The Design and Test of a Pulsed 250 MHz EPR Spectrometer by 305dJrAJ


									             10th International Workshop on Bio-Medical
                    ESR Spectroscopy and Imaging
                            Fukuoka, Japan
                            April 1-3, 2003

        The Design and Test of a Pulsed 250 MHz EPR

  Gareth R. Eaton, Richard W. Quine, George A. Rinard, and
                       Sandra S. Eaton
       Department of Chemistry and Biochemistry and
      Department of Engineering, University of Denver
  Howard Halpern and Colin Mailer, University of Chicago

NIH-funded Center for Electron Paramagnetic Resonance Imaging for in
Vivo Physiology. H. Halpern, University of Chicago, G. Eaton, University of
Denver, and G. Rosen, University of Maryland
                   Low Frequency EPR
“Low” EPR frequencies being used include
200, 220, 250, 280, 300, 600, 700, 1000, 1200 MHz

RF penetration as a function of frequency depends on
•dielectric properties
•sample shape

RF penetration increases
•the lower the RF frequency
•the more heterogeneous the sample
•deeper penetration in animals than in spheres or cylinders of water
               Pulsed 250 MHz EPR
Is being developed to
Exploit the high sensitivity of electron spin relaxation times
to O2
Gate data acquisition to physiologic motion
Detect transient phenomena

Electron spin relaxation times are at least 1000-fold faster
than the nuclear spin relaxation times important in MRI.
Consequently, we need to develop several crucial enabling
Resonators with sub-microsecond dead time following the
high-power pulse
High-power pulsed RF amplifiers
Methodology for fast data acquisition
            Comparison of CW and Pulsed EPR - 1
For a sample for which
•all of the spins in the sample form the echo
•the CW spectrum is not saturated
•the CW magnetic field modulation is approximately equal to the line

The relative CW and spin echo signal intensities are given by the ratio:

                  CW intensity CW microwave B1
                  echo intensity CW linewidth (B)

If the CW spectrum saturates, then the B1 has to be reduced below the
maximum available in the spectrometer, reducing the CW intensity.

If the magnetic field modulation is reduced to avoid distorting the CW
line shape, then the CW intensity if reduced roughly proportionately.
          Comparison of CW and pulsed EPR - 2
           If noise is thermal noise, then the noise in the signal is proportional to
 the bandwidth of the spectrometer. In a CW spectrometer the bandwidth is
 usually limited by a filter at a late stage of the detection system. Commonly a
 time constant of a few ms to 1 s is used for CW EPR, corresponding to
 bandwidths of 160 Hz to 0.16 Hz.
           The -3dB bandwidth of a single pole filter with time constant TC is
                  f 

         In a pulsed spectrometer, the filter is wider to pass all relevant frequencies,
and commonly is several MHz.
         If the bandwidths were, e.g., 10 Hz (time constant = 0.016 s) for the CW
spectrum and 10 MHz for the pulse spectrum, the noise in the pulse spectrum would
be 103 greater. However, the pulse spectrum might be acquired at a repetition of 104
per second, so in 100 seconds of signal averaging, the noise would be about the same
in the pulse spectrum as in a CW spectrum acquired as quickly as the line width
         from Rinard, Quine, Song, Eaton and Eaton, JMR 140, 69-83 (1999)
       Relative Benefits of Pulsed and CW EPR
•can be performed on many radicals
•including radicals that have wide spectra
•can be performed for large gradients, to achieve high spatial resolution

•can be more sensitive per unit time than CW for very-narrow-line spin probes
such as the trityl radicals
•is limited by dead time following the pulse
•FID detection is limited by the effective T2* caused by the gradient field

see experimental comparison at 300 MHz in:
K.-I. Yamada, R. Murugesan, N. Devasahayam, J. A. Cook, J. B. Mitchell, S.
Subramanian, and M. C. Krishna. Evaluation and Comparison of Pulsed and
Continuous Wave Radiofrequency Electron Paramagnetic Resonance
Techniques for in Vivo Detection and Imaging of Free Radicals. J. Magn.
Reson. 154, 287-297 (2002).
              Selection of Experiment - 1

The goal of the measurement dictates the conditions for

Optimum detection occurs when the line is saturated and over-
•The maximum CW signal for a single Lorentzian line occurs
when 2B12T1T2 = 1
•The line broadens when saturated. It is about 1.2 times broader at
the power that gives the maximum in the signal amplitude.
•The maximum intensity occurs when the modulation amplitude is
approximately equal to the line width, at which point the width is
increased by about a factor of 2 (depends on details of line shape).
                 Selection of Experiment - 2
 One might use less saturation and lower modulation if the goal is to
 •identify the radical
 •discriminate between radicals - also use multi-frequency, differential
 •measure viscosity
 •measure pH

However, if the goal is to measure O2 concentration, the
spectroscopy becomes more demanding.
•Small changes in line width measure O2 concentration
•Over-modulation by a factor of 1 to 2x to improve S/N is permitted if
Colin Mailer’s program is used to deconvolute the excess modulation.
(Robinson, Mailer, Reese, J. Magn. Reson. 138, 199, 210 (1999)).
•Relaxation times are sensitive to changes in O2 collision rate, so pulsed
EPR may be valuable.
                        MRI Comparison
Concentrations – 1H in vivo is about 100 M, but electron spin
concentrations for EPR typically are less than 1 mM
Relaxation Times – about 1000 times longer for NMR than EPR
Echo vs. FID – MRI uses spin echos but most pulsed EPR uses FID's
Maximum Frequency – MRI at 8T, 340 MHz (Robataille et al. J.
Computer Assisted Tomography 23:821-831 (1999).

    1.5 T       4.7 T          8T
             MRI – Field Distributions

Robataille et al., J. Computer Assisted Tomography
23:821-831 (1999).
C. M. Collins, S. Li,
M. B. Smith, Magn.
Reson. Med. 40:
847-856 (1998).
                Absolute Signal-to-Noise

  The EPR signal voltage, Vs, is given by

     Vs   " ( )Q Z 0 P

C"() is the imaginary component of the effective RF
 is the filling factor
Q is the loaded quality factor of the resonator
Zo is the characteristic impedance of the transmission line
P is the RF microwave power to the resonator
          2        1/ 2
       Q    0  d
          8       
                      Resonator B1

                   21/ 4 1/ 4  3 / 4 P
            B1                  0
                        1/ 4
d is the diameter of the LGR
z is the length of the LGR
0 is the permeability in a vacuum
 is the conductivity of the surface of the resonator
 is the resonance frequency
P is the power to the resonator
          Frequency Dependence of EPR Parameters
                       constant sample        Sample and LGR Constant sample,
                       and LGR sizes          size  1/     LGR  1/
Inductance, L                    1                     -1                      -1
Resistance, R                  1/2                    1/2                    1/2
Quality Factor, Q              1/2                   -1/2                    -1/2
Filling factor,                 1                      1                       3
EPR signal,                    3/2                    1/2                    7/2
constant P
B1/P                          -1/4                   3/4                    3/4
P for constant B1              1/2                    -3/2                   -3/2
EPR signal,                    7/4                    -1/4                   11/4
constant B1

 P is RF power, and B1 is the RF magnetic field intensity.

 G. A. Rinard, R. W. Quine, S. S. Eaton, and G. R. Eaton J. Magn. Reson. 156, 113 (2002).
      Frequency Dependence of Signal Intensity:
                Experimental Results

VHF vs. L-band FID of trityl: normalized signals agreed with
prediction within 5%
VHF, L-band, X-band resonator dimensions scaled by factors
of 6
For each pair of frequencies the CW signal at lower frequency
was predicted to be larger by factor of 1.57.
        250 MHz / 1.5 GHz was 1.52
        1.5 GHz / 9 GHz was 1.14
     Frequency Dependence of EPR S/N when
         sample loss dominates the noise

                Constant     Sample       Constant
                sample and   and LGR      sample,
                LGR sizes    size  1/   LGR  1/

S/N at           r0.5       0.5 r0.5     r0.5
constant B1
If r is         0.8         0.3         0.8
to 0.4

   r is sample resistivity
CW operation with Bruker console
Pulse operation with DU hardware
Pulse operation with Bruker console
Cross-Loop Resonator and
 Pre-Amplifier Modules
  Magnet and Gradient Coil Specifications
Working Volume      15 cm diameter

 Continuous field   90 G
 Maximum field      140 G
 Homogeneity         40 ppm
 Rapid Scanning     33 Hz, 12 G peak-to-peak
 Magnitude          10 G/cm
 Linearity           3%
 Side               15-20 cm diameter
 Axial              15-20 cm diameter
Location of gradient coils
Hall-probe/current control adapter

                a Two six-turn spiral coils of no.
                12 copper magnet wire with ~2.2
                cm mean diameter spaced ~1.25
                cm apart
                b Aluminum coil forms
                c 5 cm aluminum channel
                d Aluminum spacers
                e Fiberglass insulating spacers
                f insulated Plexiglas mount for
                Hall probe
          Magnet and cross-loop resonator
                                                       The 250 MHz crossed
                                                       loop resonator is
                                                       shown in the very
                                                       homogeneous (< 40
                                                       ppm variation over 15
                                                       cm diameter) 4-coil
                                                       resistive magnet
                                                       designed for EPR
                                                       imaging. The outer
                                                       diameter of the large
                                                       coil is ca. 1 meter.
                                                       The picture also shows
                                                       the z-gradient coils.

The x, y gradient coils have been removed to show the other coils more
clearly. The crossed loop resonator can isolate the time-domain EPR signal
from the high power RF pulse by 60 dB (typical).
Cross-Loop Resonator

            a     Sample tube resonator
            b     Cross resonator
            c,d   Brackets to adjust orthogonality
            e     Isolation adjustment screw
            f     Sample tube
            g     Sample tube holder
            h     Coupling screw for resonator a
            i     Coupling screw for resonator b
            j     Frequency tuning for resonator b

             Resonators machined from
             tellurium-copper alloy
Cross-Loop Resonator
            a. Sample tube
            b. Sample tube holder
            c. Loop (inductor) of sample
               tube resonator
            d. gap (capacitor) or sample
               tube resonator
            e. Re-entrant loop for
               sample tube resonator
            f. Loop for cross resonator
            g. Re-entrant loop for cross
            h. Sample access hole
                          Phase Noise Comparison




NLi 1     60
                                                                                  Blue – CLR
NC j 1
                                                                                  Red - LGR


     0      0
          1 .10
                                0.01                      0.1                 1
                     -3                NLi 0  NC j 0                       1
                                                                B1 in Gauss
Isolation of cross-loop resonator
Loop gap resonator   Cross-loop resonator
Q-switching circuit
          Q-switching sequence

The Q of the excitation resonator is held high and the detection
resonator Q is spoiled during the RF pulses. The Q states are
reversed during the time when the EPR signal is recorded.
Active Q-spoiling

                Without Q-spoiling
                Q1 (excitation) = 74
                Q2 (detection) = 469

                With Q-spoiling
High-power amplifier
             High-power amplifier
There was no RF amplifier commercially available that would
meet the needs of pulsed 250 MHz EPR imaging.

Two companies, Communications Power Corporation (CPC) and
Tomco, have contracted to develop and produce amplifiers with
   Output power > 400 W
   < 30 ns rise and fall times
   blanking of output noise to <10dB above thermal within 80

The Tomco amplifier (top) meets the speed specifications and
was tested at 550 W.

A 400 W prototype of the CPC amplifier is shown in the bottom
    Design Philosophy of Pulsed VHF Bridge for
               University of Chicago
oUse of a crossed-loop resonator (CLR)
oStand-alone testing of resonators
oStand-alone FID
oElectron spin echo (ESE) when controlled by a Bruker
o1 kW maximum RF pulse power
o2 kW maximum RF power
oTo be used with Bruker E540 console
oData collection via Bruker SpecJet
          Time-domain data acquisition
             methodology at 250 MHz

    The isolation of the crossed loop resonator reduces the
dead time of the system following the high-power pulse,
because the pulse power reaching the detection system is
reduced by the amount of the isolation.
    We have decreased dead time even more by active Q-
switching of the excitation and detection resonators in
synchrony with the RF pulses and detection times.
    The signal on the following slide is the free induction
decay (FID) of a sample of Nycomed triarylmethyl (trityl)
radical, whose electron spin relaxation time is a measure of
oxygen concentration in vivo. Full FIDs were digitized with
a repetition time of 100 s
Pulse Sequences
                                          FID, no gradient

                                                            FID for two 0.3 mm i.d.
                                                            tubes of 0.2 mM trityl-
                                                            CD3 separated by 1.8 cm
                                                            4096 data points,
                                                            199,680 averages
                                                            Decay time constant is
  0      5        10
                       time, ns
                                  15           20    25     2.88 s.
                                                                      6.56x10 -8
                                                            B pp                  23 mG
                                                             Zero-filled to 16k, FFT
                                                             Full-width half-height =
                                                             35 mG
                                                             Bpp = 21 mG
-0.002       0                         0.002        0.004
                 frequency (GHz)
              FID-detected projection

                                             Two 0.3 mm i.d. tubes of
                           Absolute Value
                                             0.2 mM trityl-CD3
                           Spectrum          separated by 1.8 cm.
                                             Gradient of 0.34 G/cm
                                             99,328 averages
                                             4096 points, zero-filled to
                                             Phase determined by
                            Real component
                                             starting point in the FID.
                                             Signals are sharper in the
                                             real component of the
                                             absorption spectrum than
-0.004          0
         Frequency (GHz)
                                  0.004      in the absolute value
         Comparison of spin echo and FID

                                            Same two 3 mm id tubes of
                                            0.2 mM sym-trityl
                                            separated by 1.8 cm.
                                            Gradient of 0.55 G/cm and
                                            99,328 acquisitions
                                            At this higher gradient the
                                            FID decays so quickly that
                             spin echo      the resolution in the Fourier
                                            transform is poorer than for
                                            the transform of the spin
-0.004            0                 0.004
                                            echo data.
           Frequency (GHz)
Block Diagram of DU Bridge

To top