Measurements and analysis of radiation effects in ... - Semantic Scholar
JOURNAL OF APPLIED PHYSICS
VOLUME 83, NUMBER 12
15 JUNE 1998
Measurements and analysis of radiation effects in polycapillary x-ray optics B. K. Rath Wadsworth Center, New York State Department of Health, Albany, New York 12201-0509
Lei Wang, B. E. Homan, F. Hofmann, W. M. Gibson, and C. A. MacDonalda) Center for X-ray Optics, University at Albany, State University of New York, Albany, New York 12222
~Received 1 December 1997; accepted for publication 2 March 1998! Polycapillary x-ray optics are arrays of large numbers of small hollow glass tubes which deflect x rays by successive total external reflection. These optics have growing numbers of applications in areas ranging from medical imaging to microanalysis. An accelerated radiation effects study has been performed to understand the performance limitation of these optics for medium to high intensity radiation applications, to study x-radiation damage mechanisms, and to investigate possible ways to mitigate the radiation effects on x-ray transmission efficiency. Exposures have been done in white beam bending magnet radiation with peak energies at 5 and 11 keV and focused broad band radiation centered at 1.4 keV. In situ and ex situ measurements of loss of x-ray transport efficiency have been executed at doses up to 1.8 MJ/cm2. Thin polycapillary fibers displayed noticeable bending and experienced substantial degradation of x-ray transmission. Thicker polycapillary fibers showed a linear but much slower transmission loss as a function of total dose. Annealing effectively restored the low energy ~;8 keV! transmission efficiency of the fibers. Exposure of these fibers at slightly elevated temperatures prevented any measurable loss in the low energy transmission efficiency. A variety of analytical techniques has been used to understand these results. No significant change was observed in the chemical composition of the capillary surface. Profile measurements and high energy transmission efficiency spectra, along with computer simulation studies, suggest that radiation induced bending is the primary cause of transmission efficiency degradation of the fibers. © 1998 American Institute of Physics. @S0021-8979~98!04311-4#
I. INTRODUCTION
has necessitated a detailed investigation of the radiation effects on the performance of these optics. Capillary x-ray optics use multiple total reflections to guide grazing incidence x rays. X-ray photon energies are much larger than the electron plasma energies of glasses, which are tens of electron volts. In this regime, the index of refraction of glass can be simply approximated by
Because the index of refraction of all materials in the x-ray region is slightly less than unity, x rays are barely deviated by optical lenses. This makes the task of controlling x rays different and more difficult than controlling visible light. However, there are a number of ways in which x rays can be steered. Plane crystals1 can be used for producing a monochromatic parallel beam and curved crystals2 for producing a focused beam due to Bragg reflection. Multilayers3 are also used for controlling x rays. Parabolic mirrors4 are used for obtaining a parallel beam and ellipsoidal mirrors4 for a focused beam. However, most of these optics are either spectrally selective or have a small acceptance angle. Polycapillary optics have relatively large angular acceptance and a spectral bandwidth. A typical polycapillary fiber is shown in Fig. 1. Successive reflections from the inner surface of the glass tubes deflect the incident photons by several degrees, making it possible to steer x rays over a large angular range. Polycapillary x-ray optics have shown great potential for use in x-ray diffraction,5 protein crystallography,6 x-ray astronomy,7 x-ray lithography,8 scatter rejection in mammography9 and x-ray focusing for microanalysis applications.10 The possibility of using polycapillary optics in these moderate to high intensity applications
v 2p « n 5 >12 2 , «0 v 2
where n is the index of refraction, « is the dielectric constant of the glass, v is the photon frequency, « 0 is the dielectric constant of vacuum, and v p is the plasma frequency of the material,
v p5
Nq 2 mP o
~2!
where N is the total electron density, and q and m are the charge and mass of the electron. The critical angle for total external reflection is given by
u c>
vp . v
~3!
Radiation effects in glass are many and are complicated. Glass surfaces exposed to radiation are subject to changes in their optical, electrical, and physical properties. Compaction
a!
Electronic mail: [email protected]
0021-8979/98/83(12)/7424/12/$15.00
~1!
7424
© 1998 American Institute of Physics
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
Rath et al.
7425
FIG. 2. Setup for measuring x-ray transmission efficiency of polycapillary fibers.
FIG. 1. SEM micrograph of a polycapillary fiber cross section. The fiber is approximately 0.5 mm flat to flat. Capillary channels are about 50 m m in diameter.
of glass due to neutron irradiation is more apparent, but ionization compaction in glass has also been observed due to exposure of electrons with energy as low as 2 keV,11 exposure to gamma rays,12 low energy x rays13 and ultraviolet ~UV! light.14 The compaction efficiency of x rays appears to be approximately half of that of electrons. Additionally, changes in chemical composition are also expected. These include desorption of oxygen,15 and alkali metal segregation.16 Formation of oxygen bubbles in the glass matrix due to ionizing radiation is also reported17 and the bubbles might roughen the glass surface. Surface cracking has also been observed due to electron irradiation.18 All of these radiation induced effects may decrease the surface reflectivity of the glass. Since transmission through fibers depends on the reflectivity to the nth power, where n is the number of reflections, even a slight decrease in reflectivity can be highly damaging to polycapillary optic transmission. Finally, since x rays absorbed by matter are converted to heat, large thermal gradients can be a problem for x-ray optics. In a polycapillary optic, photons incident on the open channels with an angle more than the critical angle, and those incident on the face wall of the optic, will be absorbed. Possible changes of the topology of the optic surface could be highly detrimental to the optic performance. Accelerated exposure studies in a bending magnet white beam and focused low energy beams have been performed. To investigate the resulting changes, a number of analytical techniques have been used. These include x-ray transmission spectra and source scan studies, atomic force microscopy ~AFM!, Rutherford backscattering ~RBS!, particle induced x-ray emission ~PIXE! studies, Raman spectroscopy, and scanning electron microscopy ~SEM!. II. EXPERIMENTAL OBSERVATIONS
Thin, flexible polycapillary fibers, when exposed to high intensity white beam synchrotron radiation, experienced substantial bending and resultant transmission degradation.
These fibers were observed to uncurl spontaneously upon post-exposure thermal annealing. Thicker, more rigid, fibers were more robust. In situ annealing of the thicker fibers, i.e., by placing the fibers on hot stages during exposure, prevented most of the radiation damage. A. Experimental apparatus and measurement technique
The setup used for most of the in situ and ex situ exposure measurements at the synchrotron radiation sources is shown in Fig. 2. The fiber holding plate was a half-inch thick aluminum plate on which was machined a fine straight groove. The plate was fitted with two 100 W cartridge heaters to control its temperature. The pinhole was mounted in such a way that it could be moved in horizontal and vertical directions with respect to the fiber, allowing positioning of the pinhole at any desired point of the fiber cross section. The pinhole and the plate were mounted together on a tilt/ rotation stage to avoid the relative motion of the pinhole with respect to the fiber during the alignment of the fiber with the synchrotron beam. Iron fillings were used to block x rays passing around the fiber. The fiber plate could be moved vertically with respect to the pinhole and hence can be moved out of the way of the pinhole for measuring the direct current output from the detector. The pinhole was smaller than the fiber, so that the transmission efficiency was the ratio of the photon flux with the fiber in place to that without the fiber. The pinhole was mounted on the down stream end of the fiber so that the fiber was fully exposed to the radiation, while only a small section of the beam passing through the fiber was taken for transmission efficiency measurements. This helped in reducing the beam current and resultant potential nonlinear response from the detector. A Si~111! crystal was mounted beyond the end of the fiber and positioned to reflect 8 keV photons. Although the fiber was exposed to the broad spectrum of white beam, only the change in transmission efficiency at 8 keV ~Cu K a energy! was measured. The use of the crystal also helped in reducing the potential non-linear response of the detector to an intense beam. The detector was positioned in the reflected beam from the crystal and could be moved in both horizontal and vertical directions. All the linear stages were fitted with actuators and were controlled by a Newport motion controller.
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
7426
Rath et al.
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
FIG. 3. In situ normalized 8 keV transmission as a function of the angle for a type B-4 polycapillary fiber at NSLS. The solid line is a Gaussian fit with FWHM53.8 mrad.
A small portable low-power laser was mounted on a compound stage, which gave all the degrees of freedom needed for alignment and was set behind the crystal. In measurements at Cornell University High Energy Synchrotron Source ~CHESS! and at Brookhaven National Laboratory National Synchrotron Light Source ~NSLS!, ion counters were used for detection of the x rays. The detector linearity versus total synchrotron beam current was verified. At the University of Wisconsin low energy Synchrotron Radiation Center ~SRC!, the beam was a broad-band focused beam centered at 1.38 keV, with a full width at half maximum of 0.8 keV. A photodiode was used as the detector and no crystal was used. Alignment of the fiber with the synchrotron beam was achieved in two stages. Initially, the laser beam was roughly aligned with marks made by an x-ray beam on two burn papers, one at the entrance window and the other 40 cm downstream. The fiber holding groove was then aligned with the laser and the polycapillary fiber was laid in the groove. Finally, fine alignment was done by maximizing the x-ray signal passing through the fiber. Verification of alignment and absence of leakage were done by tilt and rotation scans as shown in Fig. 3. The scan is symmetric, and the full width at half maximum ~FWHM! is about 4 mrad, approximately the critical angle of the glass. In situ measurements in the focused low energy beam at SRC were also verified by a similar technique. Vertical tilt and horizontal rotation scans
FIG. 4. In situ normalized 1.4 keV transmission as a function of the rotation angle ~squares! or tilt angle ~circles! of a type B-2 polycapillary fiber at SRC. The larger width of the rotation scan compared to the tilt scan is due to the difference in the horizontal and vertical divergence of the focused beam. The dotted line is a Gaussian fit with FWHM57.11 mrad. The solid line is a Gaussian fit with FWHM53 mrad.
are shown in Fig. 4. The larger width of the rotation scan is due to the difference in horizontal and vertical divergence of the focused beam. Ex situ exposures were done in a similar geometry but without precise alignment of the fibers with the beam. The fibers were stacked to fill the beam cross section. A small number of fibers was removed at regular time intervals to provide good statistics at a number of doses. Postexposure transmission measurements were done at an automated transmission measurement test system, which had been developed to measure the effects of radiation on transmission efficiency of the fibers reproducibly and quickly.19 The reproducibility of the measurements in the automated transmission measurement system is within 0.5%. B. Degradation of x-ray transmission efficiency
1. Flexible fibers
Exposure experiments were performed at the three synchrotron x-ray sources listed in Table I. For the white beam, the x-ray intensity was approximated by dividing the x-ray spectrum into bins, multiplying the photon flux in the bin by the photon energy at the center of the bin, and summing the bins. The beam currents were recorded during the experiment and integrated to determine total dose values. The fibers used for radiation exposure are described in Table II. Types B, C, and D were of the same composi-
TABLE I. Synchrotron radiation facilities used in this work.
Facility
Beam line
SRC
ES-2
CHESS
C-2
NSLS
X-ray energy
Characteristics
Atmosphere
Maximum beam current
Maximum x-ray intensity
1.38 keV ~centered at 1.38 and 0.8 keV FWHM! 11 keV ~critical!
Focused by a torroidal mirror to 2.533 mm
Helium
80 mA at 1 GeV
5.4 W/cm2 in the focal spot
White beam
Air
75 mA at 5.5 GeV
87 W/cm2
White beam
Air
250 mA at 2.5 GeV
13 W/cm2
X23-A3 5 keV ~critical!
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
Rath et al.
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
7427
TABLE II. Glass polycapillary fibers. All fibers were 12.5 cm long.
Designation A B-1 B-2 B-3 B-4 C D
Batch/composition Early, sodium rich borosilicate Borosilicate Borosilicate Borosilicate Borosilicate Borosilicate Borosilicate
Fiber diameter ~mm!
Channel diameter ~mm!
Open area ~%!
Initial transmission ~%!
0.40
17
60
42
0.40 0.40 0.51 0.51 5.0 0.75
17 32 10 10 10 24
58 66–69 58–60 60–62 60–62 50
58 64 58 60 52–55 50
tion. Type B fibers were 0.4–0.5 mm in outer diameter. Type D had a slightly larger outer diameter and type C considerably larger. The resulting transmission efficiency degradation as a function of x-ray dose for the thin, type B fibers is summarized in Fig. 5. These fibers displayed severe bending, as is shown in Fig. 6. The higher energy white beam ~black solid circle! and 1.4 keV SRC type B-2 ~gray diamond! data in Fig. 5 represent single fibers. Each NSLS white beam data point and SRC type B-1 data point ~squares! are an average for four and three fibers, respectively. The in situ averages were performed by interpolation, since all in situ measurements of transmission were not necessarily performed at the same dose values. Some trends are apparent. The 11 keV white beam data ~solid circle! from CHESS, which has the highest intensity and highest average photon energy, agree well with the 5
keV white beam NSLS data. The 1.4 keV results ~squares and diamonds! from SRC, which have both the lowest intensity and lowest photon energy, showed minimal damage even at the highest doses and are distinctly different from those at the other two facilities. This could be due to a dependence on dose rate, photon energy, or exposure atmosphere. Note that this is despite the plotting of the data in terms of energy flux, not photon number flux. The in situ data taken at 1.4 keV showed no change even at nearly 1.3 MJ/cm2, although ex situ measurement of that fiber at 8 keV yielded a 12% decrease. Transmission efficiency at 1.4 keV is apparently less affected by the low energy beam-induced changes in the fiber than is transmission at 8 keV. This would be expected for transmission degradation, which is dominated by shape change, because of the larger critical angle at the lower energy. Ex situ measurements were also performed ~but not plotted! for the earlier, type A fibers at the 1.4 keV SRC beam. A decrease of 22% was seen at 680 kJ/cm2, which is a larger decrease, at a lower dose, than was seen for the type B fibers. 2. Rigid fibers
Transmission efficiency degradation is shown for the thick, type C, fibers in Fig. 7. These fibers are too rigid to be
FIG. 5. Relative transmission degradation as a function of x-ray dose for thin, type B fibers. ‘‘White’’ refers to data taken at NSLS with a white beam spectrum maximum at 5 keV. The solid circle is data taken at CHESS with a white beam spectrum maximum at 11 keV. The diamonds and squares are data taken at the 1.4 keV SRC beam. Open symbols are in situ data. The open circle, off-axis in situ data were deliberately misaligned during exposure and periodically realigned for measurement. Closed symbols are ex situ data measured in the laboratory after removal from the beam. All of the data except the 1.4 keV in situ open square data were taken at 8 keV. The starred data were taken at elevated temperature.
FIG. 6. Profile of a type B-4 fiber exposed at NSLS to 1.4 MJ/cm2 ~squares!. The solid line is a polynomcial fit. The first 2.5 cm of the fiber was unsupported. Note the difference in the X and Y scales.
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
7428
Rath et al.
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
FIG. 7. Relative transmission degradation as a function of x-ray dose for thick, type C fibers. Measurement is in situ at the NSLS white beam. The fibers exposed at ambient and 200 °C were 12.5 cm long. The fiber exposed at 100 °C was 5 cm long.
easily bent by the x-ray beam. The transmission degradation at ambient temperature is less than one fifth that of the type B fibers. 3. Transient effects
a. In situ measurements. As seen in Fig. 5, all of the white data taken at ambient temperatures are fairly consistent. This represents data taken during two different beam runs, including ex situ data ~solid symbols, taken after exposure, with an 8 keV lab source!, in situ data ~open symbols! taken with the fiber exposed while aligned ~open triangles!, and in situ data taken with the fiber rotated away from alignment between measurements ~open circles!. Agreement of the on-axis in situ and off-axis ~both in situ and ex situ! results implies that damage occurs along the whole length of the fiber even in the highly parallel synchrotron beam. The agreement of in situ and ex situ data also indicates that the effects are permanent, with no observable transient effects. Transient effects, if they existed, might have affected behavior of optics in high dose rate applications. b. Beam induced heating. The polycapillary fibers are subject to beam-induced heating. Temperature rise along the length of a thick, type C fiber was measured using an Everest 3000AH infrared temperature sensor. A temperature rise of about 190 °C at the front end of the fiber was recorded during exposure to a dose rate of 11.27 W/cm2 at NSLS. Detailed measurements and calculations of the synchrotron white beam thermal loading on the polycapillary optics are reported elsewhere.20,21
FIG. 8. Relative transmission degradation per 30 minutes as a function of x-ray dose rate for thin, type B-4 fibers.
calculation. The normalized transmission efficiency decrease for each time interval versus the computed dose rate is shown in Figs. 8 and 9 for the thin and thick fibers, respectively. There is no apparent dose rate dependence in the dose rate range illustrated. There is considerable scatter in the measured values, particularly for the thin fibers, but the best linear fit to the data is a horizontal line. The scatter is typical of these measurements. The data in Fig. 5 are averages. The plots in Figs. 8 and 9 show no dose rate dependence even though the data are for constant times rather than for constant doses. This would tend to give the appearance of a dose rate effect even when none exists. The average transmission decrease for the thin type B-4 fibers when exposed to a dose rate from 8.0 to 11.0 W/cm2 is found to be 1.93% per 30 minutes. For the thick type C fibers exposed to a dose rate of 8.0–13.0 W/cm2 the loss rate is 0.5% per 30 minutes. C. Thermal annealing studies
Thin fibers exposed to white beam synchrotron radiation bent and showed a substantial decrease in transmission effi-
4. Dose rate effects
Investigation of the dose rate effect on the transmission efficiency of the fibers is quite important for predicting performance of the optics in different applications. Since the synchrotron beam current varies with time, information on dose rate effects can be extracted. Doses for different time segments were calculated and were integrated for total dose
FIG. 9. Relative transmission degradation per 30 minutes as a function of x-ray dose rate for thick, type C fibers.
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
Rath et al.
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
7429
FIG. 11. Annealing time for recovery to 90% of the initial transmission vs anneal temperature for type B-4 fibers exposed to 1.4 MJ/cm2. The solid line is a linear fit.
FIG. 10. X-ray transmission of thin, type B-4 fibers unexposed ~circle! and exposed to 1.4 MJ/cm2 ~squares! after isothermal furnace annealing at ~a! 300, and ~b! 500 °C. Note the change in the horizontal scale between ~a! and ~b!.
ciency. However, furnace annealing caused recovery of the transmission efficiency to close to the pre-exposed value. In situ annealing, exposing the fibers at elevated temperatures, reduced the radiation effects on thick, rigid fibers. 1. Post-exposure annealing
Fibers were annealed inside a small fused silica tube which acted as a sample holder. To ensure neutral environment during annealing, a continuous argon gas flow was maintained inside the furnace. The fibers were put into the furnace for annealing only after a stable temperature was reached. To ensure thermal equilibrium in the fused silica tubing and of the fiber itself, a minimum annealing time of 10 minutes was set. A systematic annealing study was performed for type B-4 fibers exposed to 1.4 and 0.04 MJ/cm2. An unexposed fiber was annealed along with the exposed fiber for comparison. Annealing temperatures varied from 300 to 500 °C. Typical plots are shown in Figs. 10~a! and 10~b!. Most of the unbending happened in the first few minutes of annealing, which resulted in a quick initial recovery of the transmission. This initial recovery and the annealing
time depend on the dose and on the annealing temperature. At 300 °C fibers exposed to 1.4 MJ/cm2 took almost 12 days and nights of annealing to recover absolute transmission efficiency up to only 37%, whereas at the same temperature fibers exposed to 0.04 MJ/cm2 recovered almost fully in about 5 hours. After the quick initial recovery, the subsequent recovery is slow until the maximum recoverable limit is reached. The time required to recover 90% of the initial transmission efficiency for type B-4 fibers exposed to 1.4 MJ/cm2 as a function of temperature is shown in the Arrhenius plot in Fig. 11. The activation energy for defect removal is estimated from the slope of the curve and is found to be 1.4 eV. An activation energy ranging from 1.3 to 3.5 eV for recovering the original density of pure vitreous silica ~Vsilica! has been reported in the literature.22
2. In situ annealing
The success of the post-exposure annealing in removing radiation-induced bending led to a study of the effect of elevated temperature during exposure. The fiber holding plate was fitted with resistive heaters, thermocouples, and insulators, which allowed the temperature of the plate to be raised up to 400 °C. Because defects are known to be removed from silicon and SiO2 at much lower annealing temperatures during exposure than the temperature required to perform post-exposure anneals, a relatively low temperature was chosen. The thin fibers still experienced bending damage, as shown in the starred data in Fig. 5. However, a thick, rigid, type C fiber irradiated with the fiber holding plate at 100 °C showed no detectable transmission decrease at doses in excess of 900 kJ/cm2, as shown in Fig. 7. A polycapillary lens, used for a later focusing beam measurement,10 displayed no damage during the measurement, which amounted to an exposure of 500 kJ/cm2 at 200 °C and 150 kJ/cm2 at room temperature.
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
7430
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
Rath et al.
III. DISCUSSION AND ANALYSIS
As will be discussed in detail below, exposure to intense x-ray radiation results in a number of physical, chemical, and electronic changes in the glass. Changes were observed in the optical and electrical properties of the glass; however, these cannot explain the loss in x-ray transmission efficiency. No changes were observed in chemical composition, and again, even large changes in composition without concomitant changes in surface roughness would not be expected to significantly affect transmission. Surface roughness and reflectivity, however, were not affected by the radiation exposure. The only significant observable change in the glass polycapillary fibers was bending. If this is accompanied by a very slight increase in surface rippling, it is sufficient to explain the loss in transmission efficiency over the whole range from 8 to 80 keV. A. Changes in optical and electrical properties
1. Formation of color centers
Glass polycapillary fibers darkened in an exposure time of less than a minute. While less observable on thin fibers, the thick fibers rapidly became nearly black. The darkening occurred all along the length of the fiber, not just at the input end, which is consistent with the observation that there was damage along the full length of the fiber. Color centers are produced due to the trapping at defect sites of electrons which are liberated by the ionizing radiation. Fibers darkened at a much lower dose than any measurable transmission efficiency degradation. This is expected, since the index of refraction, as given by Eqs. ~1! and ~2!, and therefore x-ray reflectivity, depend only on the total electron density, which the development of color centers is not expected to change. The color centers can be removed by thermal annealing, and fibers exposed at elevated temperatures did not darken.
FIG. 12. SEM picture of an unexposed ~top! and an exposed ~bottom! type B-4 fiber. The unexposed fiber had a lower electrical conductivity and demonstrated larger charging effects.
2. Changes in electrical conductivity
During the SEM measurements, the unexposed fibers displayed charging effects to a much higher degree than the exposed fibers, causing the image to blur substantially, as can be seen in Fig. 12. This suggests an increase in the electrical conductivity of the exposed fiber compared to the unexposed fiber. To verify this, a simple four point probe conductivity measurement was performed. The results showed an increase in the conductivity of the exposed fibers by about an order of magnitude in comparison to the conductivity of the unexposed fibers. However, the x-ray reflectivity depends only on the total electron density and does not depend on the electrical conductivity of the glass surface. Hence, a change in electrical conductivity alone is not expected to change the transmission efficiency of the fibers. B. Changes in chemical composition and density
1. RBS and PIXE
RBS and PIXE studies were used to measure the chemical composition of the capillary surfaces. Since RBS is sensitive to high Z elements and PIXE is sensitive to low Z elements, a combination of both is expected to give reasonable information regarding the surface chemical composition
of the exposed and the unexposed fibers. Thin, type B-4 fibers were used. The superimposed spectra of the unexposed and exposed ~1.4 MJ/cm2) fibers from the RBS studies are shown in Fig. 13. A comparison of the RBS spectra for the unexposed and the exposed fibers does not show any noticeable movement of oxygen, aluminum, potassium, lead, or silicon. The dashed line in Fig. 13 is a RUMP’90 simulation for a 110 nm thick surface layer Si0.55O2.01B0.4Li0.4K0.09Na0.05Al0.05Pb0.014C0.15 over bulk material of Si0.60O2.06B0.3Li0.4K0.09Na0.05Al0.05Pb0.004 . The presence of carbon in the surface layer is contamination, as is the 1.4% lead. To determine the sensitivity of the technique, the simulated data were used to change the concentrations of different elements. The solid line in Fig. 14 is a simulation with an increase in the concentration of potassium by 20.0%. It clearly shows that such a change should be detectable by RBS. The sensitivity of RBS for different elements in the glass sample was determined and the results are summarized in Table III. A similar comparison for PIXE spectra, shown in Fig. 15, also does not show a significant change in the surface concentration of sodium, aluminum, silicon, or potassium.
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
Rath et al.
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
7431
TABLE III. RBS sensitivity for different elements in the glass sample.
Element
Sensitivity ~at. %!
Si O Al K Na B
9.0 3.0 1.7 20.0 315.0 261
the x-ray reflectivity. The x-ray transmission simulation program was used to calculate the transmission efficiency of a borosilicate glass polycapillary for a density increase of 3.0%. The results are shown in Table V. An increase in density by 3.0% is too small to show any measurable gain in transmission efficiency. FIG. 13. Superimposed RBS spectra for a type B-4 fiber, unexposed ~circles! and exposed at 1.4 MJ/cm2 ~crosses!. The dashed line is a RUMP’90 simulation.
2. X-ray transmission simulation
A geometrical optics, ray tracing x-ray transmission simulation23 was performed to calculate the effect of change in chemical composition of the glass polycapillary material on its transmission efficiency. The results are shown in Table IV. Simulation results suggest that even pure SiO2 , pure B2O3 , pure K2O, pure Na2O, and pure Al2O3 would have the same transmission efficiency at 8 keV as that of the currently used borosilicate glass. Thus, a change in chemical composition is not responsible for the decrease in transmission efficiency of the polycapillaries. The density of glass is known to increase upon irradiation. Maximum increases in density up to 3.0% have been reported.24 An increase in density could be beneficial for the polycapillary fiber transmission efficiency, since it increases
FIG. 14. RBS spectrum for an exposed type B-4 fiber. The dashed line is the simulation of Fig. 13. The solid line simulation has 20% more potassium in the surface layer.
C. Changes in surface reflectivity and roughness
1. AFM studies
Glass surfaces exposed to radiation experience chemical, physical, and topological changes which depend on the type of radiation and the amount of exposure dose. These effects may increase the roughness of the glass surface which could result in loss of transmission efficiency. AFM tests were done to measure the roughness of the capillary channel before and after exposure and annealing. Type B-4 fibers were used for all the AFM studies. The fiber was cleaved open to expose the interior channel surface for the AFM scan. Care was taken to make the measurement at the input end of the exposed fibers, where the radiation effects should be maximum. The results are summarized in Table VI. No clear trend is visible. Measured roughnesses for all samples varied within a range of 0.4 nm. Surprisingly, the unexposed fiber, with the highest transmission efficiency, was found to have a larger roughness than the other samples. This is contrary to the initial assumption that the fibers exposed to higher dose should have the highest roughness
FIG. 15. Superimposed PIXE spectra for a type B-4 fiber unexposed ~open circles! and exposed at 1.4 MJ/cm2 ~solid line!.
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
7432
Rath et al.
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998 TABLE IV. Simulated transmission efficiency of polycapillary fibers for different compositions. Simulated transmission efficiency of polycapillaries ~%! Composition Pure Pure Pure Pure Pure
SiO2 B2O3 K2O Na2O Al2O3
8.0 keV
9.6 keV
15.0 keV
20.0 keV
30.0 keV
61.07 61.94 58.9 61.64 61.88
61.69 61.94 60.26 61.57 61.69
61.44 61.88 60.26 61.75 61.69
61.63 61.94 61.01 61.94 61.88
61.94 62.0 61.57 61.81 61.88
value. All the data presented in Table VI are for single fibers with measurements done only on one segment, so the variation may be largely statistical. AFM measurements were done over an area of 4 m m2 with a resolution of 0.1 nm. 2. SEM studies
To observe any possible microdamage such as cracking of the surface or collapsing of the channel walls, a number of type B-4 fibers were observed in the scanning electron microscope. The reported sizes of oxygen bubbles and the cracks formed in V-silica due to irradiation are in the range of micrometers.17 In the SEM measurements, no deformation of the capillary channel walls due to either oxygen bubble formation or due to surface cracking was observed. There was also no evidence of capillary wall collapse due to thermal load. The resolution at which the SEM observations were performed was about 10 nm. Similarly, none of these damage features were observed in the AFM measurements. 3. X-ray transmission spectra and source scan studies
Any change in the surface roughness, chemical composition, and/or shape of the fiber might be expected to have a TABLE V. Simulated fiber transmissions for different glass densities and photon energies. Energy ~keV! 8.0 9.6 15 20 30 40 50 59 68 75 80
Density
Transmission ~%!
2.5 2.575 2.5 2.575 2.5 2.575 2.5 2.575 2.5 2.575 2.5 2.575 2.5 2.575 2.5 2.575 2.5 2.575 2.5 2.575 2.5 2.575
61.19 61.32 61.57 61.63 61.81 61.69 61.94 61.94 61.75 61.75 61.81 61.88 61.81 61.57 61.01 60.76 56.79 57.78 55.55 55.56 51.15 51.58
noticeable impact on the transmission efficiency of the fiber for different x-ray energies. Higher energy x rays are more sensitive to the deformation in the fiber shape because of their smaller critical angle. A number of type C ~thick! and type B-4 ~thin! fibers were used to compare the transmission efficiencies of the unexposed, exposed, and annealed fibers for photon energies up to 80 keV. The transmission efficiency versus the x-ray energy for type C unexposed and exposed ~dose of 1.8 MJ/cm2 in ambient temperature! fibers is plotted in Fig. 16. The transmission for the unexposed fiber is fairly constant up to 40 keV and drops for higher energies. The transmission of the exposed fiber increases up to 15 keV and then falls off rapidly. At 80 keV a transmission efficiency of 14% was measured for the exposed fiber and 34% for an unexposed fiber. To determine whether the loss in transmission was due to blocked or bad channels or was reduced uniformly across the fiber, the pinhole, which was about 250 m m in diameter, was scanned across the fiber, keeping the source, fiber, and the detector at the aligned position. A variation of not more than 5%–8% in the transmission efficiency was measured. Source scans, which are taken by measuring the transmission efficiency of the fiber as the source is moved off axis, give significant information about the reflectivity of the capillary surface because the number of reflections increases rapidly as the angle to the source is increased. To compare the reflectivity of the unexposed fiber with that of the exposed fiber, a number of source scans at energies of 9.6, 20, 30, and 40 keV were taken and are shown in Fig. 17. No significant difference between the normalized scan curves for unexposed and exposed fibers is seen even though there was an appreciable difference in the absolute transmis-
TABLE VI. Capillary channel surface roughness measurements for five exposed and unexposed fibers taken with AFM. Capillary type B-4 B-4 B-4 B-4 B-4
Capillary history Unexposed Unexposed, annealed at 450 °C for 2.5 h Exposed to a dose of 0.08 MJ/cm2 Exposed to a dose of 1.4 MJ/cm2 Exposed to a dose of 1.4 MJ/cm2, annealed at 450 °C for 2.5 hours
Transmission efficiency ~%!
Roughness ~Å!
60 57
8.1 4.7
45.8
4.4
15
5.7
54.2
6.6
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
Rath et al.
7433
FIG. 16. Transmission efficiency of type C fibers vs photon energy, unexposed ~open squares! and exposed to 1.8 MJ/cm2 ~solid squares!.
sion efficiency. The transmission values with the source displaced are the result of a very large number of reflections, and so the width of the curves cannot be unchanged if the surface reflectivity has decreased. The lack of change in the source scan implies that the surface cannot have roughened appreciably. Transmission efficiency and the source scan measurements for the thin, type B-4 fibers are shown in Figs. 18 and 19, respectively. The unexposed type B-4 fiber showed significantly less transmission at high energies than the unexposed type C fiber. It was difficult to keep the type B-4 fiber straight in the measurement groove because it was thin and flexible. Since, as noted in Eq. ~3!, higher energy photons have smaller critical angles than lower energy photons, higher energy transmission is more sensitive to fiber bend-
FIG. 17. Source scans of transmission vs transverse x-ray source displacement at a series of x-ray measurement energies of an unexposed, thick, type C fiber ~open symbols! and a type C fiber exposed to a dose of 1.8 MJ/cm2 in ambient temperature ~closed symbols!.
FIG. 18. Transmission efficiency of thin, type B-4 fibers vs energy. Open symbols are unexposed; closed symbols are exposed. Circles are unannealed; squares and diamonds are annealed.
ing. The unexposed annealed fiber showed an additional transmission decrease above 30 keV. Since the annealing temperature, 400 °C, is far below the melting temperature of the borosilicate glass, about 570 °C, gross disorders in the structure of the glass, such as the collapse of channel walls, are not expected. However, there may be an introduction of bending due to annealing stress.
FIG. 19. Source scans of thin, type B-4 fibers for a series of different x-ray measurement energies. Squares are 9.6, circles are 20, diamonds are 30, and triangles are 40 keV. Unexposed fibers are represented by open symbols, unexposed fibers annealed at 400 °C by crossed symbols, and fibers exposed to 1.4 MJ/cm2 and then annealed up to 450 °C are shown by solid symbols.
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
7434
Rath et al.
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
FIG. 20. Transmission vs energy measurements for thick, type C fibers, unexposed ~open symbols! and exposed ~solid symbols!. Simulations, which include bending curvature, channel wall tilt ~waviness!, surface roughness, and a thin layer of glass in the channels ~overlayer! are also plotted. The best fit simulations are the thin solid line ~unexposed! and the thick solid line ~exposed!.
These thin fibers, when exposed to high radiation doses, displayed a sharp drop in transmission efficiency above 9 keV, even after furnace annealing. The high temperature annealing technique recovers the transmission efficiencies of the exposed fibers for lower energies only. As for the thick, type C, fiber source scans, the source scans for the thin, type B-4, fibers, shown in Fig. 19, did not show any significant difference in width after exposure or annealing, again suggesting the absence of any change to the surface reflectivity. D. Changes in profile: Bending and waviness
Severe bending was observed in the case of the thin, type B, fiber exposed to white beams. The profile of such a fiber is shown in Fig. 6. Assuming a two dimensional bending ~in reality there was three dimensional bending! and fitting the data with the curves shown as solid lines in Fig. 6, the x-ray transmission simulation through the fiber was calculated for the shape given by the fitting parameters. The simulation demonstrated that most of the transmission degradation could be accounted for by bending. Unbending of the fibers was observed after annealing, and was less severe for the thicker fibers. In addition to the obvious profile change, there might have been shorter wavelength changes as well. The shortest wavelength changes are described as roughness.25 In addition, mid-frequency ‘‘waviness,’’ or surface rippling, would increase the average angle of incidence and could cause further loss. The results of detailed simulation analyses26 are shown in Fig. 20. The top two simulation lines show the best fit to the unexposed fibers for simulations which include only bending and only bending and roughness. The correlation length for the roughness was set at 6 m m. Changing the correlation length alters the absolute value of the roughness heights. The thick solid line includes the effects of bending, roughness, and waviness. Both bending and waviness are
required to fit the data. The lower points and lines are data and simulations for a fiber exposed to 1.8 MJ/cm2. The radius of curvature of the unexposed fiber was 81 m, while the radius for the exposed fiber was about 50 m. This implies that even the thick type C fiber underwent slight bending, contrary to our previous assumption that it did not. However, this slight bending does not affect the transmission efficiency of the fibers for low energies (;8 keV!. The reasonably good fit of the simulation with the data for both exposed and unexposed fiber leads to the conclusion that there has not been significant surface roughness induced in the fibers due to exposure. The profile change, including both bending and waviness, is sufficient to explain the data for higher energies. The decrease in transmission efficiency of the exposed fiber for energies less than 15 keV could be explained by the addition of a 35 m m glass absorbing layer. The possible origin of glass particulates in the capillary channels is not understood. The growth of small crystallites might result in partial channel blockage, but was not observed in SEM analysis. The sharp decrease in transmission for x rays of energy less than 9 keV may be due to some other mechanism. The x-ray simulation program results show that the majority of the transmission loss is due to bending of the fibers. There are two probable causes for bending. These are stress introduced by the heat gradient and/or stress due to a density gradient developed in the fiber. Thermal modeling21 shows a negligible temperature rise in the thinner type B-4 fibers. Also, had thermal load been responsible for bending, maximum bending would have been seen when the beam intensity was maximum, i.e., at the maximum beam current. This should have been reflected in a sharp decrease in transmission in the in situ measurements at the time of maximum beam current, which was not observed. Also, there should have been less dependence of the transmission decrease on total dose and more on dose rate. The absence of such behavior suggests that the stress induced in the fiber is not due to thermal load. The in situ and ex situ data for type B-4 fibers follow pretty much the same trend, which also supports the conclusion that the bending mechanism is athermal. This leads to the assumption that the bending is due to the stress caused by a density gradient developed due to asymmetric exposure of the fiber. Annealing effectively removes the irradiation-induced stress and the fibers straighten spontaneously. An activation energy ranging from 1.3 to 3.5 eV for recovering the original density of pure V-silica has been reported in the literature.22 Our measurements show an activation energy of 1.4 eV for recovery of 90% of the initial transmission efficiency, which is in agreement with the assumption that the bending is due to density gradients developed in the fiber due to asymmetric exposure. IV. CONCLUSION
In situ and ex situ transmission efficiency degradation measurements of fibers of varying channel diameter and outer diameter have been performed for focused synchrotron 1.4 keV radiation and intense synchrotron bending magnet white radiation doses up to 1.8 MJ/cm2 at ambient and
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
Rath et al.
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
elevated temperatures. In situ measurements at 1.4 keV showed no detectable decrease in the transmission efficiency of the fibers when exposed to doses to 1.3 MJ/cm2. When exposed to the higher energy white beam lines, thinner fibers underwent bending due to exposure, resulting in severe transmission efficiency loss. This transmission efficiency loss, measured at 8 keV, was recoverable by post-exposure furnace annealing. A much smaller linear decrease was measured when thick, rigid, fibers were exposed at ambient temperature. When the thick fibers were exposed at elevated temperatures, there was no decrease in transmission efficiency. No transient effect was found during the exposure and transmission efficiency loss was found not to depend on dose rates from 8 to 13 W/cm2. RBS and PIXE studies suggest that the chemical composition of the fibers did not undergo a noticeable change due to exposure up to 1.4 MJ/cm2. Simulation results indicate that a change in chemical composition is not expected to change the transmission efficiency of the fibers. Similarly, electronic changes can explain the transmission loss. Systematic measurements of transmission as a function of angle and energy show that the surfaces of the capillaries have not roughened. Bending and surface rippling is indicated by the simulation. Roughening is also not seen in the AFM measurements. The transmission loss for high energy photons was solely driven by the change in the profile of the fibers, which is believed to be due to asymmetric stress due to irradiationinduced density gradients in the fibers.
ACKNOWLEDGMENTS
This work was partially supported by the U.S. Department of Commerce through the Advanced Technology Program under corporate Agreement No. 70NANB2H1250, by the National Institutes of Health under Grant No. R01 CA58521, and by the New York State Center for Advanced
7435
Technology. The authors are grateful for AFM measurements provided by C. M. Dozier, P. G. Burkhalter, and J. V. Gilfrich. L. G. Parratt, Rev. Sci. Instrum. 30, 297 ~1959!. G. Rosenbaum, J. Appl. Crystallogr. 7, 117 ~1974!. 3 T. W. Barbee, Jr., Proc. SPIE 563, 2 ~1985!. 4 G. N. Greaves and C. R. A. Catlow, in Applications of Synchrotron Radiation ~Chapman and Hall, New York, 1990!. 5 C. A. MacDonald, J. X-Ray Sci. Technol. 6, 32 ~1996!. 6 J. X. Ho, D. C. Carter, J. R. Ruble, R. C. Sisk, E. H. Snell, S. M. Owens, and W. M. Gibson, Acta Crystallogr., Sect. D: Biol. Crystallogr. 54, 200 ~1998!. 7 W. M. Gibson, M. V. Gubarev, F. A. Hofmann, M. K. Joy, C. A. MacDonald, S. L. O’Dell, B. D. Ramsey, C. Russell, L. Wang, and M. C. Weisskopf, Proc. SPIE 3113, ~1997!. 8 M. Vartanian, D. Gibson, R. D. Frankel, and J. Drumheller, Proc. SPIE 335, 1924 ~1993!. 9 D. G. Kruger, C. C. Abreu, E. G. Hendee, A. Kocharian, W. W. Peppler, C. A. Mistretta, and C. A. MacDonald, Med. Phys. 23, 187 ~1996!. 10 F. A. Hofmann, C. A. Freinberg-Trufas, S. M. Owens, S. D. Padiyar, and C. A. MacDonald, Nucl. Instrum. Methods Phys. Res. B 133, 145 ~1997!. 11 C. B. Norris and E. R. Eernisse, J. Appl. Phys. 45, 3876 ~1974!. 12 W. Primak and R. Kampwirth, J. Appl. Phys. 39, 5651 ~1968!. 13 W. Primak, Radiat. Eff. 45, 29 ~1979!. 14 C. Fiori and R. A. B. Devive, Mater. Res. Soc. Symp. Proc. 61, 1 ~1986!. 15 S. Thomas, J. Appl. Phys. 45, 161 ~1974!. 16 O. Puglisi, G. Marletta, and A. Torrissi, J. Non-Cryst. Solids 83, 344 ~1984!. 17 J. F. Denatale and D. G. Howitt, Nucl. Instrum. Methods Phys. Res. B 1, 489 ~1984!. 18 A. A. Kravchenko, Izv. Vyssh. Uchebn. Zaved. Fiz., 14, No. 4 ~1990!. 19 B. K. Rath, R. Youngman, and C. A. MacDonald, Rev. Sci. Instrum. 65, 3393 ~1994!. 20 B. K. Rath, D. C. Aloisi, D. H. Bilderback, N. Gao, W. M. Gibson, F. A. Hofmann, B. E. Homan, C. J. Jezewski, I. L. Klotzko, J. M. Mitchell, S. M. Owens, J. B. Ullrich, L. Wang, G. M. Wells, Q. F. Xiao, and C. A. MacDonald, Proc. SPIE 2519, 207 ~1995!. 21 B. K. Rath, F. B. Hagelberg, B. E. Homan, and C.A. MacDonald, Nucl. Instrum. Methods Phys. Res. A 401, 421 ~1997!. 22 W. Primak and H. Szymanski, Phys. Rev. 101, 1268 ~1956!. 23 Q. F. Xiao, I. Yu. Ponomarev, A. I. Kolomitsev, and J. C. Kimbal, Proc. SPIE 1736, 227 ~1992!. 24 E. Lell, N. J. Kreidl, and J. R. Hensler, Prog. Ceram. Sci. 4, 1 ~1966!. 25 J. C. Kimball and D. Bittel, J. Appl. Phys. 74, 877 ~1993!. 26 L. Wang, B. K. Rath, W. M. Gibson, J. C. Kimball, and C. A. MacDonald, J. Appl. Phys. 80, 3628 ~1996!. 1 2
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
VOLUME 83, NUMBER 12
15 JUNE 1998
Measurements and analysis of radiation effects in polycapillary x-ray optics B. K. Rath Wadsworth Center, New York State Department of Health, Albany, New York 12201-0509
Lei Wang, B. E. Homan, F. Hofmann, W. M. Gibson, and C. A. MacDonalda) Center for X-ray Optics, University at Albany, State University of New York, Albany, New York 12222
~Received 1 December 1997; accepted for publication 2 March 1998! Polycapillary x-ray optics are arrays of large numbers of small hollow glass tubes which deflect x rays by successive total external reflection. These optics have growing numbers of applications in areas ranging from medical imaging to microanalysis. An accelerated radiation effects study has been performed to understand the performance limitation of these optics for medium to high intensity radiation applications, to study x-radiation damage mechanisms, and to investigate possible ways to mitigate the radiation effects on x-ray transmission efficiency. Exposures have been done in white beam bending magnet radiation with peak energies at 5 and 11 keV and focused broad band radiation centered at 1.4 keV. In situ and ex situ measurements of loss of x-ray transport efficiency have been executed at doses up to 1.8 MJ/cm2. Thin polycapillary fibers displayed noticeable bending and experienced substantial degradation of x-ray transmission. Thicker polycapillary fibers showed a linear but much slower transmission loss as a function of total dose. Annealing effectively restored the low energy ~;8 keV! transmission efficiency of the fibers. Exposure of these fibers at slightly elevated temperatures prevented any measurable loss in the low energy transmission efficiency. A variety of analytical techniques has been used to understand these results. No significant change was observed in the chemical composition of the capillary surface. Profile measurements and high energy transmission efficiency spectra, along with computer simulation studies, suggest that radiation induced bending is the primary cause of transmission efficiency degradation of the fibers. © 1998 American Institute of Physics. @S0021-8979~98!04311-4#
I. INTRODUCTION
has necessitated a detailed investigation of the radiation effects on the performance of these optics. Capillary x-ray optics use multiple total reflections to guide grazing incidence x rays. X-ray photon energies are much larger than the electron plasma energies of glasses, which are tens of electron volts. In this regime, the index of refraction of glass can be simply approximated by
Because the index of refraction of all materials in the x-ray region is slightly less than unity, x rays are barely deviated by optical lenses. This makes the task of controlling x rays different and more difficult than controlling visible light. However, there are a number of ways in which x rays can be steered. Plane crystals1 can be used for producing a monochromatic parallel beam and curved crystals2 for producing a focused beam due to Bragg reflection. Multilayers3 are also used for controlling x rays. Parabolic mirrors4 are used for obtaining a parallel beam and ellipsoidal mirrors4 for a focused beam. However, most of these optics are either spectrally selective or have a small acceptance angle. Polycapillary optics have relatively large angular acceptance and a spectral bandwidth. A typical polycapillary fiber is shown in Fig. 1. Successive reflections from the inner surface of the glass tubes deflect the incident photons by several degrees, making it possible to steer x rays over a large angular range. Polycapillary x-ray optics have shown great potential for use in x-ray diffraction,5 protein crystallography,6 x-ray astronomy,7 x-ray lithography,8 scatter rejection in mammography9 and x-ray focusing for microanalysis applications.10 The possibility of using polycapillary optics in these moderate to high intensity applications
v 2p « n 5 >12 2 , «0 v 2
where n is the index of refraction, « is the dielectric constant of the glass, v is the photon frequency, « 0 is the dielectric constant of vacuum, and v p is the plasma frequency of the material,
v p5
Nq 2 mP o
~2!
where N is the total electron density, and q and m are the charge and mass of the electron. The critical angle for total external reflection is given by
u c>
vp . v
~3!
Radiation effects in glass are many and are complicated. Glass surfaces exposed to radiation are subject to changes in their optical, electrical, and physical properties. Compaction
a!
Electronic mail: [email protected]
0021-8979/98/83(12)/7424/12/$15.00
~1!
7424
© 1998 American Institute of Physics
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
Rath et al.
7425
FIG. 2. Setup for measuring x-ray transmission efficiency of polycapillary fibers.
FIG. 1. SEM micrograph of a polycapillary fiber cross section. The fiber is approximately 0.5 mm flat to flat. Capillary channels are about 50 m m in diameter.
of glass due to neutron irradiation is more apparent, but ionization compaction in glass has also been observed due to exposure of electrons with energy as low as 2 keV,11 exposure to gamma rays,12 low energy x rays13 and ultraviolet ~UV! light.14 The compaction efficiency of x rays appears to be approximately half of that of electrons. Additionally, changes in chemical composition are also expected. These include desorption of oxygen,15 and alkali metal segregation.16 Formation of oxygen bubbles in the glass matrix due to ionizing radiation is also reported17 and the bubbles might roughen the glass surface. Surface cracking has also been observed due to electron irradiation.18 All of these radiation induced effects may decrease the surface reflectivity of the glass. Since transmission through fibers depends on the reflectivity to the nth power, where n is the number of reflections, even a slight decrease in reflectivity can be highly damaging to polycapillary optic transmission. Finally, since x rays absorbed by matter are converted to heat, large thermal gradients can be a problem for x-ray optics. In a polycapillary optic, photons incident on the open channels with an angle more than the critical angle, and those incident on the face wall of the optic, will be absorbed. Possible changes of the topology of the optic surface could be highly detrimental to the optic performance. Accelerated exposure studies in a bending magnet white beam and focused low energy beams have been performed. To investigate the resulting changes, a number of analytical techniques have been used. These include x-ray transmission spectra and source scan studies, atomic force microscopy ~AFM!, Rutherford backscattering ~RBS!, particle induced x-ray emission ~PIXE! studies, Raman spectroscopy, and scanning electron microscopy ~SEM!. II. EXPERIMENTAL OBSERVATIONS
Thin, flexible polycapillary fibers, when exposed to high intensity white beam synchrotron radiation, experienced substantial bending and resultant transmission degradation.
These fibers were observed to uncurl spontaneously upon post-exposure thermal annealing. Thicker, more rigid, fibers were more robust. In situ annealing of the thicker fibers, i.e., by placing the fibers on hot stages during exposure, prevented most of the radiation damage. A. Experimental apparatus and measurement technique
The setup used for most of the in situ and ex situ exposure measurements at the synchrotron radiation sources is shown in Fig. 2. The fiber holding plate was a half-inch thick aluminum plate on which was machined a fine straight groove. The plate was fitted with two 100 W cartridge heaters to control its temperature. The pinhole was mounted in such a way that it could be moved in horizontal and vertical directions with respect to the fiber, allowing positioning of the pinhole at any desired point of the fiber cross section. The pinhole and the plate were mounted together on a tilt/ rotation stage to avoid the relative motion of the pinhole with respect to the fiber during the alignment of the fiber with the synchrotron beam. Iron fillings were used to block x rays passing around the fiber. The fiber plate could be moved vertically with respect to the pinhole and hence can be moved out of the way of the pinhole for measuring the direct current output from the detector. The pinhole was smaller than the fiber, so that the transmission efficiency was the ratio of the photon flux with the fiber in place to that without the fiber. The pinhole was mounted on the down stream end of the fiber so that the fiber was fully exposed to the radiation, while only a small section of the beam passing through the fiber was taken for transmission efficiency measurements. This helped in reducing the beam current and resultant potential nonlinear response from the detector. A Si~111! crystal was mounted beyond the end of the fiber and positioned to reflect 8 keV photons. Although the fiber was exposed to the broad spectrum of white beam, only the change in transmission efficiency at 8 keV ~Cu K a energy! was measured. The use of the crystal also helped in reducing the potential non-linear response of the detector to an intense beam. The detector was positioned in the reflected beam from the crystal and could be moved in both horizontal and vertical directions. All the linear stages were fitted with actuators and were controlled by a Newport motion controller.
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
7426
Rath et al.
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
FIG. 3. In situ normalized 8 keV transmission as a function of the angle for a type B-4 polycapillary fiber at NSLS. The solid line is a Gaussian fit with FWHM53.8 mrad.
A small portable low-power laser was mounted on a compound stage, which gave all the degrees of freedom needed for alignment and was set behind the crystal. In measurements at Cornell University High Energy Synchrotron Source ~CHESS! and at Brookhaven National Laboratory National Synchrotron Light Source ~NSLS!, ion counters were used for detection of the x rays. The detector linearity versus total synchrotron beam current was verified. At the University of Wisconsin low energy Synchrotron Radiation Center ~SRC!, the beam was a broad-band focused beam centered at 1.38 keV, with a full width at half maximum of 0.8 keV. A photodiode was used as the detector and no crystal was used. Alignment of the fiber with the synchrotron beam was achieved in two stages. Initially, the laser beam was roughly aligned with marks made by an x-ray beam on two burn papers, one at the entrance window and the other 40 cm downstream. The fiber holding groove was then aligned with the laser and the polycapillary fiber was laid in the groove. Finally, fine alignment was done by maximizing the x-ray signal passing through the fiber. Verification of alignment and absence of leakage were done by tilt and rotation scans as shown in Fig. 3. The scan is symmetric, and the full width at half maximum ~FWHM! is about 4 mrad, approximately the critical angle of the glass. In situ measurements in the focused low energy beam at SRC were also verified by a similar technique. Vertical tilt and horizontal rotation scans
FIG. 4. In situ normalized 1.4 keV transmission as a function of the rotation angle ~squares! or tilt angle ~circles! of a type B-2 polycapillary fiber at SRC. The larger width of the rotation scan compared to the tilt scan is due to the difference in the horizontal and vertical divergence of the focused beam. The dotted line is a Gaussian fit with FWHM57.11 mrad. The solid line is a Gaussian fit with FWHM53 mrad.
are shown in Fig. 4. The larger width of the rotation scan is due to the difference in horizontal and vertical divergence of the focused beam. Ex situ exposures were done in a similar geometry but without precise alignment of the fibers with the beam. The fibers were stacked to fill the beam cross section. A small number of fibers was removed at regular time intervals to provide good statistics at a number of doses. Postexposure transmission measurements were done at an automated transmission measurement test system, which had been developed to measure the effects of radiation on transmission efficiency of the fibers reproducibly and quickly.19 The reproducibility of the measurements in the automated transmission measurement system is within 0.5%. B. Degradation of x-ray transmission efficiency
1. Flexible fibers
Exposure experiments were performed at the three synchrotron x-ray sources listed in Table I. For the white beam, the x-ray intensity was approximated by dividing the x-ray spectrum into bins, multiplying the photon flux in the bin by the photon energy at the center of the bin, and summing the bins. The beam currents were recorded during the experiment and integrated to determine total dose values. The fibers used for radiation exposure are described in Table II. Types B, C, and D were of the same composi-
TABLE I. Synchrotron radiation facilities used in this work.
Facility
Beam line
SRC
ES-2
CHESS
C-2
NSLS
X-ray energy
Characteristics
Atmosphere
Maximum beam current
Maximum x-ray intensity
1.38 keV ~centered at 1.38 and 0.8 keV FWHM! 11 keV ~critical!
Focused by a torroidal mirror to 2.533 mm
Helium
80 mA at 1 GeV
5.4 W/cm2 in the focal spot
White beam
Air
75 mA at 5.5 GeV
87 W/cm2
White beam
Air
250 mA at 2.5 GeV
13 W/cm2
X23-A3 5 keV ~critical!
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
Rath et al.
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
7427
TABLE II. Glass polycapillary fibers. All fibers were 12.5 cm long.
Designation A B-1 B-2 B-3 B-4 C D
Batch/composition Early, sodium rich borosilicate Borosilicate Borosilicate Borosilicate Borosilicate Borosilicate Borosilicate
Fiber diameter ~mm!
Channel diameter ~mm!
Open area ~%!
Initial transmission ~%!
0.40
17
60
42
0.40 0.40 0.51 0.51 5.0 0.75
17 32 10 10 10 24
58 66–69 58–60 60–62 60–62 50
58 64 58 60 52–55 50
tion. Type B fibers were 0.4–0.5 mm in outer diameter. Type D had a slightly larger outer diameter and type C considerably larger. The resulting transmission efficiency degradation as a function of x-ray dose for the thin, type B fibers is summarized in Fig. 5. These fibers displayed severe bending, as is shown in Fig. 6. The higher energy white beam ~black solid circle! and 1.4 keV SRC type B-2 ~gray diamond! data in Fig. 5 represent single fibers. Each NSLS white beam data point and SRC type B-1 data point ~squares! are an average for four and three fibers, respectively. The in situ averages were performed by interpolation, since all in situ measurements of transmission were not necessarily performed at the same dose values. Some trends are apparent. The 11 keV white beam data ~solid circle! from CHESS, which has the highest intensity and highest average photon energy, agree well with the 5
keV white beam NSLS data. The 1.4 keV results ~squares and diamonds! from SRC, which have both the lowest intensity and lowest photon energy, showed minimal damage even at the highest doses and are distinctly different from those at the other two facilities. This could be due to a dependence on dose rate, photon energy, or exposure atmosphere. Note that this is despite the plotting of the data in terms of energy flux, not photon number flux. The in situ data taken at 1.4 keV showed no change even at nearly 1.3 MJ/cm2, although ex situ measurement of that fiber at 8 keV yielded a 12% decrease. Transmission efficiency at 1.4 keV is apparently less affected by the low energy beam-induced changes in the fiber than is transmission at 8 keV. This would be expected for transmission degradation, which is dominated by shape change, because of the larger critical angle at the lower energy. Ex situ measurements were also performed ~but not plotted! for the earlier, type A fibers at the 1.4 keV SRC beam. A decrease of 22% was seen at 680 kJ/cm2, which is a larger decrease, at a lower dose, than was seen for the type B fibers. 2. Rigid fibers
Transmission efficiency degradation is shown for the thick, type C, fibers in Fig. 7. These fibers are too rigid to be
FIG. 5. Relative transmission degradation as a function of x-ray dose for thin, type B fibers. ‘‘White’’ refers to data taken at NSLS with a white beam spectrum maximum at 5 keV. The solid circle is data taken at CHESS with a white beam spectrum maximum at 11 keV. The diamonds and squares are data taken at the 1.4 keV SRC beam. Open symbols are in situ data. The open circle, off-axis in situ data were deliberately misaligned during exposure and periodically realigned for measurement. Closed symbols are ex situ data measured in the laboratory after removal from the beam. All of the data except the 1.4 keV in situ open square data were taken at 8 keV. The starred data were taken at elevated temperature.
FIG. 6. Profile of a type B-4 fiber exposed at NSLS to 1.4 MJ/cm2 ~squares!. The solid line is a polynomcial fit. The first 2.5 cm of the fiber was unsupported. Note the difference in the X and Y scales.
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
7428
Rath et al.
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
FIG. 7. Relative transmission degradation as a function of x-ray dose for thick, type C fibers. Measurement is in situ at the NSLS white beam. The fibers exposed at ambient and 200 °C were 12.5 cm long. The fiber exposed at 100 °C was 5 cm long.
easily bent by the x-ray beam. The transmission degradation at ambient temperature is less than one fifth that of the type B fibers. 3. Transient effects
a. In situ measurements. As seen in Fig. 5, all of the white data taken at ambient temperatures are fairly consistent. This represents data taken during two different beam runs, including ex situ data ~solid symbols, taken after exposure, with an 8 keV lab source!, in situ data ~open symbols! taken with the fiber exposed while aligned ~open triangles!, and in situ data taken with the fiber rotated away from alignment between measurements ~open circles!. Agreement of the on-axis in situ and off-axis ~both in situ and ex situ! results implies that damage occurs along the whole length of the fiber even in the highly parallel synchrotron beam. The agreement of in situ and ex situ data also indicates that the effects are permanent, with no observable transient effects. Transient effects, if they existed, might have affected behavior of optics in high dose rate applications. b. Beam induced heating. The polycapillary fibers are subject to beam-induced heating. Temperature rise along the length of a thick, type C fiber was measured using an Everest 3000AH infrared temperature sensor. A temperature rise of about 190 °C at the front end of the fiber was recorded during exposure to a dose rate of 11.27 W/cm2 at NSLS. Detailed measurements and calculations of the synchrotron white beam thermal loading on the polycapillary optics are reported elsewhere.20,21
FIG. 8. Relative transmission degradation per 30 minutes as a function of x-ray dose rate for thin, type B-4 fibers.
calculation. The normalized transmission efficiency decrease for each time interval versus the computed dose rate is shown in Figs. 8 and 9 for the thin and thick fibers, respectively. There is no apparent dose rate dependence in the dose rate range illustrated. There is considerable scatter in the measured values, particularly for the thin fibers, but the best linear fit to the data is a horizontal line. The scatter is typical of these measurements. The data in Fig. 5 are averages. The plots in Figs. 8 and 9 show no dose rate dependence even though the data are for constant times rather than for constant doses. This would tend to give the appearance of a dose rate effect even when none exists. The average transmission decrease for the thin type B-4 fibers when exposed to a dose rate from 8.0 to 11.0 W/cm2 is found to be 1.93% per 30 minutes. For the thick type C fibers exposed to a dose rate of 8.0–13.0 W/cm2 the loss rate is 0.5% per 30 minutes. C. Thermal annealing studies
Thin fibers exposed to white beam synchrotron radiation bent and showed a substantial decrease in transmission effi-
4. Dose rate effects
Investigation of the dose rate effect on the transmission efficiency of the fibers is quite important for predicting performance of the optics in different applications. Since the synchrotron beam current varies with time, information on dose rate effects can be extracted. Doses for different time segments were calculated and were integrated for total dose
FIG. 9. Relative transmission degradation per 30 minutes as a function of x-ray dose rate for thick, type C fibers.
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
Rath et al.
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
7429
FIG. 11. Annealing time for recovery to 90% of the initial transmission vs anneal temperature for type B-4 fibers exposed to 1.4 MJ/cm2. The solid line is a linear fit.
FIG. 10. X-ray transmission of thin, type B-4 fibers unexposed ~circle! and exposed to 1.4 MJ/cm2 ~squares! after isothermal furnace annealing at ~a! 300, and ~b! 500 °C. Note the change in the horizontal scale between ~a! and ~b!.
ciency. However, furnace annealing caused recovery of the transmission efficiency to close to the pre-exposed value. In situ annealing, exposing the fibers at elevated temperatures, reduced the radiation effects on thick, rigid fibers. 1. Post-exposure annealing
Fibers were annealed inside a small fused silica tube which acted as a sample holder. To ensure neutral environment during annealing, a continuous argon gas flow was maintained inside the furnace. The fibers were put into the furnace for annealing only after a stable temperature was reached. To ensure thermal equilibrium in the fused silica tubing and of the fiber itself, a minimum annealing time of 10 minutes was set. A systematic annealing study was performed for type B-4 fibers exposed to 1.4 and 0.04 MJ/cm2. An unexposed fiber was annealed along with the exposed fiber for comparison. Annealing temperatures varied from 300 to 500 °C. Typical plots are shown in Figs. 10~a! and 10~b!. Most of the unbending happened in the first few minutes of annealing, which resulted in a quick initial recovery of the transmission. This initial recovery and the annealing
time depend on the dose and on the annealing temperature. At 300 °C fibers exposed to 1.4 MJ/cm2 took almost 12 days and nights of annealing to recover absolute transmission efficiency up to only 37%, whereas at the same temperature fibers exposed to 0.04 MJ/cm2 recovered almost fully in about 5 hours. After the quick initial recovery, the subsequent recovery is slow until the maximum recoverable limit is reached. The time required to recover 90% of the initial transmission efficiency for type B-4 fibers exposed to 1.4 MJ/cm2 as a function of temperature is shown in the Arrhenius plot in Fig. 11. The activation energy for defect removal is estimated from the slope of the curve and is found to be 1.4 eV. An activation energy ranging from 1.3 to 3.5 eV for recovering the original density of pure vitreous silica ~Vsilica! has been reported in the literature.22
2. In situ annealing
The success of the post-exposure annealing in removing radiation-induced bending led to a study of the effect of elevated temperature during exposure. The fiber holding plate was fitted with resistive heaters, thermocouples, and insulators, which allowed the temperature of the plate to be raised up to 400 °C. Because defects are known to be removed from silicon and SiO2 at much lower annealing temperatures during exposure than the temperature required to perform post-exposure anneals, a relatively low temperature was chosen. The thin fibers still experienced bending damage, as shown in the starred data in Fig. 5. However, a thick, rigid, type C fiber irradiated with the fiber holding plate at 100 °C showed no detectable transmission decrease at doses in excess of 900 kJ/cm2, as shown in Fig. 7. A polycapillary lens, used for a later focusing beam measurement,10 displayed no damage during the measurement, which amounted to an exposure of 500 kJ/cm2 at 200 °C and 150 kJ/cm2 at room temperature.
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
7430
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
Rath et al.
III. DISCUSSION AND ANALYSIS
As will be discussed in detail below, exposure to intense x-ray radiation results in a number of physical, chemical, and electronic changes in the glass. Changes were observed in the optical and electrical properties of the glass; however, these cannot explain the loss in x-ray transmission efficiency. No changes were observed in chemical composition, and again, even large changes in composition without concomitant changes in surface roughness would not be expected to significantly affect transmission. Surface roughness and reflectivity, however, were not affected by the radiation exposure. The only significant observable change in the glass polycapillary fibers was bending. If this is accompanied by a very slight increase in surface rippling, it is sufficient to explain the loss in transmission efficiency over the whole range from 8 to 80 keV. A. Changes in optical and electrical properties
1. Formation of color centers
Glass polycapillary fibers darkened in an exposure time of less than a minute. While less observable on thin fibers, the thick fibers rapidly became nearly black. The darkening occurred all along the length of the fiber, not just at the input end, which is consistent with the observation that there was damage along the full length of the fiber. Color centers are produced due to the trapping at defect sites of electrons which are liberated by the ionizing radiation. Fibers darkened at a much lower dose than any measurable transmission efficiency degradation. This is expected, since the index of refraction, as given by Eqs. ~1! and ~2!, and therefore x-ray reflectivity, depend only on the total electron density, which the development of color centers is not expected to change. The color centers can be removed by thermal annealing, and fibers exposed at elevated temperatures did not darken.
FIG. 12. SEM picture of an unexposed ~top! and an exposed ~bottom! type B-4 fiber. The unexposed fiber had a lower electrical conductivity and demonstrated larger charging effects.
2. Changes in electrical conductivity
During the SEM measurements, the unexposed fibers displayed charging effects to a much higher degree than the exposed fibers, causing the image to blur substantially, as can be seen in Fig. 12. This suggests an increase in the electrical conductivity of the exposed fiber compared to the unexposed fiber. To verify this, a simple four point probe conductivity measurement was performed. The results showed an increase in the conductivity of the exposed fibers by about an order of magnitude in comparison to the conductivity of the unexposed fibers. However, the x-ray reflectivity depends only on the total electron density and does not depend on the electrical conductivity of the glass surface. Hence, a change in electrical conductivity alone is not expected to change the transmission efficiency of the fibers. B. Changes in chemical composition and density
1. RBS and PIXE
RBS and PIXE studies were used to measure the chemical composition of the capillary surfaces. Since RBS is sensitive to high Z elements and PIXE is sensitive to low Z elements, a combination of both is expected to give reasonable information regarding the surface chemical composition
of the exposed and the unexposed fibers. Thin, type B-4 fibers were used. The superimposed spectra of the unexposed and exposed ~1.4 MJ/cm2) fibers from the RBS studies are shown in Fig. 13. A comparison of the RBS spectra for the unexposed and the exposed fibers does not show any noticeable movement of oxygen, aluminum, potassium, lead, or silicon. The dashed line in Fig. 13 is a RUMP’90 simulation for a 110 nm thick surface layer Si0.55O2.01B0.4Li0.4K0.09Na0.05Al0.05Pb0.014C0.15 over bulk material of Si0.60O2.06B0.3Li0.4K0.09Na0.05Al0.05Pb0.004 . The presence of carbon in the surface layer is contamination, as is the 1.4% lead. To determine the sensitivity of the technique, the simulated data were used to change the concentrations of different elements. The solid line in Fig. 14 is a simulation with an increase in the concentration of potassium by 20.0%. It clearly shows that such a change should be detectable by RBS. The sensitivity of RBS for different elements in the glass sample was determined and the results are summarized in Table III. A similar comparison for PIXE spectra, shown in Fig. 15, also does not show a significant change in the surface concentration of sodium, aluminum, silicon, or potassium.
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
Rath et al.
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
7431
TABLE III. RBS sensitivity for different elements in the glass sample.
Element
Sensitivity ~at. %!
Si O Al K Na B
9.0 3.0 1.7 20.0 315.0 261
the x-ray reflectivity. The x-ray transmission simulation program was used to calculate the transmission efficiency of a borosilicate glass polycapillary for a density increase of 3.0%. The results are shown in Table V. An increase in density by 3.0% is too small to show any measurable gain in transmission efficiency. FIG. 13. Superimposed RBS spectra for a type B-4 fiber, unexposed ~circles! and exposed at 1.4 MJ/cm2 ~crosses!. The dashed line is a RUMP’90 simulation.
2. X-ray transmission simulation
A geometrical optics, ray tracing x-ray transmission simulation23 was performed to calculate the effect of change in chemical composition of the glass polycapillary material on its transmission efficiency. The results are shown in Table IV. Simulation results suggest that even pure SiO2 , pure B2O3 , pure K2O, pure Na2O, and pure Al2O3 would have the same transmission efficiency at 8 keV as that of the currently used borosilicate glass. Thus, a change in chemical composition is not responsible for the decrease in transmission efficiency of the polycapillaries. The density of glass is known to increase upon irradiation. Maximum increases in density up to 3.0% have been reported.24 An increase in density could be beneficial for the polycapillary fiber transmission efficiency, since it increases
FIG. 14. RBS spectrum for an exposed type B-4 fiber. The dashed line is the simulation of Fig. 13. The solid line simulation has 20% more potassium in the surface layer.
C. Changes in surface reflectivity and roughness
1. AFM studies
Glass surfaces exposed to radiation experience chemical, physical, and topological changes which depend on the type of radiation and the amount of exposure dose. These effects may increase the roughness of the glass surface which could result in loss of transmission efficiency. AFM tests were done to measure the roughness of the capillary channel before and after exposure and annealing. Type B-4 fibers were used for all the AFM studies. The fiber was cleaved open to expose the interior channel surface for the AFM scan. Care was taken to make the measurement at the input end of the exposed fibers, where the radiation effects should be maximum. The results are summarized in Table VI. No clear trend is visible. Measured roughnesses for all samples varied within a range of 0.4 nm. Surprisingly, the unexposed fiber, with the highest transmission efficiency, was found to have a larger roughness than the other samples. This is contrary to the initial assumption that the fibers exposed to higher dose should have the highest roughness
FIG. 15. Superimposed PIXE spectra for a type B-4 fiber unexposed ~open circles! and exposed at 1.4 MJ/cm2 ~solid line!.
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
7432
Rath et al.
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998 TABLE IV. Simulated transmission efficiency of polycapillary fibers for different compositions. Simulated transmission efficiency of polycapillaries ~%! Composition Pure Pure Pure Pure Pure
SiO2 B2O3 K2O Na2O Al2O3
8.0 keV
9.6 keV
15.0 keV
20.0 keV
30.0 keV
61.07 61.94 58.9 61.64 61.88
61.69 61.94 60.26 61.57 61.69
61.44 61.88 60.26 61.75 61.69
61.63 61.94 61.01 61.94 61.88
61.94 62.0 61.57 61.81 61.88
value. All the data presented in Table VI are for single fibers with measurements done only on one segment, so the variation may be largely statistical. AFM measurements were done over an area of 4 m m2 with a resolution of 0.1 nm. 2. SEM studies
To observe any possible microdamage such as cracking of the surface or collapsing of the channel walls, a number of type B-4 fibers were observed in the scanning electron microscope. The reported sizes of oxygen bubbles and the cracks formed in V-silica due to irradiation are in the range of micrometers.17 In the SEM measurements, no deformation of the capillary channel walls due to either oxygen bubble formation or due to surface cracking was observed. There was also no evidence of capillary wall collapse due to thermal load. The resolution at which the SEM observations were performed was about 10 nm. Similarly, none of these damage features were observed in the AFM measurements. 3. X-ray transmission spectra and source scan studies
Any change in the surface roughness, chemical composition, and/or shape of the fiber might be expected to have a TABLE V. Simulated fiber transmissions for different glass densities and photon energies. Energy ~keV! 8.0 9.6 15 20 30 40 50 59 68 75 80
Density
Transmission ~%!
2.5 2.575 2.5 2.575 2.5 2.575 2.5 2.575 2.5 2.575 2.5 2.575 2.5 2.575 2.5 2.575 2.5 2.575 2.5 2.575 2.5 2.575
61.19 61.32 61.57 61.63 61.81 61.69 61.94 61.94 61.75 61.75 61.81 61.88 61.81 61.57 61.01 60.76 56.79 57.78 55.55 55.56 51.15 51.58
noticeable impact on the transmission efficiency of the fiber for different x-ray energies. Higher energy x rays are more sensitive to the deformation in the fiber shape because of their smaller critical angle. A number of type C ~thick! and type B-4 ~thin! fibers were used to compare the transmission efficiencies of the unexposed, exposed, and annealed fibers for photon energies up to 80 keV. The transmission efficiency versus the x-ray energy for type C unexposed and exposed ~dose of 1.8 MJ/cm2 in ambient temperature! fibers is plotted in Fig. 16. The transmission for the unexposed fiber is fairly constant up to 40 keV and drops for higher energies. The transmission of the exposed fiber increases up to 15 keV and then falls off rapidly. At 80 keV a transmission efficiency of 14% was measured for the exposed fiber and 34% for an unexposed fiber. To determine whether the loss in transmission was due to blocked or bad channels or was reduced uniformly across the fiber, the pinhole, which was about 250 m m in diameter, was scanned across the fiber, keeping the source, fiber, and the detector at the aligned position. A variation of not more than 5%–8% in the transmission efficiency was measured. Source scans, which are taken by measuring the transmission efficiency of the fiber as the source is moved off axis, give significant information about the reflectivity of the capillary surface because the number of reflections increases rapidly as the angle to the source is increased. To compare the reflectivity of the unexposed fiber with that of the exposed fiber, a number of source scans at energies of 9.6, 20, 30, and 40 keV were taken and are shown in Fig. 17. No significant difference between the normalized scan curves for unexposed and exposed fibers is seen even though there was an appreciable difference in the absolute transmis-
TABLE VI. Capillary channel surface roughness measurements for five exposed and unexposed fibers taken with AFM. Capillary type B-4 B-4 B-4 B-4 B-4
Capillary history Unexposed Unexposed, annealed at 450 °C for 2.5 h Exposed to a dose of 0.08 MJ/cm2 Exposed to a dose of 1.4 MJ/cm2 Exposed to a dose of 1.4 MJ/cm2, annealed at 450 °C for 2.5 hours
Transmission efficiency ~%!
Roughness ~Å!
60 57
8.1 4.7
45.8
4.4
15
5.7
54.2
6.6
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
Rath et al.
7433
FIG. 16. Transmission efficiency of type C fibers vs photon energy, unexposed ~open squares! and exposed to 1.8 MJ/cm2 ~solid squares!.
sion efficiency. The transmission values with the source displaced are the result of a very large number of reflections, and so the width of the curves cannot be unchanged if the surface reflectivity has decreased. The lack of change in the source scan implies that the surface cannot have roughened appreciably. Transmission efficiency and the source scan measurements for the thin, type B-4 fibers are shown in Figs. 18 and 19, respectively. The unexposed type B-4 fiber showed significantly less transmission at high energies than the unexposed type C fiber. It was difficult to keep the type B-4 fiber straight in the measurement groove because it was thin and flexible. Since, as noted in Eq. ~3!, higher energy photons have smaller critical angles than lower energy photons, higher energy transmission is more sensitive to fiber bend-
FIG. 17. Source scans of transmission vs transverse x-ray source displacement at a series of x-ray measurement energies of an unexposed, thick, type C fiber ~open symbols! and a type C fiber exposed to a dose of 1.8 MJ/cm2 in ambient temperature ~closed symbols!.
FIG. 18. Transmission efficiency of thin, type B-4 fibers vs energy. Open symbols are unexposed; closed symbols are exposed. Circles are unannealed; squares and diamonds are annealed.
ing. The unexposed annealed fiber showed an additional transmission decrease above 30 keV. Since the annealing temperature, 400 °C, is far below the melting temperature of the borosilicate glass, about 570 °C, gross disorders in the structure of the glass, such as the collapse of channel walls, are not expected. However, there may be an introduction of bending due to annealing stress.
FIG. 19. Source scans of thin, type B-4 fibers for a series of different x-ray measurement energies. Squares are 9.6, circles are 20, diamonds are 30, and triangles are 40 keV. Unexposed fibers are represented by open symbols, unexposed fibers annealed at 400 °C by crossed symbols, and fibers exposed to 1.4 MJ/cm2 and then annealed up to 450 °C are shown by solid symbols.
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
7434
Rath et al.
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
FIG. 20. Transmission vs energy measurements for thick, type C fibers, unexposed ~open symbols! and exposed ~solid symbols!. Simulations, which include bending curvature, channel wall tilt ~waviness!, surface roughness, and a thin layer of glass in the channels ~overlayer! are also plotted. The best fit simulations are the thin solid line ~unexposed! and the thick solid line ~exposed!.
These thin fibers, when exposed to high radiation doses, displayed a sharp drop in transmission efficiency above 9 keV, even after furnace annealing. The high temperature annealing technique recovers the transmission efficiencies of the exposed fibers for lower energies only. As for the thick, type C, fiber source scans, the source scans for the thin, type B-4, fibers, shown in Fig. 19, did not show any significant difference in width after exposure or annealing, again suggesting the absence of any change to the surface reflectivity. D. Changes in profile: Bending and waviness
Severe bending was observed in the case of the thin, type B, fiber exposed to white beams. The profile of such a fiber is shown in Fig. 6. Assuming a two dimensional bending ~in reality there was three dimensional bending! and fitting the data with the curves shown as solid lines in Fig. 6, the x-ray transmission simulation through the fiber was calculated for the shape given by the fitting parameters. The simulation demonstrated that most of the transmission degradation could be accounted for by bending. Unbending of the fibers was observed after annealing, and was less severe for the thicker fibers. In addition to the obvious profile change, there might have been shorter wavelength changes as well. The shortest wavelength changes are described as roughness.25 In addition, mid-frequency ‘‘waviness,’’ or surface rippling, would increase the average angle of incidence and could cause further loss. The results of detailed simulation analyses26 are shown in Fig. 20. The top two simulation lines show the best fit to the unexposed fibers for simulations which include only bending and only bending and roughness. The correlation length for the roughness was set at 6 m m. Changing the correlation length alters the absolute value of the roughness heights. The thick solid line includes the effects of bending, roughness, and waviness. Both bending and waviness are
required to fit the data. The lower points and lines are data and simulations for a fiber exposed to 1.8 MJ/cm2. The radius of curvature of the unexposed fiber was 81 m, while the radius for the exposed fiber was about 50 m. This implies that even the thick type C fiber underwent slight bending, contrary to our previous assumption that it did not. However, this slight bending does not affect the transmission efficiency of the fibers for low energies (;8 keV!. The reasonably good fit of the simulation with the data for both exposed and unexposed fiber leads to the conclusion that there has not been significant surface roughness induced in the fibers due to exposure. The profile change, including both bending and waviness, is sufficient to explain the data for higher energies. The decrease in transmission efficiency of the exposed fiber for energies less than 15 keV could be explained by the addition of a 35 m m glass absorbing layer. The possible origin of glass particulates in the capillary channels is not understood. The growth of small crystallites might result in partial channel blockage, but was not observed in SEM analysis. The sharp decrease in transmission for x rays of energy less than 9 keV may be due to some other mechanism. The x-ray simulation program results show that the majority of the transmission loss is due to bending of the fibers. There are two probable causes for bending. These are stress introduced by the heat gradient and/or stress due to a density gradient developed in the fiber. Thermal modeling21 shows a negligible temperature rise in the thinner type B-4 fibers. Also, had thermal load been responsible for bending, maximum bending would have been seen when the beam intensity was maximum, i.e., at the maximum beam current. This should have been reflected in a sharp decrease in transmission in the in situ measurements at the time of maximum beam current, which was not observed. Also, there should have been less dependence of the transmission decrease on total dose and more on dose rate. The absence of such behavior suggests that the stress induced in the fiber is not due to thermal load. The in situ and ex situ data for type B-4 fibers follow pretty much the same trend, which also supports the conclusion that the bending mechanism is athermal. This leads to the assumption that the bending is due to the stress caused by a density gradient developed due to asymmetric exposure of the fiber. Annealing effectively removes the irradiation-induced stress and the fibers straighten spontaneously. An activation energy ranging from 1.3 to 3.5 eV for recovering the original density of pure V-silica has been reported in the literature.22 Our measurements show an activation energy of 1.4 eV for recovery of 90% of the initial transmission efficiency, which is in agreement with the assumption that the bending is due to density gradients developed in the fiber due to asymmetric exposure. IV. CONCLUSION
In situ and ex situ transmission efficiency degradation measurements of fibers of varying channel diameter and outer diameter have been performed for focused synchrotron 1.4 keV radiation and intense synchrotron bending magnet white radiation doses up to 1.8 MJ/cm2 at ambient and
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
Rath et al.
J. Appl. Phys., Vol. 83, No. 12, 15 June 1998
elevated temperatures. In situ measurements at 1.4 keV showed no detectable decrease in the transmission efficiency of the fibers when exposed to doses to 1.3 MJ/cm2. When exposed to the higher energy white beam lines, thinner fibers underwent bending due to exposure, resulting in severe transmission efficiency loss. This transmission efficiency loss, measured at 8 keV, was recoverable by post-exposure furnace annealing. A much smaller linear decrease was measured when thick, rigid, fibers were exposed at ambient temperature. When the thick fibers were exposed at elevated temperatures, there was no decrease in transmission efficiency. No transient effect was found during the exposure and transmission efficiency loss was found not to depend on dose rates from 8 to 13 W/cm2. RBS and PIXE studies suggest that the chemical composition of the fibers did not undergo a noticeable change due to exposure up to 1.4 MJ/cm2. Simulation results indicate that a change in chemical composition is not expected to change the transmission efficiency of the fibers. Similarly, electronic changes can explain the transmission loss. Systematic measurements of transmission as a function of angle and energy show that the surfaces of the capillaries have not roughened. Bending and surface rippling is indicated by the simulation. Roughening is also not seen in the AFM measurements. The transmission loss for high energy photons was solely driven by the change in the profile of the fibers, which is believed to be due to asymmetric stress due to irradiationinduced density gradients in the fibers.
ACKNOWLEDGMENTS
This work was partially supported by the U.S. Department of Commerce through the Advanced Technology Program under corporate Agreement No. 70NANB2H1250, by the National Institutes of Health under Grant No. R01 CA58521, and by the New York State Center for Advanced
7435
Technology. The authors are grateful for AFM measurements provided by C. M. Dozier, P. G. Burkhalter, and J. V. Gilfrich. L. G. Parratt, Rev. Sci. Instrum. 30, 297 ~1959!. G. Rosenbaum, J. Appl. Crystallogr. 7, 117 ~1974!. 3 T. W. Barbee, Jr., Proc. SPIE 563, 2 ~1985!. 4 G. N. Greaves and C. R. A. Catlow, in Applications of Synchrotron Radiation ~Chapman and Hall, New York, 1990!. 5 C. A. MacDonald, J. X-Ray Sci. Technol. 6, 32 ~1996!. 6 J. X. Ho, D. C. Carter, J. R. Ruble, R. C. Sisk, E. H. Snell, S. M. Owens, and W. M. Gibson, Acta Crystallogr., Sect. D: Biol. Crystallogr. 54, 200 ~1998!. 7 W. M. Gibson, M. V. Gubarev, F. A. Hofmann, M. K. Joy, C. A. MacDonald, S. L. O’Dell, B. D. Ramsey, C. Russell, L. Wang, and M. C. Weisskopf, Proc. SPIE 3113, ~1997!. 8 M. Vartanian, D. Gibson, R. D. Frankel, and J. Drumheller, Proc. SPIE 335, 1924 ~1993!. 9 D. G. Kruger, C. C. Abreu, E. G. Hendee, A. Kocharian, W. W. Peppler, C. A. Mistretta, and C. A. MacDonald, Med. Phys. 23, 187 ~1996!. 10 F. A. Hofmann, C. A. Freinberg-Trufas, S. M. Owens, S. D. Padiyar, and C. A. MacDonald, Nucl. Instrum. Methods Phys. Res. B 133, 145 ~1997!. 11 C. B. Norris and E. R. Eernisse, J. Appl. Phys. 45, 3876 ~1974!. 12 W. Primak and R. Kampwirth, J. Appl. Phys. 39, 5651 ~1968!. 13 W. Primak, Radiat. Eff. 45, 29 ~1979!. 14 C. Fiori and R. A. B. Devive, Mater. Res. Soc. Symp. Proc. 61, 1 ~1986!. 15 S. Thomas, J. Appl. Phys. 45, 161 ~1974!. 16 O. Puglisi, G. Marletta, and A. Torrissi, J. Non-Cryst. Solids 83, 344 ~1984!. 17 J. F. Denatale and D. G. Howitt, Nucl. Instrum. Methods Phys. Res. B 1, 489 ~1984!. 18 A. A. Kravchenko, Izv. Vyssh. Uchebn. Zaved. Fiz., 14, No. 4 ~1990!. 19 B. K. Rath, R. Youngman, and C. A. MacDonald, Rev. Sci. Instrum. 65, 3393 ~1994!. 20 B. K. Rath, D. C. Aloisi, D. H. Bilderback, N. Gao, W. M. Gibson, F. A. Hofmann, B. E. Homan, C. J. Jezewski, I. L. Klotzko, J. M. Mitchell, S. M. Owens, J. B. Ullrich, L. Wang, G. M. Wells, Q. F. Xiao, and C. A. MacDonald, Proc. SPIE 2519, 207 ~1995!. 21 B. K. Rath, F. B. Hagelberg, B. E. Homan, and C.A. MacDonald, Nucl. Instrum. Methods Phys. Res. A 401, 421 ~1997!. 22 W. Primak and H. Szymanski, Phys. Rev. 101, 1268 ~1956!. 23 Q. F. Xiao, I. Yu. Ponomarev, A. I. Kolomitsev, and J. C. Kimbal, Proc. SPIE 1736, 227 ~1992!. 24 E. Lell, N. J. Kreidl, and J. R. Hensler, Prog. Ceram. Sci. 4, 1 ~1966!. 25 J. C. Kimball and D. Bittel, J. Appl. Phys. 74, 877 ~1993!. 26 L. Wang, B. K. Rath, W. M. Gibson, J. C. Kimball, and C. A. MacDonald, J. Appl. Phys. 80, 3628 ~1996!. 1 2
Downloaded 26 Jun 2001 to 169.226.28.24. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
