EMC & Thermal Management
Identifying optimised thermal management substrate materials for HB LEDs
The challenge facing HB LED designers, claims John Cafferkey, marketing manager at Cambridge Nanotherm, is finding substrate materials that can strike the right balance of manufacturability, performance and cost
T
he market for LEDs is growing rapidly and LED manufacturers are placing more emphasis on producing smaller and brighter LEDs. Growth in demand for LEDs is not just taking place in general lighting applications, but also in areas with more demanding requirements such as horticultural lighting, automotive headlights and UV applications such as beauty treatments and printer ink setting. The industry has seen an increase in demand for ‘high-brightness’ packaged LEDs (HB LEDs) — single LED dies (generally at over 1W) placed in a surface mountable package that can be easily ‘picked and placed’ to construct LED arrays. HB LEDs provide an excellent point source of light, making them popular for applications where a beam is required like automotive headlights. One problem with packaged LEDs is the
die is protected from the environment by a plastic lens. This acts as a thermal insulator preventing the heat generated by the inefficiency of the die from escaping into the atmosphere. Heat must be conducted out of the backside of the package (i.e. conducted through the substrate which the die sits on). Removing heat as effectively as possible is critical to ensuring the die doesn’t suffer from catastrophic failure due to overheating.
As the industry continues to package bigger and brighter die into smaller and smaller packages the need for effective thermal management substrates becomes ever more critical. However, the needs of HB LEDs are not being perfectly met by many of today’s most popular solutions.
Defining the requirements of an HD LED substrate
First, HB LED substrates need to be electrically non-conductive. A dielectric substrate enables designers to mount LED die directly on the substrate, providing an optimised thermal path for heat to conduct away from the die. Secondly, price is a major consideration.
www.cieonline.co.uk
There’s a reasonably linear relationship between substrate thermal conductivity and its price. FR4 substrates, for example, have a thermal conductivity of just 0.25 W/mk but are cheap. Beryllium oxide, at the other end of the scale, has thermal conductivity of around 220 W/mk, but is expensive. The majority of HB LED applications require thermal conductivity somewhere between these two points, and so need an approach that hits the right price/performance ratio for HB LED applications.
HB LEDs also demand high wire tracing tolerances (roughly 75µm track tolerances and 50µm gap tolerance in 75µm thick copper). This rules out reductive processes used to make most PCBs, and instead limits HB LEDs to additive/thin-film build-up processes — the only approach able to achieve such tolerances reliably. Lastly, any substrate must be able to accommodate through-hole copper vias to connect circuits on each side. These need to be placed away from the main die for reliability, so the smaller the better. They should ideally be less than 75µm to avoid an increase in substrate area and cost.
The problem with traditional ceramics
Most HB LED thermal management substrates today are made from one of two ceramic materials: Al2O3 (alumina or aluminium oxide) or AlN (aluminium nitride). These meet the most basic needs of thermal management in that they conduct heat well, are dielectric, can be easily drilled, and can undergo thin-film processing. However, these leading solutions hold considerable disadvantages in terms of their price/performance ratio. Al2O3 is relatively cheap, but has a thermal conductivity of around 20–30W/mK, which isn’t enough for many HB LED applications. AlN, on the other hand, has thermal conductivity of 170–200WmK (better than many HB LEDs will require), but is extremely expensive due to the extremely high temperatures
involved in its manufacture. It also suffers from low yield due to its brittleness, and cannot, therefore, be produced as large tiles due to risk of breakage.
Can we make ceramics more cheaply?
As the thermal performance of materials are closely wedded to their chemical makeup, there’s little chance of making Al2O3 more thermally effective. It’s more reasonable to ask if AlN can be made more cost-effectively. Efforts have been made in this direction. One has involved producing AlN powder by spraying molten aluminium into a nitrogen atmosphere. However, this solution, while cheaper, maintains the disadvantage of brittleness at a low thermal performance (of around 100W/mK). A more promising
route to reducing the cost of AlN is to reduce the sintering temperature involved in its production. This has generally involved the introduction of new sintering agents and nanomaterials. Some progress has been made, but this is still very much at the drawing board stage, with no indication of cost/performance in a product.
Nanoceramics: Offering cost and performance at the right level A new approach, however, promises to deliver thermal performance at the right level for most HB LED applications and at a price between AlN and Al2O3. The solution begins with a simple sheet of aluminium alloy — a material that’s cost-effective, readily available, familiar to most manufacturers, easily machined, and extremely thermally efficient (at around 175W/mk for an alloy). However, aluminium is an excellent electrical conductor. What’s needed is a way to prevent the circuit layer from shorting out against the aluminium core. This is achieved through a patented electrochemical oxidation process that converts the surface of the aluminium into a nanoscale dielectric layer made up of Al2O3 crystals. The thickness of this dielectric layer, and hence both the thermal impedance it presents and its breakdown
voltage, can be tightly controlled to ensure a uniform and reliable dielectric covering, including into thru-holes and vias. The key is the nanoscale thickness of the ceramic layer. By reducing the layer of alumina to a few microns (and wedding it atomically to the aluminium beneath it) the thermal path is rendered extremely efficient. This approach achieves a thermal performance of around 152W/mk, close to that of aluminium nitride but at a far keener cost, making this solution appropriate to all but the most demanding HB LED products. There are also a number of additional advantages. First, it’s extremely robust. Where ceramics fracture easily, this
nanoceramic solution can be beaten into new shapes without spalling or flaking. Secondly, it can also be applied to flexible foil solutions, opening up a whole new category of material. Thirdly, its manufacture is relatively benign. The electrochemical cell contains 99 per cent water, and is safe to be discharged into a domestic sewer.
As an additional bonus, the solution can be produced in large tiles, further reducing costs and achieving economies of scale. These tiles can also be produced in the size most appropriate to thin-film production lines, something that’s not possible with AlN. In a field where a thermal solution more closely aligned to the needs of the market is keenly sought, nanoceramics are a highly promising entrant. By combining the best of ceramics with the best of aluminium metal backed PCBs, nanoceramics present a powerful new solution for the thermal management of high-brightness packaged LEDs, and one that is now being investigated by many large LED players.
www.camnano.com Components in Electronics June 2016 31
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56