Effects of processing additives in non-fullerene organic bulk heterojunction solar cells with ef fi ciency>11%

Shenkun Xie,Jinqiu Wng,Rong Wng,Dongyng Zhng,Huiqiong Zhou*,Yun Zhng,*,Defeng Zhou,*

a School of Chemistry and Life Science,Changchun University of Technology,Changchun 130012,China

b HEEGERBeijing Research&Development Center,School of Chemistry,Beihang University,Beijing 100191,China

c CASKey Laboratory of Nanosystem and Hierarchical Fabrication,CASCenter for Excellence in Nanoscience,National Center for Nanoscience and Technology,Beijing 100190,China

Key words:Organic solar cell Solvent additive Morphology Recombination Non-fullerene

ABSTRACT Here we investigate processing additive-dependent photovoltaic performance and charge recombination in organic bulk heterojunction(BHJ)solar cells based on a polymeric donor of PBDB-T blended with a non-fullerene acceptor m-ITIC.We fi nd that PBDB-T:m-ITICsolar cells exhibit good compatibilities with the utilized additives(DIO,CN,DPE,and NMP)in optimal conditions,can have a high charge dissociation probability approaching 100%(with DIO),leading to ultimate ef fi ciency>11%.Regardless of additives,we observe a dominant 1st order monomolecular recombination with insigni fi cant bi-molecular recombination or space-charge effects in these solar cells.Despite of impressive power conversion ef fi ciency(PCE),it is of surprise that Shockley-Read-Hall recombination is identi fied to play a role in device operation.Thus,it points to the necessity to mitigate the influences of traps to further boost the ef fi ciency in non-fullerene based organic solar cells.

Organic solar cells(OSCs)are among the most promising candidates for sustainable energy resources[1–3].The main advantages in OSCs include their low fabrication cost,light weight,high fl exibility,and large-area processing capabilities using roll-toroll printing etc.In recent years,the power conversion ef fi ciency of polymer solar cells(PSCs),barely 1%fi fteen years ago,hasexceeded 13%[4–13].It is well accepted that utilizing bulk-heterojunction(BHJ)structure in which the donor and acceptor materials are blended to form bi-continuous inter-percolated networks with abundant interfacial regions w arrantsefficient exciton dissociation for desired photoconductivity[13].Despite of recognized disadvantages such as weak absorption in the visible region,high synthetic cost,and limited tunability of energy levels,fullerene derivativeswith largenonplanar sphericalcon fi gurationshavebeen predominant electron-acceptor materials in the past tw o decades[14].Very recently,non-fullerene(NF)acceptors have attracted many attentionsfor OSCapplications.The problemswith fullerenes to largeextend can be circumvented in NF-based OSCs.For example,NFs acceptors have been show n with the high tunability on energetics through chemical modification with potentially low synthetic costs.Through judicious choices or modifications of molecular skeletons and fl exible side chains,the electronic and absorption propertiesarewell tailored[15,16].Among the diversity of NFacceptors,ITICwith the fused ring structure has gained great popularity show ing successes in delivering several record ef fi ciencies in NF-OSCs[17].Another successor ITICderivative IT-M with a very small electron-donating substituent(methyl)has yielded attractive PCEs>12%[18].In addition to PCEs,in some cases ITIC-based devices can exhibit more supreme stability than their analogue fullerene-based OSCs[18].Despite of these accomplishments,so far fundamental know ledge of the influences of processing additives on photovoltaic behaviors with these emerging NF acceptor materials still lacks,which may impede the advances in improving the PCEof NF-OSCs.

Here,we use low bandgap acceptor meta-alkyl-phenyl-3,9-bis(2-methylene-(3-(1,1-dicyanomethylene)-indanone))-5,5,11,11-tetrakis(4-hexylphenyl)-dithieno[2,3-d:20,30-d0]-s-indaceno[1,2-b:5,6-b0]dithiophene(m-ITIC,Fig.1),inspired by side-chain isomerism engineering on the alkyl-phenyl substituents of ITICreported by Yang et al.[19],to blend with a medium band gap polymeric donor poly[2,6-(4,8-bis(5-(2-ethylhexyl)thiophen-2-yl)benzo[1,2-b:4,5-b0]d ith iop h en e)-co-(1,3-di(5-thiophene-2-yl)-5,7-bis(2-ethylhexyl)benzo[1,2-c:4,5-c0]dithiophene-4,8-dione)] (PBDB-T,Fig.1)[20,21]as the test bed to investigate the photovoltaic performance and relevant opto-electrical properties.poly[(9,9-bis(3 0-((N,N-dimethyl)-N-ethylammonium)-propyl)-2,7-fl uorene)-alt-2,7-(9,9-dioctyl fl uorene)]dibromide(PFN-Br,Fig.1)as interlayer.

Fig.1.Chemical structures of utilized materials.

Devices were fabricated with conventional structure of ITO/PEDOT:PSS/BHJ layer/PFN-Br/Al.The PEDOT:PSS hole extraction layer(?30 nm)was attained by spin-coating on pre-cleaned glass/ITO substrates and dried at 160?Cfor 10 min in air.BHJsolutions were prepared by solubilizing the PBDB-T:m-ITIC blends in chlorobenzene with a total concentration of 20 mg/m L with adding additives at selected concentrations.The photoactive layers were attained by spin-coating the BHJ solutions on PEDOT:PSS-coated ITO substrates at a spin-rate of 1500 rpm.The thickness of active layers was controlled at around 110 nm.After spin-coating,the BHJ fi lms were thermally annealed on a hot plate at 160?Cfor 10 min.Then the PFN-Br solution(in methanol)at a concentration of 0.5 mg/m Lwas deposited atop the active layers to form a PFN-Br cathode buffer layer with typical thicknessof?5 nm.At last,Al cathode(ca.80 nm)was thermally evaporated on the PFN-Br buffer layer at a pressure of?1.0?10?4Pa.The effective device area was 4 mm2de fined by shadow masks.Based on 4 representative processing additives,including 1,8-diiodoctane(DIO),1-chloronaphthalene(CN),diphenyl ether(DPE),and N-methyl-2-pyrrolidone(NMP),we focused on their impacts on the photovoltaic behaviors and recombination in PBDB-T:m-ITIC BHJ solar cells.Of interest,the devices show a high compatibility with the choice of additives,revealed by the overall enhancements on PCE(compared to cells without additives)with non-substantial differences in device parameters at each condition.Based on optimizing D-A ratio,additives concentration,and processing solvents,the achieve the best PCE exceeding 11%using DIO.The BHJ fi lms with the compared additives exhibit resembling surface morphologies with only slight differences in surface roughness.While visibly changed surface morphology with increased structural sizes and non-uniformity can be observed upon increasing the concentration of additives.These features to some degree reconcile the observed variations on device parameters.Based on irradiation-dependent analysis,all concerned NF-OSCs are identi fied with a dominant monomolecular recombination associated with insigni fi cant bi-molecular recombination or space charge effect.Light-dependent photovoltage measurements suggest that Shockley-Read-Hall(SRH)recombination plays a noble role in operation of studied solar cells,revealed by non-ideality factors>1,which points to the necessity to further reduce the influence of traps.Our results not only provide an effective optimization route to attain satisfactory device characteristics but also enrich fundamental insights into the impacts of additives and relevant processing parameters on the photovoltaic performance in OSCs with emerging NFelectron acceptors.

Firstly,we examined the impact of D/A ratio of BHJblends on solar cell performance.Fig.2d show s adopted conventional device architecture.To enable efficient electron extraction,a thin layer of PFN-Br was utilized as the cathode buffer layer which was found to dramatically improve the FF(fi ll factor)and ultimate PCE.The improvements are probably due to the present Br?anions that can interfacially dope the electron acceptor,leading to improved charge extraction ef fi ciency at the cathode interface[22],Fig.2b shows representative J–V characteristics of solar cells with different D/A blend ratios(w/w)under standard AM 1.5G solar irradiation(100 m W/cm2).The corresponding EQE spectra are show n in Fig.2a.Table 1 summarizes detailed device parameters with different D/A ratios.All devices compared in Table 1 have an average thicknessof?120 nm determined by a pro fi lometer.Ascan be seen,the best D/A ratio was found at the 1:1 ratio with which the highest PCEof 9.78%was achieved with an open circuit voltage(Voc)of 0.92 V,a short current density(Jsc)of 15.70 m A/cm2,and a fi ll factor(FF)of 69.30%.Fig.2b compares external quantum efficient(EQE)spectra of solar cells at different D/A ratios.The change of the spectral shape of EQEat different D/A ratios can be understood from the thin fi lm absorption of neat donor and acceptor show n in Fig.2c.At a balanced ratio if 1:1(w/w),we attained a relatively high EQEin the spectral range of 620–760 nm.The order in EQE between devices roughly follow s that of Jscshowed in Fig.2b.The deviation between the Jscdetermined based on the integration of EQEand that from J–V curves is less than 5%,assuring the accuracy of extracted device parameters.

Fig.2.(a)EQEspectra of according devices.(b)Current density versus voltage characteristics of PBDB-T:m-ITICsolar cells at different D/A blend ratios(w/w).(c)Thin fi lm absorption of neat PBDB-T and m-ITIC acceptor.(d)Device architecture of solar cells.

Table 1 Device parameters of PBDB-T:m-ITICsolar cells with various D/Ablend ratios(w/w)under standard AM 1.5G solar irradiation(100 m W/cm 2).

Morphology and phase separation of photoactive layers are important concernsfor achieving desirable devicesperformance in OSCs.Next,we examine how the D/A ratio in BHJblends affects resultant fi lm morphology(Fig.3).As seen from the surface topographic images captured by atomic force microscopy(AFM)in[13_TD DIFF]Figs.3a and d,it appears that unbalanced D/A ratios cause an increase in surface roughness with larger structures.At more balanced D/A ratios,the surface smoothness of BHJ fi lms increases(Rq=4.35 nm at 1:1 and 4.61 nm at 1:1.3,[13_TD DIFF]Figs.3b and c),compared to that(Rq=5.58 nm at 1.5:1 and 4.77 nm at 1:1.5)in[13_TD DIFF]Figs.3a and d.Of interest,the roughness determined by AFM can be correlated to the Jscin respective solar cells.The unbalanced ratios show n in[13_TD DIFF]Figs.3a and d tend to result in donor-or electron-rich surfaces that may cause accelerated recombination and thus the reduction in Jsc,consistent with the results in Table 1.However,w hen compared to the evident variation of surface morphology,the Voc,FFand PCEare barely changed(Table 1).This seems to hint that the energetics and diode quality in solar cells(that basically determine the Vocand FF)are likely unchanged at different D/A ratios.

Now,we examine the effect of solvent additives on solar cell performance and BHJ fi lm morphology.For this purpose,we chose 4 representative solvent additives,namely DIO,CN,DPEand NMP.In the optimal conditions,all applied additives yield satisfactory photovoltaic performance that can be similarly ascribed to the identi fied mechanisms for improving the PCEin fullerene devices[23–26].For simplicity,these additives were compared based on the same D/A ratio(1:1).Detailed solar cell parameters with variations on solvent additives and their concentrations are summarized in Table 2.After adding 0.5%DIO,we achieved the highest PCE of 11.09%with Voc=0.915 V,Jsc=16.98 m A/cm2and FF=0.71.Follow ing the DIOdevice,the cell using NMPalso leads to a PCE reaching 10.8%.

It is interesting to note that the additives seem to affect the solar cell operation in distinct manners,e.g.,changing the amounts of DIO or NMPwas found to mainly affect FFand Voc,while different DPEdoses result in different Jscin solar cells.Detailed mechanism for this observation is yet not fully understood at this stage.It may be related to the nanoscale phase-separated morphology,phase purity or intermolecular stacking in respective BHJ fi lms.In comparison to the other three additives,the behavior of device using CN is less sensitive to the additive concentration,possibly due to the influence of boiling points.We further examined surface morphology of BHJ fi lms based on the optimal concentrations for these additives(0.5%for DIO,NMP,and 1%for CN,DPE)with corresponding AFM images show n in [13_TD DIFF]Figs.3e–h.The main morphological differences lie in the size of structures and uniformity,which can be correlated to D/A phase segregations inside the BHJwith which charge separation and eventually the Jscare mediated.Despite of the changed solar cell parameters with different additives,resembling surface morphologies the compared BHJ fi lms are observed with slightly changed roughness.Consistently,the Rq values of BHJ fi lms can be correlated to the Jscin according devices.

To shed light on exciton dissociation and charge collection processes in m-ITICbased devices,we performed light-dependent(Plight)photocurrent/photovoltage measurements.Fig.4a show s photocurrent density(Jph)as a function of effective voltage for solar cells measured under 100 m W/cm2irradiation with the voltage sweep range of?1.5–2 V.Here Jphis de fined as Jph=JL?JD,w here JLand JDare the photocurrent densities under illumination and in the dark condition,respectively.Veffis de fined as Veff=V0?Vbiaswith V0denoting for the voltage at which Jphiszero and Vbiasfor the applied external voltage bias.With this plotting method,the charge dissociation probability Pdisscan be estimated from Jphwith respect to the photocurrent in the saturation regime(Jsat)[27].Under short-circuit condition,Pdissis approximated according to the relation by Pdiss=Jph/Jsat[27].In our case,we chose Jsatat bias=?2 V and Pdissof 96%,94%,94%,and 93%was attained using DIO,CN,DPE,and NMP additives at respective optimal concentrations.This result indicates that the additive-processed solar cells can have a high exciton dissociation rate with efficient charge collection,possibly ascribed to favorable D/A phase segregations with interpenetrated carrier transport pathw ays.This is also in line with the observed large FFs with which strong recombination seems unlikely.From Fig.4b,the EQE spectrum of DIO devices is well distinguished from those of remaining solar cells,which may be correlated to the used lower amount of DIO additive and resultant differences in absorbance and photoconductivity.The highest PCEwith DIO is mainly ascribed to the gain of FFwith the excellent diode characteristics.Besides,the champion performance with DIO may also be linked to the highest Pdiss,which can originate from the optimal balance between phase segregation,domain purity,molecular ordering and carrier mobility,eventually leading to minimizing the recombination losses[28–31].

Fig.3.AFM topographic images of PBDB-T:m-ITICBHJ fi lmsat various D/Ablend ratios(w/w)of(a)1.5:1,(b)1:1(c)1:1.3,and(d)1:1.5./Ablend ratio:1:1 added with various solvent additives at respective optimal conditions.(e)DIO,0.5%.(f)CN,1%.(g)DPE,1%,and(h)NMP,0.5%.

Table 2 Device parameters of solar cells using various solvent additives and their concentrations under standard AM 1.5G solar irradiation(100 m W/cm2).

To further clarify the impact of additives,we investigated Plightdependent Jscand Vocto gain insights into charge recombination propertieswithin the devices.Fig.4c show s Jscof OSCs with various additives as a function Plight.a slope equaling 1 indicates a dominant monomolecular recombination and the deviation from the power of unity can arise from a couple of reasons including bimolecular recombination,space-charge effect,or variations in mobility between the tw o carriers etc.[32].We note that in all cases,the power is less than unity,indicative of the presence of bimolecular recombination or even space-charges[33,34].While the slope in the DIO and NMPprocessed devices approaches more to the unique,which may implying the decrease on bimolecular recombination with these tw o additives.The characteristics of Jscversus Plightcan be affected by the balance between charge carrier transport in BHJ fi lms.

In Plight-dependent photovoltage measurements,a slope equal to thermal voltage(kT/q with k being the Boltzmann constant and q being the elementary charge)or larger than kT/q in the BHJsolar cell can respectively indicate the dominant trap-free recombination or the influence of SRH recombination[30].Fig.4d show s characteristics of the determined Vocagainst Plightfor various solar cells.The slopes are 1.46 kT/q,1.42 kT/q,1.36 kT/q,and 1.46 kT/q for devices with DIO,CN,DPE,and NMPrespectively.To our surprise,the SRH recombination tends to be present in all these devices,indicative of the influences of traps even at the optimal conditions.It can be foreseen that further boosting of PCEs may be attained through fine materials puri fi cation or device engineering with mitigated SRH recombination.At this stage,we have no clear clues if these trapsare in the bulk of BHJ fi lmsor at the cathode interface with the presence of PFN-Br buffer layer.Further investigations may be required to elaborate.

In summary,we investigate the impact of solvent additives on photovoltaic characteristics in non-fullerene PTDB-T:m-ITIC BHJ solar cells.Based judiciousoptimizations,we attain the best PCEof 11.1%with using DIO additives.The attractive performance can be mainly attributed to the complementary absorption with favorable fi lm morphology that enables to achieve high photo-harvesting and satisfactory diode characteristics(or FF).The concerned BHJ solar cells show a good compatibility with respect to the choice of additives and in the optimal conditions,high charge dissociation probabilities>93%can be achieved,possibly due to pertinent phase segregation and interpercolated transport pathw ays.The surface roughness of BHJ fi lms can be tuned by the D/A ratio.Of interest,balanced D/A ratios are found to be associated with enhanced surface smoothness,which somehow correlates to the increased Jscin solar cells.Based on comparing the effects of four representative additives,we found that the recombination in operational devices is dominated by the monomolecular type along with minor bi-molecular recombination and/or space charge effects.Of note,in these optimized solar cells,pronounced SRH recombination is still observable,indicating the role of trapsunder device operation.The presented study not only provides a systematic route to attain satisfactory photovoltaic performance but also point to the necessity to mitigate the influence of traps to further boost the PCE in organic non-fullerene solar cells.

Acknow ledgm ents

D.Zhou thanks the National Natural Science Foundation of China(No.21471022).Y.Zhang thanksthe National Natural Science Foundation of China(No.21674006).H.Zhou thanks the Chinese Academy of Science(100 Top Young Scientists Program,No.QYZDB-SSW-SLH033)and the National Key Research and Development Program of China(No.2017YFA0206600).

Fig.4.(a)Photocurrent(J ph)as a function of effective bias(see de fi nitions in text)of solar cells using various solvent additives at their optimal concentrations.(b)Corresponding EQEspectra of BHJsolar cells.(c)Light-dependentJ sc and(d)V oc of additive-processed solar cells.