InGaN/GaN µLEDS for display applications

Optical and electrical characteristics spread comprehension

Anis Daami1,*, François Olivier1, Denis Sarrasin1, Ludovic Dupré1 and François Templier 1
1
Univ. Grenoble Alpes, CEA, LETI, Minatec Campus, Grenoble, France, and III-V Lab, Grenoble, France
* E-mail : [email protected], Phone number : +33 4 38782985

Abstract
Electroluminescence and photoluminescence car-
tographies carried out on GaN-based µLEDs for display
applications show different optical spread behaviours. We
show using a simple SPICE model and Monte Carlo sim-
ulations that electroluminescence high dispersion can be
related to a variation of carrier injection due a non-uni-
formity of the contact resistance of processed µLEDs.

1. Introduction
Augmented and virtual reality tools have seen a growing
interest in this last decade. This attentiveness to these devices Fig. 1 Optical photograph of 50x50 pixels biased at 4.5V (left), pho-
has implied the need of high brightness and high resolution toluminescence intensity map at λ=440nm on a 400x400 µm2 zone.
arrays, which has drawn important research on GaN based
µLEDs [1] [2] [3]. Many results have reported GaN micro- We plot in figure 2 the optical power versus the electric cur-
displays with different pixel pitches and matrix resolutions rent taken at a voltage of 4.5V for all measured µLEDs. We
[4][5]. In this paper we report an electro-optical study on our perceive on this figure a proportionality of optical power ver-
previously reported emissive microdisplay [6] showing the sus injected current. Furthermore, it is worth noticing the high
importance of electric contacts on the uniformity of light spread of this cloud with average values of 10µA and 1.8 µW
power output. We show by a Monte Carlo SPICE based sim- for current and optical power, respectively.
ulation that optical power dispersion observed on our micro-
display is mainly due to a variation of the contact resistance
on µLEDs.

2. Results
Process fabrication
Our emissive arrays of 873x500 pixels are fabricated with
a self-aligned process and a damascene metallization on com-
mercial 4’’ epitaxial InGaN/GaN wafers grown on non-pat-
terned c-plane sapphire substrates by MOCVD. A complete
description of the process is found in [7].

Experimental measurements
Figure 1 (left side) shows the electroluminescence optical Fig. 2 Optical power versus current intensity for 300 µLEDs having
photograph taken on 2500 pixels, corresponding to an area of 8µm diameter measured on the same display matrix.
500x500 µm², from our µ-display working under a bias of
4.5V. We easily remark a rather high spreading on light in- SPICE Simulation
tensity between the µLEDs. On the other hand, the micro- We have constructed an equivalent circuit model of our
photoluminescence intensity map at λ=440 nm represented µLED describing its electrical and optical behaviour. This
on the right side of figure 1, and realized on a 400x400 µm² simple model consists of a SPICE LEVEL 1 ideal diode, a
area of the same epitaxy shows a good signal uniformity. series resistor (RS) related to electric contacts, and a parallel
We have also carried out unitary light-current-voltage meas- resistor accounting for reverse leakage. Taking into account
urements on 300 µLEDs, coming from the same emissive mi- the external quantum efficiency (EQE) of our µLEDs we have
cro-array and having a diameter of 8µm in order to under- emulated its optical output by a current-controlled voltage
stand this electroluminescence dispersion. source (H). The output of this latter source gives the optical
power (Popt), but expressed in volts:
ID hc
Popt = × EQE × (1)
q λ
where ID is the µLED current, h, c and λ are Planck’s constant, with a tail towards high values of optical power. Again, this
light vacuum celerity and the emission wavelength of our is in good agreement with the results shown on figure 2.
µLED, respectively. By consequence our simple SPICE model describes well the
In order to understand above experimental results we have in- electro-optical behaviour of the µLED and endorses the hy-
troduced a Gaussian distribution with a variation of 10% at pothesis that contact resistances plays an important on car-
3σ on the value of RS which allowed us running Monte Carlo rier’s electric injection in the µLED. Therefore the optical
simulations with our model. A uniform probability distribu- power spreading observed on our microdisplay, and shown in
tion between values of 5% and 7% has also been used for figure 1, is at first order related to a discrepancy in carriers
EQE corresponding to previously reported measurements on injection limited by µLEDs contacts.
our µLEDs [6].
3. Conclusions
We have shown that electroluminescence dispersion in our
microdisplay is primarily due to a variation of carriers injec-
tion caused by a non-uniform contact resistance. Improve-
ment of this parameter is therefore very important to achieve
high performance, homogeneous LED microdisplays.

Acknowledgements
This work was supported by the French National Research
Agency (ANR) through Carnot funding.

References
[1] Y. B. Tao et al., “Size effect on efficiency droop of blue light
emitting diode,” Phys. Status Solidi C, vol. 9, no. 3–4, pp. 616–
Fig. 3 Simulated current-voltage characteristics of 8µm diameter 619, Mar. 2012.
µLEDs. Maximum and minimum curves correspond to 3σ spread [2] F. Olivier, S. Tirano, L. Dupré, B. Aventurier, C. Largeron, and
current output. F. Templier, “Influence of size-reduction on the performances
of GaN-based micro-LEDs for display application,” J. Lumin,
We show on figure 3 the simulated current-voltage character- 2016, in press
istics of our 8µm sized µLED after 10,000 Monte Carlo runs, [3] I. Otto et al., “Micro-pixel light emitting diodes: Impact of the
chip process on microscopic electro- and photoluminescence,”
where the average, maximum and minimum output current Appl. Phys. Lett., vol. 106, no. 15, p. 151108, Apr. 2015.
limits are represented. It is obvious that RS plays an important [4] J. Day, J. Li, D. Y. C. Lie, C. Bradford, J. Y. Lin, and H. X.
role on the current level beyond the threshold voltage of the Jiang, “III-Nitride full-scale high-resolution microdisplays,”
µLED. This has as consequence an immediate effect on the Appl. Phys. Lett., vol. 99, no. 3, p. 031116, Jul. 2011.
light power emitted by the µLED. Indeed, we have repre- [5] X. Li et al., “Design and Characterization of Active Matrix
sented on figure 4 the output distribution of emitted light LED Microdisplays with Embedded Visible Light Communi-
cation Transmitter,” J. Light. Technol., vol. PP, no. 99, pp. 1–1,
power taken at a bias of 4.5V as a matter of comparison with 2016.
experimental results. [6] F. Templier et al., “GaN-based Emissive Microdisplays: A
Very Promising Technology for Compact, Ultra-high Bright-
ness Display Systems,” SID Symp. Dig. Tech. Pap., vol. 47, no.
1, pp. 1013–1016, May 2016.
[7] L. Dupré et al., “Processing and characterization of high reso-
lution GaN/InGaN LED arrays at 10 micron pitch for micro
display applications,” 2017, vol. 10104, pp. 1010422–
1010422–8.

Fig. 4 Simulated optical power distribution at a bias of 4.5V for an

8µm diameter µLED after 10000 Monte Carlo runs.

This distribution show a mean value corresponding the aver-

age optical power value represented on figure 2. Moreover,
we also observe that this spread shows a Gaussian like allure