STATE OF THE ART AND FUTURE DIRECTIONS FOR THE

ATOMIC HYDROGEN MASER

Robert F.C.Vessot
Smithsonian Astrophysical Observatory, Cambridge Massachusetts
ABSTRACT
The present stztus of technology development for atomic hydrogen masers (H-
masers) is reviewed. The limitations to frequency stability and accuracy are discussed
with emphasis on the problems associated with cavity resonator instability and the lack
of reproducibility and stability of the storage volume wall coating frequency shift. New
types of coatings developed in the Soviet Union and better, czvity resonator materials,
are expected to make possible frequency at the 10-l6 level at lo4 sec. Better control of
systematic effects should extend the long-term stability to levels better than 10-l5for
intervals beyond one day, Present use of H-masers as flywheel oscillators in timekeep-
ing systems is discussed as is the outlook for the future cryogenic and room tempera-
ture H-masers as flywheel oscillators to operate very high resolution frequency discrim-
inators based on the newly evolving technology of trapped and cooled ions and atoms.

1. INTRODUCTION
It is now 30 years since the invention of the atomic hydrogen maser1 by D. Kleppner and N.
F. Rarnsey and its technology has matured in several different directions aimed toward a variety of
uses and technical requirements.

The principal applications of the H-maser derive from its excellent short term stability. Since
the mid 1960s H-masers have been used in widely separated radio telescopes for very long baseline
interferometry (VLBI) to control local oscillators and provide timing for data recording. Signals
recorded from radio sources are later brought together for correlztion to obtain extremely precise an-
gular information of their brightness distributions.

H-masers are now almost exclusively used for tracking spacecraft. Range information is ob-
tained by measuring the time delay of time-coded signals. Range-rate information is obtained by
measuring the Doppler shifts of transponded signals. Angle information is derived using the VLBI
technique, where simultaneous measurement of spacecraft signals are made at two or more widdy
separated tracking stations.

Since the H-maser can provide frequency stability at better than 1 part in 1014over intervals of
several days, during the past decade it has proven itself as a flywheel oscillator in timekeeping appli-
I cations. With the advent of the United States' Global Positioning System and the Soviet Union's
1
GLONASS system, a highly precise local time scale can be kept by operating H-maser oscillators as
clocks and applying occasional time and frequency c&ections from observations of GPS and
b
GLONASS space clocks using the "common-view" technique.
 Form Approved
Report Documentation Page OMB No. 0704-0188

Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and
maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information,
including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington
VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it
does not display a currently valid OMB control number.

1. REPORT DATE 3. DATES COVERED

2. REPORT TYPE
DEC 1990 00-00-1990 to 00-00-1990
4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER
State of the Art and Future Directions for the Atomic Hydrogen Maser 5b. GRANT NUMBER

5c. PROGRAM ELEMENT NUMBER

6. AUTHOR(S) 5d. PROJECT NUMBER

5e. TASK NUMBER

5f. WORK UNIT NUMBER

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION

REPORT NUMBER
Smithsonian Astrophysical Observatory,60 Garden
Street,Cambridge,MA,02138
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S)

11. SPONSOR/MONITOR’S REPORT

NUMBER(S)

12. DISTRIBUTION/AVAILABILITY STATEMENT

Approved for public release; distribution unlimited
13. SUPPLEMENTARY NOTES
See also ADA239372. 22nd Annual Precise Time and Time Interval (PTTI) Applications and Planning
Meeting, Vienna, VA, 4-6 Dec 1990
14. ABSTRACT
see report
15. SUBJECT TERMS

16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18. NUMBER 19a. NAME OF
ABSTRACT OF PAGES RESPONSIBLE PERSON
a. REPORT b. ABSTRACT c. THIS PAGE Same as 23
unclassified unclassified unclassified Report (SAR)

Standard Form 298 (Rev. 8-98)

Prescribed by ANSI Std Z39-18
 The Soviet Union operates more than a hundred of H-masers as working time and frequency
standards in timing centers that serve their vast territories; these clocks are kept sycchronized and
syntonized by a variety of techniques.

Since this paper is intended to discuss the state of the art it is appropriate to define the nature
of the art. The plan of this pa r is first to look at the art of obtaining the very best frequency stability
F
over intervals up to about 10 seconds, how to cope with the systematic effects that impair maser long
term stability, followed by a brief description of the technologies now in use for operating H-masers
and, lastly, to offer a view of the fatme for H-Masers.

2. THE HYDROGEN MASER OSCILLATOR

The H-Maser is an oscillator powered by quantum transitions between two energy levels in
the hypefine structure of atomic hydrogen. At room temperature the population of hydrogen atoms is
nearly evenly distributed among four magnetic hyperfine levels designated by F=l, mF1, 0,-1 and
F=0, mp===. These energy levels depend on the relative orientations of the magnetic dipoles associ-
ated with the proton and the electron when the atom is in a magnetic field.

In the upper energy level, designated by F=l, the angular momenta of the proton and electron
are aligned and added; their magnetic dipoles are also aligned. In this state the total angular momen-
tum can orient itself with a magnetic field in three different directions and the F=l energy level splits .
into three components. The F=O energy level results from the alignment of protons and electrons that
cancel their total angular momentum and their magnetic dipoles oppose each other. The energy levels
of atomic hydrogen are shown in the upper part of Figure 1.

Figure 1 also shows a schematic diagram of the H-Maser oscillator. Molecular hydrogen at a
pressure of about 0.1 Tom is dissociated by an r.f. plasma discharge and collimated into a beam.
Atoms in two of the upper magnetic hyperfine energy levels (F=l, m ~ 1, = and 0) are selected by
passing through a highly inhomogeneous magnetic field generated by a multipole permanent magnet
which causes them to move toward the weak field near the axis of the magnet. These atoms are fo-
cussed into a storage bulb located in a resonant cavity tuned at the atomic hyperfine frequency. The
storage volume confines the atoms to a region where the oscillating magnetic field is in the same
Generally ,a 'IEoll-mode resonator is used, as shown in Figure 1.

As the atom proceed from the multipole magnet into the cavity bulb region, the magnetic field
they encounter field changes from about 9KGauss radially in the magnet to about one Gauss along the
axis of the beam. In this "drift region," the atoms remain in the F=l, mF=l, and 0 state and will be
kept in these states as they proceed along the drift region if the magnetic field they encounter is re-
duced to the level of the field in the resonator without sudden interruption or change of direction.

The all-important feature of the H-Maser is the surface coating that enables its operation as an
oscillator with a narrow resonance linewidth, or high line Q. This is achieved by storing the atoms
without appreciable loss of phase coherence from collisions with the wall surfaces or among each
other. At room temperature the atoms travel at about 2.5 W s e c , and in a typical two liter storage
vessel, whose collimator provides 2 one second storage time, a typical atom undergoes 1 6 collisions
before leaving the vessel
The frequency of the F=l, m~ =O to F= 0,r n =O~ transition that powers the oscillator depends
on the static magnetic field as Afm= 275 1 B* (where B is in Gauss). To avoid frequency shifts from
changes of magnetic field, H-Masers are operated at low magnetic fields, usually of about 1 rnilli-
gauss, To maintain these low fields and to provide a spatially uniform field, with variation at the mi-
crogauss level throughout the bulb, magnetic shields are placed about the resonator to attenuate the
outside ambient field, and a solenoid is placed within the innermost shield to provide a uniform and
controllable field
The line Q of the H-maser is defined as Q1=nf,& where f, is the oscillation frequency and
y2 is the total rate of
loss of phase of the atoms' oscillating &pole moment with respect to the phase of
the signal in the resonator. The parameter yz includes loss of phase coherence by escape from the bulb
and from recombination to form molecules, also loss of phase coherence from wall collisions, mag-
netic inhomogeneties and interatomic collisions. Maser oscillation is sustained when the energy re-
leased by the incoming atoms resulting fmm stimulated emission by the microwave fields in the res-
onator exceeds the energy lost by the resonator. The energy lost includes the the signal delivered to
the receiver.

The state of the art of microwave receiving systems for H-masers is an important topic that
will require a separate discussion outside the coztext of this paper.

2.1 FUNDAMENTAL LIMITS OF FREQUENCY STABILITY

The fundamental stability limit for the H-Maser is the same as for other oscillators.2
Following Kleppner and Rarnsey in reference 1, it is given as
(1)
) Allan standard deviation of the fundamental limits to frequency stability
where ~ ~ (is7 the
over time intervals, 7 . Q is the quality factor of the oscillating system operating at a power level, P,
k is Boltzmann's constant, and T is the absolute temperature.

From this expression we see that to obtzin the best stability we want high oscillation power
and high values of line Q. However these are incompatible situations since high power implies high
levels of atomic flux and, consequently, high levels of interatomic collisions.

The power levels normally generated in H-Maser oscillators are low, rarely more than -90
dBm and the signal-to-noise ratio of the equipment that receives the maser signal has a very signifi-
cant effect on the maser's short-term stability (z < 100 sec.). The effect of the added noise on the
Allan standard deviations is3
 Here F and B are the receiving system zoise figure and bandwidth, P as before, is the oscilla-
tor power and p is the czvity resocztor coup!i,n.g fictor that deternines the ?ewer delivered to the re-
ceiver.

T h e d noise, kT,appears in equation 1 as the noise power within the linewidth of the oscil-
lator and in equation 2 as the effective noise, FLT,within the bandwidth B of the receiver system. In
both cases there arc Llrivrul~gesto oper~tinga maser at low tempemtures.
The maser oscillator's power variation with bean flux is determined by the design of the maser and
can be characterized by a quantity q, which depends on the following raser parameted

Here h is Planck's constant, p, is the Bohr magneton, o(T)is the hydrogen atom's spin-ex-
change collisional cross section, v(T) is the average relative velocity of atoms, yd is the rate of loss of
phase coherence from the loss of atoms, yd is the rate of loss of phase coherence from all causes, in-
cluding loss of atoms but excluding the effect of interatomic collisions, and Vc/Vb is the ratio to cav-
ity to storage bulb volume, q is the "filling factor", the ratio of the average axial component of the r.f.
magnetic field squared to the square of the r.f. magnetic field averaged throughout the cavity. The
quantity I,, refers to the rate at which the total number of atoms enter the bulb and I is the rate at
which atoms in the desired F=l, m ~ 0=state enter the bulb.

Plots of the normalized output power to beam flux are shown in Figure 2, from reference 4,
for various values of q. Note that it is possible to stop oscillation by having too many atoms, that
smaller values of q provide higher power for a given flux, and that for q > 0.172, the maser will not
oscillate.

3. SYSTEMATIC EFFECTS
Figure 3 shows how the Allan standard deviation representing the frequency stability of the
H-Maser follows the behavior predicted by the combination of equations 1 and 2 up to a point where
systematic effects intrude on the maser's behavior, in this case for intervals beyond lo4 seconds. The
most serious systematic effect is the "pulling" of the oscillation frequency by the mistuning of the
cavity resonator. The frequency of oscillation of the aggregate of atoms stored in the maser storage
volume includes systematic frequency shifts induced by interatomic collisions and the loss of phase
coherence is proportional to these collisional shifts.5 Another systematic frequency shift that is re-
lated to the inter-atomic collision rate is the magnetic inhomogeneity (MI)frequency shift that occurs
when the following conditions pertain. 1) When there is an asymmetry in the population of the F=O,
m~ = +1 and -1 magnetic hypefme sublevels of the atoms entering the bulb and 2) when there exits a
magnetic field inhomogeneity over the volume of the bulb and 3) when there exists an asymmetry in
the distribution of r.f. magnetic fields over the volume of the bulb. Since the inhomogeneity and
symmetry requirements, 2 and 3 above, are difficult to fulfill, the MI shift is present in nearly all H-
Masers with magnetic state selectors that focus atoms in both the F=l,mF=l and m ~ 0=states.
 The least well understood systematic frequency shift in the atomic hydrogen maser is the ef-
fect of collisions with the sldaces of the storzge volume wall, known as the "wall shift". Many
questions still remain ur~esolvedresolved about the nature of the collision processes. These include
the collision interaction energy, its effect on the advance or retardation of the phase of oscillating
dipole moment of the atom and the amount of its phase decorrelation. Surface smoothness, which
determines the collision rate, depends on how the coating is applied. The variability of the wall shift
from maser to maser is the chief obstacle to achieving frequency accuracy.

Little work on surfaces has been done in Western countries since FEP-120 Teflon was
adopted in the mid-1960's. The most significant advance since then has been made in the Soviet
Union by Demidov et a1.,6 who report substantial improvement in the reduction of the wall shift by
using a new fluoroplastic that they designate as F-10. The eight-to-ten times smaller wallshift of F-10
that, in part, results from its highly superior sxrface smoothness, allows a correspondingly smaller
variability in the wall shift.
The question of long-term frequency stability of the wall shift, as yet, is not resolved.
Considerable insight about the present status of this question will be gained by reading the report on
ttle session on wall coatings that appears in these proceedings.

For a more thorough discussion on systematic frequency shifts please see the paper on the
physics of such shifts by E. M. Mattison7in these proceedings.

3.1 CAVITY RESONATOR MISTUNING

The shift in the output frequency Af,from cavity mistuning is given by

where AfR is the resonator's frequency offset from the atomic oscillation frequency. We note that the
penalty for reducing q by raising Qc is to make the maser more subject to cavity pulling. Typically,
Q, is about 5 x lo4 and Q1is about 2.5 x lo9, so that Afc = AfR x 2 x For fractional frequency
stability at a level of 1 part in 1015 we require hfR to be kept within 0.07 Hz. For frequency stability
at this level, without active frequency control by a servo system, we require good control of the res-
onator's temperature, and a cavity material with a very low coefficient of thermal expansion, along
with excellent mechanical stability. In terms of cavity dimensions for a typical TQll-mode resonator,
whose axial tuning rate is about 10 MHzIcm, the axial dimension must be kept constant to less than
7x cm. This is about the diameter of a hydrogen atom! If we assume that we can control
temperature at a level of 2 x loL5 we require a linear temperature coefficient smaller than 5 x lo4
O C Materials such as Zerodur,8 Cer-Vit,9 and ULElO are available that have comparable values of
thermal coefficients1*but this is not the whole story. There is also the temperature coefficient of the
dielectric constant of the storage bulb, which also causes a frequency shift. The resonator's
frequency is affected by everything with which it can interact. This includes the storage bulb within
the resonator, all the electronics coupled to the resonator, including the resonator tuning system, and
the circuits that couple power to the receiving system.
 There is no all-purpose, optimum technique for coping with the resonator frequency drift.
When predictability of the frequezcy drift is important, H- masers designed with the best available
materials and techniques for passive stability have been satisfactory. The ageing of Cer-Vit and the
other glass-ceramic materials is very predictable, once the initial settling-in of the resonator end covers
to their cylinder has taken place, a process that requires some three to six months.12 To control the
effect of cavity pulling there are limits to how far we can increase the value of QI, because of the re-
sulting decrease in the available output power and the loss of short-term stability. It is clear that some
form control, or of monitoring, of the resonance frequency of the cavity is desirable for long term
frequency stability of masers. Three types of active servo techniques have been used

1. By line Q modulation and searching for a tuning condition, such that bf, is zero.l3,14

2. By cavity resonance frequency modulation and adjusting the resonator, such that the out-
put signal amplitude is optirnized,ls

3. By introducing signals into the cavity at frequencies away from the oscillation frequency
and near the inflection points of the resonator's frequency response.16 The relative levels of
the signals that are sent through the cavity are compared and kept fixed by adjusting the center
frequency of the cavity.

Line Q modulation requires a frequency reference with stability comparable to that of the
maser being tuned during the period of modulation. Ideally, another maser is employed for this pur-
pose. There are usually sidebands at the modulation frequency in the output signal. This method
produces a signal whose frequency depends on the atomic resonance including wall collision shifts.
The H-H collisional shifts are cancelled by an offset in the resonator frequency because they are pro-
portional to the collisional line broadening.

Cavity resonance frequency modulation makes possible a H-maser with good stand-alone
qualities of frequency stability. As in system No. 1, the modulation signal will be present in the out-
put signal unless some form of compensation is employed. This method adjusts the resonator to the
atomic resonance frequency including the wall shift and the H-H co?!isional shift, which can vary
with beam flux .

The last technique mentioned above has been used successfully in masers having very small
cavity resonator-storage bulb combinations that have been designed using dielectric loading methods
or the lumped capacitor loading.17 Since using this type of resonator often requires external elec-
tronic gain to raise the resonator's quality factor and, under these conditions, the resonator frequency
is further subject to the phase stability of the Q enhancing amplifier.

Injecting the probing signals requires scrupulous avoidance of any additional noise or signal at
the maser's oscillating frequency to avoid pulling the frequency of the maser oscillation, This method
maintains the resonator at some pre-assigned frequency. Beam fluxvariations will cause changes in
output frequency depending on how the resoxitor frequency is related to the atomic oscillation fre-
quency.
 3.2 MAGNETIC FREQUENCY SHIFTS
Systematic frequency shifts from changes in ambient magnetic field result from two different
physical processes, 1) a change in the internal magnetic field in the storage volume causing changes
in Af, = 2751 B ~ ~ and : 2) a change in the MI frequency shift, 6(AfMI),resulting from time varia-
tions in the spatial uniformity of Bh,time variations in the hydrogen density in the bulb, and time
variation in the state distribution of atoms entering the bulb.

The frrst processes results from having finite magnetic shielding capability. Well annealed ar-
rays of high quality shielding material can achieve shielding factors, S = 6Bext/6Bintas high as 105.
If we assume a value for S of 5 x lo4, and operate with Brt at a level of 5 x lo4 Gauss we have that

Given that variation in BeXtcan be confined to 0.015 Gauss (or a few percent of the earth's
ambient field) we can expect systematic fractional frequency variations of about 6 x 10-16. It is clear
that more layers of magnetic shielding can be added, or that a magnetic field compensation servo sys-
tem can be used, if more severe conditions are encountered.
The second physical process causing magnetically induced Erequency shifts is far less straight-
forward than the process described above. It depends on the inequality of the population difference in
the F=l, m ~ = + and
l -1 states entering the bulb, the asymmetry of the RF magnetic fields in the stor-
age volume, and the DC magnetic gradient in the storage volume. Under conditions where these
populations are not equal and with the almost inevitable lack of RF symmetry and field uniformity in
today's equipment, the the output frequency of the maser will vary with beam flux. This causes
problems with the flux tuning process described in equation (3)' where the frequency shift owing to
spin-exchange collisions is assumed to be strictly proportional to the atomic density. (Even in cases
where there is no AfMI the cavity resonance is still offset in frequency by the amount required to
compensate for the spin' exchange collision frequency shift, which is proportional to the atomic den-
sity,) The stability of the magnetic inhomogeneity shift, AfMI,will depend on the constancy of the
state distribution of atoms and on the interatomic collision rate.

Removal of the MI shift can be done by providing a beam of atoms exclusively in the desired
F=l, m ~ state 4 or by equalizing the population of atoms in the F=l mF=+l and -1 states. The state
distribution of atoms entering the storage volume can be equalized by changing the direction of the
axial magnetic field in an alternating manner in the drift region downstream from the state selecting
magnet shown in Figure 1. When this field is in the same direction as the field in the bulb, the beam
proceeds in the normal states. When the drift region field is inverted under proper conditions, the
F=l states can be inverted. The r n =~ +1 population then appears in the r n =~-1 state. By alternat-
ing the direction of the drift region field the average population of the x n ~= +1 and -1 can be equal-
ized. This technique is used in the Soviet Chl-75 H-maser.
 A more effective, but more complicated method to provide a beam of atoms nearly exclusively
in the F=l, r n =~ 0 state is discussed in section 4.2 of this pqer. Using this system the entire popu-
lation of atoms will generate signal power with one half the rate of interatoric collisions and provide
an improvement in the line Q.

3.3 WALL COLLISION FREQUENCY SHIFTS

The most fundamental problem today with the H-maser is the lack of reproducibility and time
stability of the wall coating frequency shift. The wall shift is the principal source of maser frequency
inaccuracy. Since the wallshift is proportional to the wall collision rate multiplied by the phase shift
per collision, ideal coatings should be as smooth as possible and of a material capable of producing
the smallest possible phase shift per collision. The original masers built in 1960-1962 used coatings
of dimethyl-dichloro-silane (dri-film). In about 1962 Teflon was found to be far superior to dri-film
and Teflon has been used since that time.

Coatings of PTFE @oly-tetra-fluoro-ethylene),a long-chain fluorocarbon with a high melting

point were originally used. These coatings are difficult to apply and are rarely used today. When a
PTFE coating is slowly cooled after melting it crystallizes and has a wallshift with a small temperature
coefficient. On the other hand, when it the coating is rapidly cooled after melting, it has a larger wall-
shift and a larger dependence with temperature. It has zero wallshift at about 82 degrees C.18 PTFE
has rarely been used since 1965.

FEP fluorinated-ethylene-propylene,a branched fluorocarbon with lower melting point than

PTFE is far easier to apply that PTFE, and to date, has been the most widely used coating in masers
made in the US and Europe.

A far superior coating called "Fluoroplastic F-10"has been developed in the Soviet Union
specifically for H-masers.lg It melts at a lower temperature than PTFE, flows evenly over the surface
and has 8 to 10 times lower wallshift than PTFE and FEP Teflon. This new material should
substantially improve the accuracy capability of H-masers.

4. A SUMMARY OF TECHNOLOGIES USED IN TODAY'S

HYDROGEN MASERS
4.1 HYDROGEN MASER RESONATORS
The most often used resonator operates in the TEOllmode. Without appreciable dielectric
loading by the bulb, the resonator's typical dimensions are of a cylinder 28 cm dia. x 28 cm long.
This allows use of storage bulbs of 2 to 3 liter capacity. Smaller resonators have been made using
dielectric loading. These resonators suffer large variations of resonance frequency with temperature
owing to the thermal coefficients of dielectric constant of the loading material and require more stren-
uous autotuning or thermal control than unloaded resonators.
 Resonztors with smaller dimensions, using lumped capacitance loading to reduce dimensions,
are used in passive H-masers md regenerated-Q H-masers. This resofiator is in use in the Soviet
Chl-76 H-maser, in some masers built by the Sigma-Tau Standards Corporation (where the idea
originated) and in the small regenerated-Q spaceborne masers developed at the Hughes Corporation.

Stabilization of the resonator can be done either by passive means, using materials of high
mechanical stability and with very low coefficientsof thermal expansion, or by active means, dis-
cussed in 3.1. Rough tuning of the resonators is usually done with a mechanical device that in some
way slightly alters the internal configuration of the resonators r.f fields. Fine tuning (over a range of
a few tens of Kilohertz) is usually done with a varactor coupled to the resonator either directly, as part
of a coupling loop or probe, or via a transmission line.

4.2 ATOMIC HYDROGEN STATE SELECTION

The H-maser uses traditional molecular beams techniques with multiple magnets to concen-
trate atoms in the desired quantum states into a beam aimed into the collimator of the storage volume.
Four-pole magnets and six-pole magnets are commonly used. Small diameter four-pole magnets are
more efficient in their use of hydrogen than six-pole magnets. Six-pole magnets have focussing
properties analogous to optical lenses and can be used to focus a specific range of velocities from the
modified Maxwell distribution of velocities of atoms emerging from the source to form a real "image"
of the source aperture. While the narrowness of the velocity distribution can be less efficient of hy-
drogen use, these magnets are useful for operating a state selector that provides a beam of atoms
wholly in the desired F=l, m~ = 0 state.
A method to provide a beam of atoms nearly exclusively in the F=l, m~ = 0 state is shown in
Figure 4.m Here, the first magnet, in the usual way, focusses a beam into a second magnet of twice
the length that, in turn, refocusses the beam as an image of the source aperture at "stop 2" in the
figure. An Adiabatic Fast Passage (AFP) radiofrequency transition is made in the space between ths
magnets to invert the population. Only atoms in the the F=l, r n = ~ 0 states continue through the
magnet and on to the bulb.

4.3 MAGNETIC SHIELDS AND FIELD CONTROL

The multilnyer magnetic shields in use today appear to be adequate for controlling both the
level of magnetic felds and their gradients in the storage volume of H-masers. Systematic frequency
shifts resulting from the magnetic conditions in "normal" environments, with variations of a few mil-
liGauss, appear, for the present, to be well enough controlled in most masers. However, as progress
continues in the control of other systematic effects other than of magnetic origin, better shielding will
be required. A typical example of the shielding factor is S = AHcxt/AHinr= lo5 for f0.5 Gauss ex-
ternal variations in &Xt. A wider range of cancellation of external magnetic field variations can be
achieved with active servo methods by sensing the external field and applying a compensating field.
 The control of internal magnetic gradients by multi-section coils so far has proved adequate.
Simple degaussing techniqces using currents at powerlice r'=eqcexies to remove rexment ficids in
the magnetic shields have also been adequate.
Uses of superconducting materials for magnetic shields operated at low temperatures and for
cond~ctivecoatings of cavity resonators are being considered. The combination of these functions is
a very tempting prospect particularly if room temperature, or even low temperature, superconductors
become available. Superconducting shields are eminently useful and appropriate for H-masers
operating at very low temperatures.

4.4 HYDROGEN SUPPLY, PURIFICATION AND CONTROL

Two types of hydrogen supply systems are presently in use, high pressure gas bottle
systems and low pressure sources operating by heating some form of metallic hydride such as Li A1
H4to release hydrogen. These systems require control of the flow and purification of H2before it
proceeds to the dissociator. Both regulation of flow and purification are achieved by passing the Hz
through heated metal tubes, plugs, or diaphragms depending on the pressure of the supply.

Gas bottles can be easily vented and refilled to comply with airline transport regulations and
are a simple, reliable, low-tech devices requiring only a pressure gauge to give a measure of the gas
available.

Metallic hydrides are light in weight, small in volume, do not require a pressure vessel and are
thus ideal for spacecraft operation. A metal hydride system is used in the Soviet Chl-75H-maser
system along with a self-heated thin walled nickel tube for flow control and purification of Hz.

4.5 DISSOCIATORS TO PRODUCE ATOMIC HYDROGEN

In H-masers the dissociation of molecular hydrogen is done by r.f. plasma excitation. The
ideal choice of surfaces for the dissociator is one that will not adsorb hydrogen on its surface so as to
allow recombination with incoming colliding atoms. Surfaces of glasses that are free of metal oxides
and of pure fused silica appear to be the best.

Excitation of the plasma is done by exposure to r.f. fields, either by electrodes that prodwe
predominantly electric fields, or by a coil that produces induced displacement-current fields. The lat-
ter method causes less erosion of the glass since it is essentially an electrodeless discharge, but re-
quires a means for starting the discharge such as a spark coil or a miniscule amount of radioactive "a"
particle emitter as a "keep alive" to provide ionization to initiate the plasma discharge. The plasma in
the glassware presents a variable impedance to the excitation circuit that depends on the hydrogen
pressure. This variability can cause difficulties in matching the r.f. power to the plasma. Single
transistor oscillator circuits are commonly used, where the glass or quatrz cell containing the plasma
is included in the resonant circuit. In these circuits the resonant frequency adjusts itself to changes in
the reacta'ce of the cell wher. the plasma is started and as the hydrogen pressure is varied. These de-
vices are simple and cheap, but often temperamental and difficult to stabilize.

External r.f. power sources with a wide match to the plasma cell are less power efficient,
but more predictable, Another advantage is that external power sources can be operated at known and
fixed frequencies well away from frequencies whose harmonics, if leaked into the maser receiver,
could cause problems. The capability of monitoring of incident and reflected power to the plasma is
another useful feature.

4.6 VACUUM SYSTEMS AND HYDROGEN SCAVENGING

The use of the ion pump was probably the most sigcificant departure in the evolution of the
H-maser from the experimental laboratory environment to the marketplace. Ion pumps are still widely
used and some have been specifically designed for pumping hydrogen. For pumping hydrogen the
situation is different from the normal ion pump process where a hail of sputtered cathode material
particles bury adsorbed gas molecules on surfaces in the p u q . Since hydrogen is readily assimilated
by the titanium cathodes, in principle there is no need for ionic bombardment and sputtering.
However this process is substantially assisted by the scrubbing of the cathodes by the impacting of
hydrogen ions. Other species of ions, being heavier, cause the usual pumping by sputtering and
burial of gases.

Early ion pumps used in H-masers had cathodes in the form of plates that were attached at
their comers and were subject to bending inward and short circuiting to the anodes as the engorged
hydrogen warped the plates. This problem was easily cured by adding extra supports to the cathodes.
The principal mode of failure if ion pumps in H-masers is related to local stress caused by hydrogen
in titanium cathodes.21 The stress causes spalling and flaking with momentary arcing and eventual
short circuiting. Methods for improving the lifetime of operation incl~deannealing of the plates at
high temperatures under high vacuum and the selection of high purity titmiurn.

The use of sorption technology to scavenge hydrogen in H-masers began in 1973 in an effort
to reduce the weight of a maser for a space experiment to measure the gravitational redshift,22
Because sorption cartridges will pump hydrogen almost exclusively at room temperatures the addition
of a small ion pump was found to be necessary to pump other gases. This ion pump, which is usu-
ally operated at a lower than normal voltage to rnirnize hydrogen ion bombardment of its cathode,
serves to dissociate hydrocarbon gases allowing the freed hydrogen to be pumped by sorption. With
sufficient voltage it will operate at an acceptable speed to cope with argon, nitrogen, oxygen etc.

Sorption systems, with an ion pump backup are used in the Soviet Chl-75and Chl-76 H-
rnasers,u in the regenerated Q space masers developed at Hughes, and in the Smithsonian
Astrophysical Observatory's VLG-12 masers operating with the AFP single state selector. Because
of the sensitivity of sorption systems to contmination, use of all-metal seals is recommended in place
of elastomer seals.
 Sorption cartridges that have reached their pumping capacity can repeatedly be reactivated by
qplying power to their internal hezters and pumping away the evolved hydrogen. The activztion
temperatures zre high, near 700' C, and provision must be made to prevent damage to other nearby
systems during activation. Because this process does not require dissassembly of the maser, there is
little liklihood of contamination of the maser's storage volume wall coating during activation.

Precautions are taken in maser vacuum systems to avoid materials with ferromagnetic proper-
ties for structures within the innermost magnetic shield. Titanium, copper, aluminium and carefully
selected silicon bronze are some of the materials used inside the magnetic shields. Outside of the
magnetically sensitive regions, vacuum manifolds are usually made of stainless steel, often with cop-
per Con-Flat seals. Indium seals have been successfully used in conjunction with components made
of softer metals such as copper and aluminium.

5. A LOOK TOWARD THE FUTURE OF ATOMIC HYDROGEN

MASERS
Atomic hydrogen masers are principally used as flywheel oscillators with outstandingly g o d
short term frequency stability that reaches levels deep in the 10-l6domain over time intewals between
1 and lo5 seconds. Today, on average, the accuracy of frequency reproduction of masers is about
5x10-l4 and it is clear that H-masers do not compete with cesium beam devices as a primary stan-
dxrds.

The advent of trapped atoms and ions cooled by laser interactions has led to possibilities for
frequency discriminators capable of extremely high resolution. These devices will require oscillators
having very high frequency stability and capable of producing signals with frequency spectral densi-
ties commensurate with, or narrower than, the linewidths of the new discriminators. It is likely that
H-masers will be used to supply these signals in future primary frequency standards.

H-masers will continue to be used in applications where the best possible stability is required
for intervals beteen 1 second and lo5 seconds. The limitations imposed by systematic effects that be-
come evident beyond l ~ ~ s e c o n dare
s now better understood and, as better wall coatings surfaces
such as of the Soviet 'Fluoroplastic F-10" material are used, the fundamental limitations to frequency
stability and will be improved. Frequency stability at the level of one part in 1016should be attainable
in the near future with the best presently available coatings. Further improvement beyond this level
should result with even better wall coatings.

The most recent breakthrough in H-maser technoIogy was made in 1986 by operating R-
masers at temperatures near 0.5 X using storage volume surfaces of superfluid helium-4.
Researchers at the Massachusetts Institute of Technology,24 the University of British Columbia,25
and at Harvard University26 all succeeded in achieving sustained oscillation. These research groups
all used very different experimental configurations of cryogenic H-masers.
 Cryogenic apparatus for operating apparatus at temperatures near 0.5 K is no longer consid-
ered exotic. The use of continuously operated 3 ~ recirculating
e refrigerators or of dilution refigera-
tors for operating H-masers is defhitely within today's state of the art.

The limits to the stability of cryogenic H-masers still have to be measured. New theoretical
quantum mechanical analyses of H-H collisions have been made that predict limitations due to previ-
ously unknown processes related only to the storage time of atoms in the maser.27 Research is now
in progress to measure the properties of low temperature hydrogen collisional interactions with hy-
drogen and helium to test the new theory.

Figure 5 shows the layout of the cryogenic H-maser mentioned in reference 25. Its frequescy
stability, projected from all known effects of its design is given in Figure 6. This plot includes lirni-
tations owing to receiver noise resonator instability, and the instability of the predicted new frequency
shifts. A plot of the spectral density of the maser's spectral density of phase fluctuations is shown in
Figure 7. This figure shows the white phase noise corresponding to the T-' behavior of the stability
plot shown in Figure 6.6.

6. SOME OBSERVATIONS ON PRESENT AND FUTURE

H-MASERS
In this paper the writer has tried to shed some light on the state of the art of atomic hydrogen
masers. Since it is only recently that we, in the United States, have had the opportunity to appreciate
maser developments made in the Soviet Union, this brief survey is bound to be incomplete.

In comparison with the technological development of other types of frequency standards in the
West there has been only a few people involved in H-maser development. In the West it is only re-
cently that development is headed toward designs suitable for small scale production. On the other
hand the Soviets have placed far more reliance on H masers as working standards than we have in the
West. They have already produced many hundreds of H-masers, and are now offering both active
and passive H-masers for sale.

The market for H-masers in the West has, so far, been limited. However applications for
masers are growing in number as signals of high frequency stability rather than of high accuracy be-
come more in demand in view of the significant improvements in time transfer offered by two -way
time transfer with communication satellites, the Global Positioning System and the Soviet Uniofi's
GLONASS system.
The development of H-masers still has a long way to go to reach its full potential as a highly
stable oscillator. Improvements in Lie long term stability of room temperature H-masers are forseen
by use of new wall coatings and better control of systematic effects. Cryogenic H-masers wth stabil-
ity at the 10-l7to lovJ8level and outstandingly high spectral purity will likely serve as oscillators for
operating standards based on fume high resolution frequency discriminators using trapped and
cooled ions and atoms. The unique qualit of the H-maser as an active oscillator, with unparalleled
5Y
frequency stability for intervals up to 10 seconds and excellent long-term predictability of its fre-
quency variation, has led to its successful use as a "flywheel clock" a: the U.S. Naval Observatory
and at other thing centers.

The octstanding long-tern (year-to-yes) frequency stability of the ensernble of Soviet H-

masers reported by N. B. Kosheyaevsky and S. B. Pushkin28 and the predictability of the behavior
of H-masers discussed by Uljanov, Dernidov, Mattison, Vessot, Allan and Winkler?9 have proven
the validity of new strategies for timekeeping. Precise time transfer by operating with common view
GPS and Glonass signals makes possible the precise comparison of H-maser frequency over the
long-term and enables these devices to serve as excellent flywheel standards.

Further advances in time transfer using active H-masers in orbiting spacecraft to permit time
transfer at the few picosecond leve! will be required to keep up with the performance of future pri-
mary standards based on f i q u e ~ c ydiscriminators operating with trapped. ion and cooled atoms and
ions.

It is clear that the technology of the atomic hydrogen masers, which is now only 30 years old,
has still to reach its full maturity.

ACKNOWLEDGEMENTS
The presentation and contents of this paper have benefitted greatly by discussions with my
colleague, Dr. Fdward M Mattison, with Mr. David W. Allan of the National Institute of Standards
and Technology and with Dr. Gernot M. R. Winkler of the U. S. Naval observatory My participation
at the F"ITI was supported by the Smithsonian Institution's Research opportunities Fund.

REFERENCES
1 D. Kleppncr, H.M. Goldenberg ," Theory of the atomic hydrogen maser,"Phys. Rev. vol. 126, 603-615,
1962.
2 W A. Edson "Noise in oscillators,"F'roc. IRE ,vol. 48, pp. 1454-1456, 1960.
3 R. Vessot, L. Mueller, and J. Vanier ,'The specification of oscillator characteristics from measurements made
in the fquency domain," Proc IEEE, vol. 54, no 2, pp. 199-207,1966.
4 D. Kleppner, H. C. Berg, S.B.Crampton, N.F.Ramsey, R.F.C. Vessot, H.E. Peters and J. Vanier,
"Hydrogen-maser principles and techniques," Phys. Rev. ,Vol138, Number 4A, 1964
5 .
S B. Crampton, "Spin exchange shifts in the hydrogen maser," Phys. Rev. vol. 158, pp. 57-61, 1967.
6 N.A. Dcmidov. E.M. Yezhov. V.A, Fdorov, A.A. Ul'janov. "The scarch for ncw materials in the hydrogen
generator,"UDK 621.3 17.7.6.089.69.
7 E.M.Mattison, "Physics of systematic frequency variations in hydrogen masers," in these proceedings.
8 Zerodur is a trademark of Schott Glasswork, Inc.
9 Cer-Vit is a trademark of Owens-Illinois,Inc.
W is a .trademark of Coming, Inc.
E.M. Mattison and R.F.C.Vessot, 'Time and tempratwe stability of silver-coated cc~aiicsfor hydrogen
maser resonant cavities," Proc. 20th Annual Precise Time and Time Interval (F'ITI) Applications and
Planning Meeting. p. 313, 1988,
RP.C. Vessot and E.M. Mattison, W. J. Klepczynski, I.F. Silvera, H.P.Godfried and R.L. Walsworth,
"Results of two years of hydrogen m a r clock opedon at the U.S.Naval Observatory and ongoing research
at Harvard-Smithsonian Center for Astrophysics", Proc. 17th Annual. Precise Time and Time Interval P.T.T.I
Meeting, Washington, D.C., Dee. 3-5, 1985; pp. 413431.
J. Vanier and RP.C. Vessot, "Cavity Tuning and Pressure Dependence of Frq~encyin the Hydrogen Maser,"
Applied Physics LRtters Vol. 4, Number 7, 1964
J. Vanicr, H.E. Peters and R.F.C. Vessot, "Exchangc Collisions, Wall Interactions, and Resettability of the
Hydrogen Maser," IEEE Trans. on Imtrumentatwn and Measurement, Vol. IM-13, Number 4, pp. 185-188,
December, 1964
.
H.E. Peters, "Design and Performance of New Hydrogen Masers Using Cavity Frequency Switching Servos,"
Sigma Tau Standards Corporation, Proc. of the 39th Annual Symposium on Frequency Control, pp. 420425,
1984
C. Audoin, "Fast Cavity Auto-tuning systems for hydrogen masers," Revue Phys. Appl. Vol 16, pp. 125-130,
1981.
H.E. Peters, "Atomic Standards with reduced size and weight," U.S. Patent Number 4123777, October 31,
1978.
R.F.C.Vessot and M.W. Levine, "Studies of Hydrogen Maser Wall Shift for High Molecular Weight
Polytetraflourocthylene,"Proc. 24 th Annual. Symposium on Frequency Control, 27-29 ,U.S.Army
Electronics Command Ft. Monrnouh, NJ. pp. 270-276, 1970.
N.A. Dernidov, E.M. Yezhov, V.A. Feodorov and A.A. Ul'janov, ref. UDK62 1.317.7.4.089.691
E. M. Mattison, R.F.C. Vessot and W.Shen, "Single-state selection system for hydrogen masers," IEEE
Transactionr on Ultrasonics,Ferroelectronics, and Frequency Control. UFFC-34, No. 6 , pp. 622-628, Nov.
1987.
N.A. Dernidov, E.M.Ezhov and A.A. Ul'janov, "On the duration of hydrogen evacuation by magnetic-
discharge pumps," I m t r m n t s and Experimental Techniques, Vol. 21, pp. 739-740,1978.
R.F.C. Vessot, M.W. Levine, E.M. Mattison. T.E. Hoffman, E. A. Imbier, M. Tetu, G. Nystrom, J. J. Kelt,
Jr., H.F. Trucks, and J.L. Vaniman, "Spaceborne hydrogen maser design," Proc. of the Eighth Ann. Precise
Time and Time Interval Applications and Planning Meeting, Washington, D.C.,December; pp. 277-354.
N. A. Dernidov and A. A. Uljanov, "Design and industrial production of frequencystandards" in thcse
proceedings.
H.F.Hess. G.P. Kochanski, J. M. Doyle, T. J. Greytak, and D. Kleppncr, "Spin-polarized hydrogen
maser,"Phys. Rev. A. Vol. 34, 1602 (1986)
25 M.D.Hurliman, W.N. Hardy, A.J. Berlinsky, and R.W.Cline, "Recircukhg cryogenic hydrogen
maser,"Phys. Rev. A, Vol34. 1605 (1986).
26 RL. Walsworth, I. F. Silvera, HF. Godfried, C.C. Agosta, R.F.C Vessot and E.M. Mattison, "Hydrogen
maser at temperatures below lK," Phys. Rev. A, Vol. 34,2550 (1986).
27 J.M.V.A. Koelrnann, S..B. Crampton, H.T.R.Stoof, O.J.Luiten and B.J. Verhaar, "Spin exchange frequency
shifts in cryogenic and room-tem~raturehydrogen masws," Phys Rev. A, Vo!. 38, Number 7, pp. 3535-3547
(1988)
28 N. B. Koshalyaevsky and S. B. Pushkin "'USSRNational time unit keeping over long intervals using an
ensemble of H-masers" in bese proceedings.
29 A. A. Uljanov, N. A. Dcmidov, E. M. Mattison, R. F. C. Vcssot. D. W. Allan and G.M. R. Winklcr "A
comparison of Sovict and U. S. hydrogen masers" in thcse proceedings.
 E 0 = hv 0 4
t
H lPROGES
I HZ - lo-' TORR
FLUX COSTROL ]

R. F. DISCHARGE
DLSSOCIATOR

6 POLE (OR 1 POLE)

STATE SELECTOR
MAGSET -9000 GAUSS

mF = '1
STATES DE-FOCTSSED
m~ = 0
WlTH AXL4L
hlAGKETIC FLELD COIL

- -
1 L F = I. m F = tl STATES FOCUSSED
1 1 F=l,rn~=O

STOR4GE BULB Ts - 0.5 SEC

A E A T -h
-1 a
. a A" --AT1
AV
OMIC Q = L, -
T~ I o9
SOLENOID
FOR USIFORAI
5 x 10-4 GAUSS
QUAYTIZLKC
hLAGEiETIC
FIELD
E 0 = hvo

hlICRO\VA\'E -- \
\
RESOSATOR

OUTPUT
I
v = 1420405751.68 Hz 5 X 1om4 GAUSS

ISOLATOR

LOW NOISE
PREAhI P.
Q

Figure 1. Energy levels of atomic hydrogen and a schematic diagram of the

H-maser.
 lth

Figure 2. Oscillator power versus flux of atomic hydrogen

Prom "Hydrogen Haser Principles and Techniques1'
D. Kleppner, H.C. Berg, J.B. Crampton,
N.F. Ramsey, R.F.C. Vessot, H.E. Peters
and J. Vanier, Phys. Rev. 3, May, 19.65.
Figure 3. S t a b i l i t y of H-masers showing the e f f e c t of systematics.
 SCHEMATIC DIAGRAM OF STATE SELECTION SCHEME
TO OBTAIN HYDROGEN ATOMS ONLY IN THE
F = l , M,=O HYPERFINE STATE

HEXAPOLE HEXAPOLE MASER B U L B

ENTRANCE

STOP 2 STOP 3

INVERT
F = I STATE
POPULATION
\\

Figure 4 .
 CRYOSTAT
VACUUM 4K
ENVELOPE SHIELD
TO MIXING CHAMBER
77 K
SHIELD
t
- He INLET TUBE

TEFLON SEPTUM
AND COLLIMATOR

UNIFORM FIELD

APERTURE FOR
i
HYDROGEN BEAM

TEFLON LINED TUBE

The d i u n s i o n s of the v e e r
assembly, including its
PUMPOUT-Ad K!&E~!lSSOCdTOR
magnetic shields, are
Figure 5 . Layout of the cryogenic El-maser.
22.8 cm (9") dia. x 45.7 cm
(18")high.
Figure 6.
AVERAGING TIME INTERVAL - T (sec) (shovn i n Figure
S t a b i l i t y p r o j e c t e d f o r t h e cryogenic H-maser 5)
 -80 I I 1 1 1 1 1 1 1 1 1 1 1 1111 I 1 1 1 1 1 1 1 ~ I I I I 1 IIIt I I 1 1 1 111

-90 -
/ / / I // / I /
\
/) - - itLG-11
MASER WHITE PHASE NO ISE LEVEL AT -92 dBC
WITH OUTPUT POWER LCVEL -97 ddM
LUUP BANDWIDTH 4 HZ

-100 -
-110 --
LOOP BANDWIDTH 40 HZ
COLD MASER WHITE PHASE NOISE

-120 - LEVEL A T -118 dUC

AS CONFIGURED MAR. 1988, TB = I0 SEC
LOOP BANDWIDTH 100 HZ

PHASE SPECTRUM

L E V E L AT - 1 2 6 dUC
-130 -

-140 -

-150 I I I 1 1 1 1 1 I I I 1 I 1111 I 1 1 1 1 1 1 1 1 I I I 1 1 1 1 1 1 1 I 1 1 t l l L
1 10 100 1,000 10,000

f ( F O U R I E R FREQUENCY I N HZ)
SPECTRAL OENLITY OF PHASE FLUCTUATIONS
AT 1 0 0 MHz

Figure 7. Spectral density of phase fluctuations at 100 MHz for the cryomaser and conventional H-masers.