`
`
`PETER F. MOULTON, SENIOR MEMBER,
`
`tF,Et=.
`
`Invited Paper
`
`.soIi(l—sIuIc mcrliu /mrc
`Tultablc lasers‘ based on l()II~(/()p(.’(/
`enrwgerl from the /rlbora/ory into ('(.)IIlI7l(’I‘('i(I/ ]>mtlu<’/io/r
`in the
`/(IS! (lccude, and /raw! formrl sigiirficmtt rtpp/icalioiz in p/totontc
`sy.rtcn1.r rcsearc/r. In the future. I/icy ntay be a t't'i/icrtl L‘(1III])()Il(‘IIl
`of 5pat‘c*—/Jasea’ renmze .\‘(’It.\'lI1g systcrits, communirulions .s'_\'mvrts
`and laser /ncrlicine. In I/iris‘ article we /‘ct'r'cn= the limit" /1/z_\'.rr'('.s'
`ofttuirtblc solid~stme [[l.\‘L’l‘S and disc-tr.v.v I/ll.’ C‘/Ill/‘(I(‘!(’I'i.\'li£‘S of/lit’
`most significtz/it e$m})li.s'/zed rmrl enrarging .S’_V.\'l(‘/11.3‘.
`
`l.
`
`lNTRODUCTlON
`In the last decade tunable solid—state lasers have become
`an increasingly significant component of quantum elec-
`tronics. While they were first demonstrated in the l96t)‘s.
`they have only recently emerged from being a laboratory
`curiosity to playing key roles in variety of ClCClrt)~t)pli(.‘S
`systems. Compared to what, until recently. was the most
`widely used tunable system. the liquid—medium dye laser.
`solid~state laser media offer unlimited operating and "shelt“‘
`lifetimes along with the capability to store energy and thus
`generate high peak powers via Q—switching.
`in addition,
`for some systems one can obtain either a broader tuning
`range than a given dye or an extension to longer infrared
`wavelengths. Current applications of tunable $0ll(.l'5lZllC
`lasers include basic scientific investigations in atomic.
`molecular and solid-state spectroscopy, laboratory studies
`of new semiconductor and tiber—optic devices. generation
`of ultrashort pulses and amplification of such pulses to high
`peak powers. Future applications now under development
`include aircraft—and space—based remote-sensing lidars. sub-
`marine communication systems and laser medicine. This
`article will
`review the field by first providing a brief
`discussion of the physics of tunable solid—state lasers and
`then examining a variety of laser systems. as categorized
`by the laser—active ion.
`(To be precise. we will cover
`paramagnetic—ion lasers only; other solid—state laser media
`employing so-called color centers are considered as an
`entirely different class of laser systems.)
`
`It.
`
`PHYSICAL BACKGROUND
`
`Solid-state lasers operate on stimulated transitions be-
`tween electronic levels of ions (activators) contained in
`Manuscript received August
`lél. lW|; revised October 2‘),
`lltttl.
`The author is with Schwartz Elcctro-Optics, lnc.r C‘oncord. MA (H742.
`lF.F.E Log Number 9l(l(iS33.
`
`solid crystalline or glassy media (hosts). To date. activators
`of practical significance have been positively charged ions
`from the rare Earth or 3d transition—metal groups of the
`periodic table. While all solid—state lasers can operate over
`some range of wavelengths. and thus are tunable.
`it
`is
`common to consider “tunable" lasers as those capable of
`covering a wavelength range greater than several percent of
`the laser central wavelength.
`In the following we discuss
`some basic laser concepts and consider the mechanisms
`important for making a laser "tunable?
`
`A. Lirmwitltlr. Ct'0.rS .S'eCrion. and Lifclinre
`
`The physical characteristic determining tuning range is
`the linewidth of the laser transition. Two other quantities
`of general interest to laser design are the gain cross section
`and the lifetime of the upper laser level.
`If we establish. by optical pumping. a population density.
`N (per cm”). of ions in ripper laser level. and limit our
`discussion to four—level lasers where the population of ions
`in the lower level can be neglected. then the optical gain.
`G’.
`in a laser medium of length 1 (in cm) is given by the
`expression
`
`C.‘ : t‘X])((f(/\\),i\‘—/)
`
`where rr(/\) is the gain cross section in cmz. as a function
`of wavelength ,\. The linewidth of the laser transition is a
`measure of the range of wavelengths over which the cross
`section is large, and is usually given as the t‘ull~width at
`hall‘-mttxirnum. ie. the span between the halllgain points
`on either side of the peak wavelertgtlt. (Unfortunately.
`in
`the field of quantum electronics, wavelength and frequency
`are interchanged all
`to frequently. We will generally stay
`with wavelength as the convention for this paper when
`rcfcrring to laser radiation. except for cases where the use
`of frequency is common.)
`The population of the upper laser level. in the absence of
`laser action. decays at it rate determined by the combined
`effects of radiative (or. spontaneous) emission and other
`nonradiative processes considered below. in simple systems
`the rate is constant. which leads to an exponential decay
`oi’ the population after the pumping is turned off. and :1
`characteristic lifetime given by the inverse of the decay rate.
`Litictimes for levels used in solid—statc lasers are in the range
`
`llll l t\’~“)E'. l ‘)1’
`
`L "Sift llll
`
`‘-T2: W02 IEEE
`
`l’R(')(‘l3El)l.\l(iS 0|" THF lEEE. VOL .\’lt_
`
`.\’O 3.
`
`.\l.»\RCH [W3
`
`ASML 1018
`
`ASML 1018
`
`
`
`from 10‘8 to 10‘2 s. The upper-level lifetime is important
`in determining how much pumping power is required to
`generate the necessary population of ions for laser action.
`For a desired steady-state population, N, the pump rate, W,
`in photons/s is simply NT, where r is the lifetime.
`If the only way the upper laser level decays is by spon-
`taneous emission to the lower laser level, the relation (first
`developed by Einstein) between stimulated and spontaneous
`emission can be used to connect the gain cross section,
`lifetime and linewidth by the following expression:
`
`am = F(,\)/\5 - tsmficr / F(A)AdA]“
`where F(/\) is the measured spectral emission curve for
`the laser transition, in power per unit wavelength interval,
`nistherefractiveindexofthc|xaer¢1'!313Landcisthe
`speed of light. If F(A) can be fit to a simple function,
`such as a Gaussian distribution, the above equariorrcan be
`simplified, but the basic result of the equation is the product
`of the gain cross section and 7' is inversely proportional to
`the linewidth.
`
`large gain
`For tunable systems with large Iinewidtls,
`cross sections‘ are possible only if the lifetime is short.
`Conversely, ‘for constant lifaimes, the gain cross section
`is inversely‘-related.to the linewidth.
`If one assumes a
`fixed length of lasenmaterial andcertain level of gain
`needed-«for lasu operation, one can readily show that
`the purnping.—po~ver' required to obtain steady-state laser
`operation is proportional to the linewidth. These kinds of
`consideratiors led early investigators to search for solid-
`state laser materials with narrow; linewidths. As techniques
`for optical pumping improved, including the use of-another
`laser as the pump source, operation on systems with large
`linewidths became possible.
`
`B. Ions in Host Crystals
`A solid—state laser host provides more than just a means
`for containment of the laser-active ions. In most systems,
`the environment -surrounding the ion serves -to modify
`the ion electronic states in a favorable way, by ‘making
`possible the interaction between electromagnetic radiation
`and the electronic states»-necessary for stimulated emission.
`In terms of quantum mechanics, the environment can induce
`between electronic states a dipole moment that would not
`exist if the ion was. in free space. For example, many of
`the laser transitions we will discuss below involve electrons
`that stay in the 3d state, and in free space the electric dipole
`moment for such transitions is parity forbidden, i.e., zero.
`Another function of the host crystal is to serve as a sink
`for excess energy produced by the pumping and lasing
`cycles of the solid-state system. Typically, the laser ions are
`excited by optical pumping, and if the pump source is either
`a gas discharge or tungsten lamp, much of the pump energy
`is absorbed into ion states higher in energy than that of the
`upper laser state. Decay of the excited ion from the higher
`states into the upper state must occur rapidly, accompanied
`by -a loss of ion energy, if the pumping process is to be
`effective. In a fourrlevel laser system even more energy
`
`MOULTON: TUNABLE SOLID-STATE LASERS
`
`_
`
`must be lost after the laser transition takes place, as the
`ion needs to rapidly decay from the lower laser level into
`the ground state. The host crystal creates a possibility for
`rapid loss of energy during the lasing cycle by allowing
`“nonradiative” transitions to occur between various states
`of the‘ ion.
`1
`In a nonradiative transition, some of the ion energy is
`convened to mechanical energy in the form of vibrations
`of the atoms in the host medium, because of vibrationah
`electronic interaction, or coupling. A quanturmnechanical
`treatrnenl,-’of vibrations in a solid yields a large set of
`discrcw. vibrational states, which are represented by a set
`of quasipanicles called phonons. Phonon energy is simply
`the vibrational frequency times 75,’ Ptanck’s constant; the
`properties of the host crystal (number and types of atoms
`in the unit cell,.the overall arrangement of atoms in the
`crystal structuee and the bonding strength between atoms)
`determine the number of distinct types of phonons and
`the distribution of phonon energies. All solid media have
`some characteristic maximum vibration frequency and thus
`a maximum phonon energy. The total number of phonons
`present
`in the host medium is related to the medium
`temperature. Every time a nonradiative transition occurs
`pbonons are created or, in more classical terms, the medium
`temperature increases. The rates at which nonradiative
`transitions take place cover a wide range and depend on a
`number of variables, including the properties of the active
`ion, the host crystal, the crystal temperature and the amount
`of energy given up into phonons.
`In many systems the
`rate is greater than 10"/s, and this is rapid enough for
`most laser pumping cycles. Even when the energy gap
`between levels is greater than the maximum phonon energy,
`there exist processes. in which multiple phonons can be
`created simultaneously, although the rate for such processes
`decreases rapidly as the number of phonons increase.
`While nonradiative transitions are crucial for promoting
`the decay of levels, they present a problem if they also cause
`decay- of the upper laser level. As we will discuss later,
`the presence of-_ nonradiative decay of the upper level can
`have a major effect on the performance of some,solid—state
`tunable lasers. ln a host crystal in which all the ions are in
`the same environment, the fluorescence quantum efficiency
`for a transition is the spontaneous emission rate divided by
`theqtotal transition rate. When all the decay is by photon
`emission the fluorescence quantum efficiency is unity.
`
`C. Broad Linewidths
`
`We first consider the mechanisms giving rise to the finite
`gain linewidth of a solid-state laser. Unlike gas lasers, the
`Doppler effect due to the motion of the ions in the host is
`not of significance. Vibrational frequencies range up to 1013
`Hz, but the ions are constrained to tiny displacements 00'“
`m) by the strong forces holding the host medium together,
`and- the resultant maximum velocities and Doppler shifts
`are relatively small.
`The most fundamental cause of a finite linewidth is the
`finite lifetimes of the upper and lower laser levels, since
`the uncertainty relation requires that the linewidth of any
`349
`
`
`
`level is the inverse of the level lifetime. To invoke quantum
`mechanics again, we note that the wavcfunction for a given
`ion level has both an amplitude and a phase. in calcttlating
`the value to be used for the lifetime of the level we must
`
`consider processes which not only cause decay ottt of the
`level. bttt also processes which change the phase. For all
`but
`the lowest
`temperatures the latter occur at
`a much
`higher rate than the former, and are due to interactions
`with the phonons of the host medium.
`in many common
`fixed~wavelength solid—state lasers, the linewidth is related
`to finite level lifetimes; values at room temperature range
`from 100 CH2 and on up. Two well—known examples are
`ruby, with the transition—mctal active ion Cr" in the host
`A1303 (sapphire), and Nd:YAG with the rare—Earth active
`ion Nd“ in the garnet—structured host Y3Al_<O,3.
`Lifetime broadening is homogeneous, that is, all the ions
`in the host are subject to the same broadening mechanism.
`For disordered host materials, such as glasses, another type
`of broadening occurs because each ion is in a slightly
`different environment from another. Since the energy levels
`of the ion depend to some extent on the surroundings,
`a distribution of ion energy levels results. The enormous
`number of ions that
`interact with a typical optical beam
`effectively create a smooth distribution of energies, and
`transitions between levels have a broadened lineshape. This
`type of broadening is referred to as inhomogeneous. since
`the properties of ions vary from one to the other. Familiar
`systems with inhomogeneous broadening include various
`Nd3*~doped glasses used as active media for laser drivers
`in inertial~confinement—fusion experiments and another rare
`Earth ion, En”. doped in silica fibers and used as amplifiers
`for l550—nm. ftber—transmission systems.
`The mechanism giving rise to the large linewidths of
`nearly all broadly tunable solid—state lasers is different from
`the two mentioned above, and comes about because of
`even stronger interaction between ions and the host than
`previously described. We refer to Fig.
`1
`to illustrate the
`mechanism.
`In the figure we plot energy as a function
`of
`the distance between an individual
`laser-active ion
`and the surrounding ions of
`the host crystal,
`for
`two
`different electronic states (1 and 2) of the ion and for
`two different levels of interaction. Our energy plot
`is the
`combined electronic energy of the ion and vibrational (i.e..
`mechanical) energy of the host crystal, and is referred to
`a configuration—coordinatc diagram. The energy versus
`distance relation is in the form of a parabola, a quadratic
`approximation to the manner in which the atoms of the
`crystal react to a change in spacing. The host vibrations we
`referred to earlier occur about the lowest energy point of
`the parabola, the equilibrium position. For this discussion
`we consider a simple mode of vibration in which all
`the
`atoms surrounding the |ascr~active ion move in and out at
`the same time, the “breathing” tnode. (in real host crystals
`there are many different modes of vibration. with much
`greater complexity of motion than the breathing mode. and
`each has a particular interaction with the electronic states
`of the active ion.)
`When the active ion changes front state 1
`
`to state 2. the
`
`350
`
`ENERGY
`
`‘
`
`** DISTANCE ~'*>
`
`(b)
`(it)
`Fig. 1. Energy as a function of distance between active inn and
`surrounding ions of host crystal. for two different electronic states.
`I and 2. in (a). surrounding inns do not shift equilibrium positions
`when active ion changes electronic states. while in (b) a shit"!
`occurs.
`
`equilibrium position may not be affected at all. as indicated
`in Fig. 1(a). This is typical for the 4f energy states of rare-
`Earth ions such as Nd, which have weak interactions with
`their surroundings. On the other hand.
`if the surrounding
`atoms are affected by the active~ion electronic state‘ they
`may react by establishing a new equilibrium position. as
`shown itt Fig.
`l(b). The latter case leads to the broad
`lincwidths characteristic of most tunable solid-state lasers.
`as we illustrate in Fig. 2.
`to
`Cmsider first an electronic transition from state i
`state 2, which would result from the absorption of energy.
`An important point about
`the transition is that
`the time
`in which it
`takes place is much shorter than the inverse
`of the vibrational
`frequencies of the atoms in the host
`medium. As a result.
`the atoms are essentially "frozen"
`during the transition. Since the atoms are vibrating.
`there
`is in fact a distribution of possible positions for the atomic
`configuration at the point of an electronic transition. and as
`Fig. 2 indicates. this leads to a wide distribution of possible
`transition energies. and thus a broad linewidth. Note that it
`similar transition for the system represented in Fig. 1(a)
`would have an energy independent of the atomic position.
`as the two parabolas have exactly the same spatial function.
`Immediately after the upwards transition both the host
`crystal and the active ion are in higher-energy states. The
`first component to lose energy is the host crystal. and it does
`so when the highly localized vibrational energy around the
`ion is dissipated into the rest of the crystal. If we invoke the
`concept of phonons again, we would say that phonons local
`to the ion are “emitted” and travel away from the ion. After
`this happens the combined ion~crystal system reaches the
`“relaxed“ excited state shown in Fig. 2. in which the crystal
`is in equilibrium, while the ion is still
`in an electronic
`excited state.
`in the absence of laser action, the ion makes an electronic
`transition from state 2 to state 1 due to a combination
`oi’ spontaneous emission and nonradiative processes. With
`laser action, stimulated emission provides another channel
`
`l’ROL‘El;'DlN(iS OF Till." lFF.E. V01, 30, NO 3.
`
`.\i.»\R('H IWI
`
`
`
`
`
`"*DSTANCE "*>
`
`Fig. 2. Enery diagram similar to Fig. l(b) with details of up-
`ward (absorption) and downward (emission) electronic transitions
`shown. Absorption and emission are spread in energy due to motion
`about equilibrium position of ions. “Zero-phonon" energy,
`the
`purely electronic difference in energy between states l and 2 is
`indicated.
`
`for the transition. As in absorption, the transition linewidth
`is broad, but the distribution of emission energies peaks at a
`lower energy than that for absorption. After the downwards
`transition occurs,
`the crystal
`is again left
`in an excited
`vibrational state, and relaxation down to the equilibrium
`condition for state'1 progresses rapidly through phonon
`emission.
`
`Figure 2 illustrates how the cycle of absorption, phonon
`emission to the relaxed excited state, stimulated emission
`into the lower level and phonon emission into the relaxed
`lower state is equivalent to a four-level system of discrete
`electronic levels. The difference is that the energy levels
`are a combination of both electronic and vibrational states,
`which leads to the use of the term “vibronic” (vibrational-
`electronic) to describe the types of transitions involved.
`Otherterms applied are “phonon-terminated” and the less
`accurate “phonon—broadened,” the latter of which also ap-
`plies, as we have noted above, to transitions of the type
`shown in Fig. 1(a).
`In Fig. 2 wesliow a classical view of how smooth and
`broad lineshapes for absorption and emission result from
`the dilference in energy between two displaced parabolas.
`A more rigorous quantum-mechanical analysis of the line-
`shapes relies on a calculation of overlap integrals between
`wavefunctions of the phonons in states 1 and 2. This
`analysis predicts t:hat_the energies of vibronic transitions
`should appear as discrete values, separated by the energy
`of an individual phonon and clustered on either side of a
`“zero-phonon” energy. The latter is indicated in Fig. 2 and
`is the difference between the purely electronic energy of
`the two states. If there was no displacement ofparabolas
`the zero-phonon line would be the observed transition
`energy in absorption and emission. As the displacement
`between the parabolas increases,
`theory shows that
`the
`intensity of the zero~phonon transition decreases while that
`of vibronic transitions increases. In actuality the discrete
`vfbronic transitions predicted by quantum mechanics are
`rarely observed, for-several reasons. First, each vibronic
`
`MOULTONI TUNABLE SOLID-STATE LASERS
`
`transition has a large linewidth from lifetime broadening,
`and the individual vibronic lines often overlap enough
`to produce a broad, featureless overall lineshape. Second,
`typically many different phonons, rather than one, interact
`with the electronic transition, and the effect of the resul-
`tant distribution of phonon frequencies obscures individual
`vibronic lines.
`We have assembled, in Fig. 3, a composite of the absorp-
`tion and emission spectra for many of tunable solid—state
`laser systems we will discuss in more detail below.
`ln
`all cases we show the emission band associated with laser
`operation and the absorption band for the same electronic
`transition. For some systems we also include highenlying
`absorption bands from other electronic transitions. In the
`systems considered emission bands connected with the
`more energetic transitions are barely, if at all, observable,
`as the excitation energy decays rapidly via phonon emission
`into the lowest-lying electronic excited state.
`
`D. Energy Storage and Ezrractiorr
`We now consider some of the issues related to Q-
`switching of and short-pulse amplification by tunable solid-
`state lasers. One of the favorable attributes of solid-state
`lasers in general
`is their ability to generate high peak
`powersby the Q-switching process. in Q~switching,
`the
`laser material is excited by the pump source but laser action
`is prevented by maintaining the laser cavity in a high-loss
`(low Q) state. When a high level of excitation in the upper
`level is attained, the cavity is switched to a hign—Q state
`i.e., the cavity loss is reduced and the laser output rapidly
`builds up from noise to a point at which a good fraction
`of the energy stored in the laser medium as excitation is
`extracted. If the rate of. extraction is much faster than the
`rate at which excitation in the medium can build up again,
`the laser output falls to zero in a short time. (After the
`extraction occurs the gain in the medium becomes less than
`the loss in the laser cavity.) Thus a pulse is generated.
`Energy in the laser cavity in the absence of gain in the
`laser medium decays exponentially with a characteristic
`time -rc, related to the round-trip time for the cavity and
`the loss fraction on each round-trip. Typical values for T,
`are in the ns range and Q~switched output pulsewidths are
`some multiple of ’T¢. A crude estimate of the ratio of the
`peak_ power generated by Q-switching to the power under
`normal conditions is T/Tc. The long upper-state lifetimes
`for most solid-state laser materials lead to high Q—switched
`powers, in contrast to dye and some gas media, which have
`ns-range values for 7'.
`Let us consider how energy storage affects the generation
`of high Q-switched powers. If the gain cross section for a
`material is too high, it may be difficult to store much energy
`in the laser medium before the Q-switch changes to the low-
`loss state. Even‘ for switches with essentially infinite loss,
`such as rotating mirrors, stored energy can be lost through
`parasitic oscillations within the laser medium, Solid-state
`lasers are particularly prone to parasitics, given that
`the
`medium has a high refractive index. High indices make
`possible unwanted laser cavities formed by reflections off
`351
`
`
`
`laser media surfaces tall in the range 5-30 I/_/cnr". for Q-
`switched pulses of approximately 10-30 ns in duration. For
`Nd:YAG lasers, with an f‘}_.,., of 0.4 J/cmg. damage and
`thus energy extraction is not a major concern. but as we
`point out below, values of Em,
`for many tunable laser
`media are l0~l00;r this level, and energy extraction from
`Q-switched oscillators is damage-limited.
`For short~pulse amplifiers, as long as the output fiuence
`of the amplifier is well below Em the pulse amplification
`is linear and energy extraction is not an issue. However.
`most amplifiers are operated to extract as much possible
`stored energy as possible from the medium. and simple
`analysis shows that
`at high fraction of the stored energy
`can be obtained when the output lluence is at
`least
`twice
`Em‘. As for Q—switched systems. the high values of I-.‘..,,
`for many tunable media prevent efficient amplifier opera-
`tton.
`
`E. Exct'ted—State Absorption
`
`As a final part of our background discussion. we consider
`the issue of excited-state absorption tF.SA). a process that
`for many tunable solid—state lasers can have a profound
`effect on system performance. Consider the energy diagram
`of Fig. 4. in which we include electronic states above the
`upper energy state for a laser transition. (We have elim—
`inated the atomic displacement
`to simplify the diagram.)
`Light can be either amplified by stimulated emission. where
`the ion drops to a lower-energy state. or can be attenuated
`through ESA, in which the ion makes a transition from the
`upper laser level to the higher—lying level. The net optical
`gain in the system is simply the difference between the
`cross section for stimulated emission and the cross section
`for ESA.
`for
`than that
`is greater
`If the cross section for ES/\
`stimulated emission laser action is impossible. no matter
`how high the level of pumping. Even if
`the net gain
`is positive. ESA raises the threshold pump power
`for
`laser action and reduces the efficiency in converting pump
`power
`to laser output power. A variation on ES/\
`is
`sometimes observed, in which the pump light is absorbed by
`a transition originating on the upper laser level. The effect
`of “pump BSA" depends, among other things. on the level
`of population in the upper level under conditions of laser
`action. It is particularly detrimental when Q--switched oper-
`ation is tlesired. since the upper population can rise to high
`levels before the Q—switch is changed to the low~loss slate.
`Prediction of both the magnitude and wavelength depen-
`dence ot‘ ELSA cross sections is difficult for most all tunable
`solid-state lasers. Our ability to generate the energy curves
`shown in Fig. 2 from theory is too limited in accuracy for
`trztnsition—metal
`ions in crystals. Conventional absorption
`measurements only access transitions from the ground state
`of the ion to higher lying states, and are generally not
`indicative of the nature of transitions from the "rela\'ed“
`excited state of the upper laser level. ESA cross sections
`can be estimated by measuring the optical gain (or loss) in
`a pumped crystal as a function of wavelength.
`
`PROCEEDINGS OF THF IEEE. VOL. 50. NO. 3.
`
`.\l;\RL‘H |‘l‘l.‘
`
`Ti:sappltl\'e
`
`Alexamrite
`
`_
`
`,
`900
`
`.,
`Mon
`
`1303
`
`
`1500
`
`Co:MgFg
`
`‘v‘2_5Joo
`
`,
`
`X
`
`
`__
`ll‘
`-
`toinfiflurtrso ‘V2’U.£)“DH—"‘z2?tZl
`Wavelo.'tgth lnrnl
`
`
`zoo '
`500"
`km
`
`l j
`
`l i
`
`Fig. 3. Absorption (solid line) and emission (dashed line) line’
`shapes as a function of wavelength for several comhinuliotts oi"
`laser~active ions and host crystal. The sharp line in the alexnndrite
`emission curve is off~scale. The absorption and emission bands in
`forsterite result from both Cr” and Cr-4+ ions.
`
`the ends of the media or by internal “bounce” paths based
`on total internal reflection. Even in systems where parasitic
`oscillations have been eliminated. ASE (amplification in
`the medium of spontaneous emission from the upper laser
`level) serves to deplete upper-level stored energy,
`For amplification of short pulses, energy storage is also
`an issue.
`in a manner similar to (Q—switching,
`a short-
`pulsc amplifier is pumped up to a high excitation level and
`then the pulse to be amplified is passed through the laser
`medium. The amount of amplification is set by limits to
`the medium stored energy.
`In Ncl:YAG practical levels of energy storage are reduced
`by both parasitics and ASE to levels well below the
`theoretical maximum possible, when all of the Nd ions are
`pumped to the upper laser level. The relatively (compared
`to Nd:YAG) low gain cross sections of most tunable solid-
`statc lasers permit high levels of energy storage in both
`(._)~switching and amplification,
`in some cases close to the
`theoretical
`limit.
`
`Unfortunately, a competing issue for Q~switched sys~
`terms. the ability to extract energy, is a major problem for
`many tunable systems. An estimate of the energy tluence
`(.1/cmii)
`in El Q~switched laser cavity can be obtained
`by calculating, from the gain cross section,
`the saturation
`tluence f§_.,.,,
`tor the laser transition, given by:
`
`E5." = /tr:/(/\rI)
`
`is Planck’s constant. Note that the above expres-
`where /1
`sion is simply the energy of a laser photon divided by the
`gain cross section. The actual fluence in the laser cavity
`is some factor, proportional
`to the pumping ratio (pump
`energy/threshold pump energy) multiplied by EN...
`Typical laser—induced damage levels for coated optics and
`
`-3.t.:
`
`
`
`Pump
`excited-state ‘
`absorption
`I
`
`i Excited-state
`: absorption
`
`
`
`Laser transition
`
`E. _ _' Qfj
`
`Sitnplitied electronic energy-level diagram showing tran-
`Fin. -8.
`sitittnx ;t\\(iCill1t.’\i with laser action. excitcd~state absorption and
`pinup L-sciterl—statc absorption.
`
`lll. DlVAt.l-NT 'l‘RANstTioN METALS
`
`xi. }-Ii.rIorii'u/ 8rIr'/tgrvtirtri
`ill
`the remainder of the paper. we will discuss specific
`utnahlc lasers. For historical
`reasons, we will start our
`tliscttssion with divalent (i.e., doubly charged) laser ions
`ltom the 3d transitiommetal group of the periodic table,
`as
`they were the first
`tunable solid-state lasers to be
`demonstrated. Indeed. they were the first examples ofa laser
`with the capability for broad, continuous tuning. in l963,
`Johnson. Dietz, and Guggenheim [ii at Bell Laboratories
`reported laser operation in the i620-um wavelength region
`lrottt crystals of Ni3”—doped MgF3. cryogenically cooled
`and pumped by a llashlamp; shortly afterwards [2],
`laser
`operation from Co“-doped MgF; and '/snF3 at various
`wavelengths between l”/50 and 2160 nm was obtained.
`Several other host crystals for Ni“ and Co“ ions were
`reported {3} in 1966. along with laser operation from a
`Ni:MgF3 crystal pumped by a CW tungsten lamp. and the
`lirst
`laser operation from the V3‘ ion in MgF3, The latter
`system. oscillating at ll2() nm, was discussed in more detail
`in a latter publication [4].
`In all cases the lasers were operated with the crystals
`cooled well below room temperature, usually at
`temper-
`atures close to that of liquid nitrogen. (77K). There were
`several
`reasons for
`this, of which the most
`important
`was thought to be the fact that the majority of the laser
`transitions at room temperature had low fluorescence quan-
`tttm elliciencics. At reduced temperatures the nonradiative
`decay rates were smaller,
`if not vanishing, and pumping
`powers became reasonable. The requirement for cryogenic
`cooling and the lack of a commercial source for laser
`crystals were early barriers to the transfer of divalent
`transition—mctal
`lasers from the initial
`laboratory experi-
`ments to widespread use.
`in the latter half of the l960’s
`the demonstration oi‘ dye lasers produced a great interest in
`tunable lasers, but in liquid, rather than solid media.
`ironically. developments in dye-laser technology were
`partly responsible for
`a new investigation of divalent
`tr2tnsition—rnetal lasers. begun in the latter half of the l97()’s
`by the author and coworkers at MIT Lincoln Laboratory [5].
`The optical pumping requirements for CW operation of dye
`lasers were such that only another laser could generate the
`
`Mlll.'l.'l’()t’\‘:
`
`'I‘LlN/\l5[.[i .'3'(H.lD-S'l3\TFZ LASERS
`
`needed pumping intensity. The concept of a lascr—pumped
`laser became widely accepted in the scientific community,
`and soon not only CW but pulsed laser—pumped dye lasers
`were standard laboratory devices. Laser pumping of solid-
`state materials was also possible, and for tunable media
`was applied to color-center lasers as early as 1974 [6]. For
`lasers where cryogenic cooling was considered necessary,
`laser pumping greatly simplified the system design.
`In
`the initial studies of divalent
`transition metal systems,
`flowing cryogenic liquids were generally used to cool
`laser rods, which were pumped through the rod cylinder
`surface by lamps. With laser pumping along the direction
`of optical gain (longitudinal pumping), conduction cooling
`of the crystal
`in a simple sta1ic—liquid Dewar could he
`accomplished, Fortunately, both Ni?‘ and Co“ ions can be
`pumped by either the 1300- or 1060—nm lines of neodymium
`(Nd)-doped solid—state lasers, while the absorption bands of
`the \/3* ion allow pumping by an argon—ion gas laser, and
`thus laser—pumped laser studies of the divalent transition
`metals were straightforward.
`The results of the work at Lincoln Laboratory can be
`summarized as follows: Analysis of laser operation in
`various Nizhdoped crystals showed that ESA of the laser
`wavelengths and possibly the pump wavelengths was sig—
`nificant and limited both laser efficiency and the maximum
`temperature at which laser operation could be obtained.
`Even at
`low temperatures, because of ESA the tuning
`range of Ni“ lasers was more limited than the fluorescence
`emission data predicted. For systems with little or no
`change in quantum cfficieney from cryogenic to room
`temperature, such as Ni:Mg0, cryogenic cooling was