Subscriber access provided by University of South Dakota
Article
In-situ measurement of electric-field screening in hysteresisfree PTAA/FA Cs Pb(I Br )/C60 perovskite solar cells gives an ion mobility of ~3 x 10 cm/Vs; two orders of magnitude faster than reported for metal-oxide-contacted perovskite cells with hysteresis. .83
.17
.83
-7
.17
3
2
Luca Bertoluzzi, Rebecca A. Belisle, Kevin A. Bush, Rongrong Cheacharoen, Michael D. McGehee, and Brian C. O'Regan J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.8b04405 • Publication Date (Web): 06 Sep 2018 Downloaded from http://pubs.acs.org on September 6, 2018
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Journal of the American Chemical Society
In-situ measurement of electric-field screening in hysteresis-free PTAA/FA.83Cs.17Pb(I.83Br.17)3/C60 perovskite solar cells gives an ion mobility of ~3 x 10-7 cm2/Vs; two orders of magnitude faster than reported for metal-oxide-contacted perovskite cells with hysteresis. Luca Bertoluzzi1, Rebecca A. Belisle1, Kevin A. Bush1, Rongrong Cheacharoen1, Michael D. McGehee1, Brian C. O'Regan*2 1) Department of Materials Science and Engineering, Stanford University, 476 Lomita Mall, Stanford, California 94305, USA 2) Sunlight Scientific, Berkeley CA, USA *
[email protected] Abstract We apply a series of transient measurements to operational perovskite solar cells of the architecture ITO/PTAA/FA.83Cs.17Pb(I.83Br.17)3/C60/BCP/Ag, and similar cells with FA.83MA.17. The cells show no detectable JV hysteresis. Using photocurrent transients at applied bias we find an ~1 ms time scale for the electric field screening by mobile ions in these cells. We confirm our interpretation of the transient measurements using a drift-diffusion model. Using Coulometry during field relaxation at short circuit, we determine the mobile ion concentration to be ~1 x 1018 /cm3. Using a model with one mobile ion species, the concentration and the screening time require an ion mobility of ~3 x 10-7 cm2/Vs. As far as we know, this article gives the first direct measurement of the ion mobility and concentration in a fully functional perovskite solar cell. The measured ion mobility is 2 orders of magnitude higher than the highest estimates previously determined using perovskite solar cells and perovskite thin films, and 3 orders of magnitude higher than is frequently used in modeling hysteresis effects. We provide evidence that the fast field screening is due to mobile ions, as opposed to dark injection and trapping of electronic carriers. Introduction Organic Lead Halide Perovskites (OLHPs) have become a promising photovoltaic material. Among their positive aspects are low cost, low temperature fabrication, and light weight (due to the high absorption coefficient). The efficiencies of cells made with these materials have jumped from negligible to 19.6% on one square centimeter in less than 10 years.1-4 Amongst these materials, a current front runner is Formamidinium Lead Iodide (FAPI), improved by partial substitution of methyl ammonium (MA) or cesium on the formamidinium site and bromide on the iodide site.3,5 One possibly negative aspect of the organic lead halide perovskites is the low energy of formation and diffusion of interstitial ions and ion vacancies.6-9 Mobile ions are thought to be responsible for the large hysteresis seen in the current-voltage (JV) scans of some cells.7,10-13 The degree of hysteresis found in a given cell varies from very large to effectively zero depending on the material, fabrication, electrode composition, and age of the cell. Specifically, cells with organic hole collecting and electron collecting electrodes are frequently reported as hysteresis free.14-18 Examples include PEDOT and PTAA as hole collectors and PCBM and C60 as electron collectors. Recently some non-fullerene organic electron collection layers have also been shown to give hysteresis free cells, indicating whatever effects are present are not unique to fullerenes.19,20 Given that ion movement is thought to be a contributor to the hysteresis, it has been suggested that fullerenes and other organic molecules can block ion movement in the perovskite.13,21-24 However, in some cases, nominally hysteresis-free OLHP cells show strong hysteresis at low temperature or in the initial JVs.25-27
1
ACS Paragon Plus Environment
Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Many techniques have been used to gain insight into the possible ion motion, and its effect on cell efficiency.12,13,26,28-34 Evidence is accumulating that the mobile charged species is an iodide vacancy,31,35-37 but the case against other possibilities (e.g. iodide interstitials) is not quite closed. Because our analysis does not depend on the identity or sign of the mobile species, we will simply use "mobile ion" to identify the mobile charged atomic species. We will refer to "electronic" processes to indicate electrons and holes that move without the accompaniment of net atomic motion. We have recently developed several tools for investigating the effects of ion motion inside the cell, specifically the degree to which the moving ions accumulate near the electrodes and screen the built in electric field and any applied potential.38,39. In our previous paper we used the Step-dwellstep-probe (SDSP) photocurrent transient method to examined the time evolution of the field screening under applied forward bias in perovskite cells showing large hysteresis.38 However, at forward bias there is always a dark electronic current that flows, as well as the "displacement" current caused by the ion movement. (Details of the displacement current are given below.) We and others have found that this electronic dark current is not fixed, but can vary significantly during the ion movement.40 This renders inaccurate the standard method of finding the ionic current by subtraction of the long time "equilibrium" electronic current from the total current.41 Without measurement of the net ionic current and charge, in the previous paper we could not determine the ion mobility and could give only a lower bound for the mobile ion density. In this paper we extend the SDSP method to examine the re-equilibration of the field screening at short circuit, after the ions have equilibrated for some time at a given forward bias. This is important because at short circuit there is no dark electronic injection current, therefore we can assign all the measured current (after electrode capacitive discharge) to the ions. The new SDreSP experiment allows us to directly determine the mobile ion concentration and the mobile ion mobility. Also, with the ability to plot the change in field screening vs the ion charge moved, we can confirm that a model with one mobile ion and one fixed ion is appropriate. From these measurements on ITO/PTAA/FA.83Cs.17Pb(I.83Br.17)3/C60/BCP/Ag cells, we find an ion concentration of 1 x1018/cm3 and a mobility of 3 x10-7 cm2/Vs. As far as we know, this article gives the first direct measurement of the ion mobility and concentration in any fully functional perovskite solar cell. Our measured mobility is several orders of magnitude larger than that estimated for cells with large hysteresis8,39,40 and also much larger than mobilities derived from lateral measurements on thin films.42,43 Faster ion movement in hysteresis free cells has been previously suggested from the temperature dependence of the JVs, but without quantitative values.27,44 Experimental ITO-coated glass, 10 Ω / (Xin Yan Technology), was cleaned by sequential sonication in diluted Extran solution (EMD, EX0996-1), acetone, and isopropanol followed by UV ozone irradiation for >15 minutes. Stoichiometric perovskite solutions were made of both Cs0.17(CH(NH2)2)0.83Pb(I0.83Br0.17)3 and (CH3NH3)0.17(CH(NH2)2)0.83Pb(I0.83Br0.17)3 by combining lead (II) iodide (TCI, 99.99%), lead (II) bromide (TCI, 99.99%), methylammonium iodide (Dyesol) formamidinium iodide (Dyesol), and cesium iodide (Sigmap-Aldrich, >99.99%) in 1.2M concentrations in a 4:1 mixture of anhydrous N,N-dimethylformamide (Sigma-Aldrich, 99.8%) and dimethyl sulfoxide (Sigma-Aldrich, >99.9%) in a dry nitrogen atmosphere. All solutions were filtered through a 0.2um PTFE to remove any dust or impurities before use. A 5 mg/mL solution of poly(triaryl amine) (Solaris) in anhydrous chlorobenzene (Sigma-Aldrich, 99.8%) was deposited as a hole-transport material onto the cleaned ITO substrates. The solution was spun coat at 6000 rpm for 30 s before annealing at 100 C for 5 minutes in a dry-air environment (thickness ~15 nm). Next, the perovskite solution was deposited onto the samples via spin coating using the following
2
ACS Paragon Plus Environment
Page 2 of 21
Page 3 of 21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Journal of the American Chemical Society
protocol: 1000 rpm for 10 s; 6000 rpm for 25 s; 1000 rpm for 5 s with dry nitrogen being blown normal to the sample surface (thickness ~400 nm). After deposition samples were annealed at 60 C for 5 s and 100 C for 15 min. Electron transport layers and metal contact were deposited on top of the perovskite by thermal evaporation of 20 nm of C-60 (MER, 99.8%), 8 nm of bathocuproine (Sigma-Aldrich, 96%), and 100 nm of silver (Kurt J. Lesker). Scanning electron microscopy images were used to estimate the thickness of the perovskite layer (supplementary information photo S1). Images were collected on a FEI XL30 Sirion scanning electron microscope with a field-emission gun electron source. Current–voltage (JV) measurements were done using a Keithley 2400 sourcemeter and an Oriel solar simulator with 300W xenon lamp source. The light was calibrated for 1 suns of AM1.5G light intensity immediately before testing. Devices were illuminated through a 0.12 cm aperture to ensure accurate device area for efficiency calculations. Additional JV measurements were done under white LED illumination at ~0.7 sun equivalent intensity. For the transient measurements, the cell was kept in an airtight holder containing desiccant. Thus the cell was isolated from water but not from oxygen. The cells were mostly kept in the dark. In this environment the cells lasted about 1 month before detectable degradation occurred. Photocurrent transients and chrono-amperometry were measured on a custom built TRACER system.45 Light was provided by a selection of white or NIR (735 nm) LEDs. LED intensity was controlled GPIB controllable power supplies. LEDs were switched on and off by MOSFET switches with 0.8 will be required to re-establish an inverted potential gradient in region II, and thus a negative transient (Scheme SI_1d). Figure 2b shows that after 5 ms at 0.8 V it requires Vpr = 1.05 V to create a Jtr-6µs = 0. a 5m 2
Jtr-6µs, µA/cm
2
Jtr-6µs, µA/cm
ms
-100
µs
0
200
50
100
25 0
0 30 µs 100 µs 250 µs 500 µs 1 ms 2.5 ms 5 ms 10 ms 50 ms
s
200
100
0
-100
-200 -300 0.0
0.2
b
300
300
s 30 µ
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 8 of 21
Vo(td=0)
Vo(td=∞)
0.4 0.6 0.8 Probe Voltage, V
1.0
-200
1.2
Re-Equil. Time 0 µs 30 µs 100 µs 250 µs 500 µs 1 ms 2.5 ms 5 ms 0.0
0.2
0.4 0.6 Probe Voltage, V
0.8
1.0
Figure 3. a) Full results of an SDSP experiment carried out with a dwell voltage of 0.8 V. Legend gives the dwell time for each Jtr-6µs(Vpr) curve. b) Results of an SDreSP experiment with 10 ms dwell at 0.8 V followed by re-equilibration times at SC as shown in the legend. To get a complete picture of the movement of the ions at a given Vd, we repeat the SDSP experiment for a range of dwell times, with the full range of Vpr for each dwell time. Figure 3a shows the Jtr(Vpr) data from an SDSP experiment carried out with Vd = 0.8 V, and dwell times between 0 and 500 ms. Each data point represents Jtr-6µs for a single transient at a given dwell time and probe voltage. Lines connect points of identical dwell time, highlighting the Jtr-6µs(Vpr) curves. The Jtr-6µs(Vpr) curves for dwell times 100 to 500 ms are omitted from fig. 3 as they are identical to 50 ms. To check for cell stability, a dwell time of zero was measured at the start, middle, and end of the experiment (fig. SI_3a). The cell characteristics did not change during the experiment. As mentioned above, the shift along the x axis corresponds to a decrease in the inverted potential drop across region II, as the ions move to screen the applied potential. We can estimate the amount of inverted potential drop remaining at any dwell time, ∆VRII(td), using formula (1)
8
ACS Paragon Plus Environment
Page 9 of 21
1) ∆VRII(td) = Vo(∞) - Vo(td) where: Vo(td) is the x-intercept of the Jtr-6µs(Vpr) curve at td and Vo(∞) is the x axis intercept at long dwell times when the inverted potential drop is completely screened. Figure 4 shows ∆VRII(td) for the data in fig. 3a (red solid line) and for another similar cell (green solid line). Figure 4 inset shows the same data on a log time plot and gives the 1/2 decay times. Using fig. 4, we can see directly how fast the field screening changes inside the cell, and thus infer how fast the ions move. All PTAA/absorber/C60 cells we have measured fall in the range shown in fig. 4; mainly near the faster of the two cells. We have confirmed that the decays shown in figure 4 are independent of the pulse intensity (figure SI_3b). The decays shown are also independent of the part of the photocurrent transient (e.g. 2-4 µs, 4-8 µs, or 10-14 µs) averaged to give the Jtr magnitude used in figure 3 (figure SI_3c). Cells with MA+ instead of Cs+ show the same timescales for charges in field screening (Fig. SI_3d). Potential Change Across RII (∆VRII), V
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Journal of the American Chemical Society
Decay Halftime 0.8 ms 1.1 ms 0.22 ms 0.32 ms
0.6
0.6
0.4
0.2
0.4 0.0 10
-5
10
-4
10
-3
10
-2
0.2
0.0 0
2
4 Time, ms
6
Figure 4. Temporal evolution of the potential difference across region II after a potential step. Data from two different perovskite cells with the same composition; ITO/PTAA/FA.83Cs.17Pb(I.83Br.17)3/C60/BCP/Ag. Solid lines, decay of the inverted potential drop over RII after a step from SC to 0.8 V. Dashed lines, decay of the excess potential drop over RII at SC, after a dwell of 10 ms (red) and 100 ms (green) at 0.8 V. Y axis is the absolute value of ∆VRII(td) from equation 1. Inset: Same data plotted on a log time scale. In the inset, the reequilibration data at SC has been normalized (~10%) to start at the same ∆VRII(0) as the solid lines. In the inset, the dashed lines are single exponential fits to the data. As mentioned in the introduction, there is a disadvantage to the SDSP experiment. At an applied forward bias there is always an electronic dark current which flows in addition to the displacement current created by the moving ions. In most perovskite cells we have measured, the electronic dark current changes significantly during the time that the ions are equilibrating to the new potential. Thus, it is not possible to use the final steady state dark current as a baseline to subtract from the total current to accurately determine the ion current. Because of this, the total ionic charge moved during the field screening cannot be determined at forward bias. Without the charge, the ionic concentration and ionic mobility cannot be determined. This limitation can be overcome by first allowing the ions to equilibrate to some forward bias potential (e.g. 0.8 V) and then observing the re-equilibration of the ions after a step back to short circuit (SC). When the cell is returned to SC 9
ACS Paragon Plus Environment
Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
after a long dwell at 0.8 V, the flat potential profile across RII at 0.8 V becomes a positive slope at SC. This slope causes the ions to drift back towards their equilibrium position at SC. At short circuit there is no dark injection current. The discharge current of the electrode capacitance will dominate the first few µs after the step to SC, but after that time all of the measured current can be assigned to ion displacement current. By integrating the dark current after the electrode discharge we can measure the amount of ion charge that has moved as a function of re-equilibration time. To measure the changes in electric field screening during re-equilibration at SC we use the Step Dwell re-equilibrate Step Probe (SDreSP) experiment (Scheme II). In this experiment we first place the cell at forward bias (e.g. 0.8 V) for the chosen dwell time (tdw). We then step the cell to SC for a variable "re-equilibration" time (tre) after which we step the potential to a probe voltage and measure the photocurrent transient. Following the same logic as the SDSP experiment, we repeat the experiment many times with the same dwell time and dwell voltage, and a given reequilibration time. For each of a series of re-equilibration times, we repeat the whole experiment for a series of probe voltages. The cell is placed at dark SC for ≥ 1 minute between each repeat. A set of such Jtr(Vpr) are shown in figure 3b. For example, using 3 different dwell times, 15 reequilibration times, and 15 probe voltages, the experiment requires 625 transients, which are then automatically analyzed to create the Jtr(Vpr) curves. Figure 3b shows one set of Jtr-6µs(Vpr) curves taken as a cell re-equilibrates to SC after a 10 ms dwell at 0.8 V. Representative photocurrent transients are shown in fig. SI_4. As expected, the Jtr6µs(Vpr) curves shift back to the left for increasing re-equilibration times. The shift along the x-axis back towards the SC equilibrium is plotted as the red dashed line in fig. 4. The same data for another cell is plotted in green using a 100 ms dwell time at 0.8 V. The dwell times for each cell were chosen so that >95% of the equilibration to forward bias had occurred. We find that the reequilibration decay time is nearly independent of the preceding dwell time at forward bias, as expected for a population of equivalent mobile ions (fig. SI_4c). For each cell, the decay halftimes for equilibration to 0.8 V and re-equilibration to SC are quite similar, however the shapes of the decays are different. The re-equilibration to SC is well fit by a single exponential, whereas the equilibration to forward bias has a second slower component (fig. 4 inset.) This is true consistently, even for metal oxide/OLHP/Spiro cells we have measured. We discuss the implications of this second slower component at the end of this section. To verify that our interpretation of the photocurrent transients is reasonable, we have used a driftdiffusion model of the cell to simulate the transients in fig. 2a. The results of our simulations are shown in fig. 5. The model assumes a mobile positive defect (e.g. an iodide vacancy) and a negative fixed defect (e.g. an interstitial iodide or methylammonium vacancy). The model results are not changed by assuming a negative mobile defect and fixed positive defect. The mobile ion distribution is allowed to equilibrate to the built in potential, at short circuit in the dark, before the dwell voltage is applied. Further details are given in the supplementary information. The simulations with one mobile defect reproduce the measured transients quite well with the exception that the simulations show a longer photocurrent decay after the end of the pulse (fig. 5a). Simulations without a mobile defect do not show negative photocurrent at any point. In the measured data, the rapid post-pulse decay of the photocurrent is most likely due to charge trapping,38 which is not included in the model. Also, the model runs shown in fig. 5 use perfectly selective contacts (no surface recombination). Although these simulations indicate inverted photocurrent transients do not require imperfect contacts, they do not prove surface recombination is unimportant in the actual cell. The simulations also reproduce the Jtr-6µs(Vpr) curve quite well, including the x-axis intercept and the sloping plateau at higher forward bias (fig. 5b). However, the slope of the simulated Jtr-6µs(Vpr) at the x-intercept is significantly larger than that of the measured Jtr-6µs(Vpr). This can be explained by the polycrystalline nature of the perovskite film. The simulations predict that changing the ion concentration or the hole and electron mobilities can both shift the x-axis intercept of the Jtr-6µs(Vpr) curve (fig SI_8). It is thus possible to reproduce the
10
ACS Paragon Plus Environment
Page 10 of 21
Page 11 of 21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Journal of the American Chemical Society
slope of the measured Jtr-6µs(Vpr) with the sum of several crystallites containing different charge mobilities and/or mobile ion concentrations. An example is shown in fig. 5b using 5 different ion concentrations. Overall, the simulations show that mobile ions can be the cause of the measured nega tive transients.
11
ACS Paragon Plus Environment
Journal of the American Chemical Society
s
400
Ions at dark SC position
300
a
b
300 200
0.1 V Jtr-6µs, µA/cm
2
2
Current, µA/cm
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 12 of 21
200 0.4 V
100 0
100 0
-100
0.45 V
-100 -200
-200
0.7 V 0
20 Time, µs
Data Simulation Simulation. Avg. 5 Ion Conc.
-300
40
0.0
0.4 0.8 Probe Voltage, V
1.2
Figure 5 a) Simulated photocurrent transients at forward bias (dashed) compared to the measured data from fig. 2. b) Simulated Jtr-6µs vs. Vpr results (blue) compared to the measured data (red) from fig. 4, for the case with no dwell time at 0.8 V. An equilibrium ion concentration ~4 x 1018/cm3 and a hole and electron mobility of 12 cm2/Vs was used. The green dotted line is the weighted sum of 5 crystallites simulated with different mobile ion concentrations. The key message from figures 2-5 is that mobile ions are present in cell architectures that have been called "hysteresis free". Figure 4 shows that in PTAA/OLHP/C60 cells the mobile ions can achieve 90% equilibration to a change in voltage in as little as 3 ms. In order to see significant JV hysteresis in such a cell, the scan rate would have to be more than 500 V/s. The ions in PTAA/OLHP/C60 cell thus can move at least two orders of magnitude faster than they appear to in TiO2/OLHP/Spiro cells where seconds to minutes are required for equilibration.10,12,38,49,50 We can also use the SDreSP experiment to estimate the concentration of mobile ions in the absorber layer. During the re-equilibration of the ion distribution to SC, after equilibration to 0.8V, we can measure a dark current in the external circuit (Figure SI_5a). After the capacitive discharge of the electrodes (
