We present a device simulation of lead-halide perovskite-based thin film transistors (TFTs) containing mobile charged species to provide physical reasoning for the various experimental reports. We study the output characteristics for a range of scan duration (1/speed), average mobile ion densities, and N- and P-channel TFTs. We then directly compare our results to published data by Zeidell et al. [Adv. Electron. Mater. 4(12), 1800316 (2018)] and show that if the transistor’s measurement procedure is such that the ions’ effects are apparent, and then, our model can resolve the sign of the mobile ions in their MAPbI3−xClx TFTs (cations) and provide a good estimate of their density (∼1017 cm−3 at 200 k). Interestingly, we find that effects previously associated with channel screening are due to the ion-blocking of the charge extraction and that the incomplete saturation often reported is due to ion-induced channel shortening. Utilizing the same perovskite materials as in solar cells would allow researchers to improve their understanding of the mechanisms governing solar photovoltaics and improve their performance.

Lead-halide perovskites have attracted significant attention over the last decade,1 particularly in light-harvesting,2 with the field of halide perovskite photovoltaics rapidly growing ever since they broke the 10% efficiency barrier in solid-state solar cells.3 Since the groundbreaking works demonstrating the potential of perovskite materials for photovoltaics,3–5 the field has advanced rapidly with current record efficiencies exceeding 25% within less than a decade.6 Besides the efficiency benchmarking, there have also been advances in understanding these cells' chemistry, physics, and device chemical-physics from both the theoretical and experimental sides.7–18 Historically, the issue of mobile charge species was highlighted through the reports of hysteresis in the current–voltage characteristics as the voltage was scanned forth and back (or vice versa) and its dependence on the scan speed.9,19 Identifying these mobile species is a complex task and subject to extensive debate.20–29 Still, most researchers consider iodine the most prominent mobile species. Here, the literature is divided into those considering the motion within the crystal lattice, described as vacancy transport, or outside the crystal lattice, described as interstitial motion.13,15,30–33 In parallel to the extensive study of solar cells, lead-halide perovskite thin film transistors (TFTs) are being developed.34–49 Reports on perovskite TFTs show dominant ionic effects causing large hysteresis, gate voltage screening, device degradation, and lack of saturation in the current.34–49 Several attempts to minimize these effects are made, such as measurements in cryogenic conditions to reduce ionic conductivity or pulsed mode measurements to minimize the slow ionic response. Still, most of the experimental data on perovskite TFTs show a lack of saturation in the current.18–32 While our citation list does not cover all the papers published on perovskite-based TFTs, it captures a significant fraction.

Namely, the field is still in its infancy, with the entrance barrier being the ability to control the ion-related effects. This paper presents 2D device simulations of perovskite-based TFTs, including a single-type charged mobile species. We chose the mobile species to be anions (i.e., negatively charged) as an example of iodide. We found that the presence of the mobile species creates a difference between electron-channel (negative channel) and hole-channel (positive channel) TFTs. For example, an electron-channel will exhibit higher effective mobility if the dominant mobile ion is an anion (negative). Building on the importance of the JV scan protocol for solar cells, we devote significant attention to the role of the voltage scan speed. As we will discuss, the various effects can help decipher if the main mobile species is, indeed, an anion or perhaps it is a cation.

To analyze and predict the performance of halide perovskite field-effect transistors, we have set up a device simulation using Sentaurus Device by Synopsys®, and the mobile ions were implemented using the hydrogen diffusion model.18,50,51 The parameters needed to run such a simulation are listed in Table I, and the device structure is shown in Fig. 1(a) (L = 5 μm). The parameters of the perovskite layer used in the simulation refer to MAPbI3 (MAPI) perovskite, a commonly used perovskite compound with optimal photophysical characteristics.52 The gate oxide layer parameters refer to a bi-layer of aluminum nitride (AlN) and hafnium oxide (HfO2) insulator layers. In the simulations, we assume an ideal interface between the layers; hence, no defects or reactions are formed at the interfaces. It is worth noting that although the parameters listed in Table I are for specific materials, still the simulations are general in terms of charge redistribution, ionic behavior, and energy band offsets; thus, we believe that the conclusions addressed in the paper can be generalized for all perovskite TFTs.

FIG. 1.

(a) Schematic description of the transistor structure used in this study. (b) Output characteristics of hole-channel (VDS < 0) and electron-channel (VDS > 0) transistors that contain no ions. The threshold voltage was determined separately as −0.77 and +0.77 V for the hole and electron channels, respectively. The current is divided by the channel’s width in μm (L = 5 μm).

FIG. 1.

(a) Schematic description of the transistor structure used in this study. (b) Output characteristics of hole-channel (VDS < 0) and electron-channel (VDS > 0) transistors that contain no ions. The threshold voltage was determined separately as −0.77 and +0.77 V for the hole and electron channels, respectively. The current is divided by the channel’s width in μm (L = 5 μm).

Close modal
TABLE I.

Parameters required to reproduce the device simulations.

Perovskite layerGate oxide layer
Thickness 100 nm 25 nm 
Dielectric constant ε 25 (Ref. 5313.4 (Refs. 54 and 55
Mobility μe = μh 2 cm2  V−1 s−1 (Ref. 56⋯ 
Conduction effective density of states 7 × 1018 cm−3 (Ref. 576.3 × 1018 cm−3 (Ref. 54
Valence effective density of states 2.5 × 1018 cm−3 (Ref. 574.8 × 1020 cm−3 (Ref. 54
Bandgap, Eg 1.6 eV (Ref. 586.2 eV (Ref. 54
Ions’ average density 1018 cm−3 ⋯ 
Anion diffusion, D 10−9 cm2 s−1 (Ref. 33⋯ 
Cation diffusion, D ⋯ 
Perovskite layerGate oxide layer
Thickness 100 nm 25 nm 
Dielectric constant ε 25 (Ref. 5313.4 (Refs. 54 and 55
Mobility μe = μh 2 cm2  V−1 s−1 (Ref. 56⋯ 
Conduction effective density of states 7 × 1018 cm−3 (Ref. 576.3 × 1018 cm−3 (Ref. 54
Valence effective density of states 2.5 × 1018 cm−3 (Ref. 574.8 × 1020 cm−3 (Ref. 54
Bandgap, Eg 1.6 eV (Ref. 586.2 eV (Ref. 54
Ions’ average density 1018 cm−3 ⋯ 
Anion diffusion, D 10−9 cm2 s−1 (Ref. 33⋯ 
Cation diffusion, D ⋯ 

For easy comparison between electron and hole-channel TFTs, we ensured that the threshold voltage was similar. This was achieved by positioning the gate work function energy at the center of the perovskite gap and setting the source/drain contacts to have a 0.1 eV injection barrier to the relevant band. This small injection barrier, in addition to the ideal contact condition, eliminates the contact resistivity.

A schematic description of the transistors' structure is shown in Fig. 1(a) (L = 5 μm). As a reference to the perovskite transistors, we present in Fig. 1(b) the output characteristics of transistors without mobile species; namely, these reference transistors represent ideal crystals without ionic defects or mobile ions in the layer. We first simulated the transfer characteristics for the electron-channel and the hole-channel transistors (not shown). We extracted a threshold voltage of | V T | = 0.77 V and used it to compute the output characteristics as a function of gate bias [see Fig. 1(b)]. Since the only difference between the electron-channel and hole-channel transistors is the effective density of states of the respective band, the output characteristics are almost perfectly symmetric. We note that for | V G | = | V T | + 1, the saturation current is 1.1 × 10 7 A μ m 1 ( A μ m 1 means the current is divided by the channel's width, which is equal to 1 μm).

As mentioned above, our motivation is to provide physical reasoning for published results and to show that transistors could also be instrumental in probing ionic effects. To cover the various published experimental results as much as possible, we vary the ion density and the drain–source bias (VDS) scan rate. Varying the scan rate is important as the scan time dictates how much the slow ions can redistribute during the scan. Figure 2 presents the simulated output characteristics of electron-channel and hole-channel perovskite TFTs (i.e., ions included) for different scan durations. The different sub-figures are for varying average ion densities of 1016 (a), 1017 (b), 1018 (c), and 1019 cm−3 (d). These densities match the reported range of mobile ion densities in perovskite devices.31,59–63 As mentioned before, the mobile ions’ origin is ionic defects in the perovskite crystal; thus, the ion density varies for different perovskite layers depending on the materials forming the crystal (different defect formation energies) and the quality of the deposited perovskite layer.31 The results shown are for the case where the transistors are stabilized with the gate-source bias at ( | V G S | = | V T | + 1 = 1.77 V ), and at t = 0, the drain bias scan starts. The effect of the ion motion is demonstrated by varying the scan duration using 100 s (red), 1 s (green), 10 ms (blue), and 100 μs (orange) scan times.

FIG. 2.

Output characteristics of hole-channel (VDS < 0) and electron-channel (VDS > 0) transistors for several scan times (noted in the figure). The transistors were first stabilized at VGS = −1.77 V and VGS = +1.77 V for the hole and electron-channel TFTs, respectively. At t = 0, the drain voltage sweep was started. The threshold voltage for transistors without ions was determined separately as −0.77 and +0.77 V for the hole and electron channels, respectively. The average ion density in the perovskite layer was 1016 (a), 1017 (b), 1018 (c), and 1019 cm−3 (d).

FIG. 2.

Output characteristics of hole-channel (VDS < 0) and electron-channel (VDS > 0) transistors for several scan times (noted in the figure). The transistors were first stabilized at VGS = −1.77 V and VGS = +1.77 V for the hole and electron-channel TFTs, respectively. At t = 0, the drain voltage sweep was started. The threshold voltage for transistors without ions was determined separately as −0.77 and +0.77 V for the hole and electron channels, respectively. The average ion density in the perovskite layer was 1016 (a), 1017 (b), 1018 (c), and 1019 cm−3 (d).

Close modal

The birds-eye view of the sub-figures of Fig. 2 reveals that the average ion density of 1016 cm−3 hardly affects the TFT characteristics and, hence, can be considered low. The density of 1017 cm−3 has some effect; hence, we consider it medium density. The 1018 cm−3 ion density is considered high, as we had to enlarge the second y axis range by order of magnitude. This reasoning makes 1019 cm−3 a very high ion density. We also note that the scan duration affects the hole-channel and electron-channel TFTs oppositely. Another important observation is that as the density of mobile species goes up to 1018 cm−3 and above, the lack of saturation in the output characteristics becomes a common feature.

Interestingly, for the electron-channel TFTs and high ion density, the current of the long-duration scan goes up. It evolves into a shape resembling a Schottky contact-limited TFT, although ideal contacts are used in the simulation. The curves due to the intermediate scan speeds of 10 ms (blue) and 1 s (green) suggest that the role played by the anions is complex. In Fig. S1 in the supplementary material, we compare the results of Fig. 2(c) to those obtained for a transistor stabilized at VGS = 0 and not |VGS| = |VT| + 1. While the curves depend on the pre-bias conditions, the observations drawn above are general and not affected by the pre-bias conditions.

To shed light on the role of the anions’ transport, we plot in Fig. 3 the average charge density across the channel by averaging over the 10 nm close to the insulator. Figure 3(a) is for the reference TFTs, where we show both hole-channel and electron-channel devices using the absolute value of the drain bias. We note that the gate bias charged the channel to about 7 × 10 17 c m 3 and that in the linear regime, the drain bias reduces the average density. The density is practically fixed at the pinch-off point where the transistor enters the saturation regime ( | V D S | = | V G S | | V T |). Moving to Figs. 3(b) and 3(c) (average ion density of 1018 cm−3), we note that the hole and electron densities at the channels are very different from those found for the reference devices. Namely, the densities are not directedly determined by the gate capacitance and are, as expected, affected by the anions. However, in the curves describing the evolution of the anions' density at the channel (light blue), we see no apparent correlation to the evolution of the holes (green line) or the electrons (red line) in their respective transistor structures.

FIG. 3.

Average charge density at the 10 nm next to the insulator as a function of the drain–source bias. (a) Electron and hole average density in the respective reference devices [Fig. 1(b)]. (b) Hole and ion average density in the hole-channel perovskite TFT [Fig. 2(b)] for two scan durations. (c) Electron and ion average density in the electron-channel TFT [Fig. 2(b)] for two scan durations. The average ion density was 1018 cm−3.

FIG. 3.

Average charge density at the 10 nm next to the insulator as a function of the drain–source bias. (a) Electron and hole average density in the respective reference devices [Fig. 1(b)]. (b) Hole and ion average density in the hole-channel perovskite TFT [Fig. 2(b)] for two scan durations. (c) Electron and ion average density in the electron-channel TFT [Fig. 2(b)] for two scan durations. The average ion density was 1018 cm−3.

Close modal

The anions' role being complex implies that it is not only the anions' density at the channel that matters but rather the distribution in the entire perovskite layer. To this end, we took snapshots of the transistors' parameters at the ends of the 100 μs and 100 s scans. The results in Figs. 4–6 are arranged in a table format such that the left and right columns are for the 100 μs and 100 s scans, respectively. The top row is for the electron-channel device, and the bottom row is for the hole-channel TFT. The average ion density was 1018 cm−3. Figure 4 shows the anions' density distribution in the abovementioned four scenarios. The color coding implies that blue stands for anion density being much lower than the stationary cation density of 1018 cm−3. Namely, the blue regions are effectively n-dopant (positive ions resulting from the anions' migration from their place in the original neutral crystal) rich. The orange-red color is for the opposite case where the anions' density exceeds that of the cations, making the region effectively p-dopant rich. For example, in Fig. 4(b), almost the entire film is n-dopant rich, and we note an excess of p-dopants only close to the drain. In contrast, in Fig. 4(d) of the hole-channel device, the n-type doping seems to block the access of holes to the drain. We will return to this figure in the section titled Discussion.

FIG. 4.

Anions’ density distribution at the end of the VDS scan [Fig. 2(b)]. (a) and (b) are for the electron-channel transistor and (c) and (d) are for the hole-channel TFT. (a) and (c) are for a fast scan of 100 μs and (b) and (d) are for a slow scan of 100 s. The average ion density was 1018 cm−3.

FIG. 4.

Anions’ density distribution at the end of the VDS scan [Fig. 2(b)]. (a) and (b) are for the electron-channel transistor and (c) and (d) are for the hole-channel TFT. (a) and (c) are for a fast scan of 100 μs and (b) and (d) are for a slow scan of 100 s. The average ion density was 1018 cm−3.

Close modal
FIG. 5.

Electronic charge density distribution at the end of the VDS scan [Fig. 2(b)]. (a) and (b) show the electrons inside the electron-channel transistor, and (c) and (d) show the holes inside the hole-channel TFT. (a) and (c) are for a fast scan of 100 μs, and (b) and (d) are for a slow scan of 100 s.

FIG. 5.

Electronic charge density distribution at the end of the VDS scan [Fig. 2(b)]. (a) and (b) show the electrons inside the electron-channel transistor, and (c) and (d) show the holes inside the hole-channel TFT. (a) and (c) are for a fast scan of 100 μs, and (b) and (d) are for a slow scan of 100 s.

Close modal
FIG. 6.

Current density distribution at the end of the VDS scan [Fig. 2(b)]. (a) and (b) are for the electron-channel transistor, and (c) and (d) are for the hole-channel TFT. (a) and (c) are for a fast scan of 100 μs, and (b) and (d) are for a slow scan of 100 s.

FIG. 6.

Current density distribution at the end of the VDS scan [Fig. 2(b)]. (a) and (b) are for the electron-channel transistor, and (c) and (d) are for the hole-channel TFT. (a) and (c) are for a fast scan of 100 μs, and (b) and (d) are for a slow scan of 100 s.

Close modal

In Fig. 5, we plot the density distribution of the electronic charges. We remind the reader that the source/drain contacts were chosen to suit the appropriate band; hence, there are no holes in the electron-channel device and no electrons in the hole-channel TFT. We note that the electronic charge density in most films' volumes is significant and above 1016 cm−3. We will return to this figure too in the section titled Discussion.

Figure 6 shows the current density distribution within the films. We did not split it into vertical and horizontal components but showed the absolute value. Due to significant differences in the values, the range in Figs. 6(a) and 6(d) is an order of magnitude lower than in Figs. 6(b) and 6(c). Again, if we only look at Figs. 6(b) and 6(d), we note a strong short-channel effect for the electron transistor and contact “barrier” toward the drain contact for the holes TFT. We will expand on these in the section titled Discussion.

Most of the lead-halide perovskite-based TFT reports are, naturally, on transistors with minimal to no ionic effects. This paper expanded the range of simulated experimental conditions to present the ion-affected characteristics. Moreover, we compared electron and hole-channel transistors. Before delving into the analysis of the simulation results, we wish to establish relevance to published experimental data. Very few reports of ambipolar perovskite TFTs could allow us to compare the experimental data to our simulation in Fig. 2. Not many present clear output characteristics of both channel types, and fewer would also include all experimental details required for our comparison (mostly scan rate). Yusoff et al.35 present such data for Cs0.15(MA0.17FA0.83)0.85Pb(Br0.17I0.83)3, and the slow scan rate (10 mV/s) they used corresponds to the long-duration scans we employed. Their Figs. 1(d)1(e) are similar to our long scan results in Fig. 2(b) (red lines), where the electron-channel current is higher. As the shape of the curves is close to ideal, we cannot determine whether the differences in the extracted electron and hole mobilities are due to ionic effects. However, the fact that the difference in effective mobility that they reported was relatively small suggests that the density of mobile anions was not above 1017 cm−3.

Looking for a report of non-ideal characteristics, we found the paper by Zeidell et al.64 They report an ambipolar MAPbI3−xClx field-effect transistor with an interesting crystallization process. The results in the main article show that the hole mobility is larger, which is the opposite of what we found in Fig. 2. While it is tempting to conclude that the opposite trend is associated with the mobile species having opposite charge (i.e., cations), the shape of the curves has no clear indication of ionic effects. However, Fig. S2 in their supplementary material shows the output characteristics of an n-channel device exhibiting a strange hump [see their reproduced figure in our Fig. 7(a)]. To obtain a first clue regarding the parameters that may produce such a hump, we return to Fig. 2. Such a hump can be found in Fig. 2(c) for a P-channel transistor based on a material where the mobile species are negative anions. Since the hump reported by Zeidell et al.64 is for an N-channel transistor, it implies that the mobile species are positive cations. The clear hump in Fig. 2(c) is for a mobile species density of 1018 cm−3. Since the layer thickness of the perovskite in Ref. 64 is likely to have been thicker than 100 nm used for Fig. 2, the real density would be between 1017 and 1018 cm−3.

FIG. 7.

(a) Measured output characteristics. Reproduced with permission from A. M. Zeidell et al., Adv. Electron. Mater. 4(12), 1800316 (2018). Copyright 2018, John Wiley and Sons.64 (b) Simulated output characteristics of having the structure reported. Simulated devices, showing electron-channel (VDS > 0) with 1017 cm−3 positive ions density and short scan duration. The parameters used to reproduce the 195 k data are W/L = 400/50, scan rate = 1 V/s, me = 4 × 10−3 cm2 V−1 s−1, and Dion = 5 × 10−10 cm2 s−1. The color coding reflects that the experimental threshold voltage was higher by 10 V compared to the simulations.

FIG. 7.

(a) Measured output characteristics. Reproduced with permission from A. M. Zeidell et al., Adv. Electron. Mater. 4(12), 1800316 (2018). Copyright 2018, John Wiley and Sons.64 (b) Simulated output characteristics of having the structure reported. Simulated devices, showing electron-channel (VDS > 0) with 1017 cm−3 positive ions density and short scan duration. The parameters used to reproduce the 195 k data are W/L = 400/50, scan rate = 1 V/s, me = 4 × 10−3 cm2 V−1 s−1, and Dion = 5 × 10−10 cm2 s−1. The color coding reflects that the experimental threshold voltage was higher by 10 V compared to the simulations.

Close modal

To replicate and understand their results, we simulated the reported transistor structure assuming a 200 nm thick film (see Fig. S5 in the supplementary material). As mentioned above, to obtain the hump for an N-channel transistor, we had to assume that the mobile species are cations. The results shown in Fig. 7(b) are for a cation density of 1017 cm−3 and a diffusion constant of 5 × 10−10 cm2 s−1, and the VDS scan is at 1 V/s. The electron mobility we derived is 4 × 10−3 cm2 V−1 s−1, and considering that the measured current at 195 K was about 500 times lower than the room temperature one, it agrees with the reported values of ∼5 cm2 V−1 s−1. The low ion density and the fast scan relative to the ion diffusion are also consistent with the measurements done at 195 k. Further discussion on the origin of the hump can be found in the supplementary material.

From our results, we first found that for the ionic effects to be reflected in the measured characteristics, for a 100 nm thick layer, the free-ion density must be above 1016 cm−3 (Fig. 2). Due to the nature of the Poisson equation, thicker layers would be more susceptible to ionic effects. Examining Fig. 2 and especially Fig. 2(b), we note that an ideal shape of the transistor characteristics does not prove that there are no ionic effects. In Fig. S2 in the supplementary material, we present the results of hysteresis scans, which show that the lack of hysteresis does not guarantee that the ions do not affect the results. In Fig. 2, we showed the results for different scan durations or scan speeds. The actual times (T) stated should be understood as relative to the ion’s transport time. So, one should understand that the physical parameter is not T but P = V Dmax T 1 D ion L 2 k T / q, with D being the diffusion coefficient, which is proportional to the ion’s mobility, and L is a characteristic dimension of the transistor (as channel length). The very fast scan time of T = 10−4 s in Fig. 2 corresponds to P 5 × 10 8. At such a fast scan, the output characteristics have the ideal shape and are almost identical between electron- and hole-channel devices. At longer scans of up to 100 s (slower scans, P 5 × 10 2), the current of the hole-channel transistor will go down, and for the electron-channel transistor, it will go up. Another way to express the above is that the current will go down for the transistor, where the channel’s charge sign is opposite to the mobile ions. The current goes up for longer scans for the same charge sign between the channel and mobile ions. Namely, this is a method to determine the sign of the mobile ions. This was demonstrated in the above section, where we used our simulation model to reproduce the ion’s effect in the transistor reported by Zeidell et al., with the simulation’s parameter being P 3 × 10 6.64 

Another interesting result is the origin of the current modification. It is tempting to consider the channel as the crucial region and, thus, conclude that negative mobile ions would more effectively screen the gate field of electron-channel TFTs, and, namely, electron-channel TFTs should have presented lower effective mobility. However, the detailed simulations indicate the opposite. Examining Fig. 5(d), we note that the hole-channel seems to be ideal, but for the fact that it is disconnected from the drain. In other words, the current of the hole-channel TFT is limited by poor charge extraction caused by the n-dopants that are fixed in place and exposed due to the migration of the anions (p-dopants). Note that in Fig. 6(d), the current to the drain is weak and resembles a leakage toward the drain.

In contrast, Fig. 5(b) shows a very short n-channel perfectly connected to the drain, and, namely, the enhancement of the electron current is due to anomalous channel shortening, driven by the anion redistribution. These conclusions are supported by the current color maps presented in Fig. 6. Interestingly, the ion-induced channel shortening is so pronounced that it resulted in the Schottky-like shapes of the electron currents in Figs. 2(c) and 2(d), with the contacts not contributing to the current shape.

In the above discussion, we were concerned mainly with the charge extraction, and the occupancy of the channel (i.e., the “simple” ion gate-screening) did not play a significant role. If the effects are associated with charge extraction, one would expect that as the channel length increases and the channel resistance increases, the charge extraction effects would be less pronounced. To this end, we present in Figs. S3 and S4 in the supplementary material simulation results of a transistor having a sevenfold more extended channel (L = 35 μm). Note that compared to Figs. 1 and 2, the ion’s effects are less pronounced.

See the supplementary material for additional simulation results.

This research was supported by the Ministry of Innovation, Science and Technology Israel, the M-ERANET grant PHANTASTIC Call 2021, and the Neubauer Family Foundation. The opinions expressed in this report are those of the author(s) and do not necessarily reflect the views of the Neubauer Family Foundation.

The authors have no conflicts to disclose.

Doaa Shamalia: Investigation (equal); Writing – original draft (equal). Nir Tessler: Conceptualization (equal); Investigation (equal); Supervision (equal); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.

1.
N.
Onoda-Yamamuro
,
T.
Matsuo
, and
H.
Suga
, “
Dielectric study of CH3NH3PbX3 (X = Cl, Br, I)
,”
J. Phys. Chem. Solids
53
(
7
),
935
939
(
1992
).
2.
A.
Kojima
et al, “
Organometal halide perovskites as visible-light sensitizers for photovoltaic cells
,”
J. Am. Chem. Soc.
131
(
17
),
6050
6051
(
2009
).
3.
M. M.
Lee
et al, “
Efficient hybrid solar cells based on meso-superstructured organometal halide perovskites
,”
Science
338
(
6107
),
643
647
(
2012
).
4.
I.
Chung
et al, “
All-solid-state dye-sensitized solar cells with high efficiency
,”
Nature
485
(
7399
),
486
489
(
2012
).
5.
H.-S.
Kim
et al, “
Lead iodide perovskite sensitized all-solid-state submicron thin film mesoscopic solar cell with efficiency exceeding 9%
,”
Sci. Rep.
2
(
1
),
591
(
2012
).
6.
NREL
, see https://www.nrel.gov/pv/assets/pdfs/best-research-cell-efficiencies.20200406.pdf for “Best Research-cell Efficiency Chart” (May 2020).
7.
P.
Lopez-Varo
et al, “
Device physics of hybrid perovskite solar cells: Theory and experiment
,”
Adv. Energy Mater.
8
(
14
),
1702772
(
2018
).
8.
A.
Buin
et al, “
Materials processing routes to trap-free halide perovskites
,”
Nano Lett.
14
(
11
),
6281
6286
(
2014
).
9.
E. L.
Unger
et al, “
Hysteresis and transient behavior in current-voltage measurements of hybrid-perovskite absorber solar cells
,”
Energy Environ. Sci.
7
(
11
),
3690
3698
(
2014
).
10.
N.
Tessler
and
Y.
Vaynzof
, “
Preventing hysteresis in perovskite solar cells by undoped charge blocking layers
,”
ACS Appl. Energy Mater.
1
(
2
),
676
683
(
2018
).
11.
N. E.
Courtier
et al, “
How transport layer properties affect perovskite solar cell performance: Insights from a coupled charge transport/ion migration model
,”
Energy Environ. Sci.
12
(
1
),
396
409
(
2019
).
12.
N.
Tessler
and
Y.
Vaynzof
, “
Insights from device modeling of perovskite solar cells
,”
ACS Energy Lett.
5
(
4
),
1260
1270
(
2020
).
13.
S.
Bitton
and
N.
Tessler
, “
Electronic-ionic coupling in perovskite based solar cells: Implications for device stability
,”
Appl. Phys. Lett.
117
(
13
),
133904
(
2020
).
14.
Y.
Vaynzof
, “
The future of perovskite photovoltaics-thermal evaporation or solution processing?
,”
Adv. Energy Mater.
10
(
48
),
2003073
(
2020
).
15.
S.
Bitton
and
N.
Tessler
, “
Electron/hole blocking layers as ionic blocking layers in perovskite solar cells
,”
J. Mater. Chem. C
9
(
6
),
1888
1894
(
2021
).
16.
J. F.
Butscher
et al, “
Enhancing the open-circuit voltage of perovskite solar cells by embedding molecular dipoles within their hole-blocking layer
,”
ACS Appl. Mater. Interfaces
12
(
3
),
3572
3579
(
2020
).
17.
L.
Bertoluzzi
et al, “
Incorporating electrochemical halide oxidation into drift-diffusion models to explain performance losses in perovskite solar cells under prolonged reverse bias
,”
Adv. Energy Mater.
11
(
10
),
2002614
(
2021
).
18.
S.
Bitton
and
N.
Tessler
, “
Perovskite ionics—Elucidating degradation mechanisms in perovskite solar cells via device modelling and iodine chemistry
,”
Energy Environ. Sci.
16
,
2621
(
2023
).
19.
H. J.
Snaith
et al, “
Anomalous hysteresis in perovskite solar cells
,”
J. Phys. Chem. Lett.
5
(
9
),
1511
1515
(
2014
).
20.
J. M.
Frost
and
A.
Walsh
, “
What is moving in hybrid halide perovskite solar cells?
,”
Acc. Chem. Res.
49
(
3
),
528
535
(
2016
).
21.
S. G.
Motti
et al, “
Defect activity in lead halide perovskites
,”
Adv. Mater.
31
(
47
),
1901183
(
2019
).
22.
S. T.
Birkhold
et al, “
Interplay of mobile ions and injected carriers creates recombination centers in metal halide perovskites under bias
,”
ACS Energy Lett.
3
(
6
),
1279
1286
(
2018
).
23.
D. R.
Ceratti
et al, “
The pursuit of stability in halide perovskites: The monovalent cation and the key for surface and bulk self-healing
,”
Mater. Horiz.
8
(
5
),
1570
1586
(
2021
).
24.
C. C.
Boyd
et al, “
Understanding degradation mechanisms and improving stability of perovskite photovoltaics
,”
Chem. Rev.
119
(
5
),
3418
3451
(
2019
).
25.
C.
Li
et al, “
Real-time observation of iodide ion migration in methylammonium lead halide perovskites
,”
Small
13
(
42
),
1701711
(
2017
).
26.
J. M.
Azpiroz
et al, “
Defect migration in methylammonium lead iodide and its role in perovskite solar cell operation
,”
Energy Environ. Sci.
8
(
7
),
2118
2127
(
2015
).
27.
S.
Reichert
et al, “
Ionic-defect distribution revealed by improved evaluation of deep-level transient spectroscopy on perovskite solar cells
,”
Phys. Rev. Appl.
13
(
3
),
034018
(
2020
).
28.
P.
Fassl
et al, “
Effect of crystal grain orientation on the rate of ionic transport in perovskite polycrystalline thin films
,”
ACS Appl. Mater. Interfaces
11
(
2
),
2490
2499
(
2019
).
29.
M.
Pazoki
et al, “
Photon energy-dependent hysteresis effects in lead halide perovskite materials
,”
J. Phys. Chem. C
121
(
47
),
26180
26187
(
2017
).
30.
C.
Besleaga
et al, “
Iodine migration and degradation of perovskite solar cells enhanced by metallic electrodes
,”
J. Phys. Chem. Lett.
7
(
24
),
5168
5175
(
2016
).
31.
C.
Eames
et al, “
Ionic transport in hybrid lead iodide perovskite solar cells
,”
Nat. Commun.
6
(
1
),
7497
(
2015
).
32.
B.
Rivkin
et al, “
Effect of ion migration-induced electrode degradation on the operational stability of perovskite solar cells
,”
ACS Omega
3
(
8
),
10042
10047
(
2018
).
33.
M. H.
Futscher
et al, “
Quantification of ion migration in CH3NH3PbI3 perovskite solar cells by transient capacitance measurements
,”
Mater. Horiz.
6
(
7
),
1497
1503
(
2019
).
34.
C. R.
Kagan
,
D. B.
Mitzi
, and
C. D.
Dimitrakopoulos
, “
Organic-inorganic hybrid materials as semiconducting channels in thin-film field-effect transistors
,”
Science
286
(
5441
),
945
(
1999
).
35.
A. R.
Yusoff
,
H. P.
Kim
,
X.
Li
,
J.
Kim
,
J.
Jang
, and
M.
Nazeeruddin
, “
Ambipolar triple cation perovskite field effect transistors and inverters
,”
Adv. Mater.
29
(
8
),
1602940
(
2016
).
36.
T.
Matsushima
et al, “
Solution-processed organic–inorganic perovskite field-effect transistors with high hole mobilities
,”
Adv. Mater.
28
(
46
),
10275
10281
(
2016
).
37.
A. N.
Aleshin
et al, “
Field-effect transistors with high mobility and small hysteresis of transfer characteristics based on CH3NH3PbBr3 films
,”
Phys. Solid State
59
(
12
),
2486
2490
(
2017
).
38.
J. H. L.
Ngai
et al, “
Growth, characterization, and thin film transistor application of CH3NH3PbI3 perovskite on polymeric gate dielectric layers
,”
RSC Adv.
7
(
78
),
49353
49360
(
2017
).
39.
S. P.
Senanayak
et al, “
Understanding charge transport in lead iodide perovskite thin-film field-effect transistors
,”
Sci. Adv.
3
(
1
),
e1601935
(
2017
).
40.
N. D.
Canicoba
et al, “
Halide perovskite high-k field effect transistors with dynamically reconfigurable ambipolarity
,”
ACS Mater. Lett.
1
(
6
),
633
640
(
2019
).
41.
Y.
Gao
et al, “
Highly stable lead-free perovskite field-effect transistors incorporating linear pi-conjugated organic ligands
,”
J. Am. Chem. Soc.
141
(
39
),
15577
15585
(
2019
).
42.
T.
Matsushima
et al, “
Toward air-stable field-effect transistors with a tin iodide-based hybrid perovskite semiconductor
,”
J. Appl. Phys.
125
(
23
),
235501
(
2019
).
43.
F.
Zhang
et al, “
Two-dimensional organic-inorganic hybrid perovskite field-effect transistors with polymers as bottom-gate dielectrics
,”
J. Mater. Chem. C
7
(
14
),
4004
4012
(
2019
).
44.
H. P.
Kim
et al, “
A hysteresis-free perovskite transistor with exceptional stability through molecular cross-linking and amine-based surface passivation
,”
Nanoscale
12
(
14
),
7641
7650
(
2020
).
45.
H.
Zhu
et al, “
High-performance and reliable lead-free layered-perovskite transistors
,”
Adv. Mater.
32
(
31
),
2002717
(
2020
).
46.
B.
Jeong
et al, “
Room-temperature halide perovskite field-effect transistors by ion transport mitigation
,”
Adv. Mater.
33
,
2100486
(
2021
).
47.
Y. J.
Lee
et al, “
High hole mobility inorganic halide perovskite field-effect transistors with enhanced phase stability and interfacial defect tolerance
,”
Adv. Electron. Mater.
8
,
2100624
(
2021
).
48.
A.
Liu
et al, “
High-performance inorganic metal halide perovskite transistors
,”
Nat. Electron.
5
(
2
),
78
83
(
2022
).
49.
D.
Li
et al, “
Size-dependent phase transition in methylammonium lead iodide perovskite microplate crystals
,”
Nat. Commun.
7
(
1
),
11330
(
2016
).
50.
N.
Tessler
et al, “
Switching mechanisms of CMOS-compatible ECRAM transistors—Electrolyte charging and ion plating
,”
J. Appl. Phys.
134
(
7
),
074501
(
2023
).
52.
G.
Xing
,
N.
Mathews
,
S.
Sun
,
S. S.
Lim
,
Y. M.
Lam
,
M.
Grätzel
,
S.
Mhaisalkar
, and
T. C.
Sum
, “
Long-range balanced electron- and hole-transport lengths in organic-inorganic CH3NH3PbI3
,”
Science
342
,
344
347
(
2013
).
53.
F.
Brivio
,
A. B.
Walker
, and
A.
Walsh
, “
Structural and electronic properties of hybrid perovskites for high-efficiency thin-film photovoltaics from first-principles
,”
APL Mater.
1
(
4
),
042111
(
2013
).
54.
Y.
Goldberg
, in
Properties of Advanced Semiconductor Materials GaN, AlN, InN, BN, SiC, SiGe
, edited by
S. L. R.
Michael
,
E.
Levinshtein
, and
M. S.
Shur
(
John Wiley & Sons, Inc.
,
New York
,
2001
).
55.
J.
Fan
,
H. L.
Qianwei Kuang
,
B.
Gao
,
F.
Ma
, and
Y.
Hao
, “
Physical properties and electrical characteristics of H2O-based and O3-based HfO2 films deposited by ALD
,”
Microelectron. Reliab.
52
(
6
),
1043
1049
(
2012
).
56.
L. M.
Herz
, “
Charge-carrier mobilities in metal halide perovskites: Fundamental mechanisms and limits
,”
ACS Energy Lett.
2
(
7
),
1539
1548
(
2017
).
57.
Y.
Zhou
and
G.
Long
, “
Low density of conduction and valence band states contribute to the high open-circuit voltage in perovskite solar cells
,”
J. Phys. Chem. C
121
(
3
),
1455
1462
(
2017
).
58.
Z.
Hu
et al, “
A review on energy band-gap engineering for perovskite photovoltaics
,”
Sol. RRL
3
(
12
),
1900304
(
2019
).
59.
A. J.
Ben-Sasson
et al, “
The mechanism of operation of lateral and vertical organic field effect transistors
,”
Isr. J. Chem.
54
(
5–6
),
568
585
(
2014
).
60.
P.
Calado
et al, “
Evidence for ion migration in hybrid perovskite solar cells with minimal hysteresis
,”
Nat. Commun.
7
(
1
),
13831
(
2016
).
61.
G.
Richardson
et al, “
Can slow-moving ions explain hysteresis in the current–voltage curves of perovskite solar cells?
,”
Energy Environ. Sci.
9
(
4
),
1476
1485
(
2016
).
62.
S.
Van Reenen
,
M.
Kemerink
, and
H. J.
Snaith
, “
Modeling anomalous hysteresis in perovskite solar cells
,”
J. Phys. Chem. Lett.
6
(
19
),
3808
3814
(
2015
).
63.
S.
Reichert
et al, “
Probing the ionic defect landscape in halide perovskite solar cells
,”
Nat. Commun.
11
(
1
),
6098
(
2020
).
64.
A. M.
Zeidell
et al, “
Enhanced charge transport in hybrid perovskite field-effect transistors via microstructure control
,”
Adv. Electron. Mater.
4
(
12
),
1800316
(
2018
).