Joachim Stöhr¹, Mahesh G. Samant² and Jan Lüning^1
1Stanford Synchrotron Radiation Laboratory, P.O. Box 20450, Stanford, CA 94309
²IBM Almaden Research Center, 650 Harry Road, San Jose, CA 95120

The origin of liquid crystal (LC) alignment on polymer surfaces, a technologically important effect which underlies the manufacture of today's flat panel displays, has been debated ever since its discovery nearly one hundred years ago. In the past, the complexity of the LC-polymer system has prevented a scientific understanding that could form the basis for a next generation technology. Polari zation-dependent, surface sensitive near edge x-ray absorption fine structure (NEXAFS) measurements, carried out at the Stanford Synchrotron Radiation Laboratory (SSRL), have led to a scientific understanding and a discovery that has enabled the development of new materials and processes for LC alignment. By implementation and development of the new process, IBM has built large flat panels with the highest resolution available today.

Introduction

Fig. 1: Illustration of some key components of liquid crystal flat panel displays discussed in the text.

The light transmission from the back to the front of the display is controlled pixel by pixel by the local arrangeme nt of the LC molecules. The rubbing process aligns the LC molecules in the rubbing direction, nearly parallel to the surface, so that they form a 90^o helix from one surface to the other. By application of a small voltage to a pixel the LC molecules align along the electric field lines within the area of the pixel, perpendicular to the rubbed surfaces. The LC molecules are located between two crossed polarizers, aligned along the rubbing directions. When the LC rods are ali gned along the light direction by the electric field they do not affect the polarized light emerging from the first polarizer and the light is stopped by the second orthogonal polarizer - the display is dark at the pixel position. Without an applied voltage the helical twist of the LC molecules rotates the polarization of the locally transmitted light and it can now pass the second polarizer - the display is bright at the pixel position. In practice, each pixel is composed of three s ubpixels associated with red, green and blue color filters so that one can mix the three colors with different intensities and obtain a high resolution pixel mosaic - the final picture on the display screen.

Over the past ten years the development of alternative methods of liquid crystal alignment that overcome inherent problems associated with the rubbing process have been a great challenge of the $20 billion/year flat panel industry. The rubbing process introduces debris and may leave streaks that can affect the quality of the image. Rubbing can also produce electrostatic discharging and adversely influence the electronic circuitry just below the surface of the rubbed polyimide thin film. Also, the rubbing process does not lend itself to the manufacture of multidomain alignment layers that are necessary to increase the limited viewing angle in present displays. Larger viewing angles are deemed important for the use of LC displays in desktop and possibly television applications. Finally, the deposition process of the polymer film requires the use of organic solvents and materials which are environmentally unfriendly and the polyimide has to be cured by an additional baking step. The process is time consuming, costly and incompatible with high-tech manufacturing techniques. Despite intensive scientific and technological efforts, however, the development of new non-contact al ignment processes and high-tech alignment materials has been impeded by the fact that the origin of LC alignment has remained a mystery since its discovery in 1907.

In the past, at least two methods have been published where the rubbing process of the polymer was empirically replaced by another process. The first effort involved irradiation with polarized light [3] and the second effort used irradiation by a directional ion beam [4]. No scientific understanding of the alignment mechanism was given, however. Although suitable for LC alignment, the methods were never seriously considered for technological applications because they still relied on a polymer layer. Replacement of this low-tech process was deemed important for future developments such as larger panels with improved viewing angles, and for cost savings in manufacturing.

Toward a LC Alignment Model

Early LC alignment models were based on the existence of microgrov es at the surface of the alignment film, caused by the rubbing process [2]. Such models assume that it was the topological grooves that guided the azimuthal alignment of the LC molecules on the surface. The models, however, could not explain some key features of the observed alignment, like the unidirectional uptilt of the LC molecules from the rubbing direction, called the pretilt angle, or the LC alignment perpendicular to the rubbing direction found for polystyrene films.

Later models, based on bulk-sensitive optical measurements, emphasized the molecular orientation in polymer films [2]. Molecular alignment at rubbed surfaces was also confirmed by surface sensitive grazing incidence x-ray scattering (GIXS) [5] and NEXAFS studies [6,8]. By the mid '90s there was little doubt about the existence of molecular alignment at the rubbed polymer surface. In particular the preferential near-surface alignment of polyimide chain segm ents along the rubbing direction was linked to the preferred in-plane alignment direction of the LCs. However, the models still invoked epitaxy-like effects at the polymer surface where the LC is oriented by a preferred crystalline structure, microcrystalline nucleation sites or crystalline regions with a preferred chain orientation. The models relied on the presence of crystal-like regions at the polymer surface which align the LC molecules locally, with the rest of the LC following suit because of the long range orientational order within the LC itself. The problem with such models was the well-known fact that many of the polymers that gave rise to alignment were structurally disordered. In fact, the polyimides typically used in technology were known to lack crystalline order.

One of the most mysterious effects in LC alignment was the existence of LC pretilt angles. The pretilt angle is of key technological importance in that it defines the twist direct ion of the helix in Fig. 1. In the absence of a pretilt angle the chirality of the LC twist is not uniquely defined and domains with different chirality give rise to multi-domains that lead to streaks in the image. For lack of better knowledge the important pretilt angle has been speculated to arise from tilted main or side chain segments at the rubbed polymer surface which "guide" the LC rods. Much empirical effort has gone into control of the size of the pretilt angle by modification of functional groups within the polymer.

The origin of the LC pretilt angle and the in-plane alignment direction were naturally explained by a new model proposed in 1998 [9]. NEXAFS studies on a deliberately chosen disordered polyimide proved, that LC alignment does not require the presence of crystalline or quasi-crystalline order but only orientational order. While crystalline order is based on positional order of atomic or mole cular units, orientational order is based only on the existence of a preferred atomic or molecular alignment axis. The new alignment model, which was based on NEXAFS results for a variety of rubbed polymer surfaces [10], suggests a direct correlation between the preferred average orientation of bonds at the polymer surface and the average orientation of the LC molecules in the LC. The model predicts that LC alignment can result, in general, from orie ntational order at the surface of the alignment film alone, without the requirement of positional order or epitaxy.

In a simple picture one can imagine that the rubbed polymer surface consists of molecular units that, on average, possess a preferred direction similar to the molecules in the LC itself. For example, the phenyl rings in the polyimide chains are preferentially oriented by the rubbing process [10]. The rod like LC molecules sense the preferred average orientation and align with their rod axes parallel to the rings. Since the alignment process is governed by an energetic minimum in the LC-surface interaction, only a statistically significant orientational order needs to exist at the alignment surface. Hence in many cases the orientational order may be quite small relative to that in the LC itself.

For polymers one can easily imagine that the rubbing process, similar to combing one's hair, creates alignment into the random arrangement of the s paghetti-like polymer chains [10]. The billion dollar question that still remained was whether LC alignment could be accomplished by more versatile methods and with cheaper materials. The pursuit of this question by the flat panel industry has been like the search for the holy grail.

NEXAFS Studies of Ion Beam Treated Polymer Surfaces: A Key Discovery

The key to the puzzle came from NEXAFS studies on polyimide films that had been irradiated with a low energy (50-200eV) directional ion beam, a method that had previously been shown by Chaudhari et al. [4] to align LCs. The results of that study are shown in Fig. 2.

Fig. 2: NEXAFS spectra recorded by means of Auger and total electron yield detection for an untreated polyimide (PI) thin film, an ion beam (energy 75eV) irradiated PI film (PI + IB) and an amorphous carbon film. The IB irradiation produces an approximately 5 nm thick a-carbon layer on top of the PI, as illustrated in the inset.

The NEXAFS studies were carried out using surface sensitive Auger yield detection which only samples the first nanometer from the free surface and total electron yield detection which has a considerably longer sampling depth (of the order of 5nm). The comparison of these spectra shows clearly that ion beam irradiation converts the polyimide su rface into a thin amorphous carbon layer. The underlying polyimide is still visible in the total yield spectrum whereas the Auger yield spectrum is dominated by the a-carbon surface layer (middle row in Fig. 2).

Further NEXAFS studies showed that the a-carbon surface layer exhibited orientational order and that the preferred bond directions at the irradiated surface could explain the LC alignment observed on such surfaces [11]. A key point was the explanation of the empirically k nown opposite pretilt direction on rubbed and ion beam irradiated polymers, illustrated in Fig. 3. The in-plane polarization dependence (x-y asymmetry) of the spectra is the same while the out-of-plane one (±45^o asymmetry) is opposite. This explains the observed opposite pretilt directions for rubbed and ion beam irradiated polyimide, shown in the small illustrations on the right [11].

Fig. 3: Angle dependent NEXAFS spectra of rubbed polyimide (top row) and ion beam irradiated polyimide (bottom row). The ion beam was incident in the x-z plane, as shown in the bottom right picture. The color of the spectra is chosen to reflect the electric field vector orientation in the bottom icons (double arrow) [11]. The spectral features are also labeled. Note that the in-plane x-y asymmetry of the spectra is the same while the out-of-plane (±45^o asymmetry) is opposite. This is in agreement with the opposite LC pretilt directions for rubbed and ion beam irradiated polyimide, illustrated in the pictures on the right.

The fact that the ion beam irradiated polyimide sample whose surface consists of an amorphous carbon film aligns LCs suggests that the underlying polymer structures is not required for LC alignment. Hence it should be possible to create a surface with orientational order by irradiating an a-carbon film dir ectly, instead of a polymer film. This idea constituted the seed for a device enabling breakthrough.

Ion Beam Treated Amorphous Carbon: A Device enabling Breakthrough

The suggested experiments on ion beam irradiated amorphous carbon films indeed verified reliable LC alignment on such films and NEXAFS data showed the direct correlation of the orientational order at the surface and the LC pretilt angle . Repres entative NEXAFS spectra for ion beam irradiated a-carbon films are shown in Fig. 4. They reveal orientational order within the plane of the film (Fig. 4 (a)), setting the LC in-plane alignment direction, and out-of-plane orientational order (Fig. 4 (b)) which determines the LC pretilt. Comparison of the data for rubbed polyimide, ion beam irradiated polyimide and ion beam irradiated amorphous carbon convincingly correlated with the in-plane alignment and the opposite direction of the pretilt for rubbed and ion beam treated surfaces.

Fig. 4: Angle dependent NEXAFS spectra (same geometry as for Fig. 3) for a diamond-like carbon film, showing preferential bond alignment (top) [11]. The ion beam was incident in the x-z plane. In particular, the π orbitals of the C=C bonds are aligned more along the y than x direction in the film plane (a), and are more aligned along t he +45^o than -45^o direction in the x-z plane (b). The observed asymmetries can be explained by preferential destruction of rings with their plane perpendicular over those with their plane parallel to the ion beam direction (c). The preferred in- and out-of-plane alignment directions of the LC molecules are indicated.

The understanding of the alignment mechanism and the easy manufacturability of thin amorphous carbon layers convinced IBM Japan to develop this process for manufacturing. The NEXAFS results and the new process and material were kept confidential for about two years to allow the development of a reliable manufacturing process. During this development phase NEXAFS studies played an important role in the optimization of the process conditions and the materials.

The Manufacture of Novel Flat Panel Displays

The new process has now cleared all reliability checkpoints and panels up to a size o f 22" have been manufactured [12]. The optimized alignment material is so called diamond like carbon, an amorphous hydrogenated carbon film that can conveniently be deposited in a dry deposition process by sputtering or chemical vapor deposition. Hydrogen is added in the films to make them optically transparent. The low energy ion beam is produced by a commercially available gun. The new process has resulted in the highest resolution large displays available today and two-domain LC cells have already been produced in the laboratory that promise future LC displays with improved viewing angles [12].

References:

1. S. W. Depp and W. E. Howard, Sci. Am. 268, 90 (March 1993).

2. K.-W. Lee et al., in Polymer Surfaces and Interfaces: Characterization, Modification and Application, K. L. Mittal, K.-W. Lee, Eds. ( VSP, The Netherlands, 1996), pp. 295-315.

3. W. M. Gibbons et al., Nature 351, 49 (199 1).

4. P. Chaudhari et al., Jpn. J. Appl. Phys. Part 2 Lett. 37, L55 (1998).

5. M. F. Toney et al., Nature 374, 709 (1995).

6. Y. Ouchi et al., Physica D 208/209, 407 (1995); I. Mori et al., J. Electron Spectrosc. Rel. Phenom. 78, 371 (1996).

7. M. Samant et al., Macromolecules 29, 8334 (1996).

8. K. Weiss et al., Macromolecules 31, 1930 (1998).

9. J. Stöhr et al., Macromolecules 31, 1942 (1998)

10. J. Stöhr and M. G. Samant, J. Electron Spectrosc. Rel. Phenom. 98-99, 189 (1999)

11. J. Stöhr, M. G. Samant, J. Lüning et al., Science 292, 2299 (2001)

12. P. Chaudhari et al., Nature 411, 56 (2001).