Research Article www.acsami.org
Stress Management in Thin-Film Gas-Permeation Barriers Andreas Behrendt,† Jens Meyer,‡ Peter van de Weijer,§ Tobias Gahlmann,† Ralf Heiderhoff,† and Thomas Riedl*,† †
Institute of Electronic Devices, University of Wuppertal, Rainer-Gruenter-Str. 21, 42119 Wuppertal, Germany Philips GmbH, Philipsstrasse 8, 52068 Aachen, Germany § Philips Research, High Tech Campus 7, 5656AE Eindhoven, The Netherlands ‡
S Supporting Information *
ABSTRACT: Gas diffusion barriers (GDB) are essential building blocks for the protection of sensitive materials or devices against ambient gases, like oxygen and moisture. In this work, we study the mechanics of GDBs processed by atomic layer deposition (ALD). We demonstrate that a wide range of ALD grown barrier layers carry intrinsic mechanical tensile stress in the range of 400−500 MPa. In the application of these GDBs on top of organic electronic devices, we derive a critical membrane force (σ · h)crit = 1200 GPaÅ (corresponding to a layer thickness of about 300 nm) for the onset of cracking and delamination. At the same time, we evidence that thicker GDBs would be more favorable for the efficient encapsulation of statistically occurring particle defects. Thus, to reduce the overall membrane force in this case to levels below (σ · h)crit, we introduce additional compressively strained layers, e.g., metals or SiNx. Thereby, highly robust GDBs are prepared on top of organic light emitting diodes, which do not crack/ delaminate even under damp heat conditions 85 °C/85% rh. KEYWORDS: atomic layer deposition, thin-film encapsulation, gas diffusion barrier, organic electronics, mechanical stress
■
INTRODUCTION Among the most demanding applications of gas diffusion barriers (GDBs), there are organic light emitting diodes (OLEDs) and organo-metal halide perovskite solar cells, which are commonly quoted to require GDBs with water vapor transmission rates (WVTR) on the order of 10−6 g/(m2 day),1,2 whereas conventional barriers used in food packaging range around 1 g/(m2 day). Atomic layer deposition (ALD) has been shown to afford outstanding thin-film GDBs on large-area substrates at relatively low processing temperatures ( 50 nm). A similar level of tensile stress has previously been reported for ALD grown Al2O3 layers.8,13 It has to be noted that the tensile stress in the layers builds up due to the ALD growth mechanism and is not a result of different coefficients of thermal expansion of the substrate and the GDB. In an estimate of the stress resulting from different coefficients of thermal expansion for ALD grown Al2O3 and the Si wafer of 4.2 ppm/K17 and 3 ppm/K, respectively, we would expect stress levels of only 20 MPa for layers grown at 100 °C, taking the elastic modulus of Al2O3 (180 GPa).16,18 The level of intrinsic tensile stress (400−500 MPa) in all of our ALD layers is significantly below the typical yield stress of a wide range of ALD grown layers (e.g., for Al2O3, a critical strain εc for the onset of cracking is on the order of 0.6−1%).19 With an elastic modulus of 180 GPa for Al2O3 prepared by ALD, a critical stress σc of 1−1.8 GPa can be estimated).
Figure 3. Scheme of an OLED that is encapsulated with a thin-film GDB (a). Photograph of an OLED (active area: 11.5 cm2) encapsulated with a NL GDB (thickness: 500 nm) right after encapsulation (b), and after storage at 85 °C/85% rh for 17 h (c). Scanning electron microscopy image of a crack in the GDB in the early stages of formation with a particle defect in the center of the crack (d), and onset of delamination and rolling-up of the GDB (e). Relative delaminated area of OLEDs with NL GDBs of varied thickness, after 17 h at 85 °C/85% rh (f). 4058
DOI: 10.1021/acsami.5b11499 ACS Appl. Mater. Interfaces 2016, 8, 4056−4061
Research Article
ACS Applied Materials & Interfaces
Figure 4. Schematic representation of stress reduction in the gas diffusion barrier assembly by addition of a compressively strained layer (a). Example of reduced overall tensile stress and membrane force in a combination of 95 nm Al2O3/TiO2 NL and a 150 nm thick compressively strained SiNx layer. The resulting membrane force does not depend on layer sequence (b). The relative delaminated area of OLEDs comprising various Al2O3/ TiO2 NL GDBs without stress management ((σ · h) > (σ · h)crit) and with stress management ((σ · h) < (σ · h)crit) by introducing 600 nm of SiNx or 1.9 μm of Al (c). Details of the layer sequence of the samples studied here are shown in Figure S3. Photograph of OLEDs after 17h under 85 °C/ 85% rh with a pristine 360 nm thick NL GDB (((σ · h) > (σ · h)crit), left), and with the addition of 600 nm SiNx CSL ((σ · h) < (σ · h)crit, right) (d). Note, the OLEDs were not electrically operated in this test.
OLED after storage at 85 °C/85% rh for 17 h shows severe damage due to delamination (Figure 3c). As evidenced by scanning electron microscopy (SEM), barrier layers tend to crack at starting points associated with particles (Figure 3d). In agreement with our findings, earlier reports have identified elevated stress levels in GDBs around particles, which may thus function as centers for the onset of cracking.8 The subsequent delamination and rolling-up of the barrier layer is a result of the intrinsic tensile stress determined above. Frequently, the entire OLED stack delaminates along with the GDB. The dot-like patterns in Figure 3e point to a partial delamination of the entire OLED stack along with the barrier. Similar roll-up motives are typically found to accompany stress relaxation in strained membranes.20 For a more detailed assessment of the stress in the layers and the related delamination phenomena, we encapsulated a series of nominally identical OLEDs with NL GDBs of increasing thickness. As the stress is vastly independent of layer thickness for a wide range of thicknesses (Figure 2), the membrane force (σ · h) increases linearly with layer thickness. We measured the relative delaminated area of the OLEDs after storage at 85 °C/ 85% rh for 17 h. Note, for each GDB thickness we studied nine OLEDs with an active area of 11.5 cm2, each. As shown in Figure 3f, for NL GDBs with a thickness hf ≤ 300 nm no delamination of the barrier/OLED is observed, whereas for GDBs with hf > 300 nm a linear increase of the delaminated area with membrane force is found. From this data, a critical layer thickness of 300 nm is derived for the onset of delamination, corresponding to a critical membrane force (σ · h)crit = 1200 GPaÅ. Especially, the relatively weak van der Waals bonds between the organic layers and between the organic layers and the electrodes are expected to affect the threshold for delamination (σ · h)crit. As discussed above, we were not able to identify a set of ALD processing parameters that would allow us to lower significantly the intrinsic stress in our as grown NL barriers (Figure 2). On the other hand, a minimum thickness of the NL barrier on the order of 300 nm has been found desirable to mitigate the
detrimental impact of dust or process-related particles, typically present on top of an OLED (Figure 1). Therefore, according to the results shown in Figure 3, the (σ · h) for a 300 nm thick NL barrier is located just at the threshold of delamination, which would be a serious source of reliability issues. Therefore, we aimed at reducing the overall tensile stress in the barrier layers by the introduction of a further, compressively strained layer (CSL), as schematically shown in Figure 4a. Possible candidates for CSLs are thermally evaporated metals, e.g., aluminum (compressive stress σ ≈ −72 MPa) or silver (compressive stress σ ≈ −15 MPa), and SiNx deposited by plasma enhanced chemical vapor deposition (PECVD) with σ ≈ −80 MPa. The stress data of these materials have been measured by wafer curvature experiments, as explained above. Details can be found in the Supporting Information (Figure S2). In thin-film mechanics, the total membrane force (σ · h)total of a multilayer can be derived by adding the contributions of the individual sublayers (σ · h)i:16 (σ ·h)total =
∑ σi·hi i
(2)
In general, we were able to show that the resulting (σ · h)total does not depend on the layer sequence of GDB and CSL (Figure 4b). In the specific example, 95 nm of NL (σ = 400 MPa, tensile) are combined with 150 nm of SiNx prepared by PECVD (σ = −80 MPa, compressive). As shown in Figure 4b, a lowered overall tensile stress of 100 MPa and a total membrane force of 243−247 GPaÅ is found, in agreement with a calculation according to eq 2, predicting a total membrane force of 260 GPaÅ. Along these lines, the membrane force in thicker NL GDBs is lowered from values significantly above (σ · h)crit = 1200 GPaÅ to levels below (σ · h)crit. In a further specific example, the membrane force (σ · h) of more than 2000 GPaÅ present in a 500 nm thick NL layer, was successfully reduced to only 740 GPaÅ (significantly below (σ · h)crit) by the insertion of a 1.9 μm thick thermally evaporated Al layer (red symbols in Figure 4c). In this case, the 1.9 μm thick evaporated Al layer was sandwiched between 300 and 200 nm thick NL layers. By a 4059
DOI: 10.1021/acsami.5b11499 ACS Appl. Mater. Interfaces 2016, 8, 4056−4061
ACS Applied Materials & Interfaces
Research Article
■
similar token, the membrane force (σ · h) of a 360 nm thick NL was lowered from 1440 to 960 GPaÅ by the introduction of 600 nm SiNx, deposited by PECVD (blue symbols in Figure 4c). In any case, the layer assemblies with (σ · h)total < (σ · h)crit did not show any crack formation/delamination after 17 h at 85 °C/85% rh (Figure 4d). To verify that the improved performance of the barrier does not simply result from improved gas-permeation properties due to the additional CSL, we added 120 nm of NL to the SiNx (600 nm)/NL (360 nm) sample, which resulted in a layer sequence of NL (120 nm)/SiNx (600 nm)/NL (360 nm) with a total membrane force of 1440 GPaÅ (>(σ · h)crit, similar to that of the neat 360 nm NL without the SiNx) (green arrow and symbol in Figure 4c). In this case, in spite of the presence of SiNx, after 17 h at 85 °C/85% rh, significant crack formation and delamination was found (in a set of four encapsulated OLEDs about 25% of the total active area was delaminated), showing the great importance of the thin-film mechanics. In earlier work, it has been shown that PECVD based SiNx alone does not provide any serious gas-permeation barrier properties.8 Finally, the performance of the above-discussed GDBs is evaluated in OLEDs under operation. To this end, nine OLEDs with a total active area of 103.5 cm2 have been either encapsulated with 600 nm of SiNx, only, or with a combination of 600 nm SiNx and 300 nm Al2O3/TiO2 NL. Please note, the OLEDs in the present work are so-called bottom emitting devices, where the light is extracted via the substrate and not via the top side. That is why the optical transmittance of the encapsulation layer on top of the OLED does not affect the functionality. As the materials in the NL and the SiNx are optically transparent, they could be applied to top-emitting or semitransparent OLEDs, as well. This would not be possible in case of an opaque metal as CSL. The initial areal density of so-called dark spots was around 1−2.5 dark spots per cm2. These dark spots occurred in or immediately after the manufacturing process of the OLEDs (before encapsulation). Dark spots are a well-known phenomenon in OLEDs, which has been widely discussed in the literature.21 Typically, the origin of dark spots is related to imperfections or defects in the electrode/organic interface and/ or to particle contaminants. Note, a limited number of small sized initial dark spots may be tolerable for OLED lighting modules, as long as these dark spots do not grow in number/ size over time. However, growing dark spots will finally render the entire OLED defective. A good encapsulation layer may therefore be judged according to its ability to prevent existing dark spots from growing further. To analyze the dark spot behavior in our encapsulated OLEDs, we subjected them to damp heat conditions (Figure S4). Strikingly, while in the OLEDs encapsulated with 600 nm of SiNx about 86% of the dark spots show continuous growth, in the OLEDs encapsulated with a combination of the SiNx and the Al2O3/ TiO2 NL only about 4% of all initial dark spots increase in size. This reflects the outstanding barrier quality added by the thick Al2O3/TiO2 NL, and the success of the stress management, due to the combination of the compressively strained SiNx and the tensile strained NL, which prevented crack formation and delamination of the barrier. Note, a residual level of particle related failure probability on the order of 3% was also found in Ca tests, as discussed above (Figure 1c).
CONCLUSIONS In summary, we have shown that Al2O3/TiO2 nanolaminates grown by ALD are suitable to form excellent gas diffusion barriers even in the presence of statistically occurring particles. However, for minimized particle-related failure, the barrier layers need to be thicker than 300 nm. This requirement imposes substantial mechanical challenges, as we found an intrinsic tensile stress of σ = 400−500 MPa in a wide range of ALD grown layers. As a result of this intrinsic stress, we evidenced a critical membrane force (σ · h)crit = 1200 GPaÅ, beyond which crack formation and delamination occurred. This finding would render the application of ALD grown GDBs with a thickness ≥300 nm impossible. Therefore, we introduced additional compressively strained layers, e.g. metals or SiNx to reduce the overall membrane force to levels below (σ · h)crit. As a result, highly robust GDBs were realized on top of organic light emitting diodes, which did not crack/delaminate even under damp heat conditions of 85 °C/85% rh. Importantly, the growth of dark spots in OLEDs operated under damp heat conditions could be efficiently suppressed by the use GDBs based on these stress management concepts.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsami.5b11499. Membrane force vs thickness of various materials. Lifetime study of encapsulated OLEDs (PDF).
■
AUTHOR INFORMATION
Corresponding Author
*T. Riedl. E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We gratefully acknowledge funding by the state of North-Rhine Westfalia within the PROTECT project (FKZ: 005-11050025). We thank the German Federal Ministry for Education and Research (Grant No. 03EK3529E) for partial financial support.
■
REFERENCES
(1) Frost, J. M.; Butler, K. T.; Brivio, F.; Hendon, C. H.; van Schilfgaarde, M.; Walsh, A. Atomistic Origins of High-Performance in Hybrid Halide Perovskite Solar Cells. Nano Lett. 2014, 14 (5), 2584− 2590. (2) Burrows, P. E.; Graff, G. L.; Gross, M. E.; Martin, P. M.; Hall, M.; Mast, E.; Bonham, C. C.; Bennett, W. D.; Michalski, L. A.; Weaver, M. S.; Brown, J. J.; Fogarty, D.; Sapochak, L. S. Gas Permeation and Lifetime Tests on Polymer-Based Barrier Coatings. Proc. SPIE 2000, 4105, 75−83. (3) Carcia, P. F.; McLean, R. S.; Reilly, M. H.; Groner, M. D.; George, S. M. Ca Test of Al2O3 Gas Diffusion Barriers Grown by Atomic Layer Deposition on Polymers. Appl. Phys. Lett. 2006, 89 (3), 031915. (4) Meyer, J.; Schmidt, H.; Kowalsky, W.; Riedl, T.; Kahn, A. The Origin of Low Water Vapor Transmission Rates Through Al2O3/ ZrO2 Nanolaminate Gas-Diffusion Barriers Grown by Atomic Layer Deposition. Appl. Phys. Lett. 2010, 96 (24), 243308. (5) Abdulagatov, A. I.; Yan, Y.; Cooper, J. R.; Zhang, Y.; Gibbs, Z. M.; Cavanagh, A. S.; Yang, R. G.; Lee, Y. C.; George, S. M. Al2O3 and
4060
DOI: 10.1021/acsami.5b11499 ACS Appl. Mater. Interfaces 2016, 8, 4056−4061
Research Article
ACS Applied Materials & Interfaces TiO2 Atomic Layer Deposition on Copper for Water Corrosion Resistance. ACS Appl. Mater. Interfaces 2011, 3 (12), 4593−4601. (6) Meyer, J.; Schneidenbach, D.; Winkler, T.; Hamwi, S.; Weimann, T.; Hinze, P.; Ammermann, S.; Johannes, H.-H.; Riedl, T.; Kowalsky, W. Reliable Thin Film Encapsulation for Organic Light Emitting Diodes Grown by Low-Temperature Atomic Layer Deposition. Appl. Phys. Lett. 2009, 94 (23), 233305. (7) Meyer, J.; Görrn, P.; Bertram, F.; Hamwi, S.; Winkler, T.; Johannes, H. H.; Weimann, T.; Hinze, P.; Riedl, T.; Kowalsky, W. Al2O3/ZrO2 Nanolaminates as Ultrahigh Gas-Diffusion Barriers-A Strategy for Reliable Encapsulation of Organic Electronics. Adv. Mater. (Weinheim, Ger.) 2009, 21 (18), 1845−1849. (8) Bulusu, A.; Singh, A.; Wang, C. Y.; Dindar, A.; FuentesHernandez, C.; Kim, H.; Cullen, D.; Kippelen, B.; Graham, S. Engineering the Mechanical Properties of Ultrabarrier Films Grown by Atomic Layer Deposition for the Encapsulation of Printed Electronics. J. Appl. Phys. 2015, 118 (8), 085501. (9) Knez, M.; Nielsch, K.; Niinisto, L. Synthesis and Surface Engineering of Complex Nanostructures by Atomic Layer Deposition. Adv. Mater. (Weinheim, Ger.) 2007, 19 (21), 3425−3438. (10) Munoz-Rojas, D.; MacManus-Driscoll, J. Spatial Atmospheric Atomic Layer Deposition: A New Laboratory and Industrial Tool for Low-Cost Photovoltaics. Mater. Horiz. 2014, 1 (3), 314−320. (11) Poodt, P.; Cameron, D. C.; Dickey, E.; George, S. M.; Kuznetsov, V.; Parsons, G. N.; Roozeboom, F.; Sundaram, G.; Vermeer, A. Spatial Atomic Layer Deposition: A Route Towards Further Industrialization of Atomic Layer Deposition. J. Vac. Sci. Technol., A 2012, 30 (1), 010802. (12) Levy, D. H.; Freeman, D.; Nelson, S. F.; Cowdery-Corvan, P. J.; Irving, L. M. Stable ZnO Thin Film Transistors by Fast Open Air Atomic Layer Deposition. Appl. Phys. Lett. 2008, 92 (19), 192101. (13) Tripp, M. K.; Stampfer, C.; Miller, D. C.; Helbling, T.; Herrmann, C. F.; Hierold, C.; Gall, K.; George, S. M.; Bright, V. M. The Mechanical Properties of Atomic Layer Deposited Alumina for Use in Micro- and Nano-Electromechanical. Sens. Actuators, A 2006, 130−131 (0), 419−429. (14) Paetzold, R.; Winnacker, A.; Henseler, D.; Cesari, V.; Heuser, K. Permeation Rate Measurements by Electrical Analysis of Calcium Corrosion. Rev. Sci. Instrum. 2003, 74 (12), 5147−5150. (15) Dameron, A. A.; Davidson, S. D.; Burton, B. B.; Carcia, P. F.; McLean, R. S.; George, S. M. Gas Diffusion Barriers on Polymers Using Multilayers Fabricated by Al2O3 and Rapid SiO2 Atomic Layer Deposition. J. Phys. Chem. C 2008, 112 (12), 4573−4580. (16) Freund, L. B.; Suresh, S. Thin Film Materials: Stress, Defect Formation and Surface Evolution; Cambridge University Press: New York, 2004. (17) Miller, D. C.; Foster, R. R.; Jen, S.-H.; Bertrand, J. A.; Cunningham, S. J.; Morris, A. S.; Lee, Y.-C.; George, S. M.; Dunn, M. L. Thermo-Mechanical Properties of Alumina Films Created Using the Atomic Layer Deposition Technique. Sens. Actuators, A 2010, 164 (1− 2), 58−67. (18) Jen, S.-H.; George, S. M.; McLean, R. S.; Carcia, P. F. Alucone Interlayers to Minimize Stress Caused by Thermal Expansion Mismatch between Al2O3 Films and Teflon Substrates. ACS Appl. Mater. Interfaces 2013, 5 (3), 1165−1173. (19) Jen, S.-H.; Bertrand, J. A.; George, S. M. Critical Tensile and Compressive Strains for Cracking of Al2O3 Films Grown by Atomic Layer Deposition. J. Appl. Phys. 2011, 109 (8), 084305. (20) Mei, Y.; Huang, G.; Solovev, A. A.; Ureña, E. B.; Mönch, I.; Ding, F.; Reindl, T.; Fu, R. K. Y.; Chu, P. K.; Schmidt, O. G. Versatile Approach for Integrative and Functionalized Tubes by Strain Engineering of Nanomembranes on Polymers. Adv. Mater. (Weinheim, Ger.) 2008, 20 (21), 4085−4090. (21) Scholz, S.; Kondakov, D.; Lüssem, B.; Leo, K. Degradation Mechanisms and Reactions in Organic Light-Emitting Devices. Chem. Rev. 2015, 115 (16), 8449−8503.
4061
DOI: 10.1021/acsami.5b11499 ACS Appl. Mater. Interfaces 2016, 8, 4056−4061
