Mass fraction profiling based on x-ray tomography andits application to characterizing porous silica boules
Werner J. Glantschnig and Albert Holliday
Mass fraction profiling is a specialized tomographic technique designed to allow one to measure the concen-tration profiles of the component species of a synthetic composite material. While developed for the expresspurpose of determining the dopant, density, and porosity profiles of unsintered soot boules, the methodshould also be useful in analyzing other types of ceramic such as dried gels. Mass fraction profiling amountsto a three-step process consisting of scanning the object of interest with well-collimated beams of x rays,reconstructing the attenuation coefficient profiles for several x-ray energies, and generating the density andconcentration or mass fraction profiles. The method for determining these profiles is outlined, and somemeasurement results obtained for soot boules are presented.
1. Introduction
Computed tomography (CT) has developed into adiagnostic and analytic tool that is finding ever morediverse applications. The technique originally at-tained prominence in radiology but has lately alsofound many other scientific as well as industrial uses.By now a substantial body of literature on CT exists,and the reader is referred to a recent special issue ofApplied Optics1 which contains several articles thatgive a good flavor of the history as well as the presentmanifold uses of the technique. As the cost of compu-tation decreases further, it is safe to predict that theproliferation of applications of computed tomographywill continue.
The purpose of this paper is to present a specializedtomographic technique based on x-ray absorptionmeasurements, which shall henceforth be referred toas mass fraction profiling (MFP). This technique wasdeveloped to address the following material character-ization problem that makers of synthetic compositesare often confronted with: Given knowledge of theparticular molecular or atomic species a compositematerial consists of, what is the mass density profile ofthe composite and what are the mass or mole fraction
The authors are with AT&T Engineering Research Center, P.O.Box 900, Princeton, New Jersey 08540.
profiles of its component species? Like any othertomographic technique, the proposed method pos-sesses the desirable feature of being nondestructiveallowing analyzed samples to be further processed.
Mass fraction profiling amounts to a three-step mea-surement process. First, the object to be character-ized is scanned with a beam of well-collimated x rays,and the attenuation of the beam for each path throughthe object is measured. Then the linear attenuationcoefficient profile of the object is reconstructed for arequired number of distinct x-ray energy bands. Fi-nally, the mass or mole fractions of the constituentatomic or molecular species and the density (in abso-lute units) are determined.
MFP was specifically developed as a means for ob-taining density and dopant concentration profiles ofporous silica boules.2 These soot boules are precur-sors to solidified optical fiber preforms. They aregenerally cylindrical structures which can be grown bya variety of vapor deposition processes.3-6 In theirunsintered state they are opaque to visible light andhence cannot be analyzed by the standard opticalmethods used on fiber preforms. Current generationfibers are silica based with dopants such as germania orfluorine providing an increase or decrease, respective-ly, in the index of refraction. The impetus for devel-oping MFP was the realization that the capability ofmeasuring the dopant concentration profile in poroussoot would be helpful in studying soot deposition andeventually would provide guidance in optimizing theprocess parameters for growing boules. While it hasbeen previously suggested that there might be a way ofusing x-ray absorption measurements for characteriz-ing soot boules, a workable method has not yet beenreported.
Section II reviews the theory pertinent to MFP. Wereview x-ray attenuation by inhomogeneous amor-phous matter and how the mass fractions of the com-ponent species of a composite can be obtained frommultienergy x-ray attenuation measurements. Sec-tion III discusses the application of MFP to analyzingsoot boules. Included in this section are some exam-ples of measurement results obtained for boules grownby different soot deposition processes. There are oth-er material characterization problems to which MFPought to be applicable. For example, the methodshould be useful in analyzing simple or compositieceramics such as dried gels and alloys.
11. Theory
A. Review of X-ray Attenuation in Matter
It is well known that a beam of electromagneticradiation propagating through matter is attenuated inan exponential fashion. Attenuation arises becausefor each unit distance of propagation a certain photonfraction is removed from the initial beam by scatteringand absorption events. A measure of how efficiently aparticular material attenuates photons of energy E isprovided by its linear attenuation coefficient g(E,r).
The linear attenuation coefficient is easily measuredfor a homogeneous material in the shape of a parallelslab. If, however, the object of interest is of arbitraryshape and of inhomogeneous composition, one dealswith a typical tomographic problem. Its solution re-quires that x-ray beam attenuation data be taken alonga multiplicity of paths, and a fairly involved numericalprocedure is subsequently required to reconstruct thelinear attenuation coefficient profile in the measure-ment plane. There are several methods that can beused to reconstruct the linear attenuation coefficientprofile, some of which are reviewed in Refs. 8 and 9.The purpose of this paper, however, is to discuss thespecific information about composition and macro-scopic structure that is contained in the linear attenua-tion coefficient profile of an inhomogeneous materialand how this information can be obtained from mul-tienergy attenuation measurements.
Multienergy attenuation measurements are some-times also used in radiological work although for adifferent purpose. For example, dual-energy mea-surements can be used to correct CT images for spec-tral artifacts.10 Multienergy measurements are alsorequired for various image display enhancementschemes designed to aid in the interpretation of CTimages by either enhancing or subtracting materialsthat are extraneous to diagnosis.1
The linear attenuation coefficient is a macroscopicmeasure of absorption having units of cm-'. Since x-ray beam attenuation is the aggregate effect of manysingle photons interacting with single atoms, the linearattenuation coefficient can be expressed in terms ofthe pertinent cross sections per atom. For example,for a monatomic substance consisting of atoms of spe-cies i the linear attenuation coefficient is given by12
ui(E) = Ni (atoms f )(E) ( cm )k cm3 j=1 atom )
(1)
where a(')(E) is the cross section for a photon of energyE being involved in an interaction of type j with anatom of species i, and the sum is over all the interactionprocesses that contribute to beam attenuation.
Note that the magnitude of Ai(E) depends on howtightly these atoms are packed, since the atom densityNi is very different for the gas, liquid, or solid state.This dependency on the physical state can be eliminat-ed by dividing both sides of Eq. (1) by the density ofthe material pi. The resulting quantity
ai(E) = i(E) (2)
Pi
has units of cm2 g'1 and is usually referred to as themass attenuation coefficient.
The mass attenuation coefficient is a macroscopicmeasure of attenuation of a material per unit mass.This makes this particular parameter a useful conceptin discussing composite materials. If one multipliesai(E) for each component species by its mass fractionxi(r) and subsequently sums over all the componentspecies, one obtains an expression for the mass attenu-ation coefficient of the composite material given by12
u(E,r) p(r) (3)
Note that by letting the mass fractions of the compo-nent species be functions of position r, we have allowedfor these species to be unevenly distributed so that Eq.(3) defines the mass attenuation coefficient for aninhomogeneous composite material. Finally, the cor-responding linear attenuation coefficient is simply ob-tained by multiplying both sides of Eq. (3) by thedensity of the composite material p(r). Thus
M!,(E,r) = p(r)E ai(E)xi(r).
i=1(4)
Since ,(E,r) depends on both the density p(r) and themass attenuation coefficients a(E) of the species mak-ing up the composite, both density and compositionalinformation are encoded in the linear attenuation coef-ficient.
Up to this point the terms component and atomicspecies have been used interchangeably. However, acomponent species need not necessarily be a particularatomic species. Since molecular binding energies areonly of the order of a few electron volts, they have noeffect on the interaction processes responsible for theattenuation of hard x rays. Hence the terms under thesum sign in Eqs. (3) and (4) may represent atomicspecies, molecular species, or a combination of both.
Another point worth noting about Eq. (4) is that asingle linear attenuation coefficient profile will notnecessarily allow conclusions as to whether a compos-ite material is inhomogeneous. As can be seen fromEq. (4), the linear attenuation coefficient is the prod-uct of the density p(r) and the mass fraction weightedsum of the mass attenuation coefficients of the compo-
nent species. The presence of a heavy componentspecies in a composite tends to increase the magnitudeof both the density and sum term. Consider, however,a hypothetical object consisting of some material thatis homogeneous except for a small region which isuniformly doped with a heavy atomic species. If thatdoped region for some reason were more porous thanthe undoped region, one could have a situation wherethe increased porosity depresses the density justenough to compensate for the increased magnitude ofthe sum term due to the presence of the heavy dopant.Such an object would have a constant or flat linearattenuation coefficient profile despite it not being ho-mogeneous. This simple argument shows that a singleattenuation coefficient profile can provide misleadinginformation about porous objects.
B. Determination of the Density and Mass FractionProfiles
Before one approaches the task of determining thedensity and mass fractions one needs to know how themass attenuation coefficients of the component spe-cies can be obtained. This can be done in two ways.One can either calculate these coefficients using a se-miempirical formula developed by Victoreen,'3" 4 or,what is much more preferable, one can measure themdirectly using specially prepared pure samples of thecomponent materials.
Once the mass attenuation coefficients of the com-ponent species are known, Eq. (4) amounts to a rela-tionship between the measurable parameter A(E,r), onthe one hand, and the M + 1 parameters to be deter-mined, namely, the density of the composite materialand the M mass fractions of its component species, onthe other. To solve for these M + 1 unknowns, M + 1coupled equations are required. Such a system ofequations can be constructed by exploiting the energydependence of the various attenuation coefficients inEq. (4). By measuring the linear attenuation coeffi-cient profiles for M different x-ray energies one ob-tains a system of equations which can be representedby the following matrix equation:
al(El) ... aM(El) [x(r) (El,r)p(r) = . (5)
al(EM) ... a(EM)J xM(r) JL(EM,r)
By inverting this matrix equation a solution for thevector p(r)[xl(r),... ,xM(r)] can be found. One knowsfurthermore that at any position r the mass fractionsof the components must add up to unity so that
M ME p(r)xi(r) = p(r) E xi(r) = p(r). (6)
i~~~l ~~i=1Hence, once Eq. (5) is inverted, Eq. (6) can be used tofind p(r), and all unknowns are solved for. To obtainthe density and mass fraction profiles throughout themeasurement plane the above system of equationsneeds to be solved for each position r in the measure-ment plane.
Equation (5) is applicable if the required number oflinear attenuation coefficient profiles are measuredwith strictly monoenergetic sources such as radioac-tive sources emitting y rays of suitable wavelength.For a variety of reasons x-ray tubes make more practi-cal sources. Since x-ray tubes provide a broadbandoutput, one must either resort to monochromatizationtechniques or modify Eq. (5) to account for the poly-chromatic nature of the beam. If the latter approachis chosen, the linear and mass attenuation coefficientsin Eq. (5) need to be replaced by the respective averageor group coefficients, which for energy group j bound-ed by Ej1, and E12 can be defined as
( = E g u(Er)dE,
a j = Aai(E)dE.E E2-E 1 E1
(7)
(8)
Thus the matrix equation applicable in the case thatbroadband beams are used is given by
all ... aMl 1 xl(r) 1 F pJ(r) 1p(r) I =1 I
alM ... amm_ Lxm(r) A LM(r)J(9)
where any one row in the above matrix equation isobtained by integrating Eq. (4) with respect to E overpart of the x-ray spectrum.
Note that the second approach based on the use ofthe average attenuation coefficients presupposes spe-cial hardware. To be able to compute ,u1(r) and aijaccording to Eqs. (7) and (8) the energy dependence ofthe measured parameters ,u(E,r) and ai(E) must beprecisely known. Hence an x-ray detection systemwith very good energy resolution is required. A simplephoton counting system which suffices to measure theintensity of a monochromatic beam is not adequate inthis case.
If either one of the methods just outlined is to pro-vide accurate density and component concentrationprofiles two conditions must be satisfied. First themass attenuation coefficients of the components at thedifferent x-ray energies which make up the coefficientsof the matrix in Eq. (5) and (9) must be accurate. Ifthey are not, the density and mass fraction profiles willcontain a systematic error. According to the litera-ture15"16 and also our own experience, the values for themass attenuation coefficients computed with Victor-een's semiempirical formula agree well with experi-mentally determined values. However, we found itadvantageous to use the experimental values. This isbecause as long as one uses the same measurementgeometry, any systematic errors incurred in measuringthe attenuation coefficients of the pure componentcalibration samples and those incurred during the ob-ject scan tend to offset each other.
The second prerequisite for success is that the linearattenuation coefficient profiles of the object be mea-sured accurately. Since these profiles are not the finalresult but constitute the input data for computing the
density and mass fraction profiles, MFP is not as for-giving as conventional tomography. While a conven-tional CT image (which is but a linear attenuationcoefficient profile) tained by various measurement er-rors might still be good enough to provide qualitativestructure information about the object, the multien-ergy CT images used with MFP must be accurate andconsistent with each other.
There are several other factors that impact on howwell MFP works in a particular application. One suchfactor is the constitution of the sample to be analyzeditself. If the component species of a composite materi-al consist of elements that are far apart in terms oftheir atomic number it is easier to obtain good resultsthan if they are not. Additional difficulties with mul-tienergy measurements arise from the fact that it isdifficult to produce a beam with sufficient intensity inmultiple energy bands so that data for several linearattenuation coefficient profiles can be obtained in onescan. Furthermore, with each added energy measure-ment more noise is introduced resulting in a commen-surate increase in the noise of the mass fraction anddensity profiles. Generally, the fewer energy mea-surements necessary the better the quality of the re-sults.
Since it is advantageous to keep the number of mea-surements at different energies to a minimum, oneshould attempt to utilize additional information abouta material if such information is available. Two spe-cial cases come to mind for which M - 1 rather than Mattenuation coefficient profiles suffice for determiningMmass fraction profiles. If the material being studiedis porous, air or whatever other gas is trapped insidethe voids effectively constitutes one of the componentspecies. However, the mass fraction of the gas compo-nent is negligible compared with the mass fractions ofthe solid components. Hence the a priori assumptionthat xair(r) = 0 everywhere inside the porous sample isusually justified.
A material which can be regarded as a solid solutionfor which ideal mixing holds comprises the other spe-cial case. If ideal mixing holds, the total volume is thesum of the partial volumes of the component species sothat
M r) r)1 E xi~~~~r) . ~(10)P~)i=1 P
Since Eq. (10) provides a relationship between theparameters one wants to solve for, Eq. (10) can be usedin place of an energy-dependent equation making upone of the rows in Eq. (5) or (9). However, Eq. (10)must not be used if one deals with a porous sample.While the mass fraction of the gas in the voids isgenerally negligible, the volume fraction of the gas canbe substantial, especially for materials with large po-rosity. Hence the use of Eq. (10) is incompatible inconjunction with the assumption that xair(r) = 0.
Ill. Application of MFP to Soot Boule Characterization
MFP is ideally suited for characterizing porous silicasoot boules. A dual-energy measurement is all that is
INCIDENT BEAM WITH ATTENUATED BEAMWITH INTENSITY (Et)
BOULE ORPREFORM
Fig. 1. Schematic representation of a well-collimated x-ray beampenetrating a cylindrically symmetric object in parallel beam geom-
etry.
required to obtain complete density, porosity, anddopant concentration information about a binary sili-ca-germania boule. In addition, germanium atoms aremuch heavier than silicon atoms making for a reliabledetection of even small amounts of germania.
The CT scanner we developed is a highly collimatedsingle-beam system with a spatial resolution of '100Am. The x-ray source and detector are mounted on amovable stage driven by a stepper motor. There is acutout in the stage between the source and the detectordesigned to accommodate the boule. The boule ismounted with its axis perpendicular to the stage.During a measurement the boule is stationary and thex-ray beam is scanned across it by moving the stagewith the source, detector, and collimation hardwarefirmly mounted on it.
A. Implications of Soot Boule Symmetry
Owing to their cylindrical symmetry both the datacollection and attenuation coefficient profile recon-struction task are much less involved for boules thanfor an object with no particular symmetries. A singlebeam scan through the boule perpendicular to theboule axis suffices to obtain all the data needed todetermine the attenuation coefficient of the boule. Asa result of the cylindrical symmetry, the attenuationcoefficient is simply a function of the radial distance rrather than some specific position r in the measure-ment plane. The linear attenuation coefficient A(E,r)is given by the simple closed-form expression
,uE r d [log i(Et)] dt (11)
where io(E) and i(E,t) are the intensities of the inci-dent and attenuated beams, respectively, t is the dis-tance of impact of the incident beam from the bouleaxis, and R is the boule radius (see Fig. 1). Equation(11) is obtained by applying the Abel integral inversionprocedures to the analytical expression for the inten-sity of the attenuated beam originally incident a dis-tant t from the boule axis. This intensity is given as
(12)
Equation (1) is simply the exponential attenuation law
Fig. 2. Measured beam intensities as a function of the impactdistance t for two x-ray energies E1 and E 2 as the beams are scannedthrough the boule perpendicular to the boule axis. The measuredintensities are normalized with respect to the unattenuated beams.
as it applies to a cylindrical object with a linear attenu-ation coefficient that is a function of radius only.
B. Discussion of Sample Measurement Results
As a first example, final results as well as the resultsof intermediate measurement steps obtained from anominally step-index soot boule are presented in Figs.2-4. Figure 2 shows the normalized intensity of theattenuated beam as a function of the impact distance tas measured by scanning the x-ray beam across theboule perpendicular to the boule axis. According tothe discussion in the previous section the attenuationcoefficient profile needs to be determined for two x-rayenergies, since we are dealing with a binary silica-germania system. Hence beam attenuation was mea-sured for two different x-ray energies labeled E1 andE2, which in this case are -35 and 70 keV, respectively.Using the data shown in Fig. 2 the linear attenuationcoefficient profiles for both energies can be obtainedby employing an algorithm based on Eq. (11). Theresulting attenuation coefficient profiles are shown inFig. 3. Note that there is a clear qualitative differencebetween the high- and low-energy attenuation coeffi-cient profiles. Whereas the curve for Iu(E2,r) keepsgradually increasing toward the boule axis, there is asudden factor of 2 increase in g(El,r) at the onset of thedoped region. This is a manifestation of the strongenergy as well as atomic number dependence of photo-electric absorption. A germanium atom is a muchmore efficient absorber of 35 keV than of 70-keV pho-tons.
Finally, Fig. 4 contains the germanium concentra-tion and density profiles of the boule for which theintermediate results have been shown in Figs. 2 and 3.This boule has a hole in the center which resulted fromremoval of the mandrel on which it was grown. Thecore is seen to contain an average of -13-mol % germa-nia. The germania concentration is shown in terms ofmole rather than mass percent, since preform makersare accustomed to relating the refractive index of abinary silica-germania glass to the mole fraction of
-2.0 -1.0 0 1.0 2.0
RADIUS (cm)
Fig. 3. Linear attenuation coefficient profiles of a soot boule fortwo distinct x-ray energies E1 and E 2-
Fig. 4. Germania mole fraction and mass density profiles of anominally step-index boule with a germania doped core and puresilica cladding. The hole in the center resulted from the removal of
the alumina mandrel on which this boule was grown.
germania. The average density of this porous boule isseen to be -0.4 g cm-3 or -18% that of fully sinteredsilica. This high level of porosity is desirable for thedehydration required as part of the sintering process,which follows deposition. Noteworthy within the coreare the periodic peaks in the germania mole fractionprofile spaced about 1 mm apart. In view of the factthat the torch was backed up after each millimeter ofradial growth so as to maintain the torch-surface dis-tance about constant, this profile structure suggeststhat there are changes in the germania deposition effi-ciency associated with small changes in the torch-surface distance. The capability to observe such pro-cess subtleties in porous boules in a nondestructivefashion is very useful to the preform maker.
Fig. 5. Germania mole fraction and mass density profiles of a puresilica boule consisting of a solid silica mandrel in the center and pure
silica soot on the outside.
20
w-J0
zC:
CD
10
0
0.4
r= 0.30)
0.20.2
z, 0.1
0
t
3.0-2.0 -1.0 0 1.0 2.0 3.0
RADIUS (cm)
-J.U -. U -1 .U U 1.U e.U i.U
RADIUS (cm)
Fig. 7. Germania mole fraction and mass density profiles of a VADboule consisting of a germania doped core and pure silica soot
cladding layers deposited with multiple cladding torches.
Fig. 6. Superposition of a photograph of a boule being grown with amultiple-torch vapor-phase axial deposition system and the density
profile subsequently obtained for this boule.
Figure 5 shows the dopant concentration and densi-ty profile of a boule consisting of pure silica soot depos-ited on a solid silica rod. This boule was grown for theexpress purpose of checking the ability of MFP todiscriminate between optical density fluctuations dueto porosity on the one hand and changes in the germa-nia concentration on the other. As Fig. 5 clearly dem-
onstrates, despite a large difference between the densi-ty of the solid core and the surrounding soot, themethod provided the proper result that there is nogermanium anywhere inside this particular boule.
Figure 6 is a photograph of a boule being grown byvapor-phase axial deposition. Superimposed on thephotograph is the density profile of this boule. Notethat the shape of the density profile reflects the use ofmultiple torches. Particularly characteristic for aboule grown this way are the sharp density minima atthe cladding layer interfaces. The same multitorchsignature is evident in Fig. 7, which contains the molefraction and mass density profiles of another boulegrown by the same process but with additional torchesdepositing pure silica cladding.
Having an idea what the constitution of a boule islike the preform maker can go on to dehydration andsintering and then measure the refractive index of theglassy preform using, for example, the transverse laserbeam refraction technique.18'1 9 This way the effects ofthe various processing steps on preform compositioncan be studied and quantified separately. Therein liesthe usefulness of mass fraction profiling as applied tolightguide materials technology.
The authors are indebted to T. Miller and C. Wei forlending their boules for the sample measurements.Furthermore, the support of C. W. Draper, L. S. Wat-kins, and R. J. Klaiber is gratefully acknowledged.References1. Special Issue on Computed Tomography, Appl. Opt. 24 (1 Dec.
1985).2. W. J. Glantschnig, "Method And Apparatus For Analyzing A
3. T. Izawa, S. Sudo, and F. Hanawa, "Continuous FabricationProcess for High-Silica Fiber Preforms," Trans. IECE Jpn. E 62,779 (1979).
4. P. C. Schultz, "Fabrication of Optical Waveguides by the Out-side Vapor Deposition Process," Proc. IEEE 68, 1187 (1980).
5. A. J. Morrow, A. Sarkar, and P. C. Schultz, "Outside VaporDeposition," in Optical Fiber Communications, Vol. 1, FiberFabrication, T. Li, Ed. (Academic, Orlando, 1985); N. Niizeki,N. Inagaki, and T. Edahiro, "Vapor-Phase Axial DepositionMethod," in Optical Fiber Communications, Vol. 1, Fiber Fab-rication, T. Li, Ed. (Academic, Orlando, FL. 1985).
6. E. Potkay, "Representation of the Vapor-Phase Axial Deposi-tion Helical Layer Structure Through a 'Spiral Transforma-tion,' " J. Appl. Phys. 57, 1509 (1985).
7. H. Takahashi, S. Shibuya, and T. Kuroha, "Applicative Investi-gation of X-Ray Nondestructive Inspection Technique for Mea-surement of Core Diameter and Germanium Doping Concentra-tion Profiles of Optical Fiber Preforms," in Technical Digest,Fifth European Conference on Optical Communication, Am-sterdam (1979).
8. R. M. Levitt, "Reconstruction Algorithms: Transform Meth-ods," Proc. IEEE 71, 390 (1983).
10. J. P. Stonestrom, R. E. Alvarez, and A. Marcovski, "A Frame-work for Spectral Artifact Corrections in X-Ray CT," IEEETrans. Biomed. Eng. BME-28, 128 (1981).
11. G. Vernazza, F. Caratozzolo, S. B. Serpico, and S. Geraci, "SomeResults in Multi-Energy Digital Radiology," Proc. (Soc. Photo-Opt. Instrum. Eng.) 535, 202 (1985).
12. R. D. Evans, The Atomic Nucleus (McGraw-Hill, New York,1955), Chap. 25.
13. J. A. Victoreen, "The Absorption of Incident Quanta by Atomsas Defined by the Mass Photoelectric Absorption Coefficientand the Mass Scattering Coefficient," J. Appl. Phys. 19, 855(1948).
14. J. A. Victoreen, "The Calculation of X-Ray Mass Absorptioncoefficients," J. Appl. Phys. 20, 1141 (1949).
15. C. M. Davisson and R. D. Evans, "Gamma-Ray AbsorptionCoefficients," Rev. Mod Phys. 24, 79 (1952).
16. The International Union of Crystallography, International Ta-bles for X-Ray Crystallography (D. Reidel, Boston, 1983), Vol.3.
17. M. Bocher, An Introduction to the Study of Integral Equations(Cambridge U.P., 1926).
18. L. S. Watkins, "Laser Beam Refraction Traversely Through aGraded-Index Preform to Determine Refractive Index Ratioand Gradient Profile," Appl. Opt. 18, 2214 (1979).
19. W. J. Glantschnig, "How Accurately Can One Reconstruct anIndex Profile From Transverse Measurement Data?," IEEE/OSA J. Lightwave Technol. LT-3, 678 (1985).
Hitachi Review
This journal, in English, has just completed volume 35. The articles are written by Hitachiresearchers examining developments and trends of interest to Hitachi. The subjects vary withthe issue; August 1986, vol. 35, no. 4, for example, is devoted to optoelectronics technology.Some of the articles are on Hitachi's efforts to penetrate the fiber optics market, thedevelopment of reliable fiber-optic networks, fiber-optic data transmission equipment fordirect long distance links between computers, fiber-optic digital video transmission equipment,fiber-optic components: laser diode modules and transmission modules, 1.5-,gm DFB laserdiodes and InGaAs/lnP photodetectors, optoelectronic integrated circuits for high-speed fiber-optic transmission, and optical fiber cables. Published bimonthly by Hitachi, Ltd., Tokyo,Japan, price 900 yen per copy, the U.S. contact is Hitachi Information Center, 6 East 43rd St.,
