Statistical Paradigm for Organic Optoelectronic Devices: Normal Force

Subscriber access provided by HACETTEPE UNIVERSITESI KUTUPHANESI

Article

A statistical paradigm for organic optoelectronic devices: normal force testing for adhesion of OPVs and OLEDs Lindsay Vasilak, Silvie M Tanu Halim, Hrishikesh Das Gupta, Juan Yang, Marleen Kamperman, and Ayse Turak ACS Appl. Mater. Interfaces, Just Accepted Manuscript • DOI: 10.1021/acsami.6b15618 • Publication Date (Web): 21 Mar 2017 Downloaded from http://pubs.acs.org on March 27, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

ACS Applied Materials & Interfaces is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

A statistical paradigm for organic optoelectronic devices: normal force testing for adhesion of OPVs and OLEDs Lindsay Vasilak,† Silvie M. Tanu Halim,‡ Hrishikesh das Gupta,† Juan Yang,¶ Marleen Kamperman,¶ and Ayse Turak∗,† †Department of Engineering Physics, McMaster University, Hamilton, ON, Canada ‡School of Engineering Practice and Technology, McMaster University, Hamilton, ON, Canada ¶Department of Physical Chemistry and Soft Matter, Wageningen University ,Wageningen, Netherlands E-mail: [email protected] Phone: +1 905 525-9140 ex 23348 Abstract In this study, we assess the utility of a normal force (pull-test) approach to measuring adhesion in organic solar cells and organic light emitting diodes. This approach is a simple and practical method of monitoring the impact of systematic changes in materials, processing conditions, or environmental exposure on interfacial strength and electrode delamination. The ease of measurement enables a statistical description with numerous samples, variant geometry, and minimal preparation. After examining over 70 samples, using the Weibull modulus and the characteristic breaking strength as metrics, we were able to successfully differentiate the adhesion values between Alq3 and

1

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

P3HT:PCBM interfaces with Al, and between two annealing times for the bulk heterojunction polymer blends. Additionally, the Weibull modulus, a relative measure of the range of flaw sizes at the fracture plane, can be correlated to the roughness of the organic surface. Finite element modelling of the delamination process suggests that the out-of-plane elastic modulus for Alq3 is lower than the reported in-plane elastic values. We suggest a statistical treatment of a large volume of tests be part of the standard protocol for investigating adhesion to accommodate the unavoidable variability in morphology and interfacial structure found in most organic devices.

Keywords adhesion testing, organic electronics, interfaces, metal contact delamination, degradation, Weibull statistics

1

Introduction

Heterojunctions are inherent in and essential to all molecular optoelectronic devices. The interfacial region between the organic active layer and the inorganic electrodes play a primary role in device performance: effective charge injection/extraction can only occur at high quality interfaces at the electrodes. Additionally, a key failure mechanism for organic devices is the development of dark spots and catastrophic failure when the top or back contact metal electrode delaminates from the organic surface. 1 The adhesion between the metal contact and the organic active layers, therefore, is a measure of the quality and durability of the device, as degradation at the top contact proceeds more rapidly than any other failure process. 1 With the rapid rise and development of stretchable organic solar cells (OPVs) 2 and organic light emitting diodes (OLEDs), 3–5 this problem becomes even more significant. Electrode delamination in stretchable electronic devices is liable to occur when stress builds up in the device beyond the interfacial adhesion strength when stretched. This can limit the 2


Page 2 of 32

Page 3 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


stretchability of certain device structures dependent on the interface properties. It is crucial to have a method to rapidly assess the integrity and adhesion of this interface due to its importance to the performance and stability of organic devices. However, the investigation of interfacial adhesion in organic devices is complicated. In recent years, several approaches have been used to examine adhesion and reliability in OPVs and OLEDs, including double cantilever beam (DCB), 6,7 four point bending (FPB) 8,9 and atomic force microscopy (AFM). 10,11 The adhesion strengths values derived from these techniques vary significantly when applied to organic device materials. For example, seemingly similar preparation conditions for P3HT:PCBM/PEDOT:PSS interfaces have resulted in adhesion strength values that range from ∼ 0.4 mJ2 upto ∼ 40 mJ2 . 6,8,10 This variation of over three orders of magnitude most likely stems from the complicated nature of morphology in bulk heterojunction organic devices. The spontaneous phase separation of the immiscible donor and acceptor molecules leads to a random distribution of phase separated sections of varying shape and size, with a heterogeneous dispersion of components throughout the film. 12,13 Despite best efforts to recreate identical processing conditions, variations in morphology and interfacial structure are significant and unavoidable. 14 Additionally, organic devices always contain a distribution of flaws at the interface, due to the variability in organic film deposition and surface roughness. This results in a random distribution of dark spots 1 during device operation: the weakest spots delaminate first and form darkened regions that decrease the performance, and eventually lead to device failure. Removing and redepositing the electrode changes the location of these delaminated spots, 1 suggesting a large variation in surface structures, even with efforts to achieve identical processing conditions. A method of dealing with such wide sample variability is to adopt a statistical approach by performing a large number of tests. To effectively interpret adhesion and reliability of interfaces in organic devices, it would be useful to have a technique that is easy and quick to carry out, that applies to a wide variety of samples, that does not require special sample preparation (i.e. can be performed

3



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 32

in-situ) and that uses a statistical approach to validate the impact of changing processing conditions, changing environmental conditions, or even changing electrical field distributions over time. One such method, the normal force approach, shown in Fig.1, has wide use in engineering applications for quality control due to its low cost, simplicity of use and applicability to a wide variety of samples. 15 It is also commonly used for low Young’s modulus materials such as flexible epoxies, membranes, and polymers. 16–21 The major underlying feature of the normal force test test is that the failure load depends on the nature and distribution of the defects at the interface. 15 As described above, organic devices will always contain a distribution of flaws at the metal-organic interface. Upon loading such an interface, the weakest flaw will fail first. This point will be the initiating locus of failure, which causes the two surfaces to separate at some stress level, σyield . This so called weakest-link-in-the-chain concept 22 leads to significant variability in the measured breaking strength values. However, this type of failure is well known in mechanical testing, and the probability model for failure in a normal force pull test has been shown to follow Weibull statistics. 15,23–26 According to the Weibull model, the probability that the interface will delaminate under normal force in a stress range from 0 to σ is related to the probability that a subarea of the interface will fail in a given range σ + dσ. The total probability of failure can be expressed as a cumulative distribution function,

P (σ)cdf =

Z

σ 0

m σ − σu p(σ)dσ = 1 − exp − σ0

(1)

where σu is the stress below which failure cannot occur (typically 0), σ0 is the characteristic stress 25 (a scaling factor), and m is the Weibull modulus, which gives the degree of spreading in the data 25 and is generally thought to be related to the distribution of interfacial flaw

4


Page 5 of 32


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 32

sizes. 27

If the data follows a Weibull distribution, a double logarithmic plot of the inverse cumulative probability distribution, known as the unreliability, 22 as a function of the breaking strength will show a straight line with the Weibull modulus as the slope, and the characteristic Weibull stress as the intercept.

ln ln

1 1 − P (σ)cdf

= m ln σ − m ln σ0

(2)

Weibull analysis is a significant component of failure and reliability testing in a variety of fields. 22 In such an analysis, samples are repeatedly stressed to some failure condition (e.g. fracture, 28 dielectric breakdown, 29 loss of luminance 30 ), and the distribution of failure times or stresses is described with a Weibull distribution. The parameters extracted from a Weibull lifetime analysis can therefore be compared to assess the impact of changing a variety of factors, including processing conditions 31,32 , electrical stress, 33,34 thermal stress, 35,36 or degradation, 30,37 because the shift in the Weibull distribution indicates a change in the probability of failure. 26 Though most commonly applied in adhesion testing to normal force, scratch, and peel tests; Weibull statistics have been used in four point bending for systems where there was uncertainty about the distribution of flaws in the material. 38 As there is also considerable uncertainty in the distribution of flaws and phases in organic optoelectronic devices, Weibull statistics should apply to the analysis of organic devices. The key value derived from a normal force test, the critical force at which the surfaces pull away from each other, is in principle related to the adhesion between the surfaces. 39 This peak pull-off force or breaking strength value has no inherent meaning as the crack initiation defect and stress state in the failure plane are not known; 40 however, it is useful for comparison between similar types of systems under changing conditions. Turunun et al, 16 for example, used the relative breaking strength values to reveal the deteriorating effects of different environmental exposures, and Ge et al. 18 examined the effect of surface treatments 6


Page 7 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


on the adhesion of metallizations on photoresists. Due to the wide variability in the measured force values, the standard practice is to treat these tests statistically, with multiple tests run and data-censoring techniques applied to ferret out unwanted failure modes. 15 In this contribution, we have applied a variation of the standard normal force pull test to organic optoelectronic devices to assess its suitability as a quality control metric, able to distinguish between changing variables that would be of interest in optimizing device performance and stability. As the normal force method is prone to wide-spread data scatter, the most successful approach is to assess the system statistically, evaluating the results of large numbers of tests, using real-time video capture to censor the data appropriately. Using this approach, we were able to successfully differentiate between different metal-organic interfaces and between different processing conditions. When combined with a Weibull analysis of the breaking strength and expected failure, one could use the normal force method to examine the influence of changing processing parameters, exposure to electrical stress, or environmental conditions in a device directly, rather than in idealized samples that approximate the conditions of the sample. Adopting a statistical treatment of a large volume of tests should be part of the standard protocol for investigating adhesion, to accommodate the unavoidable variability in morphology and interfacial structure found in most organic devices.

2

Materials and Methods

2.1

Normal force measurement sample fabrication

The sandwich samples for normal force testing were produced by spin coating and physical vapour deposition on ITO (TFD, 20-32Ω/sq). The ITO substrates were machine diced into 3x3mm square samples. The substrates were cleaned with acetone and ethanol in an ultrasonic bath, rinsed with deionized water, and dried in a N2 stream. To limit sample variation as much as possible, multiple substrates were mounted together using double sided carbon tape on one platen for deposition of organic and inorganic layers. Multiple platens 7



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 32

were then mounted together for thermal evaporation (see Fig. 2(a)). There were roughly 18 samples per batch for each run of deposition. All materials were purchased from Alfa Aesar or Sigma Aldrich and have been used as received. All samples consisted of an organic layer, either 8-tris(hydroxyquinoline aluminum) (Alq3 ) or poly(3-hexyl-thiophene) and [6,6]-phenyl C61-butyric acid methyl ester ((P3HT:PCBM), sandwiched between ITO coated glass and a thermally evaporated Al layer, as shown in Fig. 2(b). (a)

(b)

v=50 +m/s

(c)

v=50+m/s 100 nm Alumunium

Tape Aluminium Organic layer ITO coated glass Tape

1 nm organic fracture plane

99 nm organic layer 100 nm ITO Encastre

500 nm

3 mm

Figure 2: (a) Schematic of multiple substrates mounted on a platens for organic and metal deposition (b) Schematic of the sample structure in normal force apparatus (c) Geometry and boundary conditions used in the Finite Element models For the Al-Alq3 samples, the deposition was performed by e-beam evaporation (NexDep, Angstrom Engineering) with a deposition rate of 0.18Å/s for Alq3 and 0.27 Å/s for Al, as determined from a calibrated quartz crystal monitor (QCM). The substrate temperature during growth was controlled by a resistive heater, with temperature oscillation less than 1o . The vacuum conditions during evaporation were typically in the range of 10−6 mbar. Thickness of each layer is 100nm. For the P3HT:PCBM samples, all sample preparation was performed in a custom OPV glovebox system (LC Technology). A 1:0.9 weight ratio (1.9 wt%) solution were dissolved in dichlorobenzene at 40o C. Dynamic spin coating (Speciality Coating Systems, SCS G3) at 2000 rpm was used to produce an ∼100nm thick film on an ITO substrate platen heated to ∼ 130o C. The samples were processed prior to deposition of the metal layer to ensure bulk heterojunction morphology and interfacial adhesion similar to that found in organic photo8


Page 9 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


voltaic devices. 41 The samples were solvent annealed for 15mins under cover immediately after deposition. Samples were then subjected to either 5 or 15mins of thermal annealing on a hotplate at ∼ 120o C (pre-annealed). 100nm of Al were deposited by vacuum thermal evaporation (MBE Komponenten) with a deposition rate of approximately 0.17Å/s, determined from a calibrated quartz crystal monitor (QCM). The vacuum pressure of the chamber attached to the glovebox was maintained at ∼ 10−6 mbar during deposition. After deposition, all samples were stored in vacuum sealed sample bags until the normal force measurements could be performed. Although the Al-Alq3 samples were exposed to air briefly after deposition, the Al-P3HT:PCBM samples were deposited and stored in sealed sample bags before being removed from the glovebox. A total of 68 P3HT:PCBM samples annealed to 15mins, 30 P3HT:PCBM samples annealed to 5mins, and 55 Al-Alq3 samples were tested. The data was censored to eliminate samples where the failure occurred at an interface other than the one of interest or where the mounting tape came into contact (see supporting information section SB), which resulted in 34, 10 and 16 measurements for the three sample types, respectively. Sample layer thicknesses from both evaporation and spin coated were calibrated using ellipsometry(M-2000 J.A. Woolam) and profilometry(Tencor Alpha Step 200), with 2% thickness errors.

2.2

Normal force adhesion testing

The normal force testing apparatus consists of a Futek strain gauge load cell (LSB 200) mounted on a linear actuator (Thorlabs Z825B) acting as a movable stage. The actuator area was larger than the sample size to ensure uniform pressure over the entire sample surface. The sample is mounted on a fixed stage below the actuator. The actuator approaches and retracts from the sample at a velocity of 50µm/s. When the pre-load force of ∼ 0.7±0.15 N is reached, the actuator dwells for 10s before retraction begins. The real-time movement of the stage is captured by a digital camera, pre- and post-delamination images were taken using an ocular microscope with digital camera (Canon IXUS 55) and adjustable light source. 9



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 32

To accommodate sample misalignment, Tesa double sided foam tape was used on both sides of the sample (Fig. 1 (a)). Real-time video, and pre and post-delamination images were analyzed using Image J 42 to determine misalignment, percent surface delamination and sample uniformity.

2.3

Scotch tape tests and electrical characterization

Al/P3HT:PCBM/PEDOT:PSS/ITO devices and samples were produced and characterized as described previously, 41 with two different pre-annealing times. Briefly, the anode consisted of ITO coated glass substrate, modified by spin-coating a diluted poly(3,4-ethylenedioxythiophene: poly(styrene sulfonate) (PEDOT:PSS, Baytron P AI 4083 from HC Starck) layer (60 nm). The P3HT:PCBM and Al metal electrode were deposited in a fashion similar to those for the adhesion tested samples. Current-voltage curves were measured using a Keithley 238 source-measure unit under simulated AM 1.5 global solar irradiation with an ORIEL sun simulator with 100 mW/cm2 to extract the basic solar cell characteristics from devices of approximately 0.14cm2 size. The spectral mismatch factor between the sun simulator and external quantum efficiency varies between 1.0 and ∼0.85, depending on the sample. Quoted values are not corrected for this mismatch. To obtain a statistical comparison of the performance, 8-12 samples for each annealing time were tested. Samples produced at the same time by the same production method were then subjected to a qualitative Scotch tape peel test (Scotch 3M) using the apparatus described in the supporting information section SE. The peel-off percentage was determined by Image J. 42 These latter samples were not masked to produce pixels, resulting in a Al coverage area on the organic active layer of roughly 1.5cm2 .

10


Page 11 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


2.4

Statistical Analysis

All statistical analyses (ANOVA, Bootstrapping, Wilcoxon rank and Weibull) were performed using the appropriate packages in the statistical computing language R. 43 The Weibull parameters, scale factor and Weibull modulus, were extracted from a maximization of the log-likelihood method, taking the linear fitting values of a double logarithm of the cumulative Weibull distribution function. 22,24,26,44 The Weibull parameters are used to extract the average expected failure stress using the Gamma function. 15,25 The 10% and 50% failure stress quantile denotes the stress value for which 10% or 50 % of the population has failed, which is also found from the log-likelihood equation. 24,26 p values reported in the text refer to the level of marginal significance within a statistical hypothesis test representing the probability of the occurrence of a given event. For all statistical calculations in this contribution, the null hypothesis was assuming that there was no difference between the measured breaking strength values. A p value

Statistical Paradigm for Organic Optoelectronic Devices: Normal Force

Recommend Documents