2.1.

Solar-Cell Fabrication

OSCs were fabricated in the inverted structure $\mathrm{ITO}/\mathrm{ZnO}/\mathrm{P}3\mathrm{HT}:\mathrm{PCBM}/{\mathrm{MoO}}_3/\mathrm{Ag}$ (Fig. 1). Glass substrates covered by prepatterned ITO (Ossila) were cleaned in acetone and isopropanol, for 30 min each, using an ultrasonic bath. Zinc-oxide layers were spin-coated from ethanol ZnO nanoparticles dispersion (Sigma-Aldrich) and were annealed at 500°C in air. Poly(3-hexylthiophene-2,5-diyl) (P3HT, Sigma-Aldrich) and phenyl-C61-butyric acid methyl ester (PCBM, Sigma-Aldrich) at mass ratio 1:0.8 were dissolved in the 1,2-dichlorobenzene ($60\mathrm{mg}/\mathrm{ml}$) by 4 h stirring at 60°C and 18-h aging of the solution at 40°C. The active layers were spin-coated at 1000 rpm under nitrogen atmosphere.

Fig. 1

Scheme of organic solar-cell stack.

Four conditions of bulk-heterojunction (BHJ) preannealing (110°C, 130°C, 150°C, and 170°C for 30 min) were applied to obtain various qualities of solar cells. Fifth group of solar cells was nonannealed (NA). Overheating of P3HT:PCBM active layer leads to crystallization of PCBM.^28–31 Using an optical microscope (model Keyence VHX-1000), the surface of P3HT:PCBM layer was observed and even for 130°C small PCBM crystals were found [Fig. 2(c)], similar to these observed by Chang et al.²⁹ Large amount of relatively small crystals (few micrometers in size) was observed for samples annealed at 150°C [Fig. 2(d)]. The largest crystals, but a small amount of them, were observed for samples annealed at 170°C [Fig. 2(e)]. Crystal size is relatively large when compared to the active layer thickness of $\sim 300\mathrm{nm}$. Thus, such a large size of PCBM crystals is expected to act as defects in the solar-cell structure. For NA and annealed at 110°C samples, the dust particles rather than PCBM crystals were observed.

Fig. 2

Optical images of P3HT:PCBM surfaces (a) NA and annealed at various temperatures (b) 110°C, (c) 130°C, (d) 150°C, and (e) 170°C. Lines in the images (a), (c), and (e) show the edges of ITO/glass.

Last two layers, i.e., ${\mathrm{MoO}}_3$ ($\sim 12\mathrm{nm}$) and silver electrodes ($\sim 120\mathrm{nm}$) were thermally evaporated under vacuum of $2\times {10}^{-6}\mathrm{mbar}$. Active area of fabricated inverted OSC was defined by intersection of electrodes and was equal $4.5{\mathrm{mm}}^2$. Samples were not encapsulated. Five groups (110°C, 130°C, 150°C, 170°C, NA) of solar cells with various quality of BHJ were fabricated and were under investigation of light-soaking effect. Solar-cell properties measured in STC (Standard Test Conditions—spectrum AM1.5, $1000\mathrm{W}/{\mathrm{m}}^2$, 25°C) using sun simulator (model #SS80AAA—Photo Emission Tech. and $I–V$ Tracer Auxiliary Unit—PV Test Solutions) and Keithley 2401 Low-Voltage Source Meter were presented in Table 1.

Table 1

As fabricated solar-cell properties (maximum, minimum, average, and standard deviation of η, Jsc, Voc, FF) measured under standard test conditions (five to six solar cells per group).

2.2.

Light-Soaking Measurement Procedure

Study of the light-soaking effect is nontrivial, due to the heating of the cells during exposure to the illumination. To allow a proper interpretation of results, the following measurement procedure was applied. Six solar cells on each substrate were measured always in the same order under the same $I–V$ sweep conditions, such as direction of $I–V$ sweep scan from $-1\mathrm{V}$ (reverse bias) to $+1\mathrm{V}$ (forward bias) and scan rate $1\mathrm{V}/\mathrm{s}$. These conditions ensure the reliable scan of $I–V$ curve for fabricated OSCs. All samples were stored in the dark in the air atmosphere between the measurements. Between the $I–V$ sweeps of solar cells from one substrate, a 20-s delay was applied for cooling down the OSCs and on the other hand for cooling down the photodetector in solar simulator, which was used for the illumination intensity monitoring during the $I–V$ sweep.

During each measurement day the first measurement of the dark-stored sample was called “before light soaking” (BLS). After that OSCs were light soaked under solar simulator (AM 1.5 spectrum, $1000\mathrm{W}/{\mathrm{m}}^2$ irradiation) for 2 min. Next, $I–V$ curves sweeps were performed after 20 s delay (cooling-down as previous). This cycle procedure was repeated up-to 10 min of total light-soaking time. Last $I–V$ sweeps (after 10 min of LS) were called as “after light soaking” (ALS) (Fig. 3). Due to a large amount of data and for the clarity of reporting results, all photovoltaics parameters are presented as an average and standard deviations of five to six solar cells for each group.

Fig. 3

Light-soaking measurement procedure performed during every measurement day.

3.1.

Improvements Due to Light-Soaking Effect

In Sec. 2, details of light-soaking measurement procedure were described. Due to various annealing conditions of OSCs, various properties of samples were obtained (efficiency in range from 0.69% up to 1.86%). LS improves the density of short-circuit current ($J_{\mathrm{sc}}$), FF, and the efficiency, while the open-circuit voltage ($V_{\mathrm{oc}}$) is constant. In Fig. 4, efficiencies of samples as a function of light-soaking time are presented as fabricated and aged solar cells.

Fig. 4

Efficiency as a function of light-soaking time for (a) fabricated and (b) aged OSCs annealed at various temperatures (each point is an average of five to six solar cells, dashed lines are guide to the eye).

After few minutes of illumination light-soaking effect is saturated for both as fabricated and aged samples. This is an evidence that total time of 10 min of LS is enough for saturation and proper analyses of LS in case of investigated solar cells. Higher photovoltaic properties were obtained for the samples annealed at low temperatures (110°C and 130°C) than for the overheated (150°C and 170°C) and the NA solar cells [see Figs. 4(a) and 4(b)]. During the first measurement day, relative efficiency improvement due to light-soaking effect was higher for the samples, which exhibit lower photovoltaic properties before being irradiated. It means that the lower the quality of the sample, the higher improvement due to LS could be expected.

3.2.

Inspection of Light-Soaking Effect on Device Lifetime Determination

Device lifetime determination for OSCs is a key issue for proper analysis of degradation data. Difference of device lifetime determined from measurements before and after LS is significant. In Fig. 5, the solar-cell efficiency as a function of degradation time measured as BLS and ALS is presented. During the degradation decreasing of all photovoltaics parameters was observed and partial recovery of $J_{\mathrm{sc}}$, FF, and efficiency was noticed due to light-soaking effect, while $V_{\mathrm{oc}}$ was constant. Lifetime is defined as a time after which the initial efficiency has fallen to 80% of this initial value ($T_{80}$).²⁷ NA OSCs and annealed at low temperatures are more stable than these samples annealed at high temperatures.

Fig. 5

Efficiency decay for samples annealed at 110°C (black) and 170°C (red) measured before (solid circle and solid upward triangles) and after 10 min LS (open circle and open upward triangles). Each point is an average of five to six solar cells divided by average at first measurement day.

For both groups of samples (low annealed and overheated) decay efficiency BLS is more rapid than for efficiency ALS. It means that lifetimes determined from the BLS decay curve are shorter compared to lifetimes determined from the ALS decay curve. Estimation of $T_{80}$ values from the BLS and ALS results were summarized in Table 2. The apparent improvement of lifetimes in ALS results means that even unintentional illumination of samples may disrupt the proper measurements and analysis of stability.

Table 2

T80 values for samples annealed at various temperatures for measurement before and after LS.

3.3.

Changes in Magnitude of the Light-Soaking Effect During the Aging of Solar Cell

As was mentioned, relative improvement due to the light-soaking effect is related to the quality and amount of defects in the solar-cell structure. With increasing degradation time, defects were created and the quality of solar cell decreased. In Fig. 6, the relationship between the ALS efficiency divided by the BLS efficiency and the degradation time was presented.

Fig. 6

Efficiency measured after 10-min LS divided by efficiency measured before LS as a function of time for annealed at 110°C (black squares), 170°C (purple downward triangles), and for NA (green diamonds) samples. Each point is an average of five to six solar cells ALS divided by average BLS.

In the first measurement day, relative improvement of efficiency due to LS was 8% and 27% for sample annealed at 110°C and at 170°C, respectively. After 100 days, these improvements increased up to 100% and 138%, respectively. During the degradation process, trap states and defects are created due to different paths of degradation, e.g., photooxidation, diffusion of atoms, reorganization of BHJ, and many others.^32–37 Some of these trap states, close to the ITO/ZnO, could be filled up during the light-soaking illumination.^19,22 On the other hand, the trap states, close to the ZnO/P3HT:PCBM, could change the induced interfacial dipoles.^23,24 Presented data do not consider which interface (ITO/ZnO or ZnO/P3HT:PCBM) plays the key role in the light-soaking effect but highlight the need for taking into account the LS effect in the analysis. The impact of light-soaking effect is increasing in time and it should be considered for proper and reliable measurements and stability analysis.

3.4.

Photodegradation Induced by Light-Soaking Measurement Procedure

Photo-induced degradation is common and well-known in P3HT:PCBM solar cells.^33–35 To investigate how the light-soaking effect and measurement procedure, which was applied, induce any additional photodegradation, one of the samples annealed at 130°C were measured only once in each measurement day and was not exposed to the illumination for LS. It minimized exposure of the solar cells to the light illumination and these measurements were called as never light soaked (NLS). Total illumination time under AM1.5, $1000\mathrm{W}/{\mathrm{m}}^2$ at the end of day #51, and $T_{80}$ values were summarized in Table 3.

Table 3

T80 values determined for measurement BLS, ALS, and NLS for solar cells annealed at 130°C.

The most rapid decay of normalized efficiency curve occurred for BLS results (Fig. 7). The most flat decay was obtained for ALS measurements. A sample that was NLS was more stable, than the sample that was light soaked and was measured as BLS. It means that on the one hand LS reduces the negative impact of degradation and on the other hand LS induces the degradation of P3HT:PCBM solar cells.

Fig. 7

Efficiency decays for samples measured before LS (black squares), after 10 min LS (red circles), and for a sample never exposure for light-soaking illumination (blue triangles). Both groups of samples were annealed at 130°C. Each point is an average of five to six solar cells divided by average at first measurement day.

3.5.

Light Soaking and Quality of Samples

In our study, the following observation could be made: if the sample is of good quality, the light-soaking effect is weak and the light-soaking effect increases with time (degradation of sample). In Fig. 8, efficiency measured after 10-min LS versus efficiency measured before LS were presented as an average of five to six solar cells from each of five groups of samples. This figure shows all results collected in 100 days and it illustrates the light-soaking effect without degradation and time aspects.

Fig. 8

(a) Efficiency ALS as a function of efficiency BLS. (b) Relative improvement of efficiency due to LS as a function of efficiency BLS. Each point is an average of five to six solar cells.

If the sample has a high BLS efficiency, the improvement due to LS will be weaker than for the sample that has lower BLS efficiency. The better the sample is the weaker light-soaking effect could be expected, and it is independent from time and degradation aspects. In Fig. 8(a), it could be observed that the slope of the graph decreases with increasing the efficiency, and thus, the relative improvement of efficiency is decreasing with the BLS efficiency [Fig. 8(b)]. It means that for the high-efficient and defects-free devices, LS is a minor effect and it could not be observed for as-prepared devices, but it could grow in time and should be considered and analyzed in the degradation studies.