**3.1. Intrinsic properties of OMHP**

**2.3. Excitation transfer**

122 Nanostructured Solar Cells

**2.4. Outlook**

ment to improve the solar cell efficiency [40].

Due to large exciton binding energy in CdSe QDs, initial photo‐generated charge species behave like excitons [39]. The motion of such excitons between the QDs occurs via Förster resonant energy transfer (FRET), which is essential for the function of optical devices with

In QD‐sensitized MOs, it is conventionally believed that electron injection only occurs from QDs directly attached to the MO surface while multilayer QD attachment would hinder the electron collection process. However, in the system of QD‐MO with multiple layer QDs attachement, it is found the excitation transfer (Förster energy transfer) also occurs within the aggregates of QDs [41]. This transfer process can be traced in TA as an additional long‐lived (5 ns) excited states depopulation (see **Figure 3**). Such energy transfer has also been reported in tandem‐layered cadmium chalcogenide QD solar cells, which can be an effective comple‐

**Figure 3.** Energy transfer between indirectly attached QDs in QD‐MO photoanodes. (a) TA kinetics of QD‐ZnO NWs with different sensitization times. The inset illustrates the long‐lived TA component (red dots) extracted from the expo‐ nential fit of the kinetics. The consistence between the TA amplitude and the number of indirectly attached QDs indi‐ cates indirect exciton depopulation with ns time scale. (b) Main photoinduced processes, including direct electron

QDs provide great opportunities for the development of optoelectronic devices. Moreover, the booming of QD solar cell research will surely put forward photo‐induced dynamics questions waiting to be revealed in the future. For example, the band alignment assembly in lead chalcogenide QD devices currently holds the highest record of solar cell efficiency. The general principles of the device have been revealed but the atomistic details are still to be understood. The conventional spectroscopic techniques used in investigating the photo‐induced dynamics are usually very different from the solar illumination conditions. These include the excitation intensity, the loading of circuit, medium conditions, etc. Systematic studies of charge carries dynamics under real solar cell functioning conditions may provide more useful reference for device application. Another issue of future research is to utilizing the "green" elements in the materials to replace the toxic Cd and Pb, which are overwhelmingly used in recent studies. All

injection and energy. Figures are reproduced from Ref. [41]. Copyright American Chemical Society, 2012.

densely packed QD films including light emitters and solar cells [40–42].

Shown in **Figure 4(a)** is the rise of transient photoconductivity of methyl ammonium lead triiodide (MAPbI3), MAPbI3/Al2O3, and MAPbI3/TiO2 measured using time‐resolved THz spectroscopy (TRTS). Notice that the rise in MAPbI3 and MAPbI3/Al2O3 is a two‐step process, one that is instrument limited (about 70% of the total amplitude), while the second is about 2– 3 ps (about 30% of the total amplitude). We note that molecular excitons that are tightly bound are neutral by definition, and therefore would not contribute to the transient photoconduc‐ tivity obtained here. However, if photo‐generated species are either loosely bound or mobile, or both, TRTS would be able to detect it. The instantaneous rise therefore means that highly mobile charges are created within the response of our instrument. The next question is how to explain the additional 2–3 ps rise. This can be understood in two ways. One, is that from the 70% highly mobile charges, these carriers become faster, that is gaining more mobility during that time scale. This would show in the technique as increase in photoconductivity. Another explanation is that there is a distribution of binding energy in the sample. The implication of this is that, there is a nonuniformity in the quality of film wherein in some parts have defect‐ free area that promotes generation of mobile charges. Area of the film with high defect density would tend to have exciton that is more tightly bounded. In fact, it has been reported that binding energy of exciton in these materials vary between few meV [43] and 50 meV [44] depending on the preparation conditions. As a result of heterogeneity in its binding energy, photo‐generated species dissociates at different rates. This explains the two‐component rise in the photoconductivity kinetics shown in **Figure 4(a)**. In this scenario, 70% are directly con‐ verted to mobile charges while the 30% are generated via exciton dissociation. For MAPbI3/ TiO2, the transient photoconductivity rises in a single‐step instrument‐limited time scale. This means that, unlike the first two samples, mobile charges are readily created. This is reminiscent of our previous results on electron injection from QD to ZnO [25] as well as the injection rate of electrons from RuN3 dye attached to TiO2 [45]. Similarly, we assign the single step rise of the transient photoconductivity as evidence of subpicosecond injection of electrons from the OMHP to TiO2. There is favorable band energy alignment between the energy of perovskite and metal oxide that helps in the separation of any bound electron‐hole pair allowing injection in the ultrafast time scale.

**Figure 4.** (a) Transient photoconductivity kinetics of neat MAPbI3, MAPbI3/Al2O3, and MAPbI3/TiO2. Normalized to 1 (*λ*pump = 400 nm, *I*exc = 1.7 × 1013 ph/cm2 per pulse). (b) Transient absorption kinetics (*λ*pump = 603 nm, *λ*probe = 970 nm, *I*exc = 6.0 × 1014 ph/cm2 per pulse). (c) Photoluminescence spectra (*λ*pump = 550 nm) of the three samples. Reprint with permis‐ sion from Ref. [46]. Copyright 2014, American Chemical Society.

To verify the above assertions, we also obtained the transient absorption kinetics of the three samples, shown in **Figure 4(b)**. For MAPbI3 and MAPbI3/Al2O3, the kinetics is characterized by two‐step decrease whose time scale is identical to that of the rise in the transient photo‐ conductivity. This means that the processes in both kinetics should be at least similar if not identical. Since the absorption kinetics is negative for the two samples, it means that something is being bleached or being emitted. Furthermore, as shown in **Figure 4(c)**, for the same two samples, the emission spectra of the steady state photoluminescence are quite high. The two‐ step stimulated emission in the transient absorption kinetics show that not all charges are created simultaneously while the bright photoluminescence in steady state PL means these charges eventually meet and recombine radiatively. In the case of MAPbI3/TiO2, the positive one‐step rise in the transient absorption kinetics supports our conclusion that there is instan‐ taneous charge generation while the very low PL count reiterates our assertion that electrons are injected into TiO2 and does not recombine radiatively with holes left in the perovskite.

We further analyzed and take the transient photoconductivity of the three samples at different excitation fluences and at up to 1 ns. Plotted in **Figure 5(a)** is the transient photoconductivity per photon absorbed per pulse for the first 40 ps. For the first two samples, MAPbI3 and the MAPbI3/Al2O3, the obtained mobility is 20 cm2 V‐1 s‐1 while for MAPbI3/TiO2, it is 7.5 cm2 V‐1 s‐1. For the MAPbI3, both electrons and holes are generated in the perovskite material. This is also true for MAPbI3/Al2O3 sample, since the alignment of band energies of Al2O3 and perov‐ skite is not favorable for charge transfer, both electrons and holes remain in the perovskite material. Therefore, 20 cm2 V‐1 s‐1 is the sum of the mobility of both charges in the perovskite. For MAPbI3/TiO2, where there is an ultrafast electron injection as discussed above, electrons are transferred to TiO2 and its mobility becomes 0.1 cm2 V‐1 s‐1 only, since it adopts the property of the accepting material. The implication is that the measured mobility of 7.5 cm2 V‐1 s‐1 in MAPbI3/TiO2 should be coming from holes left in perovskite since electrons are already in the TiO2. Now, knowing that the hole mobility is 7.5 cm2 V‐1 s‐1 and the total mobility is 20 cm2 V‐1 s‐1, one can concludes that the electron mobility should be 12.5 cm2 V‐1 s‐1 in both MAPbI3 , and MAPbI3 /Al2 O3 . This is the first report where both the electron and hole mobilities are measured in OMHP materials [46].

and metal oxide that helps in the separation of any bound electron‐hole pair allowing injection

**Figure 4.** (a) Transient photoconductivity kinetics of neat MAPbI3, MAPbI3/Al2O3, and MAPbI3/TiO2. Normalized to 1

6.0 × 1014 ph/cm2 per pulse). (c) Photoluminescence spectra (*λ*pump = 550 nm) of the three samples. Reprint with permis‐

To verify the above assertions, we also obtained the transient absorption kinetics of the three samples, shown in **Figure 4(b)**. For MAPbI3 and MAPbI3/Al2O3, the kinetics is characterized by two‐step decrease whose time scale is identical to that of the rise in the transient photo‐ conductivity. This means that the processes in both kinetics should be at least similar if not identical. Since the absorption kinetics is negative for the two samples, it means that something is being bleached or being emitted. Furthermore, as shown in **Figure 4(c)**, for the same two samples, the emission spectra of the steady state photoluminescence are quite high. The two‐ step stimulated emission in the transient absorption kinetics show that not all charges are created simultaneously while the bright photoluminescence in steady state PL means these charges eventually meet and recombine radiatively. In the case of MAPbI3/TiO2, the positive one‐step rise in the transient absorption kinetics supports our conclusion that there is instan‐ taneous charge generation while the very low PL count reiterates our assertion that electrons are injected into TiO2 and does not recombine radiatively with holes left in the perovskite. We further analyzed and take the transient photoconductivity of the three samples at different excitation fluences and at up to 1 ns. Plotted in **Figure 5(a)** is the transient photoconductivity per photon absorbed per pulse for the first 40 ps. For the first two samples, MAPbI3 and the MAPbI3/Al2O3, the obtained mobility is 20 cm2 V‐1 s‐1 while for MAPbI3/TiO2, it is 7.5 cm2 V‐1 s‐1. For the MAPbI3, both electrons and holes are generated in the perovskite material. This is

per pulse). (b) Transient absorption kinetics (*λ*pump = 603 nm, *λ*probe = 970 nm, *I*exc =

in the ultrafast time scale.

124 Nanostructured Solar Cells

(*λ*pump = 400 nm, *I*exc = 1.7 × 1013 ph/cm2

sion from Ref. [46]. Copyright 2014, American Chemical Society.

**Figure 5.** (a) Transient photoconductivity of neat MAPbI3, MAPbI3/Al2O3, and MAPbI3/TiO2 per photon absorbed. (b) Comparison TA and TRTS kinetics for neat MAPbI3 showing that similar decay rates up to 1 ns. (c) Intensity depend‐ ence transient photoconductivity of MAPbI3/Al2O3. Reprint with permission from Ref. [46]. Copyright 2014, American Chemical Society.

We then compared the kinetic traces of MAPbI3 using transient absorption and photoconduc‐ tivity up to 1 ns, at very similar excitation conditions, which are plotted in **Figure 5(b)**. From about 3 ps to 100 ps, both of the traces are flat which then started to decay at similar rate until 1 ns showing that the charge dynamics should be identical. Transient absorption monitors the population or depopulation of charge carries while transient photoconductivity measures the product of charge concentration and mobility. From 3 ps to 100 ps, where the two traces are flat means that neither the population of the charges nor the mobility of the carriers is changing. However, for time scale longer than 100 ps, the decay starts to manifest. In this case, the transient absorption kinetics shows that charges are disappearing. Similarly, the decay in transient photoconductivity should manifest the same phenomenon. Ergo, the mobility of charges in perovskite should have remained constant, at least up to 1 ns, otherwise its decay should be more substantial than just the corresponding transient absorption. This is a very important finding since mobility of charges usually decay in emerging photovoltaic materials due to the defects in the film as well as its high exciton binding energy. In the case of OMHP, it seems to suggest that there is at least less defects in these films that favor the charge to maintain its mobility up to 1 ns. For organic solar cell material, we have previously shown that mobility is 50% slower in half of nanosecond [48].

We then want to determine the influence of reduced excitation fluence to the decay of the transient photoconductivity as we have shown previously that there are nonlinear effects at high fluency in organic solar cells [48]. Shown in **Figure 5(c)** is the transient conductivity of MAPbI3/Al2O3 at three different intensities. For 7.4 × 1013 ph/cm2 per pulse, the mobility is around 5 cm2 V‐1 s‐1 and its decay started early at about 5 ps. For intensity of 1.7 × 1013 ph/cm2 per pulse, a mobility of 20 cm2 V‐1 s‐1 was obtained and the onset of decay is prolonged to around 10 ps. At the lowest excitation condition of 2.0 × 1012 ph/cm2 per pulse, the highest mobility is measured 25 cm2 V‐1 s‐1 and decay did not start until after 300 ps. This shows how essential the excitation intensity dependence measurements in transient spectroscopy. At high excitation conditions, nonlinear effects as such charge pair annihilation or second order nongeminate recombination dominates as the main channel of charge depopulation. This is the reason for the behavior of the transient photoconductivity shown above. At high excitation, mobility is low, i.e., at the earliest time scale, charges recombine right away since the resulting charge density at this condition is quite high. Moreover, the onset of decay is early. On the other hand, at the lowest excitation intensity, charges are rather sparse with each other (low charge density) and the probability of it recombining is low. This is the rationale of the high mobility at the early time scale. For the same reason, recombination is also delayed to at least 300 ps. The difference between the mobility of MAPbI3 (20 cm2 V‐1 s‐1, **Figure 5a**) and MAPbI3/ Al2O3 (25 cm2 V‐1 s‐1, **Figure 5c**) can be explained by the better film morphology in MAPbI3/ Al2O3. The metal oxide Al2O3 acts like a scaffolding creating a more continuous film than the bare MAPbI3.

Another, interesting feature of this material is the very small differences in the mobility of electrons and holes. In contrast with organic solar cells, this difference in mobility can be few orders of magnitude. Due to this, a built‐in electric field is produced since one of the charge specie, usually the electrons, arrives earlier in the electrode while the holes, being slow, is still traversing the polymer molecule, arriving later in the counterelectrode. As shown in the above results, electrons and holes in OMHP have a difference in the mobility of just about half, i.e., *µ*e = 12.5 cm2 V‐1 s‐1 and *µ*h = 7.5 cm2 V‐1 s‐1. This indicates that both charges arrive in their respective electrodes almost at the same time, avoiding the built‐up electric field. This again is advantageous for solar cell operation since it will be able to collect more charges.

#### **3.2. Injection into organic electrodes**

Similar to other emerging photovoltaic technologies, there are significant efforts on using organic molecules, like PCBM and Spiro‐OMeTAD as electrodes. In this section, we will show the mechanism and time scale of electron and hole injection in these organic electrodes. Plotted in **Figure 6(a)** is the transient photoconductivity of MAPbI3, MAPbI3/PCBM, and MAPbI3/ Spiro‐OMeTAD per photon absorbed per pulse. It can be seen that for MAPbI3 mobility obtained is 15 cm2 /Vs and remained flat for the first ns. We note that the difference in the mobility measured here with respect to the MAPbI3 sample discussed above (20 cm2 /Vs). It has been reported by Wang et al. [49] that depending on the preparation conditions, the concen‐ tration and type of defects could differ significantly. In fact, they found that thermal annealing alone should be able to shift the property of perovskite material from an *n*‐type to intrinsic to *p*‐type semiconductor. This means that no two‐perovskite samples are made identical to each other, more so, when prepared by different groups despite following the same recipe. Having said that, the transient photoconductivity measurements of all perovskite samples we studied in the past gave a rather consistent result, i.e., from 15 to 25 cm2 /Vs. We surmise that these differences do not significantly alter the interpretation of the photophysical properties of the materials we presented here.

transient photoconductivity should manifest the same phenomenon. Ergo, the mobility of charges in perovskite should have remained constant, at least up to 1 ns, otherwise its decay should be more substantial than just the corresponding transient absorption. This is a very important finding since mobility of charges usually decay in emerging photovoltaic materials due to the defects in the film as well as its high exciton binding energy. In the case of OMHP, it seems to suggest that there is at least less defects in these films that favor the charge to maintain its mobility up to 1 ns. For organic solar cell material, we have previously shown that

We then want to determine the influence of reduced excitation fluence to the decay of the transient photoconductivity as we have shown previously that there are nonlinear effects at high fluency in organic solar cells [48]. Shown in **Figure 5(c)** is the transient conductivity of

per pulse, a mobility of 20 cm2 V‐1 s‐1 was obtained and the onset of decay is prolonged to around 10 ps. At the lowest excitation condition of 2.0 × 1012 ph/cm2 per pulse, the highest

essential the excitation intensity dependence measurements in transient spectroscopy. At high excitation conditions, nonlinear effects as such charge pair annihilation or second order nongeminate recombination dominates as the main channel of charge depopulation. This is the reason for the behavior of the transient photoconductivity shown above. At high excitation, mobility is low, i.e., at the earliest time scale, charges recombine right away since the resulting charge density at this condition is quite high. Moreover, the onset of decay is early. On the other hand, at the lowest excitation intensity, charges are rather sparse with each other (low charge density) and the probability of it recombining is low. This is the rationale of the high mobility at the early time scale. For the same reason, recombination is also delayed to at least

Al2O3 (25 cm2 V‐1 s‐1, **Figure 5c**) can be explained by the better film morphology in MAPbI3/ Al2O3. The metal oxide Al2O3 acts like a scaffolding creating a more continuous film than the

Another, interesting feature of this material is the very small differences in the mobility of electrons and holes. In contrast with organic solar cells, this difference in mobility can be few orders of magnitude. Due to this, a built‐in electric field is produced since one of the charge specie, usually the electrons, arrives earlier in the electrode while the holes, being slow, is still traversing the polymer molecule, arriving later in the counterelectrode. As shown in the above results, electrons and holes in OMHP have a difference in the mobility of just about half, i.e., *µ*e = 12.5 cm2 V‐1 s‐1 and *µ*h = 7.5 cm2 V‐1 s‐1. This indicates that both charges arrive in their respective electrodes almost at the same time, avoiding the built‐up electric field. This again

Similar to other emerging photovoltaic technologies, there are significant efforts on using organic molecules, like PCBM and Spiro‐OMeTAD as electrodes. In this section, we will show

is advantageous for solar cell operation since it will be able to collect more charges.

V‐1 s‐1 and its decay started early at about 5 ps. For intensity of 1.7 × 1013 ph/cm2

V‐1 s‐1 and decay did not start until after 300 ps. This shows how

per pulse, the mobility is

V‐1 s‐1, **Figure 5a**) and MAPbI3/

mobility is 50% slower in half of nanosecond [48].

around 5 cm2

126 Nanostructured Solar Cells

bare MAPbI3.

**3.2. Injection into organic electrodes**

mobility is measured 25 cm2

MAPbI3/Al2O3 at three different intensities. For 7.4 × 1013 ph/cm2

300 ps. The difference between the mobility of MAPbI3 (20 cm2

**Figure 6.** Transient photoconductivity of (a) neat MAPbI3, MAPbI3/PCBM, and MAPbI3/Spiro‐OMeTAD per photon ab‐ sorbed per pulse (*λ*pump = 590 nm) up to 1 ns and (b) of neat MAPbI3 and MAPbI3/with 7 ns time window. Reprint with permission from Ref. [47]. Copyright 2014, American Chemical Society.

Also plotted in **Figure 6(a)** is the transient photoconductivity of MAPbI3/Spiro‐OMeTAD. Spiro‐OMeTAD is an organic hole transporting material that is widely used for highly efficient perovskite‐based solar cell material despite its reported very low conductivity, 10‐8 S/cm [50]. The measured mobility from this sample is about 5 cm2 /Vs, which is a third of that obtained from the MAPbI3 only, but it stayed constant for 1 ns. Upon photoexcitation, electrons and holes are generated in the perovskite material. In the presence of an acceptor material that has favorable band energy alignment, charges may be injected. In this case, there is a 0.5 eV difference in the band edge between the valence bands of perovskite and Spiro‐OMeTAD, leading us to conclude that there is an efficient hole injection. The injection rate is ultrafast since the mobility is reduced to three times in the earliest time scale. This is also supported by the fact that decent PCE are reported to these devices, which means that holes are really transferred from perovskite to Spiro‐OMeTAD. Moreover, since holes are injected to the Spiro‐ OMeTAD and the conductivity of this material is very small, this means that the mobility of 5 cm2 /Vs should be the mobility of electrons. This value of electron mobility is again different from the sample presented in the previous section. However, we reiterate that different preparation conditions yield different film quality manifesting as different values in their mobility, among other properties. As to the flat transient photoconductivity trace of this sample, it can be understood such that electrons remain in the perovskite material and did not encounter either defects where they can be trapped or holes which they can recombine with.

The transient photoconductivity trace of MAPbI3/PCBM is characterized by mobility similar to MAPbI3, 15 cm2 /Vs, but is decaying to almost a third in 1 ns. The initial value of the mobility implies that both electrons and holes are generated in the perovskite. The decay could then be assigned as due to second order geminate recombination. However, this type of recombination is excitation dependent. If this is the case, both the transient photoconductivity of MAPbI3 and MAPbI3/Spiro‐OMeTAD should also be decaying at the same rate since the excitation fluence used for these three samples are the same. Since this is not the case, one can discount this possibility. In addition and as shown in **Figure 6(b)**, where a higher excitation fluence is used for MAPbI3 and MAPbI3/PCBM but for a longer time window of 7 ns, the decay in MAPbI3/ PCBM is faster than MAPbI3 only, showing that there is an additional mechanism that causes the decay other than the second order recombination. It has been reported that the energy difference between the conduction bands of PCBM and perovskite is only 0.2 eV. As such, injection is still possible, but unlike in perovskite/Spiro‐OMeTAD interface, this injection can be slower since the driving force is at least two times less. This slower injection rate of electrons from perovskite to PCBM could be one of the mechanisms of the decay. Furthermore, one should also take into account the electron mobility in PCBM, which is 10‐3 cm2 /Vs [51–55]. This implies that while electrons are slowly injecting into the PCBM, the low mobility of the PCBM causes the electrons to be pinned at the interface. In this scenario, holes that are in perovskite could easily recombine with the electrons pinned at the interface of PCBM which could also have a ns time scale. We surmised that the ns decay of the transient photoconductivity shown in **Figure 6** is a convolution of the electron injection and recombination of the pinned electrons at the interface with the holes left in the perovskite material both occuring in the time scale of few ns.
