plw7070* series and plw3535* series i2led high … led die and provides advice on handling of the...
TRANSCRIPT
Page 2 of 12 Document number 294351 V3
This application note describes Plessey’s MaGICTM (Manufactured on GaN-on-Si I/C)
technology LED die and provides advice on handling of the ceramic packaged devices.
Plessey’s MaGIC LED technology uses breakthrough GaN-on-Si for low cost, high
performance LEDs. The LED die are optimised for light output in the manufacturing process;
the light extraction is further enhanced with a silicone lens.
Device Structure
Plessey’s GaN technology supports a multi-junction high voltage process. This process begins
with InGaN LED layers grown on a silicon substrate as illustrated in fig. 1(a). Buffer layers are
grown to form a suitable surface on which to grow the GaN, followed by an n-GaN layer,
multiple quantum well and p-GaN final layer. To enable both electrodes to be connected on
the top surface, an additional metal layer and insulation is added. The LED wafer is then
bonded to a handle wafer and flipped (fig. 1(b)). The original substrate is removed, and the
processing completed with a top surface patterning to enhance light extraction, and top
contacts formed for both anode and cathode.
(a)
(b) (c)
Fig. 1(a) GaN growth on silicon; (b) wafer bonding and flipping; (c) wafer finishing
The surface layers are patterned to increase the light extraction. The chip is also metallised
on the back with a nickel–silver layer (fig. 2) to allow for good die adhesion and thermal
conduction to the package.
Page 3 of 12 Document number 294351 V3
Fig.2: PLB
VH series die cross-section
Ceramic Package
Plessey’s die packaged on a ceramic header are typically finished as shown in fig. 3. The die
are mounted onto a ceramic header using an epoxy die attach. This may be replaced in
future by a silver sinter process.
Fig. 3: Section through ceramic packaged LED
The package uses aluminium nitride ceramic for good thermal performance. The ceramic is
metallised on the underside with a central thermal pad and two terminal strips. A silicone lens
is used to enhance light extraction.
Page 4 of 12 Document number 294351 V3
3. Die Handling
The LED chips are shipped in tapes on reel and are intended to be used with standard pick-
and-place machines. In addition to enhancing the light output the silicone lens protects the
wire bonds. As the lens is soft it should not be subject to lateral nor vertical forces or damage
may ensue, either to the wire bonds, chip or even for the lens to detach from the chip. To pick
up the LEDs Plessey recommends a rubber collet that will not damage the lens.
The collet should be able to pick the LED without damaging the lens. A suggested collet
dimension is shown in fig. 4:
Fig. 4: Suggested Collet Nozzle Dimensions for 7070 package
Manual handling
When picking the die up manually, it is important not to disturb the lens in any way. The LED
should be held by the ceramic only. When using tweezers, ensure that the LED is held by the
ceramic. Contact with the lens should be avoided since this may cause damage to the bond
wires; displace the phosphor coating or spoil the optical transparency of the lens itself.
Page 5 of 12 Document number 294351 V3
4. Soldering
Plessey’s ceramic packaged LEDs are provided with a thermal pad and two terminals that
provide the anode and cathode connections. Soldering should be performed in a reflow tool,
but can also be carried out manually using a hotplate. Unleaded solders should be used to
meet environmental regulations. Plessey uses a low temperature solder (Sn:B) to minimise
thermal stresses on the LED.
Solder paste should be applied uniformly to the PCB pads. A uniform coating can best be
applied using a stencil print (screen print). Standard SMD techniques can be used.
The solder thickness should target 50-100 microns consistent with minimum voiding. Voids in
the solder will increase the thermal resistance and degrade reliability. Voids can lead to
hotspots, which, if devleoped, allows a part of the LED to become hotter than the rest. This
could cause a local degradation of the LED leading to premature failure.
The solder reflow thermal cycle should follow the data sheet recommendations for a specific
LED.
5. Thermal considerations
High power LEDs, although now very efficient, still generate significant heat. In a white LED
the phosphor, which converts the blue light to other colours to make white light, also dissipates
energy due to quantum conversion losses. For a local phosphor in close proximity to the LED
this increases the die temperature. Thus the combination of LED and phosphor will develop a
lower overall efficacy compared to the LED alone as shown in fig. 5. An LED with 60% wall
plug efficiency (WPE), operating at 12V and 0.7A for example, could generate 5W of blue light.
Of this some 4W may become white light, with the phosphor dissipating 1W of heat. The
overall LED thus appears to offer an efficiency of only 47.5% even though the LED is providing
100 lm/W or more.
Page 6 of 12 Document number 294351 V3
Fig. 5: Heat sources in white LED
The excess heat should be removed with a suitable heatsink. The maximum operating junction
temperature recommended is 85°C, and the package and heatsink together should be
specified to keep the temperature of the LED at or below the recommended limit. The
technology is being assessed for higher temperature operation at 135°C.
Page 7 of 12 Document number 294351 V3
Junction temperature measurement
There is widespread interest in a commercial
system which interrogates a thermal transient and
extracts the so-called “structure function” of an
LED which provides a spectral signature of the
various time constants in an LED.
An example of the thermal resistances and
capacitances is illustrated in fig. 6. This shows that
even for the most basic case, an LED contains
several layers each with its own thermal resistance
and capacitance, and this extends up to the final
heatsink.
We have found that the transient thermal method
is problematic for LEDs because the electrical time
constant is slower than the fastest thermal time
constants, making extraction of the fine time
constants difficult.
Fig. 6 (right): Thermal resistances and
capacitances in a typical LED installation
Page 8 of 12 Document number 294351 V3
The method we propose and use is as follows. First, the LED is mounted on a suitable heatsink
with a separate power resistor also attached as shown in fig. 7. The resistor is used to heat
the heatsink. A thermocouple is placed next to the LED. An IR camera can also be used as a
check the temperature provided that the surfaces of the LED or PCB are IR emissive. Some
surfaces like copper and gold have a very low IR emissivity but can be made more so by
covering the surface with a suitable tape, or painting with white (or black) emissive paint that
can be separately characterised.
Fig. 7 (a) Pulse test setup
Fig. 7(b) Heatsink assembly
Fig. 7: Pulse test set up (a) and heatsink (b).
Fig. 7(a) shows the basic set-up. A pulse generator is used to drive the LED in current limit
mode to full current e.g. 700mA, but isothermally. This requires the pulse width to be short,
and we recommend 1 s. The duty cycle should be long, and we recommend 1ms, giving a
Page 9 of 12 Document number 294351 V3
1:1000 ratio. The forward voltage of the LED is measured while the device is on, using any
suitable high-speed data acquisition device including a digital oscilloscope.
The test is started by measuring the initial temperature of the LED, and with the low PWM duty
cycle applied, the heating resistor is powered and the forward voltage noted as a function of
the recorded temperature. With the LED thus very lightly powered, the position of the
thermocouple is not so critical as the LED and PCB should be at equilibrium, but to measure
the solder point temperature later, the thermocouple should be mounted right next to the LED
package as close as possible to, or touching, the LED solder point. The forward voltage is
then measured at several set points.
In the next step, the power to the heating resistor is switched off. When cool the LED can be
powered up using normal continuous operating conditions and allowed to stabilise. The
forward voltage can then be used to determine the actual temperature from the calibration
curve produced.
If using an IR camera to monitor the temperature of the LED and heatsink, it should be noted
that a black heatsink is a good approximation in most cases to a black body and may not
require any surface coating.
Thermal resistance measurements
The thermal resistance is defined as the rise in temperature between the junction and solder
point for a given power:
Rthj−sp =∆T
Pin − Pout
where T is the temperature difference between the junction and solder point, and Pin-Pout
the dissipated power, being the difference between the input or applied power and the output,
or optical power. The junction temperature is determined from the Vf:temperature results as
measured using the method described above, interpolating if necessary, and the measured
solder point temperature indicated by the thermocouple.
The LED shares some similarities with its tungsten lamp fore-runner in that the dissipated
power is not the same as the power supplied, or input. Some of the energy is converted to
Page 10 of 12 Document number 294351 V3
light which means that the dissipated heat power is less than the supplied input power, which
is the product of the forward voltage and current. In the case of the tungsten lamp, only about
15% of the energy is converted but for LEDs, 50 to 60% of the input can be converted to light.
This has caused some confusion as to the power to be used in the thermal resistance
calculation. One simple approach is to assume that the dissipated power is Vf x If, which
ignores the radiated light (Pout=0). This leads to an apparent, or “electrical” thermal resistance
which is about half the actual thermal resistance. This will, though, predict the correct junction
temperature, for nominal conditions, when multiplied by the applied input power. However,
there are some circumstances which require the real thermal resistance to be considered.
The light output from an LED is dependent on several factors. LEDs exhibit an efficiency drop
with increasing current, once beyond a low current threshold; an efficiency drop with
increasing temperature, and also whether there is a phosphor present in contact or not, as the
phosphor can generate heat due to conversion losses. If an “effective” thermal resistance is
calculated on the basis of a specific set of conditions, this may change if the conditions change.
The real, or “thermal” thermal resistance is determined by measuring the optical output power
from an LED under a given condition. This output power is then subtracted from the input
power to give the dissipated power, thus giving the real thermal resistance. However, to utilise
the “thermal” thermal resistance requires knowing the efficiency of an LED. Not only will this
vary with conditions, but also with lifetime. Therefore, in addition to providing a real thermal
resistance figure, a model of the light output which varies with current, temperature and time
is needed. Even this may be inadequate if, for example, a light fitting scatters or reflects some
of the light back onto the LED which will be reabsorbed, converted into heat, reducing the
effective efficiency.
As many factors need to be considered in order to use the real thermal resistance, the
datasheet figures are provided as guidelines to the junction temperature, and are correct when
multiplied by the electrical input power for nominal operating conditions. Any actual luminaire
design should be designed considering the real thermal resistance, but the junction
temperature will depend on the effective efficiency which may be affected not only by the LED
operating conditions of current, temperature and time, but also the efficiency of the luminaire.
A suitable guideline for the output power covering most of the stated conditions at worst case
(end-of-life, high temperatures and currents for example) is to assume 25% output. This is just
for thermal calculations as opposed to actual light output.
Page 11 of 12 Document number 294351 V3
6. ESD protection
Plessey’s ceramic packaged LEDs are supplied with an ESD protection diode which is
connected in reverse parallel with the LED internally. Fig. 8 illustrates the electrical circuit.
Fig. 8: ESD protection
For this reason, it is not possible to operate the LEDs in reverse bias. Circuits employing these
LEDs should be designed not to drive the LEDs in the reverse direction.
Page 12 of 12 Document number 294351 V3
Legal Notice
Product information provided by Plessey Semiconductors Limited (“Plessey”) in this document is
believed to be correct and accurate. Plessey reserves the right to change/correct the specifications and
other data or information relating to products without notice but Plessey accepts no liability for errors
that may appear in this document, howsoever occurring, or liability arising from the use or application
of any information or data provided herein. Neither the supply of such information, nor the purchase or
use of products conveys any licence or permission under patent, copyright, trademark or other
intellectual property right of Plessey or third parties.
Products sold by Plessey are subject to its standard Terms and Conditions of Sale that are available
on request. No warranty is given that products do not infringe the intellectual property rights of third
parties, and furthermore, the use of products in certain ways or in combination with Plessey, or non-
Plessey furnished equipments/components may infringe intellectual property rights of Plessey.
The purpose of this document is to provide information only and it may not be used, applied or
reproduced (in whole or in part) for any purpose nor be taken as a representation relating to the products
in question. No warranty or guarantee express or implied is made concerning the capability,
performance or suitability of any product, and information concerning possible applications or methods
of use is provided for guidance only and not as a recommendation. The user is solely responsible for
determining the performance and suitability of the product in any application and checking that any
specification or data it seeks to rely on has not been superseded.
Products are intended for normal commercial applications. For applications requiring unusual
environmental requirements, extended temperature range, or high reliability capability (e.g. military, or
medical applications), special processing/testing/conditions of sale may be available on application to
Plessey.
Contact
Customer Enquiries/Sales
+44 1752 693000 | [email protected]
www.plesseysemi.com
Plessey Semiconductors Ltd | Plymouth
Tamerton Road, Roborough
Plymouth, Devon
PL6 7BQ United Kingdom
P: +44 1752 693000
F: +44 1752 693700