- Email: [email protected]
All hot wire chemical vapor deposition low substrate temperature transparent thin film moisture barrier
Thin Solid Films 532 (2013) 84–88
Contents lists available at SciVerse ScienceDirect
Thin Solid Films journal homepage: www.elsevier.com/locate/tsf
All hot wire chemical vapor deposition low substrate temperature transparent thin film moisture barrier D.A. Spee ⁎, M.R. Schipper, C.H.M. van der Werf 1, J.K. Rath, R.E.I. Schropp Nanophotonics — Physics of Devices, Debye Institute for Nanomaterials Science, Utrecht University, High Tech Campus, Building 5, 5656 AE Einhoven, The Netherlands
a r t i c l e
i n f o
Available online 22 December 2012 Keywords: Hot wire chemical vapor deposition Thin film encapsulation Flexible electronics Moisture barrier
a b s t r a c t We deposited a silicon nitride/polymer hybrid multilayer moisture barrier for flexible electronics in a hot wire chemical vapor deposition process, entirely below 100 °C. We were able to reach a water vapor transmission rate (WVTR) as low as 5×10−6 g/m2/day at a temperature of 60 °C and a relative humidity of 90% for a simple three-layer structure consisting of two low-temperature silicon nitride (SiNx) layers and a polymer layer in between. This WVTR is low enough for organic and polymer devices. In a second experiment it is investigated how the yield of our samples increases with the number of SiNx layers, while keeping the total SiNx thickness constant. Cross sectional scanning electron microscopy images of degraded samples reveal a high structural robustness of our multilayers. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Sensitive electronic devices can easily be damaged by oxygen and water vapor permeating into their active layers. This is an important issue for devices made on flexible plastic substrates, which, contrary to glass, have a high permeability to oxygen and water. This means that permeation barrier layers on flexible substrates are required for many devices, including organic light emitting diodes, flexible solar cells, and rollable displays [1–3]. One way to create such a barrier, is to deposit an inorganic/organic multilayer in which the organic layers decouple defects in consecutive inorganic layers. In this way chances of permeation paths like pinholes or cracks propagating through the entire multilayer are decreased. A combination of SiNx and polymer is very suitable to create such a multilayer [4]. We are using poly(glycidyl methacrylate) (PGMA) as the polymer. This is a widely available and inexpensive material. SiNx is deposited by hot wire chemical vapor deposition (HWCVD) and PGMA by initiated chemical vapor deposition (iCVD), a variant of HWCVD where an initiator is dissociated into two radicals at a hot filament and starts the polymerization process [5]. HWCVD and iCVD use radical formation at heated wires, which provide a linear source of radicals. When the substrate is moved perpendicular to the wires the deposition will be homogeneous in both dimensions [6]. This makes it easy to
⁎ Corresponding author. E-mail address: [email protected] (D.A. Spee). 1 Presently at: Energy Research Centre of the Netherlands-Solliance, High Tech Campus 5, 5656 AE Eindhoven, The Netherlands. 0040-6090/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.tsf.2012.11.146
implement in a roll-to-roll or inline process. Since in the HWCVD process no high energy ions are present, layers can be deposited on delicate substrates. To be able to deposit SiNx on PGMA layers, two main requirements need to be fulfilled: (i) the PGMA should be very stable against the flux of chemical species during SiNx deposition [7] and (ii) the SiNx must be deposited at low enough substrate temperatures to be compatible with common plastic substrates and also PGMA. Here, besides discussing a low temperature hybrid moisture barrier, we will investigate how the yield of our samples increases by adding extra layers and address the robustness of the multilayer.
2. Experimental details 2.1. Layer deposition The SiNx layers were deposited in a multichamber deposition system using pure silane (SiH4) and ammonia (NH3) as source gasses. A dedicated low temperature hot wire chamber was used in which the wire-substrate distance could be as large as 20 cm. Since radiative heating is reduced drastically in this manner, process temperatures never exceeded 100 °C [8]. Thermindex TC8020 temperature indication stickers were attached on the front side of the substrate to be able to accurately determine the maximum temperature reached during deposition. The PGMA layers were deposited in a homemade iCVD reactor [9] designed after a reactor at Massachusetts Institute of Technology labs [10]. The layers were deposited at a deposition rate of 0.8 nm/s. As a monomer GMA (97%, Aldrich) was used and as an initiator tert butyl peroxide (98%, Aldrich). All samples were deposited on glass.
D.A. Spee et al. / Thin Solid Films 532 (2013) 84–88
85
Fig. 1. The development in time of 5×5 mm2 Ca squares covered with: 100 nm low temperature SiNx (top), 100 nm state-of-the-art high-temperature SiNx (middle), and a low temperature multilayer (bottom).
2.2. Characterization The permeation measurements were performed using the Ca method [11]. In this method squares of metallic calcium on glass are covered by the barrier layer. Upon oxidation the calcium becomes transparent. In this way the permeated water can accurately be determined and pinholes can be made visible. By keeping the samples in a climate chamber at a temperature of 60 °C at a relative humidity (RH) of 90%, the measurement was accelerated [2]. The Scanning Electron Microscopy
Fig. 2. Graph of the permeation through the multilayer. Fitting a linear function, a WVTR of 5×10−6 g/m2/day was found.
(SEM) images were made using a FEI (Nova Nanolab 600 Dualbeam) combined Focussed Ion Beam (FIB) and SEM system [12]. Images were made in back scattered electron mode using an operating voltage of 2.0 kV. The cross sectional SEM images were made under an angle of 52°, which means the heights in the images should be multiplied by 1/Sin(52°)= 1.3.
Fig. 3. SEM image of a multilayer, consisting of four SiNx layers and three PGMA layers, on glass. The cross section was made by the FIB method. The sample was covered by platinum to avoid electrical charging.
86
D.A. Spee et al. / Thin Solid Films 532 (2013) 84–88
Fig. 4. The development in time of arrays of 5×5 mm2 Ca squares covered with: 4 layers of 50 nm SiNx (top), 3 layers of 67 nm SiNx (middle), and two 100 nm thick SiNx (bottom), with 200 nm PGMA interlayers.
3. Results 3.1. Low temperature double layer First we developed highly transparent, compact HWCVD SiNx at a substrate temperature below 100 °C, compatible with our PGMA layers deposited by iCVD. This facilitated the deposition of transparent SiNx/ PGMA multilayers [13]. Second we tested the permeability of three different samples: a 100 nm thick low temperature SiNx layer, a 100 nm thick state-of-the-art SiNx layer [14] and a multilayer consisting of two 100 nm low-temperature SiNx layers sandwiching a PGMA layer. For the multilayer we chose a very thin interlayer of only 200 nm PGMA. On the one hand, the decoupling function of the layer will probably decrease with decreasing thickness. On the other hand, for the barrier function of a multilayer to be as high as possible, thinner interlayers are beneficial, since they create a more tortuous permeation path [1]. We chose a thin polymer layer, since Bakker et al. [10] demonstrated that PGMA layers made by iCVD are excellent planarizing layers even
Fig. 5. The yield of our samples, as a function of number of SiNx layers within the moisture barrier. For all types 36 samples were studied.
at thicknesses around 200 nm. The optical transmission of Ca squares covered by the three samples after various incremental exposure durations is shown in Fig. 1. Comparing the Ca degradation of the two single layer samples, the behavior is as expected. Higher porosity and a higher density of pinholes are expected in low temperature samples compared to the high temperature ones, since the surface mobility of precursors during deposition goes down with decreasing substrate temperature. However, a configuration of two of these layers with a PGMA interlayer shows an entirely different behavior. No significant degradation of the Ca layer is visible, even after 190 days in the accelerated measurement. In Fig. 2, the permeation graph of this sample is shown. The initial jump in the permeated water graph was attributed to residual moisture diffusing out of the as-deposited polymer layer. Therefore the data of the first day was not included in the data used for fitting. When a linear fit is applied, a water vapor transmission rate (WVTR) of 5 × 10 −6 g/ m 2/day is found, compared to 3 × 10−5 g/m2/day for a single layer of state-of-the-art SiNx. For the individual low temperature layer we were not able to determine the WVTR because it deteriorates within two days by permeation through the pinholes. The excellent barrier performance of our multilayer can be explained by the fact that our organic layer is kept relatively thin: 200 nm is probably significantly smaller than the average distance between defects in the inorganic layers, even if the number of pinholes in the low temperature SiNx is high. In this way an extremely tortuous path is created for the permeating moisture [1]. Also the good barrier performance of the double layer being partly caused by the filling up of defects in the SiNx layer by PGMA [15] cannot be ruled out. Since the WVTR of 5 × 10 −6 g/m 2/day was found at 60 °C and 90% RH, an excellent performance at room temperature can be expected [2]. The graph shown in Fig. 2 flattens over time, indicating that the extremely high barrier function is most probably a consequence of a low equilibrium permeation rather than a high lag time of the water vapor [16]. Our experiment shows the multilayers' robustness with respect to defects originating from the deposition and the substrate. It was however difficult to avoid contamination with defect-creating dust particles, since the present facilities in our laboratory are such that the samples had to be moved outside vacuum between different reactors to deposit the multilayer. This affected the
D.A. Spee et al. / Thin Solid Films 532 (2013) 84–88
yield of our samples. In practical production it is straightforward to avoid such issues by transporting the samples between subsequent reactors in vacuum.
87
increases, since the thickness of the total multilayer increases due to the increased accumulated PGMA thickness. 3.3. Robustness of the multilayer
3.2. Increasing yield by increasing the number of layers To investigate the effect of an increasing number of SiNx layers on the yield of our samples, i.e. their robustness with respect to environmental dust, a second experiment was done, in which three different configurations were tested. Here yield is defined as the fraction of the samples which does not show any form of deterioration, measured over 36 samples. We deposited multilayers consisting of two, three and four SiNx layers, keeping the total thickness of all SiNx layers constant at 200 nm: two layers of 100 nm, three layers of 67 nm and four layers of 50 nm. In between these layers were always 200 nm thick PGMA interlayers. A SEM image of a cross section of the multilayer containing four SiNx layers is shown in Fig. 3. The cross section was made by FIB. The image shows that the SiNx layers are around 50 nm, as expected. The PGMA layers however, are around 150 nm, while 200 nm was intended. If more layers are deposited on top, the surface becomes less planar. Knowing that as deposited PGMA layers exhibit a rms roughness lower than 5 nm (measured over an area of 10 ×10 μm) [7], this indicates that there is still some slight deformation of the PGMA layers during deposition of the SiNx. It is difficult to see the abruptness of the interfaces in the SEM image, since the electrons have some penetration depth and the cross section is seen under an angle. In Fig. 4 the typical development of Ca squares covered by our different barriers are shown after incremental exposures. Considering the squares that were not affected by pinholes after 30 days, we see that the WVTR goes down slightly with an increasing number of SiNx layers: 8.3 × 10−5 g/m2/day for two layers, 3.6 × 10 −5 g/m 2/day for three layers and 2.8 × 10−5 g/m 2/day for four layers, even though the total amount of SiNx is constant. The most distinct effect in the second experiment however, is an increase in the yield with an increasing number of layers. In Fig. 5 it is shown that the yield increases from 0.06, using two layers, to 0.37 using four layers, as measured over 36 samples. Probably there are two factors that contribute to this increase: the probability that pinholes in all consecutive layers are laterally in the same spot or less than roughly the interlayer thickness apart is decreasing rapidly with the number of layers. Secondly, the chance of dust particles being entirely covered
In the SEM images of Fig. 6, a dust particle causing a pinhole in a barrier coating consisting of four SiNx and three PGMA layers is shown. The particle was in the center of a typical pinhole, i.e. a pinhole through which a constant amount of moisture is permeating, causing a white spot growing linearly in time [10]. In Fig. 6 (a) it can be seen that the pinhole is caused by a dust particle, which is around 10 μm in diameter, cutting through the entire multilayer. Directly around the pinhole, large bubble-like structures have formed, indicating that the multilayer is detached from the substrate. To create a cross section of the white spot area, a rectangular piece of material was removed by FIB to a depth of 3 μm, as is shown in Fig. 6 (b). The cross section and an enlarged detail of it are shown in Fig. 6 (c) and (d). The delamination of the multilayer from the substrate is clearly a result of the deterioration of the covered calcium layer by moisture. In this way it is revealed that the adhesion of the different organic and inorganic layers and the robustness of the multilayer are high: its structure is retained even when the multilayer stands alone. 4. Conclusions In a process entirely below 100 °C we deposited a multilayer by HWCVD, consisting of only two SiNx layers with a PGMA layer in between [17]. Using iCVD to deposit PGMA we were able to create an interlayer with a thickness of only 200 nm, which is very effectively decoupling defects in consecutive SiNx layers. Although the lowtemperature SiNx itself exhibited a high porosity, this three-layer structure is pinhole free and shows a water vapor transmission rate (WVTR) as low as 5 × 10 −6 g/m 2/day at 60 °C and a relative humidity of 90%. This value is low enough for any possible electronic device. Due to the low radiation and consequently low deposition temperature and the absence of high energy particles (ions) in the process, these layers can be deposited directly onto organic of polymeric active devices. In a second experiment it was shown that the robustness of our multilayers with respect to defect-creating dust particles increases when increasing the number of SiNx interlayers while keeping the total thickness of SiNx the same. In practical production however, the issue of environmental dust can be avoided by transporting samples in
Fig. 6. SEM images of: (a) a dust particle causing a pinhole in the multilayer, (b) the rectangular area removed by FIB to create a cross section, (c) the cross section of the pinhole area and (d) an enlarged view of the multilayer being detached from the substrate due to the calcium that has reacted with water.
88
D.A. Spee et al. / Thin Solid Films 532 (2013) 84–88
vacuum. Cross sectional SEM images of a degraded sample reveal that the structural robustness of our multilayer is high, as is the adhesion between the organic and inorganic layers. Acknowledgments We thank Dr. P. van de Weijer of Philips Research Laboratories for the Ca corrosion tests and D.A.M. de Winter for the FIB-SEM imaging and we acknowledge the financial support for this research from the “Stichting Technische Wetenschappen” (project Thin Film Nanomanufacturing 10019). References [1] M. Schaepkens, T.W. Kim, A.G. Erlat, M. Yan, K.W. Flanagan, C.M. Heller, P.A. McConnelee, J. Vac. Sci. Technol. A 22 (4) (2004) 1716. [2] J. Lewis, Mater. Today 9 (4) (2006) 38. [3] C. Charton, N. Schiller, M. Fahland, A. Hollander, A. Wedel, K. Noller, Thin Solid Films 502 (2006) 99. [4] V. Verlaan, R. Bakker, C.H.M. van der Werf, Z.S. Houweling, Y. Mai, J.K. Rath, R.E.I. Schropp, Surf. Coat. Technol. 201 (2007) 9285.
[5] Y. Mao, K.K. Gleason, Langmuir 20 (2004) 2484. [6] R.E.I. Schropp, C.O. van Bommel, C.H.M. van der Werf, M. Brinza, G.A. van Swaaij, J.K. Rath, H.B.T. Li, J.W.A. Schüttauf, in: Proc. of 24th European Photovoltaic Solar Energy Conference, 2009, p. 2328. [7] D.A. Spee, R. Bakker, C.H.M. van der Werf, M.J. van Steenbergen, J.K. Rath, R.E.I. Schropp, Thin Solid Films 519 (2011) 4479. [8] D.A. Spee, C.H.M. van der Werf, J.K. Rath, R.E.I. Schropp, J. Nanosci. Nanotechnol. 11 (9) (2011) 8202. [9] R. Bakker, V. Verlaan, C.H.M. van der Werf, J.K. Rath, K.K. Gleason, R.E.I. Schropp, Surf. Coat. Technol. 201 (2007) 9422. [10] K.K.S. Lau, K.K. Gleason, Macromolecules 39 (2006) 3688. [11] G. Nisato, P.C.P. Bouten, P.J. Slikkerveer, W.D. Bennett, G.L. Graff, N. Rutherford, L. Wiese, Proc. Asia Disp. (2001) 1435. [12] M.D. Uchic, L. Holzer, B.J. Inkson, E.L. Principe, P. Munroe, MRS Bull. 32 (2007) 408. [13] D.A. Spee, C.H.M. van der Werf, J.K. Rath, R.E.I. Schropp, Phys. Status Solidi (RRL) 6 (2012) 151. [14] V. Verlaan, Z.S. Houweling, C.H.M. van der Werf, I.G. Romijn, A.W. Weeber, H.D. Goldbach, R.E.I. Schropp, Thin Solid Films 516 (2008) 533. [15] J. Affinito, D. Hilliard, in: 47th Annual Technical Conference Proceedings, Society of Vacuum Coaters, 2004, p. 563. [16] G.L. Graff, R.E. Williford, P.E. Burrows, J. Appl. Phys. 96 (2004) 1840. [17] PCT/NL2011/050601 (US provisional patent application 61/490,604).
Contents lists available at SciVerse ScienceDirect
Thin Solid Films journal homepage: www.elsevier.com/locate/tsf
All hot wire chemical vapor deposition low substrate temperature transparent thin film moisture barrier D.A. Spee ⁎, M.R. Schipper, C.H.M. van der Werf 1, J.K. Rath, R.E.I. Schropp Nanophotonics — Physics of Devices, Debye Institute for Nanomaterials Science, Utrecht University, High Tech Campus, Building 5, 5656 AE Einhoven, The Netherlands
a r t i c l e
i n f o
Available online 22 December 2012 Keywords: Hot wire chemical vapor deposition Thin film encapsulation Flexible electronics Moisture barrier
a b s t r a c t We deposited a silicon nitride/polymer hybrid multilayer moisture barrier for flexible electronics in a hot wire chemical vapor deposition process, entirely below 100 °C. We were able to reach a water vapor transmission rate (WVTR) as low as 5×10−6 g/m2/day at a temperature of 60 °C and a relative humidity of 90% for a simple three-layer structure consisting of two low-temperature silicon nitride (SiNx) layers and a polymer layer in between. This WVTR is low enough for organic and polymer devices. In a second experiment it is investigated how the yield of our samples increases with the number of SiNx layers, while keeping the total SiNx thickness constant. Cross sectional scanning electron microscopy images of degraded samples reveal a high structural robustness of our multilayers. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Sensitive electronic devices can easily be damaged by oxygen and water vapor permeating into their active layers. This is an important issue for devices made on flexible plastic substrates, which, contrary to glass, have a high permeability to oxygen and water. This means that permeation barrier layers on flexible substrates are required for many devices, including organic light emitting diodes, flexible solar cells, and rollable displays [1–3]. One way to create such a barrier, is to deposit an inorganic/organic multilayer in which the organic layers decouple defects in consecutive inorganic layers. In this way chances of permeation paths like pinholes or cracks propagating through the entire multilayer are decreased. A combination of SiNx and polymer is very suitable to create such a multilayer [4]. We are using poly(glycidyl methacrylate) (PGMA) as the polymer. This is a widely available and inexpensive material. SiNx is deposited by hot wire chemical vapor deposition (HWCVD) and PGMA by initiated chemical vapor deposition (iCVD), a variant of HWCVD where an initiator is dissociated into two radicals at a hot filament and starts the polymerization process [5]. HWCVD and iCVD use radical formation at heated wires, which provide a linear source of radicals. When the substrate is moved perpendicular to the wires the deposition will be homogeneous in both dimensions [6]. This makes it easy to
⁎ Corresponding author. E-mail address: [email protected] (D.A. Spee). 1 Presently at: Energy Research Centre of the Netherlands-Solliance, High Tech Campus 5, 5656 AE Eindhoven, The Netherlands. 0040-6090/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.tsf.2012.11.146
implement in a roll-to-roll or inline process. Since in the HWCVD process no high energy ions are present, layers can be deposited on delicate substrates. To be able to deposit SiNx on PGMA layers, two main requirements need to be fulfilled: (i) the PGMA should be very stable against the flux of chemical species during SiNx deposition [7] and (ii) the SiNx must be deposited at low enough substrate temperatures to be compatible with common plastic substrates and also PGMA. Here, besides discussing a low temperature hybrid moisture barrier, we will investigate how the yield of our samples increases by adding extra layers and address the robustness of the multilayer.
2. Experimental details 2.1. Layer deposition The SiNx layers were deposited in a multichamber deposition system using pure silane (SiH4) and ammonia (NH3) as source gasses. A dedicated low temperature hot wire chamber was used in which the wire-substrate distance could be as large as 20 cm. Since radiative heating is reduced drastically in this manner, process temperatures never exceeded 100 °C [8]. Thermindex TC8020 temperature indication stickers were attached on the front side of the substrate to be able to accurately determine the maximum temperature reached during deposition. The PGMA layers were deposited in a homemade iCVD reactor [9] designed after a reactor at Massachusetts Institute of Technology labs [10]. The layers were deposited at a deposition rate of 0.8 nm/s. As a monomer GMA (97%, Aldrich) was used and as an initiator tert butyl peroxide (98%, Aldrich). All samples were deposited on glass.
D.A. Spee et al. / Thin Solid Films 532 (2013) 84–88
85
Fig. 1. The development in time of 5×5 mm2 Ca squares covered with: 100 nm low temperature SiNx (top), 100 nm state-of-the-art high-temperature SiNx (middle), and a low temperature multilayer (bottom).
2.2. Characterization The permeation measurements were performed using the Ca method [11]. In this method squares of metallic calcium on glass are covered by the barrier layer. Upon oxidation the calcium becomes transparent. In this way the permeated water can accurately be determined and pinholes can be made visible. By keeping the samples in a climate chamber at a temperature of 60 °C at a relative humidity (RH) of 90%, the measurement was accelerated [2]. The Scanning Electron Microscopy
Fig. 2. Graph of the permeation through the multilayer. Fitting a linear function, a WVTR of 5×10−6 g/m2/day was found.
(SEM) images were made using a FEI (Nova Nanolab 600 Dualbeam) combined Focussed Ion Beam (FIB) and SEM system [12]. Images were made in back scattered electron mode using an operating voltage of 2.0 kV. The cross sectional SEM images were made under an angle of 52°, which means the heights in the images should be multiplied by 1/Sin(52°)= 1.3.
Fig. 3. SEM image of a multilayer, consisting of four SiNx layers and three PGMA layers, on glass. The cross section was made by the FIB method. The sample was covered by platinum to avoid electrical charging.
86
D.A. Spee et al. / Thin Solid Films 532 (2013) 84–88
Fig. 4. The development in time of arrays of 5×5 mm2 Ca squares covered with: 4 layers of 50 nm SiNx (top), 3 layers of 67 nm SiNx (middle), and two 100 nm thick SiNx (bottom), with 200 nm PGMA interlayers.
3. Results 3.1. Low temperature double layer First we developed highly transparent, compact HWCVD SiNx at a substrate temperature below 100 °C, compatible with our PGMA layers deposited by iCVD. This facilitated the deposition of transparent SiNx/ PGMA multilayers [13]. Second we tested the permeability of three different samples: a 100 nm thick low temperature SiNx layer, a 100 nm thick state-of-the-art SiNx layer [14] and a multilayer consisting of two 100 nm low-temperature SiNx layers sandwiching a PGMA layer. For the multilayer we chose a very thin interlayer of only 200 nm PGMA. On the one hand, the decoupling function of the layer will probably decrease with decreasing thickness. On the other hand, for the barrier function of a multilayer to be as high as possible, thinner interlayers are beneficial, since they create a more tortuous permeation path [1]. We chose a thin polymer layer, since Bakker et al. [10] demonstrated that PGMA layers made by iCVD are excellent planarizing layers even
Fig. 5. The yield of our samples, as a function of number of SiNx layers within the moisture barrier. For all types 36 samples were studied.
at thicknesses around 200 nm. The optical transmission of Ca squares covered by the three samples after various incremental exposure durations is shown in Fig. 1. Comparing the Ca degradation of the two single layer samples, the behavior is as expected. Higher porosity and a higher density of pinholes are expected in low temperature samples compared to the high temperature ones, since the surface mobility of precursors during deposition goes down with decreasing substrate temperature. However, a configuration of two of these layers with a PGMA interlayer shows an entirely different behavior. No significant degradation of the Ca layer is visible, even after 190 days in the accelerated measurement. In Fig. 2, the permeation graph of this sample is shown. The initial jump in the permeated water graph was attributed to residual moisture diffusing out of the as-deposited polymer layer. Therefore the data of the first day was not included in the data used for fitting. When a linear fit is applied, a water vapor transmission rate (WVTR) of 5 × 10 −6 g/ m 2/day is found, compared to 3 × 10−5 g/m2/day for a single layer of state-of-the-art SiNx. For the individual low temperature layer we were not able to determine the WVTR because it deteriorates within two days by permeation through the pinholes. The excellent barrier performance of our multilayer can be explained by the fact that our organic layer is kept relatively thin: 200 nm is probably significantly smaller than the average distance between defects in the inorganic layers, even if the number of pinholes in the low temperature SiNx is high. In this way an extremely tortuous path is created for the permeating moisture [1]. Also the good barrier performance of the double layer being partly caused by the filling up of defects in the SiNx layer by PGMA [15] cannot be ruled out. Since the WVTR of 5 × 10 −6 g/m 2/day was found at 60 °C and 90% RH, an excellent performance at room temperature can be expected [2]. The graph shown in Fig. 2 flattens over time, indicating that the extremely high barrier function is most probably a consequence of a low equilibrium permeation rather than a high lag time of the water vapor [16]. Our experiment shows the multilayers' robustness with respect to defects originating from the deposition and the substrate. It was however difficult to avoid contamination with defect-creating dust particles, since the present facilities in our laboratory are such that the samples had to be moved outside vacuum between different reactors to deposit the multilayer. This affected the
D.A. Spee et al. / Thin Solid Films 532 (2013) 84–88
yield of our samples. In practical production it is straightforward to avoid such issues by transporting the samples between subsequent reactors in vacuum.
87
increases, since the thickness of the total multilayer increases due to the increased accumulated PGMA thickness. 3.3. Robustness of the multilayer
3.2. Increasing yield by increasing the number of layers To investigate the effect of an increasing number of SiNx layers on the yield of our samples, i.e. their robustness with respect to environmental dust, a second experiment was done, in which three different configurations were tested. Here yield is defined as the fraction of the samples which does not show any form of deterioration, measured over 36 samples. We deposited multilayers consisting of two, three and four SiNx layers, keeping the total thickness of all SiNx layers constant at 200 nm: two layers of 100 nm, three layers of 67 nm and four layers of 50 nm. In between these layers were always 200 nm thick PGMA interlayers. A SEM image of a cross section of the multilayer containing four SiNx layers is shown in Fig. 3. The cross section was made by FIB. The image shows that the SiNx layers are around 50 nm, as expected. The PGMA layers however, are around 150 nm, while 200 nm was intended. If more layers are deposited on top, the surface becomes less planar. Knowing that as deposited PGMA layers exhibit a rms roughness lower than 5 nm (measured over an area of 10 ×10 μm) [7], this indicates that there is still some slight deformation of the PGMA layers during deposition of the SiNx. It is difficult to see the abruptness of the interfaces in the SEM image, since the electrons have some penetration depth and the cross section is seen under an angle. In Fig. 4 the typical development of Ca squares covered by our different barriers are shown after incremental exposures. Considering the squares that were not affected by pinholes after 30 days, we see that the WVTR goes down slightly with an increasing number of SiNx layers: 8.3 × 10−5 g/m2/day for two layers, 3.6 × 10 −5 g/m 2/day for three layers and 2.8 × 10−5 g/m 2/day for four layers, even though the total amount of SiNx is constant. The most distinct effect in the second experiment however, is an increase in the yield with an increasing number of layers. In Fig. 5 it is shown that the yield increases from 0.06, using two layers, to 0.37 using four layers, as measured over 36 samples. Probably there are two factors that contribute to this increase: the probability that pinholes in all consecutive layers are laterally in the same spot or less than roughly the interlayer thickness apart is decreasing rapidly with the number of layers. Secondly, the chance of dust particles being entirely covered
In the SEM images of Fig. 6, a dust particle causing a pinhole in a barrier coating consisting of four SiNx and three PGMA layers is shown. The particle was in the center of a typical pinhole, i.e. a pinhole through which a constant amount of moisture is permeating, causing a white spot growing linearly in time [10]. In Fig. 6 (a) it can be seen that the pinhole is caused by a dust particle, which is around 10 μm in diameter, cutting through the entire multilayer. Directly around the pinhole, large bubble-like structures have formed, indicating that the multilayer is detached from the substrate. To create a cross section of the white spot area, a rectangular piece of material was removed by FIB to a depth of 3 μm, as is shown in Fig. 6 (b). The cross section and an enlarged detail of it are shown in Fig. 6 (c) and (d). The delamination of the multilayer from the substrate is clearly a result of the deterioration of the covered calcium layer by moisture. In this way it is revealed that the adhesion of the different organic and inorganic layers and the robustness of the multilayer are high: its structure is retained even when the multilayer stands alone. 4. Conclusions In a process entirely below 100 °C we deposited a multilayer by HWCVD, consisting of only two SiNx layers with a PGMA layer in between [17]. Using iCVD to deposit PGMA we were able to create an interlayer with a thickness of only 200 nm, which is very effectively decoupling defects in consecutive SiNx layers. Although the lowtemperature SiNx itself exhibited a high porosity, this three-layer structure is pinhole free and shows a water vapor transmission rate (WVTR) as low as 5 × 10 −6 g/m 2/day at 60 °C and a relative humidity of 90%. This value is low enough for any possible electronic device. Due to the low radiation and consequently low deposition temperature and the absence of high energy particles (ions) in the process, these layers can be deposited directly onto organic of polymeric active devices. In a second experiment it was shown that the robustness of our multilayers with respect to defect-creating dust particles increases when increasing the number of SiNx interlayers while keeping the total thickness of SiNx the same. In practical production however, the issue of environmental dust can be avoided by transporting samples in
Fig. 6. SEM images of: (a) a dust particle causing a pinhole in the multilayer, (b) the rectangular area removed by FIB to create a cross section, (c) the cross section of the pinhole area and (d) an enlarged view of the multilayer being detached from the substrate due to the calcium that has reacted with water.
88
D.A. Spee et al. / Thin Solid Films 532 (2013) 84–88
vacuum. Cross sectional SEM images of a degraded sample reveal that the structural robustness of our multilayer is high, as is the adhesion between the organic and inorganic layers. Acknowledgments We thank Dr. P. van de Weijer of Philips Research Laboratories for the Ca corrosion tests and D.A.M. de Winter for the FIB-SEM imaging and we acknowledge the financial support for this research from the “Stichting Technische Wetenschappen” (project Thin Film Nanomanufacturing 10019). References [1] M. Schaepkens, T.W. Kim, A.G. Erlat, M. Yan, K.W. Flanagan, C.M. Heller, P.A. McConnelee, J. Vac. Sci. Technol. A 22 (4) (2004) 1716. [2] J. Lewis, Mater. Today 9 (4) (2006) 38. [3] C. Charton, N. Schiller, M. Fahland, A. Hollander, A. Wedel, K. Noller, Thin Solid Films 502 (2006) 99. [4] V. Verlaan, R. Bakker, C.H.M. van der Werf, Z.S. Houweling, Y. Mai, J.K. Rath, R.E.I. Schropp, Surf. Coat. Technol. 201 (2007) 9285.
[5] Y. Mao, K.K. Gleason, Langmuir 20 (2004) 2484. [6] R.E.I. Schropp, C.O. van Bommel, C.H.M. van der Werf, M. Brinza, G.A. van Swaaij, J.K. Rath, H.B.T. Li, J.W.A. Schüttauf, in: Proc. of 24th European Photovoltaic Solar Energy Conference, 2009, p. 2328. [7] D.A. Spee, R. Bakker, C.H.M. van der Werf, M.J. van Steenbergen, J.K. Rath, R.E.I. Schropp, Thin Solid Films 519 (2011) 4479. [8] D.A. Spee, C.H.M. van der Werf, J.K. Rath, R.E.I. Schropp, J. Nanosci. Nanotechnol. 11 (9) (2011) 8202. [9] R. Bakker, V. Verlaan, C.H.M. van der Werf, J.K. Rath, K.K. Gleason, R.E.I. Schropp, Surf. Coat. Technol. 201 (2007) 9422. [10] K.K.S. Lau, K.K. Gleason, Macromolecules 39 (2006) 3688. [11] G. Nisato, P.C.P. Bouten, P.J. Slikkerveer, W.D. Bennett, G.L. Graff, N. Rutherford, L. Wiese, Proc. Asia Disp. (2001) 1435. [12] M.D. Uchic, L. Holzer, B.J. Inkson, E.L. Principe, P. Munroe, MRS Bull. 32 (2007) 408. [13] D.A. Spee, C.H.M. van der Werf, J.K. Rath, R.E.I. Schropp, Phys. Status Solidi (RRL) 6 (2012) 151. [14] V. Verlaan, Z.S. Houweling, C.H.M. van der Werf, I.G. Romijn, A.W. Weeber, H.D. Goldbach, R.E.I. Schropp, Thin Solid Films 516 (2008) 533. [15] J. Affinito, D. Hilliard, in: 47th Annual Technical Conference Proceedings, Society of Vacuum Coaters, 2004, p. 563. [16] G.L. Graff, R.E. Williford, P.E. Burrows, J. Appl. Phys. 96 (2004) 1840. [17] PCT/NL2011/050601 (US provisional patent application 61/490,604).