Structural breakdown in high power GaN-on-GaN p-n diode ...€¦ · AlGaN/GaN high electron mobility transistors (HEMTs) correlated increased gate leakage current with defects near

J. Vac. Sci. Technol. A 38, 063402 (2020); https://doi.org/10.1116/6.0000488 38, 063402

© 2020 Author(s).

Structural breakdown in high power GaN-on-GaN p-n diode devices stressed to failure Cite as: J. Vac. Sci. Technol. A 38, 063402 (2020); https://doi.org/10.1116/6.0000488Submitted: 23 July 2020 . Accepted: 03 September 2020 . Published Online: 28 September 2020

Prudhvi Peri , Kai Fu , Houqiang Fu, Yuji Zhao , and David J. Smith

COLLECTIONS

This paper was selected as Featured

This paper was selected as Scilight

https://images.scitation.org/redirect.spark?MID=176720&plid=1225655&setID=421018&channelID=0&CID=414017&banID=519951235&PID=0&textadID=0&tc=1&type=tclick&mt=1&hc=fdf387f5fce1b8a41cc1c1cb41d5a2e44e4eb8ad&location=https://doi.org/10.1116/6.0000488https://avs.scitation.org/topic/collections/featured?SeriesKey=jvahttps://doi.org/10.1116/6.0000488https://avs.scitation.org/author/Peri%2C+Prudhvihttp://orcid.org/0000-0002-2338-441Xhttps://avs.scitation.org/author/Fu%2C+Kaihttp://orcid.org/0000-0002-9405-7512https://avs.scitation.org/author/Fu%2C+Houqianghttps://avs.scitation.org/author/Zhao%2C+Yujihttp://orcid.org/0000-0001-9199-4159https://avs.scitation.org/author/Smith%2C+David+Jhttps://avs.scitation.org/topic/collections/featured?SeriesKey=jvahttps://avs.scitation.org/topic/collections/scilight?SeriesKey=jvahttps://doi.org/10.1116/6.0000488https://avs.scitation.org/action/showCitFormats?type=show&doi=10.1116/6.0000488http://crossmark.crossref.org/dialog/?doi=10.1116%2F6.0000488&domain=avs.scitation.org&date_stamp=2020-09-28

Structural breakdown in high power GaN-on-GaNp-n diode devices stressed to failure

Cite as: J. Vac. Sci. Technol. A 38, 063402 (2020); doi: 10.1116/6.0000488

View Online Export Citation CrossMarkSubmitted: 23 July 2020 · Accepted: 3 September 2020 ·Published Online: 28 September 2020

Prudhvi Peri,1,a) Kai Fu,2 Houqiang Fu,2 Yuji Zhao,2 and David J. Smith3

AFFILIATIONS

1School for Engineering of Matter, Transport and Energy, Arizona State University, Tempe, Arizona 852872School of Electrical, Energy and Computer Engineering, Arizona State University, Tempe, Arizona 852873Department of Physics, Arizona State University, Tempe, Arizona 85287

a)Author to whom correspondence should be addressed: [email protected]

ABSTRACT

The morphology of GaN-on-GaN vertical p-i-n diode devices after reverse-bias electrical stressing to breakdown has been investigated. Allfailed devices had irreversible structural damage, showing large surface craters that were ∼15–35 μm deep with lengthy surface cracks.Cross-sectional electron micrographs of failed devices showed substantial concentrations of threading dislocations around the cracks andnear the crater surfaces. Progressive ion-milling across damaged devices revealed high densities of threading dislocations and the presence ofvoids beneath the surface cracks; these features were not observed in any unstressed devices. These results should serve as a useful referencefor future reliability studies of vertical high-power GaN devices.

Published under license by AVS. https://doi.org/10.1116/6.0000488

I. INTRODUCTION

Wide bandgap semiconductor materials are of much currentinterest for many potential applications in power electronics.GaN-based devices are of particular interest because of superior elec-trical properties such as high critical electric field, low intrinsiccarrier concentration, high thermal conductivity, large electronmobility, and high saturation velocity.1–5 Vertical GaN devices arepreferred to lateral devices for better packaging, higher efficiency,high-current, and high-voltage applications.5,6 Conventional GaNdevices grown on foreign substrates, such as Si, SiC, and sapphire,have large lattice-mismatch and differences in thermal conductivity.These differences result in GaN devices with high densities ofthreading dislocations (TDs) (108–1010 cm−2), which contribute tononradiative recombination and scattering centers, and limit the per-formance of electrical and optical devices.7 Minimizing the defectdensity is critical to maximizing device performance since defectsaffect breakdown voltage, leakage current, device reliability, high-temperature reverse bias, and operating lifetime.8 Buffer layers, suchas AlN, are commonly used to alleviate lattice-mismatch and differ-ences in thermal expansion between GaN and foreign substrates butthese layers add to overall cost and complexity. Recent developmentsin growth using hydride vapor pressure epitaxy (HVPE) have led to

the availability of freestanding GaN substrates with defect densitieslower than ∼106 cm−2.9 Epitaxial GaN-on-GaN layers can be grownwith reduced defect density (

AlGaN/GaN high electron mobility transistors (HEMTs) correlatedincreased gate leakage current with defects near gate edges,19 andHVPE-grown GaN Schottky diodes with micropipe defects werereported to show increased leakage current.20

Despite these results for GaN-based LEDs, as well as lateralHEMTs, a detailed investigation on the effect of dislocations on theperformance of vertical GaN p-n diodes/devices is lacking. Althoughit was suggested that TDs caused leakage in vertical diodes grown byMOCVD,21 microscopy-based studies do not appear to have beenreported. Such studies are critical for further device developmentsince the structural and electrical properties of devices depend sensi-tively not only on the growth method and the growth conditions butalso on the presence of dopants, impurities, and the device struc-ture.12 This present work has investigated GaN-on-GaN verticaldevices that have been electrically stressed to breakdown. A majorobjective was to gain insight into the failure mechanism and to estab-lish whether surface treatment or surface etching methods duringregrowth had any impact on the device failure.

II. EXPERIMENTAL DETAILS

Freestanding 2-in. c-plane n+ GaN substrates with a carrier con-centration of ∼1018 cm−2 were used to grow the GaN device struc-tures using MOCVD. The substrates were provided by SumitomoElectric Industries Ltd. Nine micrometers of an unintentionallydoped (UID) GaN drift layer was grown on the GaN substrates. Thesurfaces of the UID-GaN layers were treated, as described below,followed by the growth of thin UID-GaN insertion layers (∼50 nm)and overgrowth with Mg-doped p-GaN layers with thicknesses of300–500 nm. The purpose of the insertion layers was to shift the

regrowth interfaces (insertion layer/UID-GaN drift layer) well awayfrom the peak electric field in the p-n junction depletion region, thusminimizing reverse-leakage current due to any regrowth interfacecharge contributed by contaminants such as O or Si.

Two sets of devices were investigated in this study, labeled hereas A-series and L-series. Both sets were surface-treated (UV-ozone45min +HF 5min +HCl 5min) before regrowth. The UV-ozonetreatment oxidizes the UID-GaN surface, and the following treatments

FIG. 1. SEM images showing examples of devices that had been reverse-biasstressed to electrical breakdown, revealing extensive surface damage, includingdeep craters and wide cracks: (a) Device L5, (b) Device L7, (c) Device L14,and (d) Device L16.

FIG. 2. Schematic illustrating maximum crater depths of Series L devices thathad been stressed to failure.

FIG. 3. TEM cross section image of Device L14 from the lift-out locationmarked in Fig. 1(c), showing the left side of cracks with many dislocations moreconcentrated toward the crater surface and crack.

ARTICLE avs.scitation.org/journal/jva

J. Vac. Sci. Technol. A 38(6) Nov/Dec 2020; doi: 10.1116/6.0000488 38, 063402-2

Published under license by AVS.

https://avs.scitation.org/journal/jva

with HF and HCl are intended to remove surface contaminants. Inaddition, the A-series devices were etched using inductively coupledplasma (ICP) dry-etching. The ICP dry-etching procedure was asequence of etching steps intended to remove the surface layer andminimize etch damage, which was achieved by employing four con-secutive etching steps with decreasing RF-power (70, 35, 5, and 2W).22

More details about the growth and treatment procedures can befound elsewhere.23 The L-series devices showed breakdown voltagesin the range of 200–750 V with typical reverse-leakage current den-sities of 10 A/cm2 at breakdown, whereas the A-series devices exhib-ited substantially higher breakdown voltages of 1070–1270 V withtypical reverse-leakage current densities of 4 A/cm2 at breakdown.

FIG. 4. (a) TEM cross-sectional imageof unstressed devices showing theabsence of dislocations over the entirelift-out sample and (b) enlargement ofthe p-GaN/insertion layer/UID-GaNregion showing no surface-treatmentdamage or defects.

FIG. 5. Series of SEM images ofDevice L16, showing progressivecross-sectional milling across area of30 μm (length) × 25 μm (depth): (a)plan-view image showing the locationwhere milling started (black double-arrowed line); (b) cross-sectional imageafter 5 μm of milling, showing the pres-ence of voids right below the surfacecrack; (c) cluster of voids; (d) voidsextending from the surface just belowcrack to ∼16 μm deep; (e) voids andTDs; (f ) presence of TD clusters allover area; (g) and (h) TDs extendingfrom the substrate; and (i) milledalmost to the edge of device.





Samples suitable for cross-sectional observation by transmissionelectron microscopy (TEM) were prepared by focused-ion-beam(FIB) milling using an FEI NOVA 200 dual-beam system with initialthinning done at 30 keV and final thinning done at 5 keV. Scanningelectron micrographs were also recorded with the FEI NOVA 200during progressive milling across the surface craters that wereobserved in failed devices. A Philips-FEI CM-200 FEG transmis-sion electron microscope (TEM) operated at 200 keV was used forstructural imaging. All TEM images were taken at the [1�100] zoneaxis. High-resolution images were taken using phase contrastimaging without an objective aperture. Low-magnification andmedium-magnification images were taken using an objective aper-ture for better contrast.

III. RESULTS

Examples of plan-view SEM images of devices L5, L7, L14, andL16 after they were reverse-bias stressed to breakdown are shown inFig. 1. This set of devices had been regrown with an insertion layer of250 nm, followed by another 500 nm p-GaN, after the surface layer ofthe original UID-GaN had been chemically treated. All stresseddevices showed extensive, craterlike surface pits, and most also showedevidence for lengthy cracks extending across parts of the craters andeven outside the device region in some cases. Sixteen devices werestressed to breakdown and all 16 showed similar craterlike surfacedamage with varying crater depths. The measured depths of thesecraters ranged from 13 to 38 μm, as shown schematically in Fig. 2.

Lift-out with the FIB was performed across the crack onDevice L14, as indicated by the arrow in Fig. 1(c), and the corre-sponding TEM image is shown in Fig. 3. This cross-sectional imagereveals a high density of dislocations along the left side of thecrack, which appear to be more concentrated toward the GaNsurface and adjacent to the crack. For comparison, Fig. 4(a) showsa cross-sectional TEM image of an unstressed device, where thewhite arrow indicates the position of the interface between p-GaNand the insertion layer. No dislocations or cracks are visible and nostructural damage between the insertion layer and the UID-GaNlayer due to surface treatment can be observed. The enlargement inFig. 4(b) shows the p-GaN/insertion layer/UID-GaN area wherethere are no visible defects or surface damage, although the topp-GaN layer shows evidence for Mg precipitates.

In order to appreciate the full extent of defect formation in thefailed devices, Device L16 was progressively milled in cross sectionacross the surface crater and then repeatedly imaged in situ with theSEM after another roughly 5 μm had been milled away. Figure 5(a)shows a plan-view SEM image of the original device after failurebefore commencement of trenching. Milling started at the top ofcrater at the position indicated by the arrow and then proceededprogressively upward. Figures 5(b)–5(h) are a series of SEM imagesshowing the progression of this milling over an area of roughly30 × 25 μm2. Figure 5(b) reveals a collection/cluster of voids that arepresent in the GaN substrate beneath the crack. Figures 5(c)–5(e)show parts of the same void cluster after the additional material,about 5 μm at a time, had been removed. Figures 5(c)–5(f) show a

FIG. 6. Plan-view SEM image of Device A1 after reverse-bias breakdown at∼1.27 kV, showing surface pits and crack. White arrowed lines indicate thelocation of cross-sectional milling and double-arrowed line indicates the locationof TEM lift-out.

FIG. 7. Cross-sectional TEM imagesof Device Al lift-out across the surfacecrack from location marked withdouble-arrowed line in Fig. 6: (a) low-magnification image showing large4-μm-deep crack, smaller cracks, andthe presence of dislocations concen-trated near the cracks and surface and(b) high-magnification image showingsmall 2-μm-deep crack and surround-ing defects.





cluster of threading dislocations at the center of image, right belowthe crack, and away from the surface crack. Figure 5(e) shows athreading dislocation extending from the crater surface to deepwithin the substrate over a vertical distance of almost 80 μm.

In addition to the chemical etching used for the L-series,Devices A1 and A2 of the A-series were ICP dry etched, followed bygrowth with 50 nm of the GaN insertion layer and overgrowth with500 nm of p-GaN. Device A1 suffered electrical breakdown at areverse bias of ∼1.27 kV. It was then observed for surface anomalies,as shown by the SEM image in Fig. 6. The surface shows pits nearthe center of the circular contact with a long crack that extends acrosspart of the device to outside the device area. The crater and the cracksurface morphology appear similar to that observed for the L-seriesdevices despite the considerable difference in breakdown voltages.

A cross-sectional sample suitable for TEM observation waslifted-out across the visible crack of Device A1, as marked in Fig. 6 bythe double-headed arrow. Figure 7 shows corresponding TEM images.Figure 7(a) reveals that the entire top surface region here is riddled withdislocations, not unlike the L-series devices, and a crack is also visiblethat penetrates more than 4 μm from the crater surface. Figure 7(b)from an adjacent area shows another crack that extends almost 2 μmdownward from the crater surface. Dislocations here are concentratedclose to the crack and along the top surface, as clearly visible inFig. 7(b). The GaN top surface is not flat/even because of the irregularloss of surface material that occurred during the device breakdown.

Figure 8 shows a sequence of cross-sectional SEM images ofDevice A1 after progressive milling had been done along the directionindicated by the arrowed lines in Fig. 6. Figures 8(a) and 8(c) revealthe presence of TDs all across the milled area, while Fig. 8(b) shows aTD that extends almost 49 μm downward from the crater surface.Figures 8(d) and 8(e) show that the surface crack is almost 15 μmdeep, and that TDs are present everywhere. For comparison, Fig. 9shows TEM images of the equivalent Device A2, which was not sub-jected to any electrical tests. Figure 9(a) shows that no dislocations or

FIG. 8. Sequential series of SEMimages of Device A1 after progressivecross-sectional milling across area of40 μm (length) × 25 μm (depth) fromthe location marked with arrowed whitelines in Fig. 6: (a) TDs all over milledarea; (b) high-magnification imagefocused on TD that extended ∼49 μmdown from the crater surface; (c) voidsconcentrated nearer to cracks; (d)surface crack that appears to be∼15 μm deep extending into thedevice; (e) TDs near and away fromcracks and extending deep from thesubstrate; and (f ) milling stopped atthe edge of device and surface crackextending outside devices is stillobserved.

FIG. 9. Cross-sectional TEM images of Device A2 which was not subjected toany electrical tests: (a) low-magnification image shows no cracks or dislocationsand (b) high-magnification image of the p-GaN/insertion layer/UID-GaN areashows no visible etch damage or defects.





cracks are present in the lifted-out area. The enlargement in Fig. 9(b)shows the p-GaN/insertion layer/UID-GaN area, and no etching-based damage or structural defects are visible.

IV. SUMMARY

This paper described the irreversible structural changes observedin two sets of GaN-on-GaN high-power devices after they had beenreverse-biased to the point of electrical breakdown. Surface-treateddevices (Series L) developed deep craters and lengthy cracks whenfailure occurred. The craters extended across much of each faileddevice and had depths of tens of micrometers. The failed devices alsoshowed the presence of cracks, clusters of voids, and TDs under thecraters. These cracks extended typically about 6–15 μm downward.The voids were present in the substrate region, whereas TDs werepresent across entire devices extending deep into the substrate.Devices that had been etched and then overgrown (Series A) showedsimilar morphology after failure, although their breakdown voltageswere considerably higher. Cracks were again observed on the crateredsurfaces of the failed devices, while cross-sectional TEM imagesshowed the formation of threading dislocations that were concen-trated close to the cracks and near the crater surfaces. Similar disloca-tions were never observed in unstressed devices subjected to identicaletch-regrowth treatment. The presence of pre-existing defects indevices before regrowth does not seem to be an important factorin device failure since the Series L and Series A devices weregrown on similar substrates, but they had significantly differentbreakdown voltages. The lower breakdown voltages for the surface-treated devices can most likely be attributed to surface oxidationduring the UV-ozone treatment. Overall, these results provide highlyuseful information about the reliability of vertical power electronicsand should contribute to the design of better future devices.

ACKNOWLEDGMENTS

This work was supported by an ARPA-E Award (No.DE-AR0000868). The authors acknowledge the use of facilitieswithin the John M. Cowley Center for High Resolution ElectronMicroscopy at the Arizona State University.

REFERENCES1F. Zeng, J. X. An, G. Zhou, W. Li, H. Wang, T. Duan, L. Jiang, and H. Yu,Electronics 7, 377 (2018).2K. Harafuji and K. Kawamura, Jpn. J. Appl. Phys. 44, 6495 (2005).3T. Sugahara et al., Jpn. J. Appl. Phys. 37, L398 (1998).4S. Fujita, Jpn. J. Appl. Phys. 54, 030101 (2015).5J. Chen, W. Yi, T. Kimura, S. Takashima, M. Edo, and T. Sekiguchi, Appl. Phys.Express 12, 051010 (2019).6Y. Saitoh et al., Appl. Phys. Express 3, 081001 (2010).7H. Y. Shih, M. Shiojiri, C. H. Chen, S. F. Yu, C. T. Ko, J. R. Yang, R. M. Lin,and M. J. Chen, Sci. Rep. 5, 13671 (2015).8I. C. Kizilyalli, A. P. Edwards, O. Aktas, T. Prunty, and D. Bour, IEEE Trans.Electron Devices 62, 414 (2015).9K. Motokia et al., J. Cryst. Growth 237–239, 912 (2002).10H. Fujikura, T. Yoshida, M. Shibata, and Y. Otoki, Proc. SPIE 10104, 1010403(2017).11M. Lee, D. Mikulik, M. Yang, and S. Park, CrystEngComm 19, 2036(2017).12Y.-Y. Wong, E. Y. Chang, T.-H. Yang, J.-R. Chang, J.-T. Ku, M. K. Hudait,W.-C. Chou, M. Chen, and K.-L. Lin, J. Electrochem. Soc. 157, H746(2010).13P. Kozodoy, J. P. Ibbetson, H. Marchand, P. T. Fini, S. Keller,J. S. Speck, S. P. Denbaars, and U. K. Mishra, Appl. Phys. Lett. 73, 975(1998).14D. S. Li, H. Chen, H. B. Yu, H. Q. Jia, Q. Huang, and J. M. Zhou, J. Appl.Phys. 96, 1111 (2004).15X. A. Cao, J. A. Teetsov, F. Shahedipour-Sandvik, and S. D. Arthur, J. Cryst.Growth 264, 172 (2004).16M. Lee, H. U. Lee, K. M. Song, and J. Kim, Sci. Rep. 9, 970(2019).17J. W. P. Hsu, M. J. Manfra, R. J. Molnar, B. Heying, and J. S. Speck, Appl.Phys. Lett. 81, 79 (2012).18J. W. P. Hsu, M. J. Manfra, D. V. Lang, S. Richter, S. N. G. Chu,A. M. Sergent, R. N. Kleiman, L. N. Pfeiffer, and R. J. Molnar, Appl. Phys. Lett.78, 1685 (2001).19M. R. Johnson et al., J. Vac. Sci. Technol. B 30, 062204 (2012).20A. Y. Polyakov, E. B. Yakimov, N. B. Smirnov, A. V. Govorkov,A. S. Usikov, H. Helava, Y. N. Makarov, and I.-H. Lee, J. Vac. Sci. Technol. B32, 051212 (2014).21K. H. Lee, S. J. Chang, P. C. Chang, Y. C. Wang, and C. H. Kuo,J. Electrochem. Soc. 155, H716 (2008).22K. Fu, H. Fu, X. Huang, H. Chen, T. H. Yang, J. Montes, C. Yang, J. Zhou, andY. Zhao, IEEE Electron Device Lett. 40, 1728 (2019).23K. Fu et al., IEEE J. Electron Devices Soc. 8, 74 (2020).




https://doi.org/10.3390/electronics7120377https://doi.org/10.1143/JJAP.44.6495https://doi.org/10.1143/JJAP.37.L398https://doi.org/10.7567/JJAP.54.030101https://doi.org/10.7567/1882-0786/ab14cbhttps://doi.org/10.7567/1882-0786/ab14cbhttps://doi.org/10.1143/APEX.3.081001https://doi.org/10.1038/srep13671https://doi.org/10.1109/TED.2014.2360861https://doi.org/10.1109/TED.2014.2360861https://doi.org/10.1016/S0022-0248(01)02078-4https://doi.org/10.1117/12.2257202https://doi.org/10.1039/C7CE00246Ghttps://doi.org/10.1149/1.3392365https://doi.org/10.1063/1.122057https://doi.org/10.1063/1.1763234https://doi.org/10.1063/1.1763234https://doi.org/10.1016/j.jcrysgro.2004.01.031https://doi.org/10.1016/j.jcrysgro.2004.01.031https://doi.org/10.1038/s41598-019-38664-xhttps://doi.org/10.1063/1.1490147https://doi.org/10.1063/1.1490147https://doi.org/10.1063/1.1356450https://doi.org/10.1116/1.4766303https://doi.org/10.1116/1.4895658https://doi.org/10.1149/1.2959117https://doi.org/10.1109/LED.2019.2941830https://doi.org/10.1109/JEDS.2020.2963902https://avs.scitation.org/journal/jva

Structural breakdown in high power GaN-on-GaN p-n diode devices stressed to failureI. INTRODUCTIONII. EXPERIMENTAL DETAILSIII. RESULTSIV. SUMMARYReferences

Documents

Structural breakdown in high power GaN-on-GaN p-n diode ...€¦ · AlGaN/GaN high electron mobility transistors (HEMTs) correlated increased gate leakage current with defects near