A Refractive Lens for X-Rays

(From: Joseph Dunphy.) This is an X-ray optics notion that didn't pan out. Some while ago, I posted a half formulated idea, that while unworkable, produced an interesting response:

A question about some old news (found it on sci.astro.research). I know that a lot of what will follow is half baked. So, please be patient with me. I'm just asking questions.

(From: AIP listserver (physnews@aip.org).)

Until now impractical because lenses absorb too much or refract too little at X-ray wavelengths, one has been developed by scientists at the European Synchrotron Radiation Facility (ESRF) in Grenoble. The compound lens consists simply of a series of closely-spaced 0.6- mm holes drilled in a piece of aluminum. With this lens, a 14- keV beam of x rays was focused to a an 8-micron spot size. (A. Snigirev et al., Nature, 7 November 1996.)

Kwiat, Weinfurter and Zeilinger ("Interaction-Free measurement" in Physical Review Letters, Vol.74, no.24, pp. 4763-4766, June 12, 1995, co-authored with Herzog and Kasevich; "Quantum Seeing in the Dark", Scientific American, November 1996) made reference to the theoretical possibility of reducing the intensity of x-rays used in medical photography without sacrificing the quality of the image produced.

(In case someone reading hasn't seen the article yet, the question it poses is how one could tell if an object was present in a chamber without any interaction occurring (probably).)

The approach used in "interaction free measurement", is to perform a modified double slit experiment give a photon a choice of pathways, with the object to be detected placed along one of the pathways (if it is present at all). If it is present, and opaque, an individual photon passing through can't interfere with itself.

I'm wondering: If the object is translucent, will we see a contribution to the wave function at the other end of the apparatus, from the obstructed pathway, whose amplitude will be proportional to the probability that absorption won't occur ?

 Yielding an intermediate interference pattern
 consisting of a dot (as one would see if no
 self-interference occurred), at which a photon
 has a probability of (1/2) (prob. of absorption)
 of being detected at, superimposed with a normal
 interference pattern whose amplitude is multiplied
 by (1 - prob(absorption)), (1/2) (prob. of 
 absorption) being the probability that the photon
 will go down the obstructed pathway (1/2 chance)
 and then get absorbed by the target  ?

 Allowing one to determine the absorption probability
 by looking at the intensity of the dot on the
 screen ?

  The practical difficulty they foresaw lay in the nonexistence of x-ray
  optics. I was wondering if this this difficulty seemed less
  insurmountable, now ? For what they wanted, it would be necessary to
  produce an optical component capable of rotating polarized x-ray

  In particular, I was wondering if one could produce a high resolution
  CAT scanner that would do less damage to the patient.

  Producing holes drilled in a thin sheet, to sub micrometer accuracy,
  would actually be well within the current capabilities of
  microfabrication technology. (An individual microdevice on a chip, the
  last time I checked, could be as small as .55 microns, a few years ago.
  By now, it's almost certainly smaller. Needless to say, photolithography
  was NOT the technique used - that gets you down to about 1 micron).

  Meaning that one could conceivably produce an extremely regular hole,
  though the process would be expensive, and slow. Microfabrication
  techniques, naturally, have generally been used on thin films in the
  past. Meaning, that you might be in for a LONG wait while those holes
  form. But it COULD be done. Given the obvious problem of bubble
  formation if wet etching is attempted on a metal substrate (acid on
  metal), plasma etching using a reactive gas such as Fluorine would seem
  a natural choice. As for finding a type of resist that will adhere to
  aluminum, the electrical connections on a chip tend to be made out of
  deposited aluminum, so whichever brand of photoresist is used to pattern
  those connections might serve nicely. Now, if it turns out that the
  brand you obtain doesn't stand up to plasma etching conditions very
  well, you might still be able to make due if you can find a second
  variety which does, and adheres to the first variety, which ends up
  being reduced to playing the role of glue. Probably, though, there will
  be a simpler, cheaper, easier to work with material available to promote
  adhesion in this fashion. 

  I was also wondering if reflecting optics might not be possible. In my
  extremely naive introduction to Computational Electromagnetics course
  (Maxwell's laws only, no quantum theory of radiation involved)
  I remember that one could show that an idealized perfect conductor would
  reflect incident electromagnetic radiation. Would a superconductor be
  close enough to being one of those to reflect x-rays ? If so, how
  intense a flux of x-rays could such a reflector handle before
  superconductivity broke down. (Excessively strong EM fields destroy
  superconductivity, right ?) Would it be high enough, for practical x-ray
  photography to be done with a beam of such low intensity ?

  While even high temperature superconductors require liquid nitrogen
  temperatures (right ?) this problem might not be insurmountable.
  Perhaps one might set up sort of a thermos arrangement, with one
  monocrystalline quartz container inside the other, the x-ray mirror
  inside the inner container, the connection between the two containers
  and the liquid nitrogen hookup to the inner container passing behind the
  mirror, the hookup consisting of a many thin, flexible sheets of
  material rather than a few bolts (to accommodate thermal expansion -
  the outside container is at room temperature), the containers being
  properly silverized (this is a thermos, after all), a good hard vacuum
  being present between the containers.

    Like I said, though, very half baked. I'm not even sure which phase
    transitions SiO2 undergoes as it gets that cold, one of them might be 
    a destructive one. While SiO2 is generally polycrystalline in
    industrial use, a modification of the Czochralski method has been used
    to produce monocrystalline Gallium Arsenide (a layer of slag is left
    over the melt in order to prevent evaporation), so maybe it would be
    possible to do so with another nonatomic substance, without getting
    little flecks of elemental silicon embedded because some of the oxygen
    has diffused out of the melt. A monocrystalline container's
    contribution to the image produced by a collimated x-ray beam passing
    through it would be a regular, predictable diffraction pattern which 
    one might be able to remove from the final image.

    The technology already exists to produce a very regular parabolic 
    reflector (the same used to produce telescope mirrors). So the idea,
    again, half baked and untested, I have in mind is to form a parabolic
    surface, and then chemical vapor deposit a superconducting film on it.
    This, incidentally, has already been done, though on a flat
    surface, with a relatively high temperature superconductor. In this
    case, one would have to do the deposition quite slowly, of course, 
    because one will have to change the orientation of the parabolic
    surface a good many times to get anything resembling an even coating,
    as the thickness should be dependent on the inclination of the surface
    during deposition. Slow, and expensive. If it works, at all.

    Put a point source of x-rays inside the inner chamber, and project
    onto the mirror, with a cup around one side of the source, blocking
    the departure of x-rays which aren't reflected off the mirror. The
    source is placed at the focus of the paraboloid defining the shape
    of the reflector. High school geometry - after reflection, the
    x-rays will follow parallel paths out of the mirror. Again, old
    technology reapplied. What you've done is build a searchlight, only
    with x-rays, instead of visible light. Same design principle, though.
    The high accuracy attainable by surface grinding translates into
    extremely good performance in producing parallel x-rays. You can see,
    now, why we wanted as even a layer of superconducting material
    deposited as possible. Irregularities in the deposition will scatter
    x-rays in undesired directions. The shadows wouldn't be as crisp, if
    you will, because the lighting is more diffuse.

    Result : collimated x-rays, and a crisper image ? Allowing for
    increased resolution if applied to computer aided tomography ?

    So, the question.....am I out of my mind on this ? Is it feasible ?
    What should be changed ? I hope the description I gave isn't too

    My own background : thesis stage PhD Math, MS Electrical Engineering,
   but a terminal Bachelor's in Physics, alas.

    So, I might be of assistance in answering fabrication questions,
    maybe; would be of more help on computational ones, and definitely
    need your help on theoretical physics questions. As I'm sure you could

    E-mail is appreciated, but followups even more. 


Newsgroups: sci.electronics.design
Organization: The University of Iowa

Joseph Dunphy (stats@typhoon.xnet.com) wrote:

>   The practical difficulty they foresaw lay in the nonexistence of x-ray
>   optics.

I believe my father has a patent on X-ray and gamma ray optics made using
lead (that's Pb) phase-plates.  He was working fairly hard with various
radiologists on using these for radioisotope imaging, but new technologies
like CAT scans and PET scans came along and people lost interest in his

What's a phase plate?  It's a set of precisely dimensioned concentric
rings; an appropriate phase plate will diffract light to a focus in
exactly the same way a lens refracts it.  A phase plate that actually
brings gamma rays to a focus would require infinitesimal dimensions, but
my father's trick was to use a practically sized phase plate to record
an X-ray or gamma ray hologram that could be viewed with visible light.
The phase plate dimensions depended on the difference in wavelength
between the light used to record the hologram and the light used to view
it.  An interesting trick, and the results were subject to most of the
problems we associate with monochromatic holograms (grainy image,
difficulty in viewing, etc).

				Doug Jones


Don't worry about the mental oven being set too cool (i.e. half-baked
ideas)--what you ask has some MAJOR physical limitations, but at least you
aren't talking about why we aren't extracting gold from Mars.....You do
have interesting ideas, and deserve better than the 'does anybody read
this' snottiness of someone who posted on the thread.

The most serious limitation to the proposal you make is in the 'punch' of
an Xray photon and the availability of electronic states in solids. In
reflection of visible light, there is little opportunity for the light to
cause serious electronic phenomena to occur in the metal (i.e. you are
usually WAAAAY under the photoelectric threshold). With an Xray (say, a C
xray at ~286eV), you have plenty of energy to cause core-electron
excitations. Even if the excitation doesn't cause photoemission, you will
have essentially lost the wavelength 'identity' of your incident Xray.
Much of your light (Xrays) gets lost  in causing electronic excitation in
the mirror, leaving little for your output beam.
This is not to say that xray focussing is not possible. At the synchrotron
sources (I'm most familiar with the IBM line at Brookhaven), Xray
focussing down to ~300Å is performed with Fresnel Zone Plates (similar to
those used in camera eyepieces), with the proviso that this is a focus
point, not a collimated beam. Other focussing techniques include use of
diffraction mirrors, such as are used in the PHI Q2000 ESCA spectrometer.
In this case, the Xrays diffract (not reflect) from the mirror, and there
are losses in the optics due to photoelectron formation (can't make an
omelette without breaking a FEW eggs).

David Neiman

Pump it down, Bake it out, Drop the beam, Or do without!


From: Johannes Ullrich
To: Joseph Dunphy 
Subject: Re: Some questions about x-ray optics, and CAT

Joseph Dunphy wrote:

> So, if you are reading this, and are posting from a faculty account,
> or
> some other account that would indicate professional credentials, could
> you
> give an indication that you saw this, even if you don't feel like
> commenting on it ? It would be nice to know that I'm not just tilting
> at
> windmills, here.


I am not usually reading this newsgroup. However, I am scanning
newsgroups for certain keywords (e.g. x-ray optics) using DejaNews. I
found this technique very effective in cutting down on the noise level
usually found in newsgroups.

I subscribe to a few newsgroups which are more specific to my interests
and have less trafic.

I am currently working as a researcher for 'X-Ray Optical Systems
(www.xos.com)' and have a Ph.D. in physics (thesis work related to x-ray

A few comments on your post (the article from the AIP newsserver and the
CAT scanner idea):

First of all: There are many different kinds of x-ray optics. The
special difficulty is the nonexistence of a material that is transparent
to x-rays and has a index of refraction that is significantly different
from 1 (vacuum). Glasses as they are used for visible light, have
transperancies of over 99 % for many meters of thickness. Typically, the
index of refraction is larger than 1.5 maybe even as high as 3 (just to
give some numbers, I am not that aware of actual specifications for
different numbers).

Moreover: Visible light interacts with outer electrons. This allows the
variation of the index of refraction and absorbtion by slightly changing
the composition of a glass. X-rays interact with inner core electrons
which are only slightly effected by the chemical environment of the

A refractive lens for x-rays would have to be very thick (meters) and
all the x-rays would be absorbed within the lens. The article below
shows an experiment that uses a series of lenses which overcome this
problem (partially). It works because the material (Al) does not absorb
x-rays too badly and the energy used (17 keV) is high enough to
penetrate Al with reasonable losses.

Now something very special about x-rays. The index of refraction is
actually smaller than one. A typical index of refraction is 1-10^(-6)
(e.g. almost 1 but a little bit less).

The second article: Sounds kind of strange to me and I have to think
more about it. But one problem with x-rays is the short coherence length
of typically a few micron (depends on the source).

> :   The practical difficulty they foresaw lay in the nonexistence of
> x-ray
> :   optics. I was wondering if this this difficulty seemed less
> :   insurmountable, now ? For what they wanted, it would be necessary
> to
> :   produce an optical component capable of rotating polarized x-ray
> :   radiation. Could their approach be modified to produce such an
> optical
> :   component ?

There are x-ray polarizers (crystals) I am not sure if magnetic
multilayers can be used to rotate the polarization.

> :   reflect incident electromagnetic radiation. Would a superconductor
> be
> :   close enough to being one of those to reflect x-rays ? If so, how
> :   intense a flux of x-rays could such a reflector handle before

X-rays would not 'see' the superconductor. However, x-rays can be
reflected at grazing incidence and this is done a lot. The one problem
is the small critical angle for total reflection (typically about  0.2

X-ray mirrors are know to be subject to radiation damage. This occurs
mostly at high intensity synchrotron sources.

X-rays can be reflected at larger angles from crystalline material
(x-ray diffractions). But only one particular energy will be reflected.

> :   While even high temperature superconductors require liquid
> nitrogen
> :   temperatures (right ?) this problem might not be insurmountable.

Many x-ray detectors use liquid nitrogen. Not a technical problem. There
are some cooled x-ray mirrors at synchrotron. It is tricky but can be

> :
> :     Result : collimated x-rays, and a crisper image ? Allowing for
> :     increased resolution if applied to computer aided tomography ?

Overall: The problem of x-ray imaging is not a parallel bea, but the
contribution of scatter. X-rays are scattered inside the patient.
Scatter grids (lead strips) are used after the patient to remove some of
that scattered radiation.

The fan beam as obtained from a point source is very adequate for most
imaging. One problem is the size of the source. the smaller the source
the crisper the image.

Hope this help. Let me know if you want to know more.

Subject: Re: producing x-rays with old vacuum tube

RE: Homemade X-ray machine.  The potential hazards here cannot be
overemphasized.  Any device that produces ionizing radiation has the
potential for long term harm.  Whenever a vacuum tube type device and
voltages over 10kV or so are put together there is the potential to
produce X-rays.  Old TVs with tubes in the HV section had shielding
to protect against X-rays from the HV rectifier and regulator.  The CRT
including moderm CRTs are constructed with leaded glass for a similar
I am not saying that such experimentation should be avoided - just
that you understand all of the ramifications and take appropriate precautions.
Radiation exposure is something you cannot use an Undo command on.

Having said all that. if you are really want more info on these sorts
of projects, I suggest you contact:

From: Steve Hansen 
Newsgroups: sci.electronics
Subject: Vacuum Newsletter #2 - Generating X-Rays with Receiving Tubes

*  -----------------------------------                 ----------------  *  
*   the Bell Jar (electronic version)                  #2 (October 1994) *  
*  -----------------------------------                 ----------------  *  
*                                                                        *
*  This newsletter contains material which has been extracted from the   *
*  hard-copy newsletter of the same name. Devoted to the vacuum          *
*  experimenter, the intent of "the Bell Jar" is to broaden interest     *
*  in vacuum technology through useful discussions of theory and         *
*  technique, and to present ways in which a variety of apparatus may    *
*  be assembled using common and inexpensive materials. Information      *
*  on "the Bell Jar" may be obtained by sending email to the editor,     *
*  Steve Hansen, at hansen35@delphi.com or by writing to 35 Windsor Dr., 
*  Amherst, NH 03031. Please feel free to circulate this electronic      *
*  version, intact, to others who might be interested in the subject     *
*  matter. New numbers will be mailed at approx. quarterly intervals.    *
*  Email subscriptions are free and may be obtained by contacting the    *
*  editor. Comments, contributions and criticisms are always welcome.    *
*  Copyright 1994, Stephen P. Hansen.                                    *