The past five years have resulted in an explosion in LED applications, driven primarily by growth in the backlighting market for LCDs in TVs, laptops and handheld devices. However, the bust has been about as hard as the boom. In 2011 and 2012, sluggish sales of these devices have resulted in Chinese and Korean makers to slash chip prices by as much as 30 percent, with many factories operating at as low as 50 percent capacities, resulting in the shutdown of hundreds of small to mid-size Chinese factories. Meanwhile the larger factories, supported by government subsidies, are doubling down on the next wave of growth promised by the general lighting market.
The benefit of the recent downturn to the sign industry has been a large drop in the cost of LEDs for sign applications, especially white LED modules for illuminated signs that have served as an outlet for excess chip capacity. Today, the actual semiconductor die or chip has become a smaller and smaller cost of the overall device, resulting in the ability to increase the total number of lumens per module at fractional increases to the overall price. Current LED technology development is focused on the fundamental manufacturing process and how to “get more with less,” so let’s explore what that means for the LED industry and specifically the signage sector.
Figure 4: Red LED die mounted onto a PCB and tested prior to packaging.
First, we should understand how an LED is made. Light emitting diodes (LEDs) are made in a cleanroom and in batch process, very much like baking a high-tech cake. The process involves depositing and growing crystalline layers on a substrate in a process called epitaxial growth.
Thin layers of semiconductor materials such as gallium nitride are mixed together in a chamber at high pressure and temperatures. Once the materials are mixed uniformly, a rod is placed into the chamber and is pulled out slowly. As the solution cools, a crystal forms on the rod. The crystal is then sliced into thin wafers about the thickness of a piece of sign vinyl.
Next, the wafers are cleaned and polished smooth in order to accept more layers of semiconductor material. Multiple layers of semiconductor materials with different dopants are then grown step by step on the surface of the wafer, resulting in a semiconductor stack. From there the metal contacts are added in a repeatable pattern. The wafer is then sawed or cleaved into individual “die”. The size of the die depends upon the application and desired light output.
The die are then individually tested and “mapped” according to their optoelectronic properties. Individual dies are then packaged into a discreet lamp (often just referred to as an “LED” in our industry). Packaged LEDs can have a range of optics, outputs, and efficiencies. For white LEDs, a phosphor is coated over the die to produce broad-spectrum white light. The point is that LEDs are made one at a time, in a multi-step batch process and depend on carefully controlling the processing conditions. This fundamental manufacturing process is the main driving force in the cost to produce an LED.
Reducing Manufacturing Costs
The most obvious way to reduce LED costs is to bake a bigger cake (i.e. simply make bigger wafers). After all, in a batch process it takes about the same amount of time to make a cake that feeds four as it does to feed 24. For example, if you can grow a one-inch wafer in a single step and produce 2,000 LED die, going to a two-inch wafer allows you to produce 8,000 die of the same size with the same number of steps.
The problem is that baking these big cakes can result in the edges being burned and the center still soft and gooey—to use an analogy. Ongoing research, however, has allowed LED makers to grow semiconductors on larger and larger wafers at higher and higher yields, resulting in a lower and lower cost to produce. Figure 1 shows the growth of LED wafer size over time (i.e. baking larger cakes). These processing improvements have been and will continue to be a major driver in reducing LED die costs over the next five years.
The second approach is to get more out of the cake that you just baked (i.e. improvements to the materials that yield more light and can handle higher electrical current). It turns out that crystal growth is only as good as the starting substrate. The substrate serves as a template to build the epitaxial layer. Today, 90 percent of LEDs are built on synthetic sapphire substrates (i.e. aluminum oxide Al2O3). The problem with sapphire is that it is expensive (duh) and limits the efficiency of the LED due to what is called a lattice mismatch—the crystal structure is different that the gallium nitride (GaN) semiconductor material. This results in more cracks or “defects” and a lower yield and performance.
Today, sapphire is basically the “PC” of LED substrates. Cree semiconductor has long built their “Mac” platform on Silicon Carbide substrates, which provides a lower lattice mismatch and better heat dissipation (however, at a higher cost to produce). Other companies like BridgeLux and Osram are trying to use silicon substrates (really low cost), with limited success due to large lattice and huge thermal mismatches.
Figure 7: Deep LED cabinet illuminated with LEDs. (Photo courtesy Bitro Group.)
The most promising approach seems to be coming from start-up Soraa and giant Hitachi Cable who are both using GaN templates to grow wafers (GaN on GaN technology). The advantage is that there is zero lattice mismatch, theoretically resulting in fewer defects (i.e. higher yield) and the ability to operate at very high electrical currents. If this proves out, it could mean a tenfold improvement in light output per unit area, resulting in brighter die at the same size or the ability to cut the wafer into smaller die and achieve the same output. This is great news for LED luminaires like MR-16 lamps and LED light bulbs where the name of the game is getting lots of light in a small area and should open the doors for LEDs to better compete with traditional forms of lighting.
What This Means to Sign Guys
What does this mean for the sign industry? We have established that the die is becoming a smaller fraction of the overall cost. The more electronics, drivers, etc., that are attached to the product, the less impact these improvements have on the ultimate price of the product. So, if you’re in the electronic message center (EMC) business don’t expect lower prices. If you buy LEDs for illuminated signs, don’t expect lower prices necessarily, but the LED modules you do purchase will have better heat performance and higher outputs (exceeding 250 LM/ft.) at costs comparable to today’s entry level LED modules.
This is great news for large-format, deeper cabinets and channel letters, where there is more “space” to allow the light to mix or distribute. However, we are seeing more sign manufacturers making thinner and lower-profile signage. In this case, brighter is not necessarily better and the name of the game is to distribute the light evenly across a large surface. This will require the use of specialty optics and other innovative designs by providers of LEDs to meet the needs of smaller, low-profile signs.
The take-home message is to expect technology advances over the next few years to result in improvements in performance per dollar more so than overall price for LED sign modules. This is good for big, deep signs, but not necessarily for small, shallow ones. Innovation in this area will depend on unique application designs not driven by fundamental LED costs.