasing use of materials with micro- and nanoscale features such as polycrystals in perovskite films, bulk heterojunction networks in organic semiconductors, and nanotextured light-trapping layers.
With its nanoscale spatial resolution, the atomic force microscope (AFM) provides complementary information to other imaging techniques2 and tools that probe a whole device. Moreover, its ability to measure both structure and functional response enables deep insight into relations between structure, properties, processing, and performance (Figure 1). Here, we explore the power of today’s AFMs, such as the Cypher™ and MFP-3D™ from Oxford Instruments Asylum Research, for characterizing PV materials. Although discussion is limited to perovskites and organic semiconductors, other PV systems can benefit from AFM characterization in analogous ways.
Figure 1. Nanoscale map of short-circuit current ISC overlaid on topography for a methylammonium lead triiodide (CH3NH3PbI3 or MAPbI3) film under ~0.07 W/cm2 illumination. The map was obtained by acquiring images of current with photoconductive AFM (pcAFM) at different bias voltages ranging from 0 to +1 V. The images were then combined to form an I-V curve for each pixel, from which values of ISC were determined. Scan size 3 μm; acquired on the MFP-3D-BIO AFM. Adapted from Ref. 3.
Solar cell technology based on hybrid organic-inorganic perovskites is viewed with great excitement due to rapid gains in efficiency (already at >22%)1,4. In addition, perovskite solar cells can be manufactured using relatively simple, inexpensive solution-processing techniques such as spin coating.
As seen below, AFM techniques can supply important information for current research on fundamental properties and long-term stability5.
Understanding Grain Structure
Evaluating microstructure helps with both fundamental and practical inquiries, such as the highly sensitive dependence of PV response on crystal grain size or how the perovskite crystallizes out of the precursor state during large-scale manufacturing. 
Such questions can be answered with AFM images of three-dimensional surface height, or topography (see examples in Figures 1-4). Topography images reveal film coverage and uniformity, and allow rapid calculation of surface roughness and other metrics. Topography images are acquired in either tapping or contact mode and typically resolve vertical features well below a nanometer. In fact, some current-generation AFMs like the Cypher can achieve vertical resolution down to a few tens of picometers, enabling lattice-scale imaging of crystals and molecules.
Figure 2. The image shows Kelvin probe force microscopy (KPFM) surface potential overlaid on topography for a polycrystalline MAPbI3 (CH3NH3PbI3) film. Correlation with transmission electron microscopy data on crystallographic orientation (not shown) revealed that boundaries between grains with large potential differences (eg, ∆) had higher angles than those with small potential differences (eg, , ). CAFM I-V curves showed strong dark-current hysteresis at the high-angle boundaries but very little at low-angle boundaries. (Blue and red arrows indicate directions of increasing and decreasing voltage, respectively.) The results indicated that current migration was much faster and more dominant at grain boundaries than within grains. Scan size 2 µm; acquired on the MFP-3D AFM. Adapted from Ref. 5.
When perovskites are exposed to ambient conditions, irreversible changes in microstructure or other properties can occur through oxidation or other chemical reactions. Such degradation can be prevented by performing AFM experiments in a purified, inert-gas environment with humidity control using specialized measurement cells. Even more stringent environmental control can be achieved by enclosing the entire AFM in a glovebox for complete atmospheric isolation (see Figure 3 below).
Figure 3. Topography of a MAPbI3 film prepared by solvent annealing reveals micrometer-sized crystalline grains with a terraced structure. (right) Corresponding vertical PFM amplitude image acquired near resonance, revealing domains of regularly spaced stripes. The color-coded PFM line sections show the stripe periodicity varied from approximately 100 to 350 nm. Results suggest the film was ferroelastic, with a structure highly dependent on film texture and thus the specific preparation route. Scan size 7 μm. Acquired on the MFP-3D AFM in a glovebox with nitrogen. Adapted from Ref. 6.
Figure 4. Maps of relative photoluminescence (PL) intensity, topography, CAFM injected current, and FM-KPFM surface potential were acquired for a methylammonium lead tribromide (CH3NH3PbBr3 or MAPbBr3) film deposited on glass/indium tin oxide/poly(3,4-ethylenedioxythiophene) polystyrene (glass/ITO/PEDOT:PSS). The dashed, dotted, and solid curves indicate regions with dim, intermediate, and bright PL response but high, intermediate, and low injected current, respectively, despite similar topography. Moreover, the surface potential image lacked any correlated behavior. Comparison to results for MAPbBr3 on bare glass suggest the heterogeneities arose from effects at the electrode-film interface, not within the film. Scan size 7 μm; acquired on the MFP-3D AFM. Adapted from Ref. 7. 
Sensing Electrical and Functional Response
Of course, most PV mechanisms require data on electrical and related behavior for full understanding, and the polycrystalline structure of perovskite films impels measurements on micro- to nanoscales. Many such processes, including charge transport, trapping, and recombination, can be examined using the AFM’s capabilities for functional imaging.
For example, local variations in photo-induced conductivity and carrier mobility can be mapped with conductive AFM (CAFM), also called photoconductive AFM (pcAFM) when performed under illumination. In CAFM and pcAFM, a conducting tip senses the current flowing from the sample under an applied DC bias voltage.
Images are created by scanning in contact or using a fast-force-mapping approach (see “Imaging Nanoscale Photoresponse” below). Varying bias voltage, tip load, or illumination parameters such as intensity, wavelength, or polarization can impart additional knowledge. Today’s AFMs can perform measurements with high sensitivity and low noise over currents ranging up to six orders of magnitude (picoamperes to microamperes). Photo-induced artifacts can also be reduced by deactivating the AFM’s detection laser during pcAFM measurements.
CAFM and pcAFM can also be used to obtain current-voltage (I-V) curves with nanoscale resolution. The tip is placed in contact at a user-defined position, and the bias voltage is ramped while the current is measured. The resulting I-V curves lend insight into charge generation and injection, contact resistance, and the effects of annealing or other process variables (Figure 2).
Electrostatic force microscopy (EFM) and Kelvin probe force microscopy (KPFM) are other modes to evaluate photoelectric behavior, even within a single grain or grain boundary. Both operate in tapping mode and approximately represent open-circuit behavior. EFM senses electric field variations due to long-range electrostatic force gradients and is often convenient for obtaining qualitative contrast quickly and easily. To minimize crosstalk from topography variations, a dual-pass scanning approach to EFM can be implemented. 
KPFM senses the contact potential difference between the tip and sample (Figure 2 and cover image). A key benefit is its ability to quantitatively measure work function, an underlying cause of potential variations in many PV systems. KPFM images can therefore elucidate band bending, dopant density, and other effects. KPFM is commonly performed in a dual-pass, amplitude-modulated (AM) approach similar to EFM but can also be operated in a single-pass, frequency-modulated (FM) mode. FM-KPFM often gives higher spatial resolution and contains additional information from the cantilever’s higher-harmonic response.
Ferroelectric behavior can also affect perovskite performance, for instance by separating electron-hole pairs more efficiently or creating additional conduction pathways. Piezoresponse force microscopy (PFM) is a powerful technique for characterizing ferroic properties on scales that match domain and grain sizes. It can interrogate both static and dynamic behavior including domain structure, growth, and polarization reversal. On thin films, applying a PFM voltage sufficiently low to avoid polarization switching or even breakdown can yield sub-optimal signal-to-noise ratios. One measurement solution involves operating near the cantilever’s contact-resonance frequency to achieve higher sensitivity at lower drive voltage (Figure 3).
A deeper understanding of PV materials can also be reached with multimodal and correlated approaches. Here, complementary data are obtained with multiple AFM modes and/or other characterization tools such as Raman spectroscopy, scanning and transmission electron microscopy (SEM and TEM, Figure 2), and photoluminescence (PL, Figure 4).
Engineering Interfacial Layers
To improve performance, solar cell designs may include additional layers beyond the basic configuration of an absorber sandwiched between two electrodes. AFM methods are invaluable for characterizing these layers either individually or in combination. Measurements can be performed on entire integrated devices and under realistic operating conditions (illumination, applied bias, etc.) if desired. Devices can be examined in cross section, or in plan view with a conducting tip used as the top contact.
Imaging nanoscale topography of interfacial layers gives information on surface roughness, which affects layer-to-layer adhesion, and reveals morphological features such as phase segregation and dispersion of organic films. CAFM and pcAFM measurements are also useful, for instance to assess conduction uniformity or identify areas of charge trapping or recombination. KPFM characterization is particularly beneficial due to its sensitivity to surface contact potential and work function. Because interfacial layers are often used to create a more favorable route for carriers away from the absorber and toward an electrode, they should improve the alignment of energy levels at each interface. KPFM imaging of spatial variations in band bending and work function can supply helpful feedback for this purpose (Figure 5).
Figure 5. Surface potential images for a MAPbI3 film on NiOx before and after addition of phenyl-C71-butyric acid methyl ester (PC70BM) and rhodamine 101 (Rh) layers. The Rh layer significantly reduced spatial variations in potential by passivating defects at the perovskite grain boundaries. The graph of induced surface photovoltage (ie, light-to-dark difference in surface potential) shows the extra layers decreased surface potential and reduced band bending at the interface between the electron transport layer and cathode. Scan size 1 μm; acquired on the MFP-3D AFM in dual-pass KPFM mode. Adapted from Ref. 8.
Polymers and small organic molecules are also promising candidates for next-generation PV technology. Materials are widely available and relatively eco-friendly, and device fabrication involves inexpensive techniques such as solution processing or vapor deposition. Because commercially-viable efficiencies (>10%) have already been achieved, increasing device lifetime from years to decades is now considered of paramount importance9. Critical to this effort is understanding how performance degrades due to light, heat, and other factors, and AFM characterization can clarify these issues10.
Mapping BHJ Morphology
Organic solar cells typically employ a bulk heterojunction (BHJ) photoabsorber with a nanostructured network of donor and acceptor materials. Efficiency depends strongly on the network’s phase segregation and connectivity, but predicting the specific structure formed by a given processing route remains challenging. Morphological characterization is therefore essential, but options such as electron microscopy often incur sample damage.
AFM topography images reveal the size and dispersion of BHJ components and allow exploration of process variable effects (Figure 6). Topography is usually imaged with tapping mode, which can apply extremely gentle tip-sample forces that minimize sample damage and increase spatial resolution. In fact, forces as low as sub-piconewtons can be resolved and controlled using very small cantilevers on newer, fast-scanning AFMs. As an additional benefit, these AFMs feature automated routines that automatically calculate optimal tapping-mode imaging parameters to make imaging simpler and more reproducible.
Figure 6. Films containing blends of PC70BM and four polymers (PF-0 without fluorine, PF-1a and PF-1b with intermediate fluorine and different regioselectivity, and PF-2 with the most fluorine) were solution-processed using different amounts of the solvent additive DIO (see Ref. 12 for more details). These tapping-mode topography images for the PF-1a blend indicate that low amounts of DIO increased phase separation and thus improved efficiency, but higher amounts yielded sub-optimum morphology. The graph reveals that roughness generally increased with fluorine content in all four blends, likely due to enhanced aggregation. Scan size 5 μm; acquired on the MFP-3D AFM. Adapted from Ref. 11.
BHJ morphology may also be characterized with AFM modes that sense mechanical properties. For example, phase imaging in tapping mode often reveals contrast between blend components and resolves fine structural details. Force curve techniques can also be used to investigate phase separation and dispersion by mapping elastic modulus (Figure 7). Other nanomechanical modes offer both qualitative imaging and quantitative mapping of elastic and viscous response. In particular, newer bimodal tapping techniques such as AM-FM mode enable very fast mapping with high spatial resolution13.
Figure 7. Force-curve modulus maps and topography (insets) of diketo-pyrrolopyrrole-thienothiophene polymer (PDPP4T-TT) and phenyl-C61-butyric acid methyl ester (PCBM) blends with different polymer chain number average molecular weight. Lower and higher modulus values correspond to the PDPP4T-TT and PCBM phases, respectively. Large PCBM domains in a film with intermediate molecular weight indicated a PCBM-rich surface created by vertical segregation during spin casting, which could explain the unusually low series resistance of a transistor made with this film. In other films the phases appeared well intermixed. Scan size 3 μm; acquired on the Cypher AFM. Adapted from Ref. 12.
Imaging Nanoscale Photoresponse
Understanding charge injection, transport, trapping, and recombination in organic semiconductors remain key research priorities to increase efficiency and reduce performance deterioration. AFM imaging of nanoscale photoresponse serves to elucidate these processes and pinpoint where in the BHJ each one occurs.
Imaging organic semiconductors with CAFM and pcAFM allows nanoscale visualization of photocurrent and charge transport networks in the donor-acceptor blend. These images help to determine the role that structural anisotropy, light intensity, and other parameters play in photoconversion. However, the relatively delicate nature of these materials makes them prone to damage from lateral forces when using conventional contact-mode CAFM. Moreover, tip wear from contact-mode scanning can affect the measured current and complicate image interpretation. 
Recently developed fast current mapping techniques can prevent such issues. Fast current mapping uses a fast-force-curve approach, in which the cantilever is moved vertically in a continuous sinusoidal motion while it is also scanned laterally. Current is measured during high-speed acquisition of force curve arrays to obtain spatially correlated topographical and electrical results, often taking <10 min for 256×256 pixels. These techniques also present many options for data analysis, providing that complete curves of current and deflection versus time are saved.
Electrical characterization with EFM and KPFM also gives insight into device performance and long-term stability. The noncontact nature of these modes minimizes energy-barrier effects created by the tip’s work function, so that measurements represent the system’s open-circuit response. EFM and KPFM images acquired with dual-pass scanning typically take a few minutes to acquire and are thus suited to studying processes occurring over hours or days. Faster processes with timescales of milliseconds to seconds can be examined with modes such as cantilever ringdown imaging, time-resolved EFM, and heterodyne KPFM14,15. Although not standard on commercial AFMs, these techniques highlight the power of open software platforms such as those on all Asylum AFMs.
Photoelectrical experiments are both simplified and enhanced on AFMs equipped with sample illumination capabilities. Pre-integrated optical components make it virtually effortless to explore light intensity, polarization, and wavelength effects. A turnkey photovoltaics research package for the MFP-3D Infinity AFM facilitates customizable bottom-side sample illumination for immense flexibility in designing experiments (Figure 8).
Figure 8. The sample was a layer of poly(3-hexylthiophene) and phenyl-C61-butyric acid methyl ester (P3HT:PCBM) bulk heterojunction annealed on an indium tin oxide (ITO) substrate. During current imaging at -1 V bias, the 530-nm illumination source was turned on and off while increasing the intensity in 1% increments (full power ~0.9 W/cm2). The vertical section through the image reveals the dependence of measured current on intensity and demonstrates high sensitivity to small changes in intensity. Acquired on the MFP-3D Infinity AFM with the Photovoltaics Option and CAFM holder.
Optimizing Interlayers
Organic solar cells often incorporate extra layers for extracting and receiving charge and controlling surface recombination. AFM characterization of these layers can prove beneficial, for instance, to evaluate BHJ morphological changes induced by addition of interlayers, which can impact carrier recombination efficiency16. Likewise, EFM and KPFM imaging across interfaces can inform the design of interlayers that better align energy levels from the photoabsorber to the electrodes.
Layers may also improve long-term device stability, for instance by inverting geometries or by complete encapsulation. In stability and lifetime studies, environmental control is often desirable or even critical. It permits exposed devices to be surrounded by an inert gas and allows experiments to be performed under realistic or enhanced humidity. Temperature is also controlled by the user. Newer AFMs offer specialized sample stages that provide stable, precise temperature control up to several hundred degrees.
Photovoltaic technologies are already helping to meet ever-increasing energy demands, and devices based on perovskites and organic semiconductors offer further promise. Realizing a future of plentiful, low-cost renewable energy is within reach but requires improved characterization of next-generation PV materials. Today’s AFMs feature numerous techniques for evaluating the nanoscale structure and functional response of PV systems, both in the dark and under variable illumination. Combined with higher spatial resolution, faster imaging, and greater environmental control, these benefits make AFMs indispensable for enlightening and empowering photovoltaics research.
1. A. Polman, M. Knight, E. C. Garnett, B. Ehrler, and W. C. Sinke, Science 352, aad4424 (2016).
2. E. M. Tennyson, J. M. Howard, and M. S. Leite, ACS Energy Lett. 2, 1825 (2017).
3. Y. Kutes, Y. Zhou, J. L. Bosse, J. Steffes, N. P. Padture, and B. D. Huey, Nano Lett. 16, 3434 (2016).
4. J. Li, B. Huang, E. N. Esfahani, L. Wei, J. Yao, J. Zhao, and W. Chen, npj Quantum Materials 2, 56 (2017).
5. Y. Shao, Y. Fang, T. Li, Q. Wang, Q. Dong, Y. Deng, Y. Yuan, H. Wei, M. Wang, A. Gruverman, J. Shield, and J. Huang, Energy Environ. Sci. 9, 1752 (2016).
6. I. M. Hermes, S. A. Bretschneider, V. W. Bergmann, D. Li, A. Klasen, J. Mars, W. Tremel, F. Laquai, H.-J. Butt, M. Mezger, R. Berger, B. J. Rodriguez, and S. A. L. Weber, J. Phys. Chem. C 120, 5724 (2016).
7. D. Moerman, G. E. Eperon, J. T. Precht, and D. S. Ginger, Chem. Mater. 29, 5484 (2017).
8. J. Ciro, S. Mesa, J. I. Uribe, M. A. Mejia-Escobar, D. Ramirez, J. F. Montoya, R. Betancur, H.-S. Yoo, N.-G. Park, and F. Jaramillo, Nanoscale 9, 9440 (2017).
9. J. R. O'Dea, L. M. Brown, N. Hoepker, J. A. Marohn, and S. Sadewasser, MRS Bull. 37, 642 (2012).
10. M. Pfannmoeller, W. Kowalsky, and R. R. Schroeder, Energy Environ. Sci. 6, 2871 (2013).
11. J. Yuan, M. J. Ford, Y. Zhang, H. Dong, Z. Li, Y. Li, T.-Q. Nguyen, G. Bazan, and W. Ma, Chem. Mater. 29, 1758 (2017).
12. A. Gasperini, X. A. Jeanbourquin, and K. Sivula, J. Polym. Sci., Part B: Polym. Phys. 54, 2245 (2016).
13. M. Kocun, A. Labuda, W. Meinhold, I. Revenko, and R. Proksch, ACS Nano 11, 10097 (2017).
14. R. Giridharagopal, P. A. Cox, and D. S. Ginger, Acc. Chem. Res. 49, 1769 (2016).
15. J. L. Garrett, E. M. Tennyson, M. Hu, J. Huang, J. N. Munday, and M. S. Leite, Nano Lett. 17, 2554 (2017).
16. T.-H. Lai, S.-W. Tsang, J. R. Manders, S. Chen, and F. So, Materials Today 16, 424 (2013).

Figure 6. Films containing blends of PC70BM and four polymers (PF-0 without fluorine, PF-1a and PF-1b with intermediate fluorine and different regioselectivity, and PF-2 with the most fluorine) were solution-processed using different amounts of the solvent additive DIO (see Ref. 12 for more details). These tapping-mode topography images for the PF-1a blend indicate that low amounts of DIO increased phase separation and thus improved efficiency, but higher amounts yielded sub-optimum morphology. The graph reveals that roughness generally increased with fluorine content in all four blends, likely due to enhanced aggregation. Scan size 5 μm; acquired on the MFP-3D AFM. Adapted from Ref. 11.

Website developed by S8080 Digital Media