Issue
Eur. Phys. J. Appl. Phys.
Volume 89, Number 1, January 2020
Disordered Semiconductors: Physics and Applications
Article Number 10303
Number of page(s) 6
Section Thin Films
DOI https://doi.org/10.1051/epjap/2020190300
Published online 07 April 2020

© EDP Sciences, 2020

1 Introduction

Specializing in amorphous silicon devices in a lab that has been active for over 30 years, our team has focused on electroformed thin film silicon-based p-i-n diodes for the last decade since the consequences of the electroforming process are both scientifically and technologically interesting. Simply electroforming of hydrogenated amorphous Si (a-Si:H) based structures can be described as the rapid crystallization induced by Joule heating under the application of electric field [1]. Electroforming of amorphous silicon based p-i-n diodes was observed by various groups in various structures [24]. Kocka’s group found that their homojunction p-i-n diode was electroformed under high electric field and the diode started to exhibit strong electroluminescence (EL) in the near infrared and visible regions [2,5,6]. They used Raman measurements to show that the electroformed parts of the diode contained Si nanocrystallites. Cabarrocas’s group showed that their polymorphous silicon carbide based p-i-n diodes were electroformed under high electric field accompanied by a strong orange EL [3]. Our group studied the electroforming of both a-Si:H based homojunction [7] and hydrogenated amorhous silicon nitride (a-SiNx:H) based heterojunction [8,9] p-i-n diodes. As we also observed enhanced EL in both type of diodes, it seems that strong light emission is the common consequence of electroforming in p-i-n diodes.

Beside being a light source, we recognized that the electroformed device also exhibits resistive memory switching effect making it a potential candidate for the next generation memory cells, i.e. light emitting memory (LEM) [10]. This LEM device is interesting by means of its scientific aspects as the underlying physics of this phenomenon seems to be quite sophisticated. In order to account for all current-voltage, EL, XRD and capacitance-voltage-frequency characteristics, it was concluded that silicon nanocrystallites were formed within the silicon nitride film during electroforming. Finally, we developed a model based on a temporary sub-band transport level formed during the reverse bias regime.

So far, we studied the structural properties of the electroformed diode using indirect tools. Although those indirect tools were useful to understand the main changes in the diode structure imposed by electroforming, there is still a gap to be filled by direct measurements such as SEM and TEM analyses. In this paper, we focus on the SEM, EDX and real-time microscopy results, which showed that our previous model mostly remains valid while some points of it needs to be modified.

2 Experimental part

The structure of the thin film diode used in this work can be described as p-i-n layers sandwiched between a 170 nm Cr-coated glass substrate and a 230 nm ITO top electrode of 1 mm diameter (Fig. 1). For this purpose, 90 nm p+ nanocrystalline Si (p+ nc-Si:H), 30 nm intrinsic a-SiNx:H and 90 nm n+ nc-Si:H films were grown consecutively by an Oxford Instruments PECVD system using the deposition recipe given before [9]. The surface images of the ITO top electrodes of both as-deposited and electroformed diodes were obtained by Carl Zeiss Ultra Plus Gemini FE-SEM. Chemical element analyses of the diode surface and its cross-section were performed with a Bruker XFlash 6|10 EDX detector connected to SEM device. Then, SEM cross-sectional images of the as-deposited and electroformed diodes were obtained at the closest level allowed by the microscope. Finally, real-time optical microscopy of the ITO surface was performed during electroforming and visible light emission.

thumbnail Fig. 1

(a) Schematical cross section of the a-SiNx:H based p-i-n diode. (b) Tilted cross-sectional SEM image of an electroformed diode. Note that the ITO top electrode seems to almost completely escape from the surface during electroforming.

3 Results and discussions

3.1 SEM top-view and elemental surface mapping of partially electroformed diode

In order to better understand the phenomena that occur during electroforming, the structural differences between the electroformed and the non-electroformed regions should be examined. For this reason, a partially electroformed diode was selected and the SEM image was taken to capture the top-view (Fig. 2). Figure 2 shows that approximately half of the diode surface area became rough during electroforming. Looking at the elemental mapping signals from the smooth surface region (non-electroformed), it is seen that the In and Sn signals are densely received from the top electrode ITO, whereas very weak signalscome from Si which is the main material of the semiconductor layers below ITO. Looking at the rough (electroformed) areas on the diode surface, very weak signals come from In and Sn, whereas Si signals are very strong. In other words, In and Sn in the rough areas, and meanwhile, the ITO electrode is reduced in quantity during electroforming. Considering that the Si signals coming from the circular surface of the diode are weaker than those coming from the region outside the diode, it is noticed that In and Sn may not be completely removed from the structure during electroforming. Also it should be noted that in the case of completely electroformed diodes, the entire surface has a rough character.

Elemental mapping was performed also along a line during the acquisition of SEM images from the top (Fig. 3). Similar tothe qualitative analysis shown in Figure 2, the In and Sn elements were significantly reduced on the roughened diode surface in Figure 3, where the EDX signals were mostly from the Cr bottom electrode. A slight increase in the number of Si elements was observed when the signal was collected from this roughened region. At some points, the In and Sn signals fall completely to zero, where it can be thought that the ITO electrode completely escaped from the structure during electroforming. A small amount of O may have come from both the ITO and the glass substrate under the Cr bottom electrode. During this measurement, the N elements in the intrinsic layer of the diode could not be measured because they were relatively much lower in quantity than the other elements.

thumbnail Fig. 2

EDX elemental surface mapping of a partially electroformed diode showing: (a) original SEM top-view image, (b) In mapping,(c) Sn mapping and (d) Si mapping.

thumbnail Fig. 3

(a) Top-view of a partially electroformed a-SiNx:H p-i-n diode and (b) relative amounts of elements along a line.

3.2 Cross-sectional SEM and EDX study of partially electroformed diode

After examining the top-views of the diodes, cross-sectional images were also analyzed. Cross-sectional images were obtained from several electroformed and non-electroformed regions (not all of the images are shown in this work). In order to get an image through the cross-section, the glasses onto which the diodes were coated were broken along a line drew by a diamond pen. During the breaking process, it was found that some regions were damaged in such a way that they were not suitable for taking SEM cross-sectional images. In order to perform SEM cross-sectional image analysis in a healthy way, we focused on the areaswhere the least damage was done during the breaking process. SEM cross-sectional images showed that the thin film layers were coated with a uniform thickness distribution (Fig. 4). The films coated successively on the glass substrate are shown in Figure 4 by arrows. The film thickness values obtained from the SEM cross-sectional image are consistent with the expected values calculated by the respective deposition rates of each individual layer. Although these cross-sectional images provide information about the macro-structure of the films, TEM measurements are necessary to obtain information about the order and/or irregularity at the atomic level. We will also publish another paper focusing on the TEM analysis of the electroformed diodes.

In one of the cross-sectional images of the electroformed p-i-n diode, the ITO upper electrode almost completely went away from the structure during electroforming (Fig. 1). In contrast, no change in the planar uniformity of the other thin film layers was noted (Fig. 4). Thus, it can be concluded that there was no mechanical damage to Si-based thin films during electroforming.

When the SEM cross-sectional images are examined, it is seen that the ITO electrode at certain points of the diode surface sometimes partially, sometimes completely escaped from the structure during electroforming, This result is in agreement with the SEM top-view analysis (Fig. 3). Since the EL efficiency of the LEDs increased significantly after electroforming, the phenomenon (or phenomena) that occurred during electroforming and caused the removal of the ITO from the structure may be expected to have led to similar changes in the Si-based thin films. In our previous model [4,8], we suggested that structural changes occurring during electroforming, leading to enhanced EL efficiency, might have been caused by Joule heating. It was shown previously [8] that ITO, which was coated by physical vapor deposition at room temperature, crystallizes at high local temperature caused by the Joule effect during electroforming. Considering the disordered structure of ITO at the time of initial coating, the elimination of the disordered parts might have led ITO to have a more compact form during this crystallization. Therefore, it is possible that the interface between the ITO and n+ nc-Si:H film experiences a high stress whichmay result in the partial or complete dispatching of the ITO from some random parts of the diode surface.

While the SEM cross-sectional images were taken, elemental mapping was also performed along the sections from appropriate regions. Thegreen arrow in Figure 5 shows that the mapping measurement was performed from the glass bottom to the ITO top electrode. Since the measurement was made in a tilted manner, it was observed that even after the EDX scan was passed from glass to film layers, Si and O elements in the glass still showed signals. The mapping measurements from cross-section were also taken from non-tilted images, but there was insufficient sharpness in elemental changes in transitions between the film layers. For example, during the transition from glass to Cr, it is expected that the signals from the elements Si and O suddenly decrease and those from the element Cr suddenly increase; this was not observed. That is, regardless of whether thecross-sectional view was fully perpendicular to the cross-sectional plane or slightly tilted, while the signal was expected to be collected from one individual layer, it was also received from the neighboring layers. However, the change in the behaviors of the elements during the transitions from one layer to the other gives some idea about the structure of the films. When the mass ratios of the elements in Figure 5 are examined along the direction of measurement, it is seen that there is an increase in Cr element and then a decrease (Cr bottom electrode), the signals from the Si element pass through a hill in the middle regions (Si-based thin films), signals fromthe In element are observed to increase gradually due to the effects of oblique measurement and the size of the EDX measuring point (ITO top electrode). Similar behaviors were observed in the elemental changes during elemental mapping from the cross-sectional images that were selected to be exactly perpendicular or slightly tilted. Due to the limited measurement resolution and the absence of a mechanical change in Si-based films, it was not possible to conclude that ITO, or especially the In element, diffused into the Si-based thin films during electroforming.

After SEM analysis was performed from this electroformed diode, the regions from which a more detailed TEM analysis could be performed were determined (TEM study will be published separately). During the sample preparation step for TEM analysis, a 100 nm-thick slice from the diode was extracted using focused ion beam (FIB) technique. SEM image of that slice is given in Figure 6. One of the noteworthy points in this image is that ITO was almost unaffected in some regions, while in some other regions it was observed to completely escape from the structure. In addition, whether the ITO remained or escaped, the Si-based films, Cr bottom electrode and glass substrate seems to be at least mechanically unaffected during electroforming process.

An image analysis of the top-view SEM image of the electroformed diode captured during the FIB process (not shown here) revealed that the lateral dimensions of the craters were few micrometers, the mean value being ~2.5 μm. Although the size distribution is narrow, craters of sizes as small as 1 μm and as large as 4 μm are also present. The shapes of the craters are mostly circular. Similar blisters and craters were reported in the literature for a-Si:H films to be formed during both thermal [1113] and laser [14] annealing processes. Considering one of the observations [11] that “nearly perfectly circular discs pop off the samples” and “their diameters scatter for a given sample less than a factor of 1.5”, the geometry of the craters on the electroformed diodes of this work is very similar to those on the annealed films of the previous works. Also, it was reported that blisters/craters were formedon the surface of a-Si:H films of high hydrogen content when the annealing temperature reached 300 °C, whereas no blisters appeared on the surface of low hydrogen content (<%1) films even if the annealing temperaturewas kept above 500 °C for extended times [12]. Therefore, the mechanism of blistering/pitting in both electroformed and annealed systems must be jointly associated with the “explosive H evolution” [14].

Figure 6 does not contradict the inferences of the partial or complete dispatching of the ITO from the structure due to the stressthat might occur after its crystallization. However, surface roughening may be related also to the nanocrystallization of the Si-based films. It is well known that dehydrogenation of a-Si:H structure begin at temperatures above 300 °C [15] which isknown to be exceeded during nanocrystallization [4,8]. Therefore, a high rate of hydrogen outburst can be expected from the structure during electroforming. This phenomenon may explain the roughened surface of the ITO which includes many crater-like points of distributed size (Fig. 3).

thumbnail Fig. 4

Uniform thickness distribution of the a-SiNx:H p-i-ndiode with Si-based thin films sandwiched between the Cr buttom and ITO window electrodes.

thumbnail Fig. 5

(a) Cross-sectional view and (b) elemental mapping of an electroformed a-SiNx:H p-i-n diode. The indicated line (green arrow) shows the direction of scan.

thumbnail Fig. 6

Cross-sectional SEM image of a partially electroformed a-SiNx:H p-i-n diode taken during the FIB process.

3.3 Real-time optical microscopy analysis of the electroforming process

a-SiNx:H p-i-n diode was examined under a microscope during the electroforming process (Fig. 7). The initially smooth diode surface (Fig. 7a) began to get roughened at random points (Fig. 7b) after applying a certain level of forward bias and initiating the electroforming. After a few seconds, the roughened region spread throughout the whole diode surface and the diode was considered to be mostly electroformed (Fig. 7c). Then, we turned-off the microscope light and captured the image of that mostly electroformed diode surface while the diode was forward biased. It was seen in the high contrast image that the light was emitted from many point-like regions on the diode surface (Fig. 7d) where the ITO windows electrode still existed and rendered efficient electric field and charge injections.

thumbnail Fig. 7

Real-time optical microscope images of a-SiNx:H p-i-n diode surface (a) before and (b) during electroforming process. (c) The surface of the mostly electroformed diode. (d) High contrast microscope image of the mostly electroformed diode while it is emitting light.

4 Conclusion

The electroformed a-SiNx:H light emitting memory device was studied by SEM, EDX spectroscopy and optical microscopy analyses. In addition to the model that we previously developed using indirect measurements related to the electrical transport and luminescence phenomena, this work revealed some of its suspicious parts owing to the direct measurements on the structure. SEM top-view images showed that the smooth surface of the as-deposited diodes became rough after the electroforming. Elemental mapping taken from the top of the electroformed diode showed that In and Sn elements more or less removed from the roughened points of the surface, whereas at other points, ITO seems to have remained unaffected. Cross-sectional images of the electroformed diode showed that, regardless of whether ITO was unaffected or completely escaped the structure, the Si-based thin films always had very uniform thickness along the lateral direction; the films under ITO-free points did not exhibit any mechanical damage. We did not detect any clue of the diffusion of any of the ingredients of the ITO electrode through the Si-basedfilms. Although EDX analysis along the cross-section gave an idea about the general behavior of the structure, it could not give a detailed result because of the limited measurement resolution. A more detailed cross-sectional analysis using TEM images will be published in another work. The electroforming process was also followed by real-time optical microscopy. Initially smooth surface of the diode was shown to become rough during electroforming starting at a random point and spreading throughout the whole diode area within few seconds. The light emission from this LEM device was captured by the microscope which demonstrated that there were many bright spotson the diode surface identified as the regions where the ITO windows electrode still existed after electroforming.

Author contribution statement

All the authors were involved in the preparation of the manuscript. All the authors have read and approved the final manuscript.

Acknowledgements

This work was financially supported by Karabuk University Scientific Research Projects Coordination Unit with the Project Numbers KBUBAP-18-DS-030 and KBUBAP-18-DS-146.

References

  1. D.H. Kim, W.E. Hong, J.S. Ro, S.H. Lee, C.H. Lee, S. Park, Thin Solid Films 519, 5516 (2011) [Google Scholar]
  2. I. Pelant, P. Fojtik, K. Luterova, J. Kocka, K. Knizek, J. Stepanek, Thin Solid Films 383, 101 (2001) [Google Scholar]
  3. I. Stenger, A. Abramov, C. Barthou, Th. Nguyen-Tran, A. Frigout, P. Roca i Cabarrocas, Appl. Phys. Lett. 92, 241114 (2008) [Google Scholar]
  4. T.A. Anutgan, M. Anutgan, I. Atilgan, B. Katircioglu, Electrochem. Solid-State Lett. 14, H330 (2011) [CrossRef] [Google Scholar]
  5. I. Pelant, P. Fojtik, K. Luterova, J. Kocka, A. Poruba, J. Stepanek, Appl. Phys. A 74, 557 (2002) [CrossRef] [Google Scholar]
  6. P. Fojtik, K. Dohnaleva, T. Mates, J. Stuchlik, I. Gregorova, J. Chval, A. Fejfar, J. Kocka, I. Pelant, Philos. Mag. B 82, 1785 (2002) [CrossRef] [Google Scholar]
  7. M. Anutgan, T. Anutgan, I. Atilgan, B. Katircioglu, Appl. Phys. A 109, 197 (2012) [CrossRef] [Google Scholar]
  8. M. Anutgan, T.A. Anutgan, I. Atilgan, B. Katircioglu, IEEE Trans. Electron Dev. 58, 2537 (2011) [CrossRef] [Google Scholar]
  9. M. Anutgan, T.A. Anutgan, I. Atilgan, B. Katircioglu, Philos. Mag. B 93, 3332 (2013) [CrossRef] [Google Scholar]
  10. T. Anutgan, M. Anutgan, I. Atgan, Appl. Phys. Lett. 111, 053502 (2017) [Google Scholar]
  11. H. Shanks, C.J. Fang, L. Ley, M. Cardona, F.J. Demond, S. Kalbitzer, Phys. Status Solidi B 100, 43 (1980) [CrossRef] [Google Scholar]
  12. H. Shanks, L. Ley, J. Appl. Phys. 52, 811 (1981) [Google Scholar]
  13. J. Wang, L. Wu, X. Chen, W. Zhuo, G. Wang, Sens. Actuators A 276, 11 (2018) [CrossRef] [Google Scholar]
  14. P. Mei, J.B. Boyce, M. Hack, R.A. Lujan, R.I. Johnson, G.B. Anderson, D.K. Fork, S.E. Ready, Appl. Phys. Lett. 64, 1132 (1994) [Google Scholar]
  15. J. Shi, M. Boccard, Z. Holman, Appl. Phys. Lett. 109, 031601 (2016) [Google Scholar]

Cite this article as: Mustafa Anutgan, Tamila Anutgan, Ismail Atilgan, SEM, EDX spectroscopy and real-time optical microscopy of electroformed silicon nitride-based light emitting memory device, Eur. Phys. J. Appl. Phys. 89, 10303 (2020)

All Figures

thumbnail Fig. 1

(a) Schematical cross section of the a-SiNx:H based p-i-n diode. (b) Tilted cross-sectional SEM image of an electroformed diode. Note that the ITO top electrode seems to almost completely escape from the surface during electroforming.

In the text
thumbnail Fig. 2

EDX elemental surface mapping of a partially electroformed diode showing: (a) original SEM top-view image, (b) In mapping,(c) Sn mapping and (d) Si mapping.

In the text
thumbnail Fig. 3

(a) Top-view of a partially electroformed a-SiNx:H p-i-n diode and (b) relative amounts of elements along a line.

In the text
thumbnail Fig. 4

Uniform thickness distribution of the a-SiNx:H p-i-ndiode with Si-based thin films sandwiched between the Cr buttom and ITO window electrodes.

In the text
thumbnail Fig. 5

(a) Cross-sectional view and (b) elemental mapping of an electroformed a-SiNx:H p-i-n diode. The indicated line (green arrow) shows the direction of scan.

In the text
thumbnail Fig. 6

Cross-sectional SEM image of a partially electroformed a-SiNx:H p-i-n diode taken during the FIB process.

In the text
thumbnail Fig. 7

Real-time optical microscope images of a-SiNx:H p-i-n diode surface (a) before and (b) during electroforming process. (c) The surface of the mostly electroformed diode. (d) High contrast microscope image of the mostly electroformed diode while it is emitting light.

In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.