Research and Advances
Architecture and Hardware

Phase Change Recording

Burning data onto and off of the storage medium is such a reliable read/write method, the CD is now the world's most popular means of distributing and exchanging prerecorded digital content.
  1. Introduction
  2. The Search for Recordable Media
  3. Conclusion
  4. References
  5. Authors
  6. Figures

Heating a metallic alloy to its melting point, then letting it solidify to an amorphous state, is a process usually associated with a blast furnace. Yet this is what happens in a matter of nanoseconds when one records a bit on a rewritable CD in a CD writer, or burner—a process called “phase change recording.” To understand why CD engineers opted for phase recording for CD-RW rewritable media and its successors, instead of magneto-optical or magnetic recording, it is necessary to analyze the factors that have made the CD such a commercial and technological success over the past 20 years.

Hard disks use magnetic fields to read and write data. Optical disc systems use a focused beam of light. Magnetic fields can be concentrated within soft magnetic materials serving as a flux guide. However, because a free-space “magnetic lens,” which would focus the magnetic field, is fundamentally impossible, magnetic recording is intrinsically a near-field technology; the magnetic head used to write or read bits has to be in close proximity to the magnetic information layer. The minimum bit length that can be written on a hard disk therefore depends on the separation between the read/write head and the disk. In contrast, light can be focused to an extremely small spot at a distance far from both the light source and the objective lens.

In CD systems, a semiconductor laser generates a convergent light beam that enters a 1.2mm-thick transparent plastic substrate on the surface opposite the one carrying the information layer. The substrate protects the information layer from scratches and wear. Since the laser beam is out of focus at the disc’s entrance surface, the system is insensitive to dust and fingerprints on the disc. This technique, called “substrate incident readout,” is the crucial advantage of optical disc storage over hard disks, because it enables media removability and ease of handling. In addition, the large lens-disc separation (more than 1mm in a CD drive) guarantees that disc optical-pickup crashes are avoided at all times. These advantages were well understood when the CD Digital Audio system, designed for distributing prerecorded digital music, was conceived almost 25 years ago by researchers at Philips Electronics in The Netherlands and at Sony in Japan [2].

A complication resulting from substrate-incident optical recording is that the propagation of the focused laser beam through the 1.2mm-thick plastic disc introduces distortions in the optical wave front. As a result, the shape of the focused laser beam used to read the information on the disc is distorted, leading to a loss of resolution and possibly misread information. Spherical aberration of the laser beam can be corrected through a special lens system designed for the 1.2mm-thick substrate. To reduce drive complexity, the preferred solution is a single aspherical objective lens.

Additional aberrations are introduced if the disc is tilted slightly with respect to the optical axis. Unfortunately, these added aberrations are unavoidable, because the removable plastic disc may be slightly warped and is only loosely clamped in the drive. This problem is addressed by assigning a tolerance budget as determined by the CD design team to both the disc and the drive. However, the definition of these tolerances makes sense only if all disc and drive manufacturers adhere to the same specifications. Optical storage therefore needs stable worldwide standards dealing with each and every disc’s basic physical parameters, including storage capacity and data rate, and with specifications, including reflectivity and the light’s modulation level. Also specified is the disc format, including channel modulation code, error-correction method, and addressing format.

Beside substrate incident readout, low-cost mass-manufacturing of media by replication technology is a critical factor in how many systems the market ultimately buys. The pits representing the digital information on the discs are embossed. Replication is done through an injection-molding technique and a stamper carrying the negative of the desired pit pattern.

As a result of these advantages, the CD emerged in the 1980s as the preferred medium for distributing and exchanging prerecorded digital audio and in the 1990s for digital data. The CD can be viewed as one of the most successful examples of the “digital convergence” between the worlds of consumer electronics and computing.

Back to Top

The Search for Recordable Media

Various recordable versions of the CD have been proposed since the 1980s [8]. An early magneto-optical version of the CD failed due to its incompatibility with the original standardized read-only CD. Subsequently, a write-once 12cm disc with 650MB storage capacity, compatible with the CD, was designed. Known as the recordable compact disc (CD-R), it was introduced to the world of computing and consumer electronics in the 1990s; the worldwide market for CD-R discs today is estimated by market researchers at three billion discs per year. A CD-R can be used for many applications, including photo archiving, data backup, and copying read-only discs for personal use.

A rewritable CD based on phase change media was introduced commercially by Philips and other vendors in 1996 under the standardized name CD-RW. The drives used for recordable and rewritable CDs are similar to their read-only counterparts, although they require a higher-power semiconductor laser, as well as additional electronics for generating the write strategies.

CD-RW is the first really successful disc format based on phase change recording. After recording is finalized, a CD-RW disc meets all specifications of the CD-ROM (or CD-Digital Audio) standard, with the exception of the reflection level, which is lower than that of a CD-ROM disc. Full read-compatibility with playback equipment therefore required some modification to the preamplifier settings in the drives, now standardized as the “multi-read” function; the result is that all CD-type discs can be played back. Its use is widespread worldwide in new drives in PCs and in home audio-video equipment.

CD-RW could become the dominant worldwide successor to the floppy disk within 10 years. Both CD-R and CD-RW have enhanced the role of the CD as a digital-convergence product. Consumers appreciate the convenience of downloading content to and from their PCs and subsequently using it for playback in their cars and homes and on personal portable players of all types. The breakthrough of CD-R and CD-RW has also proved that compatibility with the existing read-only format for the distribution of digital content is a key factor in introducing a recordable format for optical storage.

Principle of phase change recording. Phase change recording is based on the fact that some alloys can exist in different phases [8]. Microscopically, these phases correspond to different degrees of atomic order. In the crystalline phase, the atoms are arranged in a periodic 3D lattice. In the amorphous phase, the material has about the same density but follows no long-range order throughout the medium; atoms are arranged randomly. The term “amorphous,” or without shape, refers to the fact that a macroscopic piece of such material lacks the regular facets that are characteristic of crystals.

The crystalline state is the equilibrium state, with the lowest possible energy—and is why we say diamonds are forever. The amorphous phase is a meta-stable state. At ambient temperatures, because atomic motion is effectively blocked, this state can persist for a long time. However, when the amorphous material is heated to a few hundred degrees Celsius (see Figure 1, right-hand side), the atoms are able to find positions that are energetically lower. The crystalline order is thereby restored and preserved upon cooling the material back to ambient temperature.

To prepare the material in its amorphous phase, laser power is turned up, heating the alloy even further, to beyond its melting point (around 600 degrees Celsius), as in Figure 1, left-hand side. In the liquid state, there is no trace of crystalline order. When the alloy is cooled rapidly (nanoseconds), the atoms have no time to find their preferred crystalline positions; they are locked into a random arrangement. Physicists developing this technology speak of “quenching” the material to the amorphous state.

On a phase change disc, information is recorded as amorphous marks embedded in a polycrystalline layer; these marks can be viewed through an electron microscope (see Figure 2). A blank disc carries a stack of thin films, one of which is the alloy in its polycrystalline state. In a recorder, such as the CD-rewriter, the amorphous marks are written by locally melting the alloy on the disc with a pulsed focused laser beam, with its power adjusted to a high level (the write level). The length of the amorphous marks is adjusted by choosing the appropriate number of write pulses, each creating a small amorphous dot on the rotating disc.

Direct overwrite. A notable advantage of phase change recording over magneto-optical recording is that erasing previously written marks can be done on the fly, without additional means. The drive electronics have to adjust only the laser power to a slightly lower value, so the material remains just below its melting temperature (the “erase level”). The crystalline state is then restored. In this way, a previously written track can be directly overwritten. Finally, after many rewrites by the user, the material deteriorates, limiting the number of times a phase change disc can be rewritten; for CD-RW, rewrites can be executed at least 1,000 times. The pattern of amorphous marks on the disc is read out using the same laser beam, adjusted to the even lower read level.

On track and on time. Writing, erasing, or reading the small marks imposes a rather tight tolerance on the laser beam being focused onto the phase change layer. The same holds for the accuracy of the focused beam kept on the center of the track on which the amorphous marks are positioned. Focusing and tracking error signals are derived from the reflected laser beam. For CD-ROM or CD-Audio, the tracking signal is derived from the embossed pits. In rewritable and recordable discs, tracking is also necessary while writing on a blank disc on which the tracking-error signal is generated from a pre-embossed groove on the surface of the plastic substrate.

That groove can also be used to generate timing signals by providing a small-amplitude modulation on its radius of curvature—called a “wobble” by drive engineers. If the modulation has a sufficiently high spatial frequency, the focused beam cannot follow the wobble, due to the finite bandwidth of the electromechanical tracking actuator on which the objective lens is mounted. However, the detector that generates the tracking-error signal is still able to detect the wobble. In this way, the system clock can be locked to the wobble frequency, ensuring appropriate synchronization with the disc velocity of the laser pulses used to write or erase the amorphous marks on the disc. The groove on the disc may be interrupted by short intervals of pre-embossed pits carrying addressing information. This technique is employed in disc formats with headers, including DVD-RAM and digital video recording (DVR). Alternatively, the addressing information may be coded in the wobble.

Land and groove recording. The laser beam can be focused onto the grooves or at the open space between the grooves, called “lands.” Thus it is possible to write the amorphous marks only in the groove (as in CD-RW or DVD+RW) or on the lands as well to increase storage capacity (called “land-groove recording”), a method employed in, for example, DVD-RAM. Unfortunately, the gain in density is much less than a factor of two, because, in land-groove recording, the track pitch has to be considerably wider to avoid crosstalk between adjacent tracks. The net density gain is typically only 15%–20%.

The X-game. The earliest CD-ROM drives had a user data-rate of 1.2Mb/s, or equal to that of CD-Audio. That data rate, in turn, was dictated by the audio quality required for a particular application. For computer applications, users want higher data rates, and in recent years, the CD-ROM industry has been in a speed race, with data rates now approaching 40 times the original CD rate. Product marketers speak of a “40X” drive and of the “X-game.”

High-speed CD-RW drives need more powerful lasers and more light-efficient optics. Also, with increasing writing speeds, the transitions of amorphous to crystalline areas tend to exhibit ever-stronger deviations (called “jitter”) from their intended position. Beyond a certain point, the data contains too many errors and is therefore unreadable. The transitions can be controlled much better through a more elaborate write strategy. Creating a system for guaranteeing such control calls for high-precision laser-driver electronics.

The write data rate for CD-RW is primarily determined by the media. The transformation from amorphous to crystalline states takes a minimum amount of time, or “crystallization time,” as in Figure 1. Corresponding to the time for the amorphous material to first form nuclei on which crystallites can grow, it depends on which types of phase change material is being used. For a 1X CD-RW disc, it is on the order of 0.5ms. Faster crystallization can be achieved through different materials and clever design of the stack of thin films surrounding the phase change layer. Crystallization times of less than 100 nanoseconds, allowing 8X CD-RWs, represent the state of the art, and 10X–12X CD-RW (12–15Mb/s) will be achieved in the laboratory within the next two years; market introduction could begin in 2003.

DVD+RW and video recording. In 1996, leading electronics companies introduced a new read-only optical system called the Digital Versatile Disc (DVD) with enough capacity to hold 135 minutes of compressed video, or more than 90% of all feature-length movies. A smaller focused-spot diameter was achieved by using a shorter-wavelength laser (red) and higher-numerical aperture optics (an objective lens with numerical aperture of 0.60; a CD uses 0.45). Combined with the tightening of all system margins, these technological improvements led to a storage capacity of 4.7GB per single layer of a DVD. The 1.2mm-thick plastic substrate through which data is accessed in a CD would introduce too many errors in a DVD due to the stronger optics used. A thinner substrate—only 0.6mm thick—was chosen by the DVD Forum, which was responsible for setting DVD standards. Two such thin substrates are glued back to back, producing a DVD disc as rigid as a CD.

Like the CD 15 years earlier, the DVD started out in a read-only format, but soon after completion of the DVD standard in 1996, it became clear to the main players in the optical disc industry that rewritable versions would again be needed for video-recording and data-recording applications. What followed was the emergence of a range of competing rewritable system proposals, each with its own special characteristics. Initially, systems with less storage capacity than DVD-ROM were introduced: DVD-RAM, with 2.6GB data capacity, and DVD+RW at 3.0GB. More recently, upgraded standard versions of DVD-RAM (lead by Panasonic within the DVD Forum) and DVD+RW (lead by Seagate and a consortium including Philips and Sony) store 4.7GB. Making matters more confusing, a 4.7GB-format called DVD-RW was independently proposed by Pioneer (also within the DVD Forum). The way the data is arranged on DVD-RAM makes it suitable for PC use, whereas the DVD-RW format is optimized for consumer-type video recording. Moreover, neither disc plays back in a standard DVD-video or DVD-ROM player, so both formats are incompatible with DVD.

To address the need for a truly compatible rewritable DVD format for both consumer electronic and PC use worldwide, a vendor consortium including Philips, Sony, and Thompson Multimedia are now developing a 4.7GB DVD+RW [3]. The DVD+RW disk that will soon be introduced commercially uses the same phase change recording materials as CD-RW. It also employs a wobbled groove arranged in a continuous spiral; because the wobble is used for address information, instead of headers as in DVD-RAM, it is possible to write files continuously, as in DVD-video discs. Compatibility of DVD+RW and DVD-Video and DVD-ROM was achieved by using the same physical parameters as those in the existing read-only DVD media. When writing on a DVD+RW disc, tracking follows the wobbled groove. During playback, tracking uses the written marks, as in DVD-ROM and DVD-Video.

Another issue influencing compatibility is the continuity of the data stream on the DVD+RW disc. In rewritable systems, new blocks of data are usually entered between existing blocks with only a limited amount of precision. Dummy areas are needed to link together the old and the new blocks; the blocks are separate entities in the file structure. For DVD+RW, a new technology had to be developed by Philips engineers, called “loss-less linking,” to permit bit-accurate positioning of the new data stream. The precise location of each bit stored on the disc has helped produce a disc on which the organization of the data and the file structure (long file format) is compatible with existing DVD-ROM and DVD-Video media.

DVD+RW also supplies the “record button” on the DVD player, meeting a critical user expectation. DVD is already penetrating the worldwide market for prerecorded video (still dominated by the VHS tape format). With DVD+RW, we expect that in the next two to five years the DVD video recorder will be the natural replacement for the VHS recorder in millions of homes worldwide. Philips and other consumer electronics vendors view the opportunities for DVD+RW in the video market as quite promising.

DVD+RW is also the natural successor to CD-RW in the PC market, when the storage capacity of the CD-RW format becomes insufficient for the most popular home and office applications. Again, a speed race is expected. For the first generation of DVD+RW, beginning 2001, phase change materials are being developed to support a 2.5-times DVD data rate, or 26Mb/s. Because a constant rotational frequency mode (1,600rpm) is employed in data applications, these new phase change materials ensure a 1X data rate on the inner diameter.

The third generation: DVR. Philips and Sony have designed the technology for such a system, basing it on the blue laser invented by the engineer S. Nakamura of Nichia Corp. in Japan and an objective lens with a numerical aperture of 0.85, allowing the laser beam to be focused to a very small-diameter spot. With this system, 22GB can be recorded on a single layer of a 12cm disc, or storage capacity of about 35 CDs on a single disc with the same physical dimensions as a regular CD. We expect the most important commercial application for this system to be the recording of high-definition digital video, and is why we refer to it by the initials DVR, for Digital Video Recording [11].

The broadcasting of high-definition TV (HDTV) has begun in Japan and the U.S. (Europe uses the digital video broadcast, or DUB, system for standard-definition video recording.) The data rate of the high-definition video stream is typically about 22Mb/s, so the new system’s 22GB storage capacity guarantees a recording time exceeding two hours. The system may also be used to download movies at a lower data rate and correspondingly longer recording time. Data applications for computers are expected to emerge as well. Read-only versions of the DVR system will be possible, and ultimately, it may be able to replace DVD, just as DVD can be viewed as the successor to the CD (see Figure 3).

In DVR, the information layer is addressed through a 0.1mm cover layer [4]; the substrate is just a mechanical carrier. The reason for this change from the DVD’s 0.6mm is the need for sufficient “disc tilt margin,” or the tolerance of disc features and orientation in the drive, that otherwise would be reduced to unacceptably low values due to the use of a high-numerical aperture lens.

A key challenge for materials scientists developing the DVR system is to make phase change media supporting the capacity expected by optical designers, due to the new optics [1, 5, 7, 9, 12]. Developing this media is in no way trivial, because the attainable density in a phase change disc is limited by thermal phenomena. For example, data written in a track may be erased by repeated writing in the adjacent track, a phenomenon called “cross-erase” that puts a lower limit on the track pitch (see Figure 4). Thermal effects and the microcystalline structure of the phase change material also limit the disc’s linear density. It is not self-evident that these unwanted phenomena would scale with the optical spot size.

On top of all this, the same system designers, as well as the casual PC user and home video viewer, would also like to achieve a much higher data rate (up to 80Mb/s) to allow multiple-stream recording, playback, and editing of HDTV content. Fortunately, the scaling of the attainable density and data rate is possible for a certain class of phase change materials, called “fast growth materials.”

The spot-size effect. Rewritable phase change materials can be divided in two classes: “nucleation-determined materials” and “growth-determined materials.” In both, writing an amorphous mark happens in the same way: The laser spot heats the material above the melting temperature, after which it cools very fast, freezing-in the amorphous state. Producing a crystalline mark proceeds differently, depending which material is being addressed. One type of material (nucleation-determined) crystallizes through the formation of many crystalline seeds, or nuclei, that then grow to form small crystallites. In the other type of material—fast-growth materials (FGM)—nucleation is difficult, but growth of a crystalline area, once a seed is present, is quick. In practice, crystallization of an amorphous written mark proceeds by growth from its edges inward. CD-RW and DVD+RW use an FGM as phase-change material, whereas DVD-RAM employs nucleation-determined materials.

The crystallization time determines the data rate. In FGM, the amorphous marks crystallize from the edge inward. Therefore, the size of the mark determines the minimum crystallization time; the smaller the marks, the shorter the time and the faster the media. Higher densities in FGMs imply higher achievable data rates (see Figure 5). In nucleation-determined materials, because crystallization time is not dependent on the amorphous mark size, we do not see a strong dependency between data density and data rate.

The maximum data rate points are determined by the maximum acceptable jitter level (9% timing uncertainty, called the “system clock period”). This result incorporates crosstalk from neighboring tracks. Using improved codes and optimization of the materials and write strategies, data rates of 70–80Mb/s should be possible within two years [5].

Back to Top


Although magneto-optical recording is characterized by the potential advantage of efficient storage techniques, including super resolution, and of intrinsically higher data rates, it has not penetrated the high-volume mass market (with the exception of Sony’s Mini Disc in portable music applications). This situation is unlike that of CD-RW, which is based on phase change recording and is experiencing a market-acceptance breakthrough worldwide, especially for storage-hungry PCs.

The key reason for the success of phase change recording is playback compatibility with ROM drives. Other success factors are simple disc structure (typically four layers), use of simple optical detection, signal strength (allowing a high read-data rate), and lack of a magnetic coil (required for magnetic field modulation direct overwrite in optical drives). These advantages outweigh by far the drawbacks of phase change recording, such as the limited number of direct overwrite cycles (only, say, 10,000). They are also why DVD+RW and eventually DVR will succeed in the mass market worldwide.

Researchers have also begun to seek options for a fourth-generation optical recording medium that could be based on near-field optics (with a numerical aperture of 1.4 or even 2). A single-layer capacity of 60–100GB might be possible [6]. For near-field recording, the objective lens is mounted on a slider, similar to the magnetic head in a hard disk. However, these researchers must still establish whether the key success factors of optical disc storage (such as removability) can be maintained in this ultimate density regime.

Back to Top

Back to Top

Back to Top


F1 Figure 1. Principle of phase change recording. The amorphous state is formed after rapid cooling from the molten state (left). The polycrystalline state is formed on cooling below the crystallization temperature, preceded by annealing at a temperature below the melting point. Crystallization should be completed before the laser spot has passed.

F2 Figure 2. A phase change disc viewed through an electron microscope. The grooved structure is required for tracking during recording and reading. The amorphous marks show up as gray regions without a visible microstructure. The marks are surrounded by polycrystalline material consisting of a large number of small randomly oriented crystallites whose facets show up as sharp boundaries between the crystallites. Direct overwrite is done by adjusting the laser power to an erase level; erased marks show up as regions with smaller crystallite size.

F3 Figure 3. Three generations of optical disc systems. Progress in “areal density,” or bit count per unit area, takes big steps; a CD holds 650MB, a DVD 4.7GB, and a DVR 22GB) by reducing the spot size through a shorter wavelength and stronger objective lens (with higher numerical aperture). The electron micrographs show read-only discs with replicated pit patterns.

F4 Figure 4. This view, seen through an electron microscope, shows the cross-erase phenomenon. A single tone is written in the central track (a sequence of short marks). After writing the longer marks in the adjacent tracks, the short marks nearby are partly erased.

F5 Figure 5. The AginSbTE phase-change material is a “fast growth material” resulting in higher data rates for smaller spot sizes. In nucleation-determined materials, the growth rate is independent of the mark size [7].

Back to top

    1. Borg, H., Blom, P., Jacobs, B., Tieke, B., Wilson, S., Ubbens, I., and Zhou, G.–F. AgInSbTe materials for high-speed phase change recording. In Proceedings of SPIE (Koloa, HI, 1999), 191.

    2. Brouwhuis, G., Braat, J., Huijser, A., Pasman, J., van Rosmalen, G., and Schouhamer Immink, K. Principles of Optical Disc Systems, Adam Hilger, Bristol, U.K., 1985.

    3. Buma, C., Brondijk, R., and Stan, S. DVD+RW: Two-way compatibility for video and data applications. In Proceedings of the IEEE International Conference on Consumer Electronics (Los Angeles, June 13–15, 2000).

    4. Decre, M. and Vromans, P. Cover layer technology for the high-numerical-aperture digital video recording system. Jpn. J. Appl. Phys. 39, 2B (2000), 775.

    5. Dekker, M., Pfeffer, N., Kuijper, M., Ubbens, I., Coene, W., Meinders, E., and Borg, H. Blue phase change recording at high data densities and data rates. In Proceedings of the Optical Data Storage Conference (ODS2000) (Whistler, B.C., 2000).

    6. Ichimura, I., Kishima, K., Osato, K., Yamamoto, K., Kuroda, Y., and Saito, K. Near-field phase-change optical recording of 1.36 numerical aperture. Jpn. J. Appl. Phys. 39, 2B (2000), 962.

    7. Ichimura, I., Maeda, F., Osato, K., Yamamoto, K., and Kasami, Y. Optical disk recording using a GaN blue-violet laser diode. Jpn. J. Appl. Phys. 39, 2B ( 2000), 937.

    8. Johnson, G. and Thomas, G., Eds. Information Storage Materials, Ullmann's Encyclopedia of Industrial Chemistry, 6th Ed. John Wiley & Sons, Weinheim, Germany, 1999.

    9. Kasami, Y., Kuroda, Y., Seo, K., Kawakubo, O., Takagawa, S., Ono, M., and Yamada, M. Large-capacity and high-data-rate phase change disks. Jpn. J. Appl. Phys. 39, 2B (2000), 756.

    10. Kurokawa, K., Naito, M., Yasuda, K., Kashiwagi, T., and Kawakubo, O. In Proceedings of SPIE (Koloa, HI, 1999), 197.

    11. Narahara, T., Kobayashi, S., Hattori, M., Shimpuku, Y., van den Enden, G., Kahlman, J., van Dijk, M., and van Woudenberg, R. Optical disc system for digital video recording. Jpn. J. Appl. Phys. 39, 2B (2000), 912.

    12. Tieke, B., Dekker, M., Pfeffer, N., van Woudenberg, R., Zhou, G.-F., and Ubbens, I. High data-rate phase-change media for the digital video recording system. Jpn. J. Appl. Phys. 39, 2B (2000), 762.

Join the Discussion (0)

Become a Member or Sign In to Post a Comment

The Latest from CACM

Shape the Future of Computing

ACM encourages its members to take a direct hand in shaping the future of the association. There are more ways than ever to get involved.

Get Involved

Communications of the ACM (CACM) is now a fully Open Access publication.

By opening CACM to the world, we hope to increase engagement among the broader computer science community and encourage non-members to discover the rich resources ACM has to offer.

Learn More