Low-temperature polysilicon

In the active-matrix liquid-crystal displays (AMLCDs)now typically found in laptop computers and an increasing number of monitors, each pixel is driven by an amorphous-silicon transistor that is fabricated directly onto the glass.

However, these amorphous-silicon transistors must be driven by standard silicon integrated circuits.

For small-area 'video-capable' displays, where the cost and size of these driver ICs becomes significant, Philips Research has developed a semiconductor process that could result in these driver ICs disappearing and the whole surface of the glass becoming an active layer in which new display architectures are implemented.

This article describes the effect that low-temperature polysilicon (LTPS) technology could have on creating high-performance low-power displays for the next generation of mobile phones and PDAs.

Low-temperature polycrystalline silicon(polysilicon) starts off life as amorphous silicon, applied to large-area glass sub-strate panels by production techniques that have become commonplace in the LCD industry.

The problem with this amorphous silicon is its low charge-carrier mobility and its inherently poor stability, which, although good enough to produce transistors to switch individual display pixels on and off, is not sufficient to handle the logic and mixed-signal functions necessary to drive the display.

Hence the need for silicon chips in current AMLCD panels. What Philips Research has developed is a way of heating up this thin layer of amorphous silicon to a temperature above its melting point, such that when it cools, it crystallises into polysilicon.

The whole process is achieved using an excimer laser, which by locally heating the amorphous silicon for only a few nanoseconds, keeps maximum process temperatures below the 450°C limit for glass substrate handling.

The process is similar to the recrystallisation phenomena at the heart of modern CD-RW optical storage technologies.

Compared with amorphous silicon, charge-carrier mobility in polysilicon is much faster. In fact, Philips Research has already demonstrated excimer-laser-controlled micron-scale zone-refining techniques that can increase charge carrier mobility in polysilicon to around 90% of that for the silicon used in IC manufacture.

Polysilicon is not only faster. Unlike amorphous silicon, it should also allow the fabrication of CMOS transistors, enabling all the IC industry's existing expertise in analog, digital and mixed-signal circuit design to be applied to polysilicon structures.

Higher integration

The row and column driver ICs necessary to drive a 1.6-inch-diagonal,amorphous-silicon AMLCD currently occupy around a third the total area of the display assembly.

Situated along two sides of the display, these ICs have to be mounted on a special glass ledge that pushes the display window off-centre, limiting design flexibility when it comes to mounting the display centrally in an enclosure or when positioning user interface controls. In addition, they can typically require over 500 connections to be made between the ICs and the surface of the glass, which can affect long-term reliability.

And whenever the display is alive, these ICs must also be alive, continuously refreshing the image with the contents of the display's frame store memory.

Even in standby mode, when the image is not changing, this consumes battery power. The steppers that handle large-area glass substrates only allow 3 micron design rules for integrated polysilicon circuitry (a far cry from the 0.13 micron design rules for state-of-the-art silicon wafer CMOS process technologies).

But even with this relatively large feature size there is more than enough area in the 1.5 mm wide seal line around a small AMLCD to accommodate most of the required drive circuitry if it is integrated into the polysilicon layer.

Connections to this circuitry are via the same metallisation patterns used for the display's row and column electrodes, so there are no IC bonding problems to worry about.

And the bond pads for the remaining 70 or so off-glass connections that are required to feed the display with data, can be made much larger and more robust.

Clearly, polysilicon has the potential to take what is required today to produce an AMLCD, compress it into a smaller footprint to allow greater freedom in aesthetic and ergonomic design and make the assembly more robust and reliable.

It can also cut standby power consumption. However, it remains a more expensive process than amorphous silicon, typically requiring three additional mask steps.

This means that where the physical size and cost of conventional driver ICs are less significant in relation to overall panel size and cost, amorphous silicon is still likely to predominate.

At the moment, the estimated price cross-over point is a display diagonal of between eight and 10", giving polysilicon a competitive edge in AMLCDs up to PDA size, and obviously including displays for very-high-volume markets such as mobile phones.

Intelligent displays

In addition to improving the price/performance ratios of current small-footprint AMLCDs, polysilicon technology may also enable a whole new generation of 'intelligent' and low-power displays.

With further development work under way to reduce the threshold voltage of polysilicon transistors, these new displays will be able to interface directly to the low-voltage, low-power CMOS, system-on-chip solutions that now dominate the consumer electronics market. At the same time they can reduce power dissipation in the display's integrated polysilicon electronics.

More complex polysilicon circuitry could also start to move from the periphery into the pixel structure of the display itself.

For example, with smaller design rules, a complete frame store could be constructed directly behind the pixel array, replacing the external frame store memory that is required in existing display systems.

Even a partial frame store built into the display would mean that when you are displaying a simple static image, such as in the standby mode of a mobile phone, all the display interfacing and multiplexing circuitry can be turned off, providing dramatic power consumption improvement (about 10 times less).

A 1 bit memory cell behind each pixel in a colour display gives you the ability to display eight colours.

Such displays may not be that far away. Philips Research has already demonstrated a 384 column (128 x 3 RGB pixels) x 160 row reflective colour AMLCD that incorporates all the row and column driver circuits and 4 bit DACs, plus a 1 bit/pixel frame-store memory behind the pixel array.

Also featuring integrated DC/DC converters to generate drive voltages for the display, this display operates from a 3 V supply and interfaces to 3 V logic systems. Static images can be displayed in eight colours with a total power consumption of much less than 1 mW.

PolyLED on polysilicon

While polysilicon will probably first be seen driving AMLCDs, it could also hold the key to driving a new and emerging display technology that is also being worked on at Philips Research.

PolyLED (polymer light-emitting diode) is an emissive-display technology that is capable of producing high-brightness, high-contrast colour displays that are viewable under virtually all lighting conditions and have a near-180° viewing angle.

Being emissive devices similar to LEDs, PolyLED pixels must be current-driven rather than voltage-driven.

Existing PolyLED displays, such as those already demonstrated by Philips Research, employ passive-matrix addressing that results in each pixel only being illuminated for a very short time.

To achieve the required overall display brightness, the intensity of the light pulse from each pixel must therefore increase as the display size (the number of pixels) increases. For display diagonals above a few inches, the PolyLED current, and the associated resistive power losses in the display electrodes, become excessive.

Philips Research is therefore studying active-matrix PolyLED displays, in which each pixel remains illuminated for the entire frame period.

This reduces PolyLED cur-rents to levels at which polysilicon transistors (but not amorphous-silicon transistors) can switch them.

Polysilicon also holds the promise of being able to build high-performance current sources (so called 'current mirrors') into each pixel, allowing very precise control of the PolyLED current.

Polysilicon on plastic

Polysilicon has a bright future on glass, but it may also have a bright future on plastic. Experimental results already indicate that process temperatures can be reduced to 275°C or less, making it possible to produce polysilicon layers on plastic films. This could open up a complete new world of flexible displays.

Reprinted from 'Philips Research Password' magazine.