Should I Use Phototransistors or Photodiodes for Sensing?
Q. In my mistreatment of the LEDs, I ended up with one which puts out no light at 10mA, but will light up OK at 100mA!
A. You have probably 'cratered' the ball-bond.
Another poster mentioned 'dark line defects'. This is, indeed, one mechanism for declining output. However, many of the more modern LED processes are not prone to DLDs. Here are some other defect types:
1) Cratering: A crack develops under the ball bond metallization zone. If you pull on the bond wire (not possible in encapsulated LEDs), a chip pulls out, leaving a 'crater', hence the name. Symptoms: If you apply pressure to the ball (sometimes just by pressing on the top of the LED, light output momentarily increases (or perhaps, is restored, in the case of total failure). Decapsulate and test bond/chip integrity. If you check the VI characteristic with a curve tracer, you may see 'breakover' characteristic if you apply sufficiently high bias Voltage.
Causes (singly or in combination): a) incorrect ball bonding parameters such as too much pressure, bad capillary, contaminated pad, etc. b) tension on bondwire, related either to incorrect looping, vibration, or shock c) power density of input pulses exceeds device capabilities
2) Die attach migration shunts junction and/or reduces optical transmission: This is more likely to happen with LEDs that use silver-filled epoxide die attach materials, as opposed to eutectic (solder) attachment, but it can occur in either case. The silver can creep up the side of the die, eventually shorting it out. Symptoms: Can be seen using appropriate visual inspection techniques. Usually shows up looking like a parallel resistance on curve tracer VI characteristic, i.e., current starts to flow at low bias voltage, whereas the healthy junction shows little current flow until the bias is near the threshold for the material, usually between 1.1 and 1.8 Volts for LEDs.
Causes: The mfgr. Is using too much die attach material if this happens. However, the problem is aggravated by high temperatures and pulse energy levels. Cures: Get vendor to control process properly. Reduce drive levels and/or temperature.