ArticlePDF Available

Reliability & failure analysis methodology of optoelectronics and its extendibility for future technologies

Authors:
Jack Jia-Sheng Huang at Source Photonics
Jack Jia-Sheng Huang
  • Source Photonics
Download full-text PDF
Read full-text
Download citation

Abstract

We review the state-of-the-art reliability and failure analysis methodology of the optoelectronic devices. We will discuss the feasibility and extendibility of applying those established techniques to the future technologies such as nanotechnologies and renewable energies.
Content uploaded by Jack Jia-Sheng Huang
Author content
All content in this area was uploaded by Jack Jia-Sheng Huang on Nov 22, 2017
Content may be subject to copyright.
RELIABILITY & FAILURE ANALYSIS METHODOLOGY OF
OPTOELECTRONICS AND ITS EXTENDIBILITY FOR FUTURE
TECHNOLOGIES
Jia-Sheng Huang
Emcore, Fiber Optics Division, 2015 W. Chestnut Street, Alhambra, CA 91803
Abstract: We review the state-of-the-art reliability and failure analysis
methodology of the optoelectronic devices. We will discuss the feasibility and
extendibility of applying those established techniques to the future technologies
such as nanotechnologies and renewable energies.
1. Introduction and Background
Reliability and failure analysis are the crucial elements for integrated circuits (IC) and optoelectronics
industries to safegurad their products before and after field deployment. With ever increasing complexity
and novelty of device development in the past 40-60 years, reliability and electrostatic discharge (ESD)
studies continue to be indispensable laboratory-controlled means to ensure the product quality and
reliability prior to shipment [1-2]. Albeit seemingly not a fancy topic, reliability is perhaps the most
persistent and critical area of investment for the electronics and photonics industries to maintain, even
during the severe 2008-09 economic downturn. On the other hand, failure analysis (FA), also referred as
failure mode analysis (FMA), is often an aftermath firefighting. FA not only provides early in-line
detection during device manufacturing, but also helps understand the failure mechanism of the device after
degradation. Despite of the numerous reliability qualification data, companies often rely on the FA results
to help themselves regain customer confidence of their products in the events of field return. As visible and
sensitive as it turns out, FA has always been a critical part of customer service.
Due to the recent concern of global climate change and terrorist threat, new emerging technologies
such as nanotechnologies, renewable energies, microelectromechanical systems (MEMS) and smart devices
have drawn increasing attention. Nanotechnologies promise to offer great opportunities not only for its
nanoscale feature size, but also for its unique novel electronic/optical properties [3,4]. Many phenomena
beyond imagination have been observed in the nano-world. For example, CdS and InP nano-wires (NW) as
integratable laser sources [5], as well as carbon nano-tubes [6] and GaN nano-wires [7] for field effect
transistors (FET) have been demonstrated. In this paper, we review the state-of-the-art reliability
methodologies and FA techniques of ICs and optoelectronics devices. We also discuss their applicability
and extendibility for the future technologies.
2. Reliability
2.1. Reliability of optoelectronics
Due to the demand to increase bandwidth using the existing fiber networks, coarse/dense wavelength
division multiplexing (CWDM/DWDM) continues to receive increasing market interest. Figures 1(a)-(c)
show the reliability performance and wavelength stability data of O-band CWDM lasers with 20nm
wavelength spacing in the range of 1271-1371nm. Distributed feedback (DFB) lasers with low threshold
current and excellent linearity for CWDM network were demonstrated. All lasers showed robust long-term
reliability behavior with small degradation in the threshold current (Ith). The Ith change was only <5% after
>6000hr aging, compared to the 50% failure criterion. The wavelength stability also met the stringent
DWDM requirement (<=0.1nm).
O-band CWDM lasers
100C, 175mA
Aging tim e (hour)
0 2000 4000 6000 8000
Relative Ith change (%)
0.9
1.0
1.1
1.2
1.3
1.4
1.5
Fig.1: (a) Optical spectra (b) reliability data showing relative Ith trend chart and (c) wavelength drift data of O-band CWDM lasers.
59
12:15 PM – 12:45 PM
ThB4 (Invited)
978-1-4244-5313-9/10/$26.00 ©2010 IEEE
2.2. Reliability of future technologies
Although the future technologies such as nano-devices have different dimension, design and geometry
from the present ICs and optoelectronics, the testing methodology and underlying physics have some
commonality. For example, the CdS nanowires with diameters of 80-200nm could be stressed in a similar
way to the semiconductor lasers where the ridge widths are typically 2-10um except that the stress current
and temperature need to be scaled for the nanowires based on their dimension and thermal impedance. The
physics of degradation driven by defect growth and propagation may need to be modified from the existing
model, depending upon the material and process. Test fixture modification and addition of new test
parameters may be needed to provide proper acceleration that is relevant to device operation [8].
3. Failure analysis
3.1. FA of ICs and Optoelectronics
Analytical tools play a very important role in device characterization. Scanning electron microscopy
(SEM), transmission electron microscopy (TEM), scanning transmission electron microscopy (STEM) and
focused ion beam (FIB) have been used to reveal the device morphology. Atomic force microscopy (AFM)
has been widely used to map the surface topography. Secondary ion mass spectrometry (SIMS) and Auger
electron spectroscopy (AES) have been utilized to determine the doping profile.
Figure 2 shows the STEM images of Ti/Pt/Au/Cr/Au metal contact to p-InGaAs epitaxial layer,
revealing the interdiffusion reaction during the annealing process. The layer thickness and morphology
were unfolded with this STEM technique [9]. Figure 3(a) shows the SEM-FIB images of electromigration
(EM)-induced voids in the IC multilevel interconnects. Figure 3(b) shows the SEM images of the Ni/p+Si
contact where ultra-fast silicide formation at a diffusion rate close to liquid phase was observed [10].
Fig.2: STEM of p-InGaAs/Ti/Pt/Au/Cr/Au
contact after annealing at 430C.
Fig.3: (a) EM-induced voids in IC interconnects and (b) electrically-driven NiSi2
silicide formation.
3.2. FA of future technologies
The analytical techniques established for electronic and photonic industries offer good foundations that
are likely extendable for future technologies. Several cutting-edge technology studies have been performed
by using the existing analytical techniques. For example, AFM was employed to study the surface
morphology of 7nm thick polymer solar cell [11]. TEM has been used to study the nanoscale devices such
as nano-eFUSE and nano-gate oxide [12].
4. Conclusions
We have reviewed the reliability and failure analysis methodology of the ICs and optoelectronics. We
discuss its applicability for the future technologies and show that the establishment of the present
techniques offers great opportunities for future studies.
6. References
[1] J.S. Huang, et al, IEEE Tran. Device Mater. Reliab., vol. 5, no. 4, p.665-674, 2005.
[2] J.S. Huang, et al, IEEE Tran. Device Mater. Reliab., vol. 7, no. 3, p.453-461, 2007.
[3] C.M. Lieber, Sci. Am., vol. 285, p.58-64, 2001.
[4] N. A. Melosh, et al, “Ultrahigh-density nanowire lattices and circuits”, Science, vol. 300, p.112-115, 2003.
[5] X. Duan, Y. Huang, Y. Cui, J. Wang and C.M. Lieber, Nature, vol. 409, p.66-69, 2001.
[6] S.J. Tans, R.M. Verschueren and C. Dekker, Nature, vol. 393, p.49-52, 1998.
[7] Y. Huang. X. Duan, Y. Cui, L.J. Lauhon, K.H. Kim and C.M. Lieber, Science, vol. 294, p.1313-1317, 2001.
[8] J.S. Huang, IEEE Tran. Device Mater. Reliab., vol. 5, no. 1, p.150-154, 2005.
[9] J.S. Huang and C.B. Vartuli, J. Appl. Phys., vol. 93, no. 9, p.5196-5200, 2003.
[10] J.S. Huang, C.N. Liao, K.N. Tu, S.L. Cheng and L.J. Chen, J. Appl. Phys., vol. 84, p.4788-4796, 1998.
[11] A. Kumar, G. Li, Z. Hong and Y. Yang, Nanotech., vol. 20, p.165202, 2009.
[12] C.H. Tung, K.L. Pey, R. Ranjan, L.J. Tang and D.S. Ang, Appl. Phys. Lett., vol. 89, issue 22, p.221902, 2006.
60
A research on optoelectronic coupler storage reliability by accelerated degradation testing
Conference Paper
  • Jan 2016
In this paper, the storage failure mode and mechanism of optoelectronic couplers are studied by Accelerated Degradation Testing (ADT) and a series of following analyses. Firstly, the type of optoelectronic coupler to be studied is determined in the paper. Then, an ADT is conducted, which ends after 4176 hours. Then it is found that all the parameters show no degradation except for Current Transfer Ratio (CTR) and the degradation rates of CTR between stress levels follow the Arrhenius law. However, different from the previous research, a phenomenon of phased degradation is discovered in the highest stress level. To determine the degradation causes of the two phases, a series of follow-up analyses is conducted. By excluding the possibility of integrated photosensitive circuit degradation and Light Emitting Diode (LED) contamination, we identify the active layer of LED as the degradation region and ascribe the CTR degradation to an improvement of deep-level defects. To describe the phased degradation, we also establish a phased model. In the first phase, the latent defect sources transform into deep-level defects. In the latter phase, deep-level defects form non-radiative carrier-recombination centers, and the non-radiative carrier-recombination centers increase the amount of deep-level defects in return. For a further verification, a simulation with Atlas is also conducted by injecting different amount of deep-level defects into the active layer of LED and observing the degradation of LED luminance. With the fixed transmission ratio between the luminance and the output current, the degradation trend of CTR is acquired. A good fit between ADT and simulation results is shown, which could verify the correctness of the storage degradation model. The established model can be utilized to assess and improve the storage reliability of AlGaAs/GaAs LED based optoelectronic couplers.
View
Human-Body-Model Electrostatic-Discharge and Electrical-Overstress Studies of Buried-Heterostructure Semiconductor Lasers
Article
Full-text available
  • Oct 2007
Optoelectronic components such as laser diodes, light-emitting diodes, and photodiodes are susceptible to electro-static discharge (ESD) and electrical overstress (EOS). Human-body model (HBM) is the most widely adopted method for the characterization of the ESD performance. In this paper, we report a comprehensive study of the ESD and EOS characteristics of buried-heterostructure (BH) semiconductor lasers using the HBM. Threshold current, optical power, optical spectrum, and reverse-bias current are characterized during the ESD study. We show that the ESD-failure thresholds depend upon the polarity. The chip can sustain the highest ESD stress under forward bias and the lowest one under forward/reverse bias. We also show that the BH lasers exhibit two types of ESD-degradation behavior. The soft degradation is characterized by a gradual increase in the threshold current, whereas the hard degradation is identified by a sudden jump in the threshold current during the ESD voltage ramp. The ESD-degradation behavior seems to be influenced by the cavity length. The failure-analysis results show that about 27% of the ESD failure is related to facet damage. The damage regions occur at the upper laser mesa structure and form preferentially on the bond-pad side. The preferential formation of the facet damage is suggestive of current-crowding effect. We have also found that the ESD-degradation behavior is a function of the facet damage. The soft-degradation failure shows a stronger correlation with the facet damage than the hard-degradation one. Finally, we demonstrate that the ESD performance of the laser can be improved by adding a protection diode. Index Terms-Electrical overstress (EOS), electrostatic discharge (ESD), failure analysis (FA), optical materials, scanning electron microscopy (SEM), semiconductor lasers.
View
Scanning transmission electron microscopy study of Au/Zn/Au/Cr/Au and Au/Ti/Pt/Au/Cr/Au contacts to p-type InGaAs/InP
Article
Full-text available
  • May 2003
We studied the interfacial reaction of Au/Zn/Au/Cr/Au and Au/Ti/Pt/Au/Cr/Au contacts to p-InGaAs/p-InP using scanning transmission electron microscopy. We found that the alloying morphology was distinctly different in the two contact systems. For Au/Zn/Au/Cr/Au, significant interdiffusion between the metal and InGaAs contact layer occurred. Two types of compound were formed: one was rich in Au and the other was rich in Ga and As. Another interesting observation was that a significant amount of As has outdiffused into the Cr layer after alloying. For the Au/Ti/Pt/Au/Cr/Au, only interfacial layers were involved in the reaction. Compounds of Au-Ga-In, Ti-As, and Au-Ga were formed, and the Cr layer remained intact. The mechanisms of compound formation are discussed.
View
Abnormal electrical behavior and phase changes in implanted p +- And n +-Si channels under high current densities
Article
Full-text available
  • Nov 1998
The p+- and n+-Si channels were prepared by implanting a dose of 5×1015 ions/cm2 of BF2+ and As+ at 40 keV, respectively, into the n-Si and p-Si substrates, followed by a 900 °C-30 min annealing. Nickel contacts (cathode and anode) to these channels were electron-gun evaporated. Electrical current was applied gradually to 80 mA to these channels, corresponding to current densities of 106–107 A/cm2. For the p+-Si, the resistance responded by increasing to a maximum, then decreasing until a precipitous drop took place. For the n+-Si, the resistance increased by a less amount compared to the p+-Si, but also dropped abruptly. The resistance drop is permanent in the p+-Si channels, but not in the n+-Si channels. Mechanisms responsible for these resistance changes in terms of Joule heating, high field effect and junction leakage are proposed. For the precipitous drop in channel resistance of the p+-Si, scanning electron microscopy and transmission electron microscopy showed that a NiSi2 line bridging the cathode and anode contacts had been formed. No silicide line formation in the n+-Si channels was observed. © 1998 American Institute of Physics.
View
Room-Temperature Transistor Based on a Single Carbon Nanotube
Article
Full-text available
  • May 1998
The use of individual molecules as functional electronic devices was first proposed in the 1970s (ref. 1). Since then, molecular electronics(2,3) has attracted much interest, particularly because it could lead to conceptually new miniaturization strategies in the electronics and computer industry. The realization of single molecule devices has remained challenging, largely owing to difficulties in achieving electrical contact to individual molecules. Recent advances in nanotechnology, however, have resulted in electrical measurements on single molecules(4-7). Here we report the fabrication of a field-effect transistor-a three-terminal switching device-that consists of one semiconductings(8-10) single-wall carbon nanotube(11,12) connected to two metal electrodes. By applying a voltage to a gate electrode, the nanotube can be switched from a conducting to an insulating state. We have previously reported(5) similar behaviour for a metallic single-wall carbon nanotube operated at extremely low temperatures. The present device, in contrast, operates at room temperature, thereby meeting an important requirement for potential practical applications. Electrical measurements on the nanotube transistor indicate that its operation characteristics can be qualitatively described by the semiclassical band-bending models currently used for traditional semiconductor devices. The fabrication of the three-terminal switching device at the level of a single molecule represents an important step towards molecular electronics.
View
Indium phosphide nanowires as building blocks for nanoscale electronic and optoelectronic devices
Article
Full-text available
  • Feb 2001
Nanowires and nanotubes carry charge and excitons efficiently, and are therefore potentially ideal building blocks for nanoscale electronics and optoelectronics. Carbon nanotubes have already been exploited in devices such as field-effect and single-electron transistors, but the practical utility of nanotube components for building electronic circuits is limited, as it is not yet possible to selectively grow semiconducting or metallic nanotubes. Here we report the assembly of functional nanoscale devices from indium phosphide nanowires, the electrical properties of which are controlled by selective doping. Gate-voltage-dependent transport measurements demonstrate that the nanowires can be predictably synthesized as either n- or p-type. These doped nanowires function as nanoscale field-effect transistors, and can be assembled into crossed-wire p-n junctions that exhibit rectifying behaviour. Significantly, the p-n junctions emit light strongly and are perhaps the smallest light-emitting diodes that have yet been made. Finally, we show that electric-field-directed assembly can be used to create highly integrated device arrays from nanowire building blocks.
View
View
Ultrahigh-Density Nanowire Lattices and Circuits
Article
  • Apr 2003
We describe a general method for producing ultrahigh-density arrays of aligned metal and semiconductor nanowires and nanowire circuits. The technique is based on translating thin film growth thickness control into planar wire arrays. Nanowires were fabricated with diameters and pitches (center-to-center distances) as small as 8 nanometers and 16 nanometers, respectively. The nanowires have high aspect ratios (up to 10^6), and the process can be carried out multiple times to produce simple circuits of crossed nanowires with a nanowire junction density in excess of 10^(11) per square centimeter. The nanowires can also be used in nanomechanical devices; a high-frequency nanomechanical resonator is demonstrated.
View
Nanometal-oxide-semiconductor field-effect-transistor contact and gate silicide instability during gate dielectric breakdown
Article
  • Nov 2006
  • APPL PHYS LETT
Contact and gate silicide migration induced by gate dielectric breakdown transient has been observed in various silicide materials, including Ni, Co, and Ti silicides, used across several generations of metal-oxide-semiconductor field effect transistor (MOSFET) process technology nodes with different device geometries, gate dielectric materials, and stress conditions. This failure mechanism occurs during gate dielectric breakdown transient on both anode and cathode sides and on both n- and p-MOSFETs. The wear-out mechanisms are discussed and believed to be responsible for the instability and transformation observed.
View
An atomistic simulator for thin film deposition in three dimensions
Article
  • Oct 1998
We describe an atomistic simulator for thin film deposition in three dimensions (ADEPT). The simulator is designed to bridge the atomic and mesoscopic length scales by using efficient algorithms, including an option to speed up surface diffusion using events with multiple diffusion hops. Sputtered particles are inserted and assigned ballistic trajectories with angular distributions appropriate for magnetron sputtering. Atoms on the surface of the film execute surface diffusion hops with rates that depend on the local configuration, and are consistent with microscopic reversibility. The potential energies are chosen to match information obtained from a database of first principles and molecular dynamics (MD) calculations. Efficient computation is accomplished by selecting atoms with probabilities that are proportional to their hop rates. A first implementation of grain boundary effects is accomplished by including an orientation variable with each occupied site. Energies and mobilities are assigned to atoms in grain boundaries using values obtained from MD. In this article we describe simulations of the deposition of aluminum. The film atoms occupy face centered cubic lattice sites of a single crystal. Three-dimensional films deposited on vias and trenches up to 0.05 μm in size have been simulated. We discuss these films in the context of step coverage issues, and scaling to macroscopic sizes. Texture development is discussed in the light of simulations showing anisotropic crystal growth. Future versions of ADEPT will include a more complete model of grain development in thin films. © 1998 American Institute of Physics.
View
The Incredible Shrinking Circuit
Article
  • Oct 2001
Researchers Have Built Nanotransistors And Nanowires. Now They Just Need To Find A Way To Put Them All Together
View