The Use of Laser Stereolithography to Produce Three-Dimensional Tactile Molecular Models for Blind and Visually Impaired Scientists and Students

ABSTRACT

Laser stereolithography, a rapid prototyping process that produces three-dimensional plastic models from the images created in certain computer aided design (CAD) programs, has been used to fabricate tactile molecular models for blind and visually impaired individuals. The process uses a computer-controlled laser to cure and solidify a light-sensitive, liquid polymer in the shape of the image. The models can be customized and used for educational and research purposes. Several models built using four different scales are described. Surface textures are varied to allow atom types to be distinguished.

INTRODUCTION

Several reports describing the status of disabled individuals with respect to careers in science and engineering have been published (Reddon, Davis, & Welsh-Brown, 1978; U.S. Government Report, 1989). They indicate that disabled individuals are seriously under-represented in both areas.It has been suggested that attitudinal and physical barriers exist that can make pursuit of careers in technically oriented disciplines less attractive for people with physical or learning disabilities. Some of the negative attitudes may arise from a lack of information about the capabilities of people with disabilities, or about the technology available to help disabled individuals meet physical and perceptual requirements. For those who are blind or who have visual impairments (and some people with specific learning disabilities), one major physical barrier is access to the vast amount of information that is part of scientific disciplines. With the growing use of electronically stored data and adaptive technologies, more information is being made accessible to this segment of the population. Textual material can be acquired by use of recorded media, computer files can be read using speech synthesis, display enhancement, or Braille output.

Laboratory instrumentation and classroom demonstrations can similarly be adapted by the use of alternative outputs to visual displays (Crosby, 1981; Flair & Setzer, 1990; Hinchliffe & Skawinski, 1983; Lunney & Morrison, 1981; Morrison, Lunney, Terry, Hassell, & Boswood, 1984; Reese, 1985; Shimizu, 1986; Smith, 1981; Tombaugh, 1981). In some cases simple two-dimensional tactile graphics can be produced and some attempts to represent depth have also been made (Dietrich & Seufert, 1986). One area that is not easily addressed is the presentation of three-dimensional graphics in a form that can be perceived directly by blind and visually impaired individuals.

Sophisticated, three-dimensional computer graphics techniques are now commonly used in all scientific disciplines to produce representations of experimental and calculated data. Images of physical phenomena and mathematical graphs can be displayed in a number of ways to emphasize specific features and enhance scientific communication. Spatial relationships are critical in many areas of chemistry, physics, and biology. Mathematical relationships in three dimensions are utilized in many other disciplines. Blind and visually impaired scientists, students, and teachers of science and mathematics however, encounter serious barriers to acquiring important information that is presented in the form of three-dimensional images, and the use of tactile models can be essential for a full understanding of these concepts.

In recent years several systems for rapid prototyping have been developed which allow engineers and designers to create an image within a computer-aided design (CAD) program and automatically translate that image into an accurate plastic model (Hull, 1988). These systems fabricate the models layer-by-layer from bottom to top, and are capable of accurately producing intricate structures. We describe here, the application of one such rapid prototyping system, laser stereolithography, to the preparation of tactile models of molecular structures. Since these models can be designed to represent any of a variety of images, they are also useful to sighted researchers and students. Models can be made by this method which cannot be made using standard molecular model techniques and can thus be designed to fit the specific needs of researchers, students and teachers. The specific aspects of this work applied to chemical research problems are described elsewhere (Skawinski, Busanic, Ofsievich, Venanzi, Luzhkov & Venanzi, 1994). We emphasize the aspects of this work particularly interesting to blind and visually impaired individuals.

The first models to be fabricated were molecular models to assist one of us (WJS), a blind chemist, to perceive the detailed structures of several molecules. Standard molecular model kits were inadequate for this purpose since they are limited in the range of diameters and angles they can represent. Stereolithography is virtually unlimited in this respect and is highly accurate. A brief description of the stereolithography process is presented, followed by a detailed description and evaluation of the tactile models.

METHODS

The stereolithography apparatus (SLA) used was the model SLA-250 from 3-D Systems, Valencia, California. The system consists of a cubic tank approximately ten inches on a side, which is filled with a liquid resin. Within the tank is a table that can be positioned vertically. At the start of the process the table is elevated to a level that allows a thin layer of the resin to lie above the table surface. The computer-controlled laser then traces out the shape of the first, (lowest), slice of the object being built. The interaction of the laser with the resin converts the liquid into a solid plastic. The table then moves downward until another thin layer of liquid lies above the first solidified slice. The laser then traces out the shape of the second slice of the object in the liquid simultaneously solidifying it and bonding it to the first slice. This process is repeated until the entire object is built slice-by-slice from the bottom up.

In order to control the SLA process data files defining the object to be built, it must first be prepared within a CAD program that supports rapid prototyping. Several CAD systems are capable of preparing files for use by the SLA. I-DEAS, a CAD software package produced by SDRC Inc., Milford, Ohio, was used for this work.

A short program was written to input the molecular structure data (i.e., positions and radii of atoms and atom type) into I-DEAS. This allowed the CAD program to automatically assemble the image of the molecule. Once the image was obtained, several processing steps are carried out in order to produce the data files which are then used to control the SLA. The data files containing the information describing the CAD image were processed into a format which represents that image as a series of horizontal slices. Thus, when these slices are stacked, they form an exact model of the original screen image. In some cases temporary support structures are required during the building process to attach parts that are disconnected temporarily while the slices are being fabricated. These supports can either be designed manually from within the CAD program or designed automatically with appropriate software. After the model is completed these supports can be readily removed. A program called Bridgeworks, from Solid Concepts, Valencia, California, was used to automatically design these supports for several of the models described here.

Atoms would normally be depicted as colored spheres in video displays or physical models, with different colors being assigned to each atom type to allow identification. For the purpose of these tactile models we used faceted rather than smooth surfaces for each atom. These facets are essentially planar faces which form a polyhedron rather than a perfectly smooth sphere. The number of facets used to define an atom, horizontally and vertically, can be specified. The number of facets for each atom type were: hydrogen -- 8 x 8; carbon -- 8 x 8; oxygen -- 12 x 12; nitrogen -- 16 x 16; chlorine -- 20 x 20. Since the hydrogen and carbon atoms differed significantly in size, the number of facets used for each could be the same without confusing the atom types. Models of 12 molecular structures were built using four different scales. These scales produced carbon atoms that were approximately 0.3 inch, 1.0 inch, 1.2 inch, or 1.4 inch in diameter. The dimensions of all the atoms were obtained by taking the vander Waals radii and setting the model radius to that value in inches in the CAD program and multiplying the result by factors of 0.08, 0.25, 0.30, or 0.35, respectively, for each scale.

The molecular structures modeled were the natural amino acids glycine, L-alanine, and L-serine; two amino acid derivatives; the drug amiloride in several forms; two derivatives of amiloride in non-planar conformations; the carbohydrate macrocycle beta-cyclodextrin; and the transition state complex of beta-cyclodextrin and phenyl acetate. These were selected because of their importance in several investigations being carried out in this laboratory. The results, however, will be described in general terms so that no knowledge of their structures will be required for an understanding of the method and results. These relatively complex molecular models should be considered as examples of the range of shapes that can be fabricated and the power and flexibility of this method.

RESULTS AND DISCUSSION

The time required for the SLA to build the five amino acid models simultaneously was eight hours, three amiloride models were built in 12 hours and the beta-cyclodextrin and beta-cyclodextrin-phenyl acetate models required 24 and 31 hours respectively. Normally the models require an additional one to two hours in an ultraviolet oven to complete the curing process.

The models of amiloride and its derivatives were built using the second smallest scale. They were approximately three inches long and two and a half inches wide. Amiloride is planar while two of the derivatives are twisted slightly out of planarity and this distinction was easily perceived tactually using the models. At this scale it was possible to tactually locate and identify all 23 to 26 atoms in these structures. It is possible to probe the entire surface of the model and detect the different textures produced by the varying number of facets for each atom type.

The amino acids and their derivatives were built on a slightly larger scale, approximately 20% larger in linear dimensions. These models were approximately two inches across. As expected, the features of these models were more easily recognized. These structures are not planar but consist of a central carbon atom with four other groups of atoms projecting from it at various angles. The spatial relationships of these projecting groups, which are important for understanding their nature, are easily perceived in these models.

The beta-cyclodextrin models were built on the largest scale, approximately 40% greater in linear dimensions than the amiloride models. This model can best be described as being funnel shaped. The large end was approximately six inches in diameter compared with five inches for the smaller end. The model was three inches high. The sides are composed of seven sugar molecules, consisting of 147 atoms, linked together to form a ring. Thus the molecule has a large internal cavity. This was built on the larger scale in order to allow easy examination of the atoms within that cavity. The groups of atoms along each opening of the cavity were easily identified. Another model of this molecule was made with a molecule of phenyl acetate within the cavity. This was done to represent a transition state structure which occurs during a reaction involving these two molecules. The important aspect of this second model was the positioning of the smaller molecule within the cavity and this was clearly depicted in the model. Tactual examination of the smaller molecule within the cavity was facilitated by the larger scale.

A second model of beta-cyclodextrin was made using the smallest scale yet attempted. It was approximately one and a half inches in diameter and three-quarters of an inch high. Even at this scale many of the atoms along the openings of the cavities were still distinguishable.

Since the same number of facets was used for each atom type in all these models, it was noted that as the scale decreases the atoms would feel more spherical because the facets would be smaller. Even in a given model that contains both hydrogen and carbon atoms the hydrogen atoms would feel more spherical because the facets on their surfaces were smaller. Also as the number of facets increased to as many as 20 x 20 in the chlorine atoms, these most nearly approached a spherical structure. The model of the beta-cyclodextrin made using the smallest scale contained the most spherical atoms. As a result this technique allowed facile discrimination between atom types even to some extent in the very small model.

Since these models were constructed for specific research purposes the radius of each atom was chosen to model selected properties. As a result the spheres representing the atoms overlapped to a large extent yielding models that appear as spheres with bumps protruding from their surfaces. Though these were well-suited for the investigations being carried out, they might not be appropriate for educational purposes at an elementary level. In those cases it might be desirable to make the individual atoms more distinct. This is readily done by decreasing the radii of the spheres while maintaining their locations in space. This would result in less overlap and more distinct atoms.

It is also possible to produce shapes based on mathematical equations using stereolithography (Peterson, 1991). We are currently working on methods of building plastic models of physical phenomena which have specific spatial distributions. For example, it is often useful to know the way in which electrical charge is distributed in the space surrounding a molecule. Using data on this property we are attempting to build a useful model which would allow blind and visually impaired individuals to perceive this information. In a similar manner any property that can be described by a mathematical function in three dimensions could conceivably be modeled in this way.

We are also planning the fabrication of an entire protein molecule on a small scale to allow examination of the general topology of the molecule. Though the size of any single model fabricated is limited by the ten-inch cube of this particular system, if the image can be created in the CAD program it can then be cut into several pieces which can be fabricated separately and assembled afterward. Additionally, the model SLA-500 is available which has a tank approximately 24 inches on a side.

Fabrication of models using SLA can become expensive as the size and intricacy of the models increases. The primary component of the cost is the resin. As a result we have made several models that are hollow in order to consume less material. They have proven to be sturdy and have been dropped on a hard floor without being damaged. Fabricating hollow models when possible also utilizes significantly less laser time which is another consideration when determining the cost. Since the models are made of uncolored translucent plastic, they may be painted to allow persons with some vision to perceive important features. This is readily done using common paints appropriate for plastics.

Methods have been developed to produce castings from these models (Imamura, Meng, & Nakagawa, 1993) which might be used for mass production. Several other rapid prototyping methods have been developed (Wohlers, 1991), including laser sintering and fused deposition modeling, which could also extend the use of such systems for producing custom tactile models, and as these methods are refined the cost of models should decrease.

Though simple models might be more easily obtained for educational purposes that adequately convey the desired information, this method extends the range of models that can be made in a reasonable time and for a reasonable cost for blind and visually impaired students and professionals. The most important aspect is that they can be custom-made to suit the application for which they are required. Images from chemistry, physics, biology or any other discipline could be modeled in this way. Virtually any image that can be produced in a CAD program can be translated into plastic.

The process of stereolithography can be applied to the fabrication of accurate three-dimensional tactile models for use by blind and visually impaired individuals. Models can be made of virtually any CAD image that can be custom-designed to suit the requirements of the user. Currently, sighted operators are still required to carry out the complete model-building process, but procedures are generally straightforward once the CAD image has been produced. Though the equipment and materials used in this process might be prohibitively expensive for most institutions to purchase, some commercial enterprises exist which will produce models for clients. The most effective use for this process is the production of models that can not be readily made by any other method. As the use of these systems become more widespread, costs may decrease allowing more extensive use by institutions with limited resources.

ACKNOWLEDGMENTS

This work was funded by a grant to C.A.V. and W.J.S. from the National Science Foundation. W.J.S. thanks Paul Strauss, Texas Instruments Corp., for bringing the stereolithography technique to his attention as a potential aid for blind and visually impaired individuals. The authors thank Alan Bondhus for his invaluable technical assistance and Ram Reddy for programming assistance.

REFERENCES

(1989). Changing America: The New Face of Science and
Engineering. Final Report, The Task Force on Women,
Minorities, and the Handicapped in Science and Technology,
established by the U.S. Congress in public law, 8.

Crosby, G. A. (1981). Chemistry and the Visually
Handicapped. _J.Chem. Educ._, _58(3)_, 206-8.

Dietrich, H. F., & Seufert, W. D. (1986). The translation of
the dimension of depth into grey values. _J. Imaging
Technol_,_12(6)_, 311-15.

Flair, M. N., & Setzer, W. N. (1990). An olfactory indicator
for acid-base titrations: a laboratory technique for the
visually impaired. _J. Chem. Educ._, _67(9)_, 795-6.

Hinchliffe, L. V. & Skawinski, W. J. (1983). Hearing is
Believing. The modified Spectroscope. _The Science
Teacher_,_50_, 53-55.

Hull, C. (1988). StereoLithography: Plastic prototypes from
CAD data without tooling. _Mod. Cast._, _78(8)_, 38.

Imamura, M., Meng, Y., & Nakagawa, T. (1993). Application of
laser stereolithography to castings. _Seisan Kenkyu_,
_45(6)_,385-92.

Lunney, D., & Morrison, R. C. (1981). High technology
laboratory aids for visually handicapped chemistry students.
_J. Chem.Educ._, _58(3)_, 228-31.

Morrison, R. C., Lunney, D., Terry, R. J., Hassell, J., &
Boswood, G. (1984). Voice-operated microcomputer-based
laboratory data acquisition system to aid handicapped
students in chemistry laboratories. _J. Chem. Inf. Comput.
Sci._, _24(4)_, 271-5.

Peterson, I. (1991). Plastic Math, Growing Plastic Models of
Mathematical Formulas. _Science News_, _140_, 72-3.

Reddon, M. R., Davis, C., & Welsh-Brown, J. (1978). Science
for Handicapped Students in Higher Education. _AAAS
publication_,_19910_, Sm177-19910.

Reese, K. M. e. (1985). Teaching Chemistry to Physically
Handicapped Students. _American Chemical Society

Shimizu, Y. (1986). Tactile display terminal for visually
handicapped. _Displays_, _7(3)_, 116-20.

Skawinski, W. J., Busanic, T. J., Ofsievich, A. D. Venanzi,
T.J., Luzhkov, V. B., & Venanzi, C. A. (1994) The
Application of Stereolithography to the Fabrication of
Accurate Molecular Models. Manuscript submitted to Journal
of Molecular Graphics.

Smith, D. (1981). Teaching aids for visually handicapped
students in introductory chemistry courses. _J. Chem.
Educ._, _58(3)_,226-7.

Tombaugh, D. (1981). Chemistry and the visually impaired:
available teaching aids. _J. Chem. Educ._, _58(3)_, 222-6.

Wohlers, Terry T. (1991). Make Fiction. _Manuf.
Eng.,__106(3),_ 44-49.