a r X i v :q u a n t -p h /0106093 v 1 18 J u n 2001Algorithmic Cooling and Scalable
NMR Quantum Computers
P.Oscar Boykin ,Tal Mor ,Vwani Roychowdhury ,Farrokh Vatan ,and Rutger Vrijen 1.Electrical Engineering Department,UCLA,Los Angeles,CA 90095,USA.2.Electrical Engineering Department,College of Judea and Samaria,Ariel,Israel.3.Jet Propulsion Laboratory,California Institute of Technology,4800Oak Grove Drive Pasadena,CA 911094.Sun Microsystems Laboratories,Mountain View,CA,USA *To whom correspondence should be addressed.Email:talmo@cs.technion.ac.il Abstract We present here algorithmic cooling (via polarization-heat-bath)—a powerful method for obtain-ing a large number of highly polarized spins in liquid nuclear-spin systems at ?nite temperature.Given that spin-half states represent (quantum)bits,algorithmic cooling cleans dirty bits beyond the Shan-non’s bound on data compression,by employing a set of rapidly thermal-relaxing bits.Such auxiliary bits could be implemented using spins that rapidly get into thermal equilibrium with the environment,e.g.,electron spins.Cooling spins to a very low temperature without cooling the environment could lead to a break-through in nuclear magnetic resonance experiments,and our “spin-refrigerating”method suggests that this is possible.The scaling of NMR ensemble computers is probably the main obstacle to building useful quantum computing devices,and our spin-refrigerating method suggests that this problem can be resolved.
1Introduction
Ensemble computing is based on a model comprised of a macroscopic number of computers,where the same set of operations is performed simultaneously on all the computers.The concept of ensemble com-puting became very important recently,due to the fact that NMR quantum computers[1]perform ensemble computing.NMR quantum computing has already succeeded in performing complex operations involving up to7-8qubits(quantum bits),and therefore,NMR quantum computers are currently the most successful quantum computing devices.
In NMR quantum computing each computer is represented by a single molecule,and the qubits of the computer are represented by the nuclear spins embedded in a single molecule.A macroscopic number of identical molecules is available in a bulk system,and these molecules act as many computers performing the same computation in parallel.To perform a desired computation,the same sequence of external pulses is applied to all the molecules/computers.Finally,a measurement of the state of a single qubit is performed by averaging over all computers/molecules to read out the output on a particular bit on all computers.Due to the use of a macroscopic number of molecules,the output is a noticeable magnetic signal.It has been shown that almost all known quantum algorithms designed for the usual single-computer model,can be adapted to be implemented on ensemble computers[2],and in particular,these ensemble computers can perform fast factorization of large numbers[3]and fast data-base search[4].
Unfortunately,the wide-spread belief is that even though ensemble quantum computation is a powerful scheme for demonstrating fundamental quantum phenomena,it is not scalable(see for instance[5,6,7]). In particular,in the current approaches to ensemble computing,identifying the state of the computer requires sensing signals with signal-to-noise ratios that are exponentially small in,the number of qubits in the system.We refer to this well-known problem as the scaling problem.The origin of the scaling problem is explained in the following.
The initial state of each qubit,when averaged over all computers(a macroscopic number),is highly mixed,with only a small bias towards the zero state.At thermal equilibrium the state is
(1)
where the initial bias,,is mainly determined by the magnetic?eld and the temperature,but also depends on the structure and the electronic con?gurations of the molecule.For an ideal system,one has
leading to,meaning that the state is with probability one,and it is with probability zero.For a totally mixed system,,hence the probabilities of and are both equal to half.We also de?ne to be the initial error probability.Typically,is around for the liquid NMR systems in use[1],and can probably be improved(increased)a great deal in the
2
near future.Especially promising directions are the use of liquid crystal NMR for quantum computing[8], and the use of a SWAP operation for the nuclear spin and the electron spin known as ENDOR technique[9]. The state of an-qubit system in the ideal case is with(a tensor product of single qubit states).In general,the initial state of an-qubit liquid NMR system can be represented as a tensor product of states of the individual qubits:
(2)
This state can also be written as,a mixture of all states—the basis vectors of the system, and(for qubits)is a-bit binary string.E.g.,for two qubits,.In fact,the initial bias is not the same on each qubit[10],but as long as the differences between these biases are small we can ignore this fact in our analysis.The analysis we do later on is correct if we replace all these slightly different initial biases by their minimum value and call this value.
Currently,researchers use the so-called“pseudo pure state(PPS)”technique to perform computations with such highly mixed initial states.In this technique,the initial mixed density matrix is transformed to a state
(3)
which is a mixture of the totally-mixed state
The exponential advantage of quantum computers over classical ones[3]is totally lost in these NMR computing devices since an exponential number of molecules/computers is required for the computation, and therefore the scaling problem must be resolved in order to achieve any useful NMR quantum com-puting.This scaling problem(plus the assumption that quantum computing requires entanglement,and cannot rely on pseudo-entanglement)has led several researchers to suggest that the current NMR quantum computers are no more than classical simulators of quantum computers[7].Actually,the important con-tribution of[7]is the result that in some neighborhood of the totally-mixed state,all states are separable; hence,some pseudo-entanglement states contain no entanglement.But this work does not prove(and does not claim to prove)that current NMR quantum computers do not perform quantum computation.We,in contrast,conjecture that the PPS technique and the work of[7]form the?rst step in proving that quantum computing without entanglement is possible.
The?rst important step in resolving the scaling problem is to understand that the scaling problem is not an inherent characteristic of ensemble computers but is an artifact of the existing PPS methods.In fact,the original highly mixed state contains a great deal of information,and this can be seen by rotating each qubit separately and?nally measuring the qubits.However,the existing methods of transforming the highly mixed state into the PPS cause the scaling problem,by losing information on purpose.Furthermore, it is important to mention that for any,there is a range of bias,,not close to zero,where the currently existing methods for creating PPS work just?ne.In order to be in that range,the state of each qubit must be almost pure:,where,the error probability,satis?es(actually is already useful).Then scales well:
problem.Our“information-theoretic”puri?cation is totally classical,hence the density matrices are treated as classical probability distributions,and no explicit quantum effects are taken into consideration.In earlier work,Schulman and Vazirani[11]already demonstrated novel compression-based(and not PPS-based) alternative NMR computing,which does not suffer from the scaling problem.Their scheme is based on information theoretic tools,and it leads to Eq.(6).However,the Shannon bound on the puri?cation ability prevents purifying any reasonable fraction of bits for small values of:
2.1Shannon’s bound
Let us brie?y describe the puri?cation problem from an information theoretic perspective.There exists a straightforward correspondence between the initial state of our-qubit system,and a probability dis-tribution of all-bit binary strings,where the probability of each string is given by the term,the probability of the state in the mixed state described by Eq.(2).A loss-less compression of a random binary string which is distributed as stated above has been well studied.In an optimal com-pression scheme,all the randomness(and hence,the entropy)of the bit string is transferred to
bits,while with extremely high probability leaving bits in a known deterministic state,say the string .The entropy of the entire system is with
measured in bits.Any loss-less compression scheme pre-serves the entropy of the entire system,hence,one can apply Shannon’s source coding bound on to get.Simple leading-order calculation shows that is bounded by(ap-proximately)
and the adjusted bit is thrown away since it got dirtier.In both cases,the supervisor bit has a reduced bias(increased entropy),hence it is thrown away.However,before being thrown away,the supervisor bit is used as a control bit for a SWAP operation:if it has the value“0”,then it SWAPs the corresponding adjusted bit at the head of the array(say to the left),and if it is“1”it leaves the corresponding adjusted bit at its current place.In either cases the supervisor bit is then SWAPped to the right of the array.[Note that we use here a hybrid of English and symbol languages to describe an operation,such as SWAP or CUT.] As a result,at the end of the BCS all puri?ed bits are at the?rst locations at the left side of the array,the dirty adjusted bits are at the center,and the supervisor bits are at the right side of the array.Thus the dirty adjusted bits and the supervisor bits can be thrown away(or just ignored).
Starting a particular BCS on an even number of bits with a bias,at the end of the BCS there are(on average)puri?ed bits with a new bias.The new length and new bias are calculated as follows. The probability of an adjusted bit being in the puri?ed mixture,i.e.,,is obtained by a direct application of Bayes’law and is given by:
(7)
The number of puri?ed bits,,with the new bias is different on each molecule.Since,for each pair,one member is kept with probability,and the other member is thrown away,the expected value of the length of the puri?ed string is
(that is,less than)operations of controlled-SWAPs and SWAPs are performed to conditionally put the adjusted bit in the?rst location at the left of the array,or leave it in its current location:?rst,a controlled-SWAP is performed with the supervisor bit as a control,the adjusted bit and the bit to its left as the target.Then the SWAP operation is performed on the supervisor bit and the bit which is one location to its left.Then a controlled-SWAP is again performed to conditionally swap the two bits at the left of the supervisor bit,and again the supervisor bit is SWAPped one location to the left.These SWAP and controlled-SWAP operations are then repeated until the adjusted bit is conditionally SWAPped all the way to the?rst location of the array[the supervisor bit is at the third location in the array when this?nal controlled-SWAP is performed].Finally,the supervisor bit is SWAPped till it reaches the previously used supervisor bit.At the end of these operations all used supervisors are at the right of the array,all puri?ed adjusted bits are at the left of the array,and all the
7
adjusted bits which got dirtier are to the right of the puri?ed adjusted bits.Considering controlled-SWAP, SWAP,and CNOT as being a single operation each(hence one time step each),we obtain a total of
(9)
time steps for a single BCS operation.Actually,even if each controlled-SWAP is considered as two time steps this bound still holds,once a more tight bound is calculated.
A full compression scheme can be built by repeating the BCS several times,such that the?rst appli-cation,,puri?es the bits from to,and the second puri?cation,,acts only on bits that were already puri?ed to,and puri?es them further to.The application,,puri?es bits from to.Let the total number of BCS steps be and let the?nal bias achieved after applications of the compression be.By iterating equation(7)we calculate directly when starting with or,and after application of BCS.Then,using, and Eq.(5)with puri?ed bits,we estimate the number of bits for which a scalable PPS technique can be obtained.The results are summarized in Table1.The?rst interesting cases within the table are ;,or;,allowing up to bits.We refer to these possibilities as a short term goal.As a long term goal,up to200bits can be obtained with;,or;
.We consider cases in which the probability of the pure state is less than as unfeasible.
When only the BCS is performed,the resulting average?nal length of the string is
,so that the initial required number of bits,,is huge,but better compression schemes can be designed[11],which approach the Shannon’s bound.However,for our purpose,which is to achieve a“cooling via polarization-heat-bath”algorithm,this simplest compression scheme is suf?cient.
3Algorithmic Cooling via Polarization-Heat-Bath
3.1Going Beyond Shannon’s Bound
In order to go beyond Shannon’s bound we assume that we have a thermal bath of partially polarized bits with a bias.More adequate to the physical system,we assume that we have rapidly-reaching-thermal-relaxation(RRTR)bits.These bits,by interaction with the environment at some constant temperature, rapidly return to the?xed initial distribution with bias of(a reset operation).Hence,the environment acts as a polarization heat bath.
In one application of the BCS on bits at a bias of,some fraction(satisfying)is puri?ed to the next level,while the other bits have increased entropy.The supervisor bits are left with a reduced bias of,and the adjusted bits which failed to be puri?ed are changed to a bias,that is, they now remain with full entropy.
8
To make use of the heat bath for removing entropy,we swap a dirtier bit with an RRTR bit at bias, and do not use this RRTR bit until it thermalizes back to.We refer to this operation as a single“cooling”operation[14].In a nearest-neighbor gate array model,which is the appropriate model for NMR quantum computing,we can much improve the ef?ciency of the cooling by assuming that each computation bit has an RRTR bit as its neighbor(imagine a ladder built of a line of computation bits and a line of RRTR bits). Then cooling operations can be done in a single time step by replacing dirty bits with RRTR bits in parallel.
By applying many BCS steps and cooling steps in a recursive way,spins can be refrigerated to any temperature,via algorithmic cooling.
3.2Cooling Algorithm
For the sake of simplicity,we design an algorithm whereby BCS steps are always applied to blocks of exactly bits(thus,is some pre-chosen even constant),and which?nally provides bits at a bias. Any BCS step is applied onto an array of bits at a bias,all puri?ed bits are pushed to the head of the array(say,to the left),all supervisor bits are swapped to the back of the array(say,to the right),and all unpuri?ed adjusted bits(which actually became much dirtier)are kept in their place.After one such BCS step,the bits at the right have bias of,the puri?ed bits at the left have a bias,and to their right there are bits with a bias zero.Note that the boundary between the puri?ed adjusted bits and the dirtier adjusted bits is known only by its expected value
of bits is obtained,from which the?rst bits are de?ned as the output bits with,and the rest are ignored.If an additional puri?cation is now performed,only these?rst bits are considered as the input for that puri?cation.We refer to as the“cooling depth”of the cooling algorithm[13].
The algorithm is written recursively with puri?cation-steps,where the puri?cation step cor-responds to purifying an initial array of bits into a set of bits at a bias level of,via repeated compression/cooling operations described as follows:In the puri?cation step we wish to obtain bits with a bias.In order to achieve this we SWAP bits with RRTR bits,which results in cooling operations performed in parallel.The number of bits required for is.In one puri?cation step (with)we wish to obtain bits with a bias.In order to achieve this goal we apply puri?cation steps,each followed by a BCS applied to exactly bits at a bias.First,is applied onto bits,yielding an output of bits at a bias.A BCS is then applied onto these bits,yielding a string of expected length
tion of the entire string.[In the case of,of course,there are no BCS operations within.] At the end of that second application,a BCS is applied to bits at a bias(at locations till ),purifying them to.The puri?ed bits are pushed all the way to the left,leading to a string of expected length
bits puri?ed to.For we are promised that,and a CUT operation,,de?nes the?rst bits to be the output of.
The total number of bits used in is bits,where the bits are the ones used at the last step,and the bits are the ones previously kept.The output of
is de?ned as the?rst bits,and in case is to be performed,these bits are its input.Let the total number of operations applied at the puri?cation step,,be represented as.Note that, meaning that bits are SWAPped with RRTR bits in parallel.Each application of the BCS has a time complexity smaller than for a near-neighbor connected model(9).When the cooling is done(with )the number of additional steps required to(control-)SWAP the adjusted bit at the top of the array is less than.Thus we get.Hence,for all,
A full cooling algorithm is and it is performed starting at location.A pseudo-code for the complete algorithm is shown in Figure5.For any choice of,one can calculate the required(mini-mal)such that,and then bits(cooled as desired)are obtained by calling the procedure COOLING,where.We actually use in the rest of the paper(although is suf?cient when the block’s size is very large)in order to make sure that the probability of a successful process does not become too small.[The analysis done in[11]considers the case in which goes to in?nity,but the analysis does not consider the probability of success of the puri?cation in the case where does not go asymptotically to in?nity;However,in order to motivate experiments in this direction,one must consider?nite,and not too large blocks,with a size that shall potentially be accessible to experimen-talists in the near future.In our algorithm,the case of does not provide a reasonable probability of success for the cooling process,but does].
3.3Algorithmic Complexity and Error Bound
3.3.1Time and Space Complexity of the Algorithm
We now calculate,the number of bits we must start with in order to get puri?ed bits with bias .We have seen that and,and in particular
is a constant depending on the purity we wish to achieve(that is,on)and on the probability of success we wish to achieve(that is, on).For reasonable choices,in the range and in the range,we see that is in the range .To compare with the Shannon’s bound,where the constant goes as,one can show that here is a function of.
As we have seen in Section3.2,the total number of operations applied at the puri?cation step, ,satis?es
With the short-term goal in mind we see that can be achieved(for)with, ,steps,and bits,or with,,steps,and bits.Increasing to50only multiplies the initial length by,and multiplies the time steps by .Thus,this more interesting goal can be achieved with,,steps,and bits,or with,,steps,and bits.
Concentrating on the case of and,let us calculate explicitly the timing demands. For bits,we see that the switching time must satisfy
in order to allow completion of the puri?cation before the system spontaneously relaxes.Then,with time steps for each BCS operation,the relaxation time for the RRTR bits must satisfy
,if we want the RRTR bits to be ready when we need them the next time.As result,a ratio of is required in that case.The more interesting case of demands
,,and.Note that choosing increases the size by a factor of,and the time by a factor of.We shall discuss the possibility of obtaining these numbers in an actual experiment in the next section.
3.3.2Estimation of error
Since the cooling algorithm is probabilistic,and so far we have considered only the expected number of puri?ed bits,we need to make sure that in practice the actual number of bits obtained is larger than with a high probability.This is especially important when one wants to deal with relatively small numbers of bits.We recall that the random variable is the number of bits puri?ed to,after the round of puri?cation step–each followed by.Hence,prior to the CUT we have bits with bias,where the expected value,and we use.Out of these bits we keep only the?rst qubits,i.e.,we keep at most a fraction
(14)
12
The probability of success is given here for several interesting cases with(and remem-ber that the probability of success increases when is increased):For and we get success of the algorithm.For and we get success of the algorithm .This case is of most interest due to the reasonable time scales.Therefore,it is important to mention here that our bound is very conservative since we demanded success in all truncations(see details in appendix C),and this is not really required in practice.For instance,if only bits are puri?ed to
in one round of puri?cation,but bits are puri?ed in the other rounds,then the probability of having bits at the resulting process is not zero,but actually very high.Thus,our lower bound presented above should not discourage the belief in the success of this algorithm,since a much higher probability of success is actually expected.
4Physical Systems for Implementation
The spin-refrigeration algorithm relies on the ability to combine rapidly relaxing qubits and slowly relaxing qubits in a single system.T1lifetimes of atomic spins in molecules can vary greatly,depending on the degree of isolation from their local environment.Nuclei that are positioned close to unpaired electrons,for example,can couple strongly to the spin of these electrons and decay quickly.Identical nuclei that are far removed from such an environment can have extremely long lifetimes.Many examples of T1varying over three orders of magnitude,from seconds to milliseconds,exist in the literature.One example is the C nuclear relaxation rate,which changes by three orders of magnitude depending on whether the C atom is part of a phenoxyl or triphenylmethyl radical[9].By combining these different chemical environments in one single molecule,and furthermore,by making use of different types of nuclei,one could hope to achieve even a ratio of.
Another possible choice is to combine the use of nuclear spins and electron spins.The coupling be-tween nuclei and electrons that is needed to perform the desired SWAP operations,has been well studied in many systems by the Electron-Nuclear Double Resonance(ENDOR)technique[9].The electron spins which typically interact strongly with the environment could function as the short lived qubits,and the nuclei as the qubits that are to be used for the computation.In fact,the relaxation rate of electrons is commonly three orders of magnitude faster than the relaxation rate of nuclei in the same system,and an-other order of magnitude seems to be easy to obtain.Note also that the more advanced TRIPLE resonance technique can yield signi?cantly better results than the ENDOR technique[15].
This second choice of strategy has another advantage—it allows initiation of the the process by SWAPing the electron spins with the nuclear spins,thus getting much closer to achieving the desired initial bias,,which is vital for allowing reasonable time-scales for the process.If one achieves ,and,then a switching time of allows our al-
13
gorithm to yield20-qubit computers.If one achieves,and, then a switching time of allows our algorithm to yield50-qubit computers.
5Discussion
In this paper we suggested“algorithmic cooling via polarization heat bath”which removes entropy into the environment,and allows compression beyond the Shannon’s bound.The algorithmic cooling can solve the scaling problem of NMR quantum computers,and can also be used to refrigerate spins to very low temperatures.We explicitly showed how,using SWAP operations between electron spins and nuclear spins,one can obtain a50-qubit NMR quantum computer,starting with350qubits,and using feasible time scales.Interestingly,the interaction with the environment,usually a most undesired interaction,is used here to our bene?t.
Some open questions which are left for further research:(i)Are there better and simpler cooling algorithms?(ii)Can the above process be performed in a(classical)fault-tolerant way?(iii)Can the process be much improved by using more sophisticated compression algorithms?(iv)Can the process be combined with a process which resolves the addressing problem?(v)Can one achieve suf?ciently different thermal relaxation times for the two different spin systems?(vi)Can the electron-nuclear spin SWAPs be implemented on the same systems which are used for quantum computing?Finally,the summarizing question is:(vii)How far are we from demonstrating experimental algorithmic cooling,and how far are we from using it to yield20-qubit,30-qubit,or even50-qubit quantum computing devices?
References
[1]N.A.Gershenfeld,I.L.Chuang,Science,275,350(1997).D.G.Cory,A.F.Fahmy,T.F.Havel,
Proc.Natl.Acad.Sci.94,1634(1997).J.A.Jones,M.Mosca,R.H.Hansen,Nature,393,344 (1998).E.Knill,https://www.doczj.com/doc/ff12191724.html,?amme,R.Martinez,C.-H.Tseng,Nature,404,368(2000).
[2]P.O.Boykin,T.Mor,V.Roychowdhury,F.Vatan,“Algorithms on Ensemble Quantum Computers,”
(LANL e-print,quant-ph/9907067,1999).
[3]P.Shor,SIAM https://www.doczj.com/doc/ff12191724.html,puting,26,1484(1997).
[4]L.Grover,in Proceedings of28th ACM Symposium on Theory of Computing(1996),pp.212–219.
[5]W.S.Warren,Science,277,1688(1997).
[6]D.DiVincenzo,Nature,393,113(1998).
14
[7]S.L.Braunstein et al.,Phys.Rev.Lett.,83,1054(1999).
[8]C.S.Yannoni et al.,Appl.Phys.Lett.75,3563(1999).
[9]H.Kurreck,B.Kirste,W.Lubitz,in Methods in Stereochemical analysis,A.P.Marchand,Ed.(VCH
Publishers,Inc.,1988).
[10]Individual addressing of qubits requires a slightly different bias for each one,which is easily achiev-
able in practice.
[11]L.J.Schulman,U.Vazirani,in Proceedings of31st ACM Symposium on Theory of Computing(1999),
pp.322-329.
[12]We refer to these qubits as bits since no quantum effects are used in the cooling process.
[13]An algorithm with replaced by(different numbers of repetitions,depending on the bias-level)
could have some advantages,but will not be as easy to analyze.
[14]Actually,adjusted bits which failed to purify are always dirtier than the RRTR bits,but supervisor
bits are dirtier only as long as.Therefore the CUT of the adjusted bits which failed to purify, (which is explained in the next subsection)is the main“engine”which cools the NMR system at all stages of the protocol.
[15]K.Mobius,M.Plato,W.Lubitz,Physics Reports,87,171(1982).
15
cut
Figure1:The puri?cation step.
16
cut
cut
Figure2:The puri?cation step.The details of the operations of on the second level are shown.
17
procedure COOLING
call BCS depth
begin
Apply the BCS to the bits starting at location,and push the puri?ed bits always to the location(where,).
end
procedure SWAP
0.1 6.4unfeasible
30.1672 1.1
0.922 4.5 3.7
00.495unfeasible
0.010.5657.4unfeasible
70.0718 2.4
A PPS technique and the scaling problem
To illustrate the PPS scheme,let us?rst consider the case of qubits,with an arbitrary bias.The initial state is given by equation(1)[with replacing],
(15)
where the second form is already in the form of a PPS,.
Let us now consider the case of qubits.Since the initial state of each qubit is given by equation (15),the density matrix of the initial thermal-equilibrium state of the two-qubit system can be represented as:
(16)
For the purpose of understanding the PPS it is legitimate to ignore the difference between the of the two spins,but in practice they must differ a bit,since the only way to address one of them and not the other is by using accurate?elds such that only one level splitting is on resonance with that?eld.
For the purposes of generating the PPS,it is instructive to represent the initial state as
(17) where,and is being a binary string.The coef?cient of each,say,is the probability
of obtaining the string in a measurement(in the computation basis)of the two qubits.Thus, ,,and.In order to generate a pseudo-pure state,let us perform one of the following three transformations:(the identity),
with equal probability,so that each molecule(each computer)is subjected to one of the above-mentioned transformations.This transformation can be carried out in an experiment by applying different laser pulses to different portions of the liquid,or by splitting the liquid into three portions,applying one of the transfor-mations to each part,and mixing the parts together.These permutations map to itself,and completely
20
44瓦超高功率808 nm半导体激光器设计与制作 仇伯仓，胡海，何晋国 深圳清华大学研究院 深圳瑞波光电子有限公司 1. 引言 半导体激光器采用III-V化合物为其有源介质，通常通过电注入，在有源区通过电子与空穴复合将注入的电能量转换为光子能量。与固态或气体激光相比，半导体激光具有十分显著的特点：1）能量转换效率高，比如典型的808 nm高功率激光的最高电光转换效率可以高达65%以上[1]，与之成为鲜明对照的是，CO2气体激光的能量转换效率仅有10%，而采用传统灯光泵浦的固态激光的能量转换效率更低, 只有1%左右；2）体积小。一个出射功率超过10 W 的半导体激光芯片尺寸大约为mm3, 而一台固态激光更有可能占据实验室的整整一张工作台；3）可靠性高，平均寿命估计可以长达数十万小时[2]；4）价格低廉。半导体激光也同样遵从集成电路工业中的摩尔定律，即性能指标随时间以指数上升的趋势改善，而价格则随时间以指数形式下降。正是因为半导体激光的上述优点，使其愈来愈广泛地应用到国计民生的各个方面，诸如工业应用、信息技术、激光显示、激光医疗以及科学研究与国防应用。随着激光芯片性能的不断提高与其价格的持续下降，以808 nm 以及9xx nm为代表的高功率激光器件已经成为激光加工系统的最核心的关键部件。高功率激光芯片有若干重要技术指标，包括能量转换效率以及器件运行可靠性等。器件的能量转换效率主要取决于芯片的外延结构与器件结构设计，而运行可靠性主要与芯片的腔面处理工艺有关。本文首先简要综述高功率激光的设计思想以及腔面处理方法，随后展示深圳清华大学研究院和深圳瑞波光电子有限公司在研发808nm高功率单管激光芯片方面所取得的主要进展。 2.高功率激光结构设计 图1. 半导体激光外延结构示意图
实验八脉冲式灯泵浦YAG激光器被动调Q实验 实验目的 （1）掌握被动调Q Y AG激光器的工作原理与调试方法。 （2）测量脉冲与连续泵浦Y AG激光器的静态输出特性。 （3）分析被动调Q率被动调Q Y AG激光器的动态输出特性。 （4）在被动调Q理论分析的基础上，通过实验研究，针对相应的运转条件和应用需求，设计被动调Q Y AG激光器的光学参数。 实验原理 1．固体Nd:Y AG激光器的工作原理。 （1）Nd:Y AG晶体的性质 Nd3+:YAG是以三阶钕（Nd3+）离子部分取代Y3Al45O12晶体中Y3+离子的激光工作物质，称为掺钕钇铝石榴石（简称Nd3+:YAG）。它以Nd3+离子作为激活粒子。图8-1给出了Nd3+:YAG晶体中Nd3+离子的与激光产生过程有关的能级图。处于基态4I9/2的钕离子吸收光泵发射的相应波长的光子能量后跃迁到4I5/2，2H9/2和4F7/2，4S3/2能级（吸收带的中心波长是810nm和750nm，带宽为30nm），然后几乎全部通过无辐射跃迁迅速降落到4F3/2能级。4F3/2能级是一个寿命为0.23ms的亚稳态能级。处于4F3/2能级的Nd3+离子可以向多个较低能级跃迁并产生辐射，其中几率最大的是4F3/2至4I11/2的跃迁（波长
为1064nm）。 图8-1 Nd3+:YAG激光的激发机理 （2）静态运转特性分析 （a）脉冲运转→驰豫振荡（尖峰效应）暂态过程。 （b）连续运转→阈值条件（增益饱和）稳态过程。 按“激光原理与技术”中有关章节的分析，结合实验得出：仅仅依靠增加泵浦能量与功率，不能获得窄脉宽，高峰值功率的激光脉冲的结论。 2．Cr:YAG饱和吸收被动调Q原理 自饱和被动式调Q激光器由于器件结构简单，对激光器无电磁干扰，应用十分广泛，但由于通常的染料调Q介质，导热率极低，使其应用范围受到局限，只能用于低重复率的脉冲调Q激光器中。近年来，由于激光晶体技术的进步，我国已生产出可用于高重复率调Q的多掺Y AG晶片，制成了被动式的Q开关器件，兼备声光和染料调Q的长处，在激光医疗、激光打标和非线性光学等领域获得广泛的应用。 在Y AG晶体内，除了Nd+3之外，某些分凝系数较大的金属氧化物也能以原子或离子的形式长入晶体之中，这些异价离子将在其近邻产生具有反号电荷的离子空位或点阵空位，当一个自由电子或空穴被俘获时，就形成了吸收光的色心（Color Center）。多掺杂的YAG晶片在0.9-1.2μm波长范围内有很强的吸收，可将该晶片置于Nd:Y AG激光谐振腔内，其透过率将随腔中振荡激光（1.06μm波长）的强度变化而变化，并改变着谐振腔的Q值，因此起着被动式的Q开关作用，使激光器产生巨脉冲。 多掺杂的Y AG晶片是一种非线性吸收介质，在强光作用下由于吸收饱和而对激光
高性能全固态大功率绿光激光器立项投 资融资项目 可行性研究报告 (典型案例〃仅供参考) 广州中撰企业投资咨询有限公司
地址：中国〃广州
目录 第一章高性能全固态大功率绿光激光器项目概论 (1) 一、高性能全固态大功率绿光激光器项目名称及承办单位 (1) 二、高性能全固态大功率绿光激光器项目可行性研究报告委托编制单位 (1) 三、可行性研究的目的 (1) 四、可行性研究报告编制依据原则和范围 (2) （一）项目可行性报告编制依据 (2) （二）可行性研究报告编制原则 (2) （三）可行性研究报告编制范围 (4) 五、研究的主要过程 (5) 六、高性能全固态大功率绿光激光器产品方案及建设规模 (6) 七、高性能全固态大功率绿光激光器项目总投资估算 (6) 八、工艺技术装备方案的选择 (6) 九、项目实施进度建议 (6) 十、研究结论 (7) 十一、高性能全固态大功率绿光激光器项目主要经济技术指标 (9) 项目主要经济技术指标一览表 (9) 第二章高性能全固态大功率绿光激光器产品说明 (15) 第三章高性能全固态大功率绿光激光器项目市场分析预测 (15) 第四章项目选址科学性分析 (15) 一、厂址的选择原则 (15) 二、厂址选择方案 (16) 四、选址用地权属性质类别及占地面积 (17) 五、项目用地利用指标 (17) 项目占地及建筑工程投资一览表 (18)
六、项目选址综合评价 (19) 第五章项目建设内容与建设规模 (19) 一、建设内容 (19) （一）土建工程 (20) （二）设备购臵 (20) 二、建设规模 (21) 第六章原辅材料供应及基本生产条件 (21) 一、原辅材料供应条件 (21) （一）主要原辅材料供应 (21) （二）原辅材料来源 (21) 原辅材料及能源供应情况一览表 (21) 二、基本生产条件 (23) 第七章工程技术方案 (24) 一、工艺技术方案的选用原则 (24) 二、工艺技术方案 (25) （一）工艺技术来源及特点 (25) （二）技术保障措施 (25) （三）产品生产工艺流程 (25) 高性能全固态大功率绿光激光器生产工艺流程示意简图 (26) 三、设备的选择 (26) （一）设备配臵原则 (26) （二）设备配臵方案 (27) 主要设备投资明细表 (28) 第八章环境保护 (28) 一、环境保护设计依据 (29) 二、污染物的来源 (30) （一）高性能全固态大功率绿光激光器项目建设期污染源 (30)
文章编号:0253 2239(2003)04 0469 03 20W 腔外倍频全固态Nd YAG 绿光激光器* 冯 衍 毕 勇 张鸿博 姚爱云 汪家升 许祖彦 (中国科学院物理研究所,北京100080) 摘要: 研制了一台高光束质量的全固态N d YAG 激光器,采用KT P 腔外倍频,获得了大于20W 准连续绿光输出,峰功率可达1.15 105W,光束质量因子M 2值约为4。关键词: 全固态激光器;532nm 绿光;K T P 晶体;腔外倍频中图分类号:T N 248.1 文献标识码:A *中国科学院知识创新工程项目!大功率全固态激光器系统技术研究?资助课题。 E mail:biy ong92184@https://www.doczj.com/doc/ff12191724.html, 收稿日期:2002 03 15;收到修改稿日期:2002 06 03 1 引 言 大功率全固态准连续绿光激光器因其在工业、娱乐、军事等方面的重要应用而得到广泛的重视。国际上报道的绿光平均功率最高为美国利弗莫尔实验室的315W [1] (采用KTP 和LBO 腔内倍频)。国内多家单位在进行这方面的研究,华北光电研究所采用KTP 腔内倍频获得大于50W 的绿光,天津大学获得40W [2]。但所有这些均没有模式指标。而在实际应用中,亮度通常是决定性指标,它与表征光束质量的参量M 2成反比,与平均功率成正比,因此光束质量与平均功率是两个同样重要的因素。我们采用腔内倍频,获得了大于20W 的准连续绿光输出,但光束质量不理想。为获得高功率、高光束质量激光,一般采用主振且放大的方式(M OPA),腔外倍频也成为产生绿光的最佳方案[3,4] 。TRW 公司的Pierre 等[4] 采用双通板条放大及KTP 腔外倍频,获得了175W,1.1倍衍射极限的输出。另外,在打标等具体应用中,针对不同材料和加工要求,需要不同波长的激光,因此,多个波长间的方便切换是很重要的。显而易见,腔外倍频与腔内倍频相比容易做到切换波长时的即插即用。 我们研制了一台高光束质量的全固态准连续Nd YAG 激光器,重复频率1~50kH z 可调,平均功率可达75W 。用KTP 对其进行腔外倍频,我们获得了高平均功率、高峰值功率、高光束质量的绿光输出。 2 实验装置 谐振腔采用双棒串接配置。两个激光头为侧面抽运的Nd YAG,二极管列阵从三向对称抽运,每个激光头所用列阵最大输出功率为180W;Nd YAG 棒长64mm,直径3mm,掺杂质量分数为 0.006。Q 开关是NEOS 公司生产的,中心频率27.12MHz,1~50kHz 连续可调。两个激光头之间加90#旋光片用来补偿热致双折射。根据最大抽运时激光晶体中的热透镜大小,在设计谐振腔参量时可使最大抽运条件下的模体积较大,因此没有用光阑限模,不损失激光功率,从而获得较好的模式。在10kHz 时,1064nm 最大输出功率为75W,脉宽小于80ns,M 2?3.9。测量M 2的同时,可以计算得到输出镜处激光腰斑直径约为0.8mm 。 倍频采用KT P %类临界相位匹配( =90#, =23.5#),晶体尺寸分别为3mm 3mm 8mm 、3mm 3mm 16mm 、3mm 3mm 20mm,两端镀1064nm 和532nm 增透膜,未经透镜聚焦,直接置于输出镜后。旋转KTP 晶体使绿光输出水平偏振,旋转半波片使绿光输出最大。经过KTP 的剩余基频光与倍频光由布氏棱镜分光后测量(图1)。 3 结果与分析 图2是不同重复频率下,三块KT P 晶体在相同抽运条件下的平均输出功率。在15kHz 时,20m m 晶体的输出最大。但在较低频率,10kHz 以下,长晶体的效率明显地下降。这种现象可能是由温度效应引起的相位失配造成的。低重复频率时,脉冲峰值功率高,倍频输出达到饱和需要的作用长度降低,相位失配引起的反过程提前来临。 第23卷 第4期 2003年4月 光 学 学 报 ACTA OPT ICA SIN ICA V ol.23,No.4Apr il,2003
半外腔He-Ne 激光器的调试及参数测量 1. 引言 虽然在1917年爱因斯坦就预言了受激辐射的存在，但在一般热平衡情况下，物质的受激辐射总是被受激吸收所掩盖，未能在实验中观察到。直到1960年，第一台红宝石激光器才面世，它标志了激光技术的诞生 按工作物质的类型不同，激光器可以分成四大类：固体激光器、气体激光器、液体激光器和半导体激光器。He-Ne 激光器是继红宝石激光器后出现的第二种激光器，也是目前使用最为广泛的激光器之一。因此有必要通过实验对He-Ne 激光器作全面的了解。 2. 实验目的 1） 了解He-Ne 激光器的构造。 2） 观察并测量He-Ne 激光器的功率、发散角、横模式等性能参数。 3） 调整谐振腔一端的反射镜，观察谐振腔改变后He-Ne 激光器性能参数的变化。 3. 基本原理 3.1 He-Ne 激光器结构 He-Ne 激光器由光学谐振腔（输出镜与全反镜）、工作物质（密封在玻璃管里的氦气、氖气）、激励系统（激光电源）构成，如下图 He-Ne 激光器激励系统采用开关电路的直流电源，体积小，重量轻，可靠性高，并装有散热风机，可长时间运行。 激光管的布氏窗与输出镜、全反镜之间用模具成型的耐老化的硅胶套封接。避免了因灰尘、潮气污染布氏窗、输出镜、全反镜而造成的激光输出功率下降。输出镜、全反射调节采用差动螺丝，粗调调节范围大，可锁定。细调调节范围小，调节时不易出差错。在激光管的阴极、阳极上串接着镇流电阻,防止激光管在放电时出现闪烁现象。激光器外壳接地，手碰激光器外壳无静电感应的刺痛感。 放电毛细管内充的氦氖混合气体的压强比约为7:1，总压强在100Pa 至400Pa 。放电管两端贴有用水晶片制成的布儒斯特窗。窗口平面的法线与放电管轴向间的夹角也恰好等于水晶的布儒斯特角，约56°。安装布儒斯特窗口可以使激光器输出的激光为在纸面内振动的偏振光，沿该方向振动的偏振光通过布儒斯特窗时不会反射，因此有利于减少损耗，提高输出功率。 3.2 He-Ne 激光器谐振腔与激光横模 光学谐振腔的两个反射镜构成腔的边界，他对腔内的激光场产生约束作用，使激光场的分布以及振荡频率都只能存在一系列分离的本征状态，每一个本征态称为一种激光模式。激光模式有两类：一类称为纵模，它是指可能存在于腔内得每一种驻波场，用模序数q 描述沿腔轴线的激光场的节点数。另一类是横模，指可能存在于腔内的每一种横向场分布，用模序数m 和n 描述。如果谐振腔由两面方形孔径的反射镜组成，则m 和n 分别表示沿镜面直角坐标系的水平和竖直坐标轴的激光场节线数。如果谐振腔由两面圆形孔径反射镜组成，则m 和n 分别表示沿镜面极坐标系的角向和径向的激光场节线数。因此每一个激光模式可以用三个独立的模序数表示，记成n m q TEM ,,。单独表示横模时可记成n m TEM ,。如00TEM 表示基
模块化掺饵光纤宽带光源驱动电路设计 李栋李流超黎志刚 中国电子科技集团公司第三十四研究所，广西桂林541004 摘要：为了提高模块化宽带光源的稳定性，采用自动温度控制ATC电路和自动功率控制APC 电路驱动泵浦激光器。实验结果表明，光源驱动电路可靠，输出光谱和光功率稳定，达到了预定的技术指标要求。该电路集成度高、体积小，能够满足宽带光源模块化需求。 关键词：掺饵光纤；宽带光源模块；泵浦激光器；驱动；功率控制； 1、引言 掺饵光纤宽带光源是一种相干性低的光源，具有输出功率高、光谱宽、温度稳定性高、使用寿命长等特点。由于这些特点，掺饵光纤宽带光源广泛应用在光通信、光纤传感、光器件测试及光谱分析等领域。随着超高速、大容量光纤通信系统和光传感系统的发展，对宽带光源在功率、带宽、稳定性及体积方面提出了更高的要求。泵浦激光器的驱动电路作为宽带光源的一个组成部分，电路的稳定性将直接影响掺饵光纤宽带光源的光谱输出质量。近年来，高稳定的模块化掺饵光纤宽带光源是一个研究热点。 本文将针对高稳定的模块化掺饵光纤宽带光源中的泵浦激光器的驱动电路展开设计，通过采用高集成度的自动温度控制ATC电路和自动功率控制APC电路，对泵浦激光器进行驱动，实现了光源光谱宽度和功率的高稳定输出。该设计电路具有体积小，稳定性高等特点，对研制模块化宽带光源具有一定指导和参考意义。 2、模块化掺饵光纤宽带光源驱动电路设计 2.1驱动电路总体设计 掺饵光纤宽带光源中，除了激光器的泵浦需要电光转化外，其余均为无源光路，所以泵浦激光器的可靠驱动是整个光路设计稳定的一个不可或缺的保证。模块化掺饵光纤宽带光源驱动电路设计包括：电源电路、泵浦激光器及其保护电路、APC电路、ATC电路。 整个驱动电路采用外置输入+5V(2A)电源供电，内部对输入电压进行滤波和稳压处理，保证电源的稳定性。由于内部驱动电路单元均采用+5V电压系统，所以内部不再需要电压变化处理。 2.2泵浦激光器及其保护电路 掺饵光纤宽带光源中的泵浦激光器采用980nm泵浦激光器，型号为LC96A74P-20R。该激光器模块输出光纤集成了光纤光栅，波长稳定性高。激光器最大输出功率可以到360mW,尾端带有保偏光纤。在激光器模块内部集成有热敏电阻、监控光电管、致冷模块，便于对模块进行自动功率控制和自动温度控制。 在使用中应注意一定不要超出激光器的极限值，同时操作还应注意静电保护，焊接时要断电焊接，保证良好接地。在激光器的驱动端并联滤波电容和反向偏置二极管，可以对激光器形成很好地保护。在电源输入端，增加电容避免电源不稳，对激光器造成冲击。 2.2泵浦激光器APC及泵浦激光器驱动电路设计 要使泵浦激光器输出光具有较强的稳定性，首先要有功率自动控制电路和良好的电流驱动电路。APC控制原理：将驱动电流经过电阻形成电压，将电压信号连接到驱动源的反向端形成反馈，对输出光功率进行很好地控制。另一方面，由于温度、湿度及器件内部老化造成的驱动波动，也可以通过APC电路改变反馈电压，从而稳定驱动激光器的电流，最终稳定光源功率输出。详细设计电路如图1：
高稳定全固态绿光激光器的研究 -------------------------------------------------------------------------------- 来源: 雅信通 霍玉晶何淑芳林彦 北京清华大学电子工程系固体激光与光电子技术研究所 前言 全固态绿激光器具有能量转换效率高、功率大、光束质量好、体积小、寿命长、使用方便等优点，在彩色显示、激光医疗、水下通讯等方面有重要的应用，是国内外光电子技术研究与开发的热点之一。全固态腔内倍频绿激光器通过高效率的腔内倍频获得绿激光输出，它的结构紧凑，转换效率高，是获得绿激光的主要器件之一。但是此类绿激光输出功率波动一般都很大。腔内倍频绿光激光器输出功率的稳定性成为目前全固态绿光激光器的研究重点课题之一[1,2]。 本文报到我们研制的全固化微型全固态绿光激光器和分离元件的全固态脉冲绿光激光器，以及提高它们输出功率稳定性的最新结果。 1、全固化的微型全固态连续绿光激光器 在本文中，“全固化”是指激光头的全部元件被用环氧树脂粘合为一个整体，整个器件中没有任何需要调整的元件，这种器件的使用十分方便。本文研制的全固化的微型全固态连续绿光激光器采用LD端面抽运Nd:YVO4激光晶体得到1064nm近红外激光，再用KTP晶体作腔内倍频得到532nm绿激光输出。其原理如图1所示。 它采用北京半导体研究所生产的808nm的1W的LD作为抽运光源。由于所用LD的输出功率比较小，因此提高器件的光－光转换效率是提高器件性能的关键。为此，采用半导体制冷器作为温度控制元件，对LD的温度进行控制，以使其工作波长和激光晶体的吸收波长峰值准确重合，提高对抽运光的吸收效率。在本实验中把LD的工作波长调整到808.5nm。采用中科院福建物构所生产的Nd:YVO4晶体作激光介质，掺杂浓度为3at％，a轴方向切割，长度为1mm，横截面尺寸为3×3mm2。采用端面同轴抽运方式使LD的抽运光束和所产生的1.064mm激光光束在空间上更好地耦合，以提高抽运效率。采用焦距为3mm的非球面透镜把LD的光束聚焦到Nd:YVO4晶体内部，焦点处光斑直径为100mm，抽运光从LD到Nd:YVO4晶体内部的总耦合效率为92％。采用平凹稳定型谐振腔提高器件的效率和稳定性。为了提高抽运光利用率，采用凹面后反射镜和平面输出镜组成激光谐振腔。Nd:YVO4晶体的入射端面是曲率半径为50mm球面。后反射镜直接镀制在此面上。它对1064nm的反射率>99.8%，对808nm光的透射率>99.5%。Nd:YVO4晶体的另一端为平面，镀同时对1064nm和532nm 的双波长增透膜，剩余反射率<0.25%。平面输出镜对1064nm光的反射率>99.8%，对532nm 光的透射率>97%。平面输出镜的背面镀有对532nm光的增透膜，剩余反射率<0.25%。采用北京人工晶体研究院生产的KTP倍频晶体作腔内倍频器，按类临界相位匹配角切割，它的横截面为3×3mm2，通光方向长5mm，两个通光面同时镀对1064nm和532nm的增透膜，每个面的剩余反射率<0.25%。为了提高器件的稳定性，用环氧树脂把全部元件粘接为一个整体，同时用一个半导体致冷器对它们进行整体控温。HSH型微型绿激光器的外形尺寸如图2所示。可以看到，它的体积很小，使用是很方便的。 用滤光器滤除输出光束中的808nm的抽运光和1064nm的基频光后，用Ocean Optics, Inc. 公司的HR2000CG-UV-NI光谱仪测量绿激光的波长，用清华大学研制的HH型F－P共焦球面扫
氦氖激光器调试实验 一、实验目的 1、了解He-Ne 激光器的工作原理和基本结构； 2、掌握外腔式He-Ne 激光器的F-P 腔调节技术； 3、分析放电电流对激光输出功率的影响。 二、实验仪器 外腔式He-Ne 激光器、准直光源，光学导轨，激光功率计，光阑，腔镜。 三、实验原理 一、激光原理概述 1 普通光源的发光——受激吸收和自发辐射 普通常见光源的发光（如电灯、火焰、太阳等的发光）是由于物质在受到外来能量（如光能、电能、热能等）作用时，原子中的电子就会吸收外来能量而从低能级跃迁到高能级，即原子被激发。在没有外界作用时会自发地向低能级（E1）跃迁，跃迁时将产生光（电磁波）辐射。辐射光子能量为 12E E h -=ν 2受激辐射和光的放大 受激辐射的过程大致如下：原子开始处于高能级E2，当一个外来光子所带的能量h υ正好为某一对能级之差E2-E1,则这原子可以在此外来光子的诱发下从高能级E2向低能级E1跃迁。这种受激辐射的光子有显著的特点，就是原子可发出与诱发光子全同的光子，不仅频率（能量）相同，而且发射方向、偏振方向以及光波的相位都完全一样。于是，入射一个光子，就会出射两个完全相同的光子。这意味着原来光信号被放大，这种在受激过程中产生并被放大的光，就是激 E 2 E 2 E 1 ν h (a) 自发辐射 E 2 E 2 E 1 ν h (b) 受激吸收 νh E 2 E 2 E 1 νh ν h νh ν h (c) 受激发射 高能态原子 低能态原子 双能级原子中的三种跃迁
电子碰 撞 激 发 波数 1 3390nm 632.8nm 1150nm 自发辐射 管壁效应驰豫 3P （525p p ） 1S （523p s ） 2S （5 24p s ） 共振转移 共振转移 电子碰撞激发 Ne 10S He 17 1 6 15 14 13 12 11 3S （525p s ） 2P （523p p ） 光。 3 粒子数反转 一个诱发光子不仅能引起受激辐射，而且它也能引起受激吸收，所以只有当处在高能级的原子数目比处在低能级的还多时，受激辐射跃迁才能超过受激吸收，而占优势。由此可见，为使光源发射激光，而不是发出普通光的关键是发光原子处在高能级的数目比低能级上的多，这种情况，称为粒子数反转。但在热平衡条件下，原子几乎都处于最低能级（基态）。因此，如何从技术上实现粒子数反转则是产生激光的必要条件。 4 激光器的结构 激光器一般包 括三个部分：激光工作介质；激励源； 谐振腔 二 He-Ne 气 体激光器工作原 理： 气体激光器的种 类很多，He-Ne 气体 激光器是目前应用最广泛的气体激光器。由于它的发散角小、单色性和方向性极好、稳定性高，故 在准直、计量、全息、 检测、导向、信息处 理、医疗等技术中得到了广泛的应用。但He-Ne 气体激光器的输出功率较小， He-Ne 气体激光器的输出功率只有 1 100mW ，最常用的25cm 的激光管，放电电流为5mA,高压为1500V ，输出功率为1.5mW ，效率仅为0.02％。制作He-Ne 气体激光器时，为了在有限的腔长内，尽可能获得较大的功率输出，要选择最佳的放电条件。所谓最佳放电条件是指一定管径和管长的He-Ne 气体激光器在适当的总气压、气体配比和放电电流下运转，以获得最大功率的激光输出。 四、 实验内容与步骤： 1、将准直光阑放在准直用的氦氖激光器和半外腔氦氖激光器之间，取掉其
半导体激光器 摘要：由于三五族化合物工艺的发展与半导体激光器的多种优点，近几十年来，半导体激光器发展十分迅速，而且在各个领域发挥着越来越重要的作用。本文将介绍半导体激光器的基本理论原理、相关发展历程、研究现状以及其广泛的应用。 1.引言 自1962 年世界上第一台半导体激光器发明问世以来, 半导体激光器发生了巨大的变化, 极大地推动了其他科学技术的发展, 被认为是二十世纪人类最伟大的发明之一[1], 近十几年来, 半导体激光器的发展更为迅速, 已成为世界上发展最快的一门激光技术[2]。激光器的结构从同质结发展成单异质结、双异质结、量子阱（单、多量子阱）等多种形式，制作方法从扩散法发展到液相外延(LPE)、气相外延(VPE)、分子束外延(MBE)、金属有机化合物气相淀积(MOCVD)、化学束外延(CBE) 以及它们的各种结合型等多种工艺[3]。由于半导体激光器的体积小、结构简单、输入能量低、寿命较长、易于调制及价格低廉等优点, 使得它目前在各个领域中应用非常广泛。 2.半导体激光器的基本理论原理 半导体激光器又称激光二极管(LD)。它的实现并不是只是一个研究工作者的或小组的功劳，事实上，半导体激光器的基本理论也是一大批科研人员共同智慧的结晶。 早在1953年，美国的冯·纽曼（John Von Neumann）在一篇未发表的手稿中第一个论述了在半导体中产生受激发射的可能性；认为可以通过向PN结中注入少数载流子来实现受激发射；计算了在两个布里渊区之间的跃迁速率。巴丁在总结了这个理论后认为，通过各种方法扰动导带电子和价带空穴的平衡浓度，致使非平衡少数载流子复合而产生光子，其辐射复合的速率可以像放大器那样，以同样频率的电磁辐射作用来提高。这应该说是激光器的最早概念。 苏联的巴索夫等对半导体激光器做出了杰出贡献，他在1958年提出了在半导体中实现粒子数反转的理论研究，并在1961年提出将载流子注入半导体PN结中实现“注入激光器”，并论证了在高度简并的PN结中实现粒子数反转的可能性，而且认为有源区周围高密度的多数载流子造成有源区边界两边的折射率有一差值，因而产生光波导效应。1961年，伯纳德和杜拉福格利用准费米能级的概念推导出了半导体有源介质中实现粒子数反转的条件，这一条件为次年半导体激光器的研制成功提供了重要理论指导。 1960年，贝尔实验室的布莱和汤姆逊提出了用半导体的平行解理面作为产生光反馈的谐振腔，为激发光提供反馈。 回顾这些理论发展历程，可以总结半导体激光器的基本理论原理：在直接带隙半导体PN结中，用注入载流子的方法实现伯纳德—杜拉福格条件所控制的粒子数反转；由高度简并的电子和空位复合所产生的受激光辐射在光学谐振腔内震荡并得到放大，最后产生相干激光输出[4]。 3.半导体激光器发展历程 在上述理论的影响下，以及1960年产生的红宝石激光器的刺激下，美国和苏
浙江大学 硕士学位论文 980nm泵浦激光器及其它半导体激光器和透镜光纤耦合的研究 姓名：赵发英 申请学位级别：硕士 专业：光学工程 指导教师：刘旭;张全 2003.3.1
９８０ｒｉｍ泵浦激光器及其他半导体激光器和透镜光纡耦合的研究 摘要譬《３３６ｓ８ ＼／、／ 关键词：９８０ｈｍ毫遭婆＆型缀，１３１０ｎｍ激光器，１５５０ｎｍ激光器，单模髡纤，耦合臌率，契形柱透镜光缪，锥形球透镜光纤，横向偏移容忍度，纵向偏移。 ＼／、，，——＿————‘容忍度，角向对准容忍度 ｆ，一√ ｒ随着高速度大容量光纤通信和光电子器件的迅猛发展，半导体激光器（ＬＤ）＼ 与单模光纤（ＳＭＦ）的高效耦合日益受到人们的重视。它不仅直接影响光纤传输的中继距离，而且对改善ＬＤ泵浦掺铒光纤放大器（ＬＤＰ—ＥＤＦＡ），ＬＤ泵浦固体半导体激光器（ＬＤＰ—ＳＳＬ）、半导体光放大（ＳＬＡ）等光通信器件及系统的性能上，在提高他们的性价比有着十分重要的意义。 按照模式耦合理论ｔ，Ｚ．３ｇＬＤ到单模光纤的耦合，实质上是两者之间的模场匹配。如果用平端光纤直接与ＬＤ耦合，由于严重的相位和模场半径不匹配，耦合效率很低，仅１０％左右。因此有必要采用适当的耦合系统对ＬＤ模场进行变换。 如何提高激光器和单模光纤之间的耦合效率４’５’“，最近已经有了相关的报道。但是报道的技术中多是针对圆对称光束或近圆对称光束。但是对于像９８０ｎｍ泵浦激光器发出的激光来说，由于特殊的物理结构，使得激光光束具有高度的椭圆性。 对于椭圆度高的激光光束和单模光纤耦合的报道不是很多。本文从理论上和实验上呈现了一个实用的耦合技术，即在没有使用分立球形光学元件的情况下将高度椭圆化的激光光束通过透镜光纤耦合到单模光纤中。本文研究的透镜光纤主要包括契形桩透镜光纤７以及锥形球透镜光纤。它们的作用是修正激光光束波前和从光纤出来光束之间的相位失匹。本文给出了不同的情况下的耦合效率公式，主要包括不考虑倾角和偏移时的耦合效率，以及考虑了倾角和偏移时候的耦合效率。 本论文在研究９８０ｎｍ激光器与契形柱透镜光纤耦合的同时，为将这种模型推广， 进一步的研究了１３１０ｎｍ激光器和１５５０ｎｍ激光器与锥形球透镜光纤的耦合情况。 并且通过实验，验证了这些模型的正确性。这方面的研究非常具有实用性。
文章编号:025322239(2004)07292524 104W 内腔倍频全固态Nd ∶YAG 绿光激光器 3 徐德刚 姚建铨 郭 丽 周 睿 张百钢 丁 欣 温午麒 王 鹏 (天津大学精仪学院激光与光电子研究所光电信息技术科学教育部重点实验室,天津300072) 摘要: 报道了一台高功率内腔倍频全固态Nd ∶Y A G 绿光激光器,针对KTP 晶体热效应和激光热稳定腔,采取了对KTP 晶体进行低温冷却的优化措施,以便减少KTP 晶体的热效应导致的相位失配,同时兼顾了Nd ∶Y A G 棒的热致双折射效应和KTP 晶体热透镜效应,设计了热稳定谐振腔;实验中采用80个20W 激光二极管阵列侧面抽运 Nd ∶Y A G 棒和Ⅱ类相位匹配KTP 晶体(在27℃时相位匹配角为<=23.6°;θ=90°,尺寸为7mm ×7mm ×10mm ) 内腔倍频技术,谐振腔腔长为530mm ,KTP 晶体的冷却温度为4.3℃,抽运电流为18.3A 时,实现平均功率达104 W 、脉冲宽度为130ns 的532nm 激光输出;其重复频率为20.7kHz 。光光转换效率为10.2%。 关键词: 激光器;Nd ∶Y A G 激光器;532nm 绿光;KTP 晶体;内腔倍频;低温冷却中图分类号:TN248.13 文献标识码:A 3国家863计划(2002AA311190)、教育部天津大学、南开大学科技合作项目、天津市光电子联合科学研究中心项目 (013184011)资助课题。 E 2mail :xudegang @https://www.doczj.com/doc/ff12191724.html, 收稿日期:2003207202;收到修改稿日期:2003212202 104W L as e r Di ode 2P u mp ed I nt r aca vi t y F r eq ue nc y 2Do u bled N d ∶YA G Gr ee n L i g ht L as e r Xu Degang Y ao J ianquan Guo Li Zhou Rui Zhang Baigang Ding Xin Wen Wuqi Wang Peng (Key L abor a tor y of Op to 2Elect ronics I nf or m a t ion a n d Tech nical Scie nce Mi nis t r y of Ed uca t ion ,I ns t i t u te of L asers a n d Op toelect ronics ,College of Precision I ns t r u me n t a n d Op toelect ronics Engi neeri ng ,Ti a nj ui n U ni versi t y ,Ti a n ji n 300072) (Received 2J uly 2003;revised 2Decembe r 2003) A bs t r act : An ave rage 104W green beam ope ration by int racavit y f requency doubling of Nd :Y AG lase r ,which is p umped by eight y 20W high 2powe r lase r diodes is reported.A t ype Ⅱp hase matched KTP cr ystal (<=23.6°,θ=90°unde r t he condition of 27℃,its size is 7mm ×7mm ×10mm )is applied for f re quency doubling of Nd ∶Y AG lase r.The len gt h of resonator is 530mm.The KTP cr ystal is placed in t he special cooling device.Unde r t he p umping current of 18.3A ,KTP crystal is cooled to 4.5℃.A maximum green powe r of 104W was gene rated at 20.7kHz repetition rate and 132ns p ulse widt h when p umping current of lase r diodes is 18.3A ,leadin g to 10.2%of optical -optical conve rsion efficienc y.Key w or ds : lase rs ;Nd ∶Y AG lase r ;green beam at 532nm ;KTP cr ystal ;int racavit y f requency doubling ;st rong cooling 1 引 言 高功率绿光激光器在可调谐激光器的抽运源、 流场显示、海洋探测、光电对抗、污染检测、大功率大 能量的激光加工以及军事应用(激光雷达,激光制导等)等科研和工业领域中得到了广泛的应用。利用激光二极管抽运Nd ∶YA G 晶体棒内腔倍频技术是获得高效高功率稳定绿光光源的重要途径之一。发达国家在这方面已经进行了很多的研究[1～5],国内由于国外高功率半导体激光器禁运、价格等条件限制,一直落后于发达国家。近年来,随着国产半导体 第24卷 第7期 2004年7月 光 学 学 报ACTA OPTICA SIN ICA Vol.24,No.7 J uly ,2004
半导体泵浦激光器说明书 目的及意义 半导体泵浦0.53μm绿光激光器由于其具有波长短，光子能量高，在水中传输距离远和人眼敏感等优点。效率高、寿命长、体积小、可靠性好。近几年在光谱技术、激光医学、信息存储、彩色打印、水下通讯、激光技术等科学研究及国民经济的许多领域中展示出极为重要的应用, 成为各国研究的重点。 半导体泵浦0.53μm绿光激光器适用于大学近代物理教学中非线性光学实验。本实验以808nm半导体泵浦Nd：YVO4激光器为研究对象，让学生自己动手，调整激光器光路，产生1064nm激光。在腔中插入KTP晶体产生532nm倍频光，观察倍频现象，测量倍频效率、相位匹配角等基本参数。从而对激光原理及倍频等激光技术有一定了解。
一． 激光原理： 光与物质的相互作用可以归结为光与原子的相互作用，有三种过程：吸收、自发辐射和受激辐射。 如果一个原子，开始处于基态，在没有外来光子，它将保持不变，如果一个能量为hv21的光子接近，则它吸收这个光子，处于激发态E2。在此过程中不是所有的光子都能被原子吸收，只有当光子的能量正好等于原子的能级间隔E 1-E 2时才能被吸收。 激发态寿命很短，在不受外界影响时，它们会自发地返回到基态，并放出光子。自发辐射过程与外界作用无关，由于各个原子的辐射都是自发的、独立进行的，因而不同原子发出来的光子的发射方向和初相位是不相同的。 处于激发态的原子，在外的光子的影响下，会从高能态向低能态跃迁，并两个状态间的能量差以辐射光子的形式发射出去。只有外来 hv E 2 1 (a) 2 1 (b) 2 E 1 (c) 光与物质作用的吸收过程 E 2 1 (c) E 2 E 1 (a) 2 1 (b) 光与物质作用的自发辐射过程
光纤打标机和半导体及灯泵浦激光打标机三者 主 要 性 能 比 较 武汉百一机电工程有限公司
光纤激光打标机与灯泵浦激光器性能对比光纤激光打标机设备型号及性能 “武汉百一”的BY-YLP光纤激光打标机在激光打标应用方面具有许多独特的优势。 与传统的固体激光器使用晶体棒作为激光介质不同，光纤激光器的激光介质是很长的掺镱双包层光纤，并被高功率多模激光二极管所泵浦。 BY-YLP系列光纤激光打标机使用特点 1、光束质量极好，适用于精密、精细打标 BY-YLP系列光纤激光打标机光束质量比传统的灯泵浦固体激光打标机好得多，为基模（TEM00）输出，发散角是灯泵浦激光器的1/4。尤其适用于要求高的精密、精细打标。 2、体积小巧、搬运方便、实现便携化 BY-YLP采用光纤传输，由于光纤具有极好的柔绕性，激光器设计得相当小巧灵活、结构紧凑、体积小。其重量和占地面积分别是灯泵浦泵浦激光打标机的1/10和1/4，节省空间，便于搬运。且采用光纤传输决定了其能适应加工地点经常变换的要求，实现产品的便携化。 3、激光输出功率稳定、设备可靠性高 能量波动低于2%，确保激光打标质量的稳定；平均无故障使用时间可达10万小时以上，灯泵浦激光打标机的氪灯的使用寿命在800小时左右。4、效率高、能耗低、节省使用成本 电光转换效率为30%（灯泵浦激光打标机为3%），设备功率仅500－1000W，日均耗电10度，是灯泵浦激光打标机的1/10左右，长期使用可为用户节省大量的能耗支出。 5、自主知识产权的操作软件，操作简便、功能强大 可以标刻矢量式图形、文字、条形码、二维码等，可升级实现在线打标，自动打标日期、班次、批号、序列号，支持PLT、PCX、DXF、BMP等文件格式，直接使用SHX、TTF字库。 激光打标机系统组成
全固态激光器全固态激光器（DPL）具有体积小、重量轻、效率高、性能稳定、可靠性好、寿命长、光束质量高等优点，市场需求十分巨大。全固态激光技术是目前我国在国际上为数不多的从材料源头直到激光系统集成拥有整体优势的高技术领域之一，具备了在部分领域加速发展的良好基础。引言高功率、小型化的全固态蓝绿激光器在海洋探测、水下通信等军事领域或者医学方面都具有重要的地位,这些应用一般都需要高功率蓝绿激光。目前,常用的1064 nm Nd∶Y AG激光器的倍频效率一般只有50%左右[1~4],因此通过提高倍频效率来提高整机的电光效率显得非常重要。如何提高非线性光学频率变换的效率一直是激光技术界的研究热点。David Eimerl[5]提出了正交频率变换的概念受到关注,他们按照正交频率变换的方式使用两块KD*P晶体,对于基波是Nd∶YLF激光输出经掺Nd磷酸盐玻璃放大器放大后的1053 nm激光脉冲,在基波功率密度为200 MW/c…半导体激光泵浦的全固态激光器是20世纪80年代末期出现的新型激光器。全固态激光器的总体效率至少要比灯泵浦高10倍,由于单位输出的热负荷降低，可获取更高的功率，系统寿命和可靠性大约是闪光灯泵浦系统的100倍,因此，半导体激光器泵浦技术为固体激光器注入了新的生机和活力，使全固态激光器同时具有固体激光器和半导体激光器的双重特点，它的出现和逐渐成熟是固体激光器的一场革命，也是固体激光器的发展方向。并且，它已渗透到各个学科领域，例如：激光信息存储与处理、激光材料加工、激光医学及生物学、激光通讯、激光印刷、激光光谱学、激光化学、激光分离同位素、激光核聚变、激光投影显示、激光检测与计量及军用激光技术等，极大地促进了这些领域的技术进步和前所未有的发展。这些交叉技术与学科的出现，大大地推动了传统产业和新兴产业的发展。全固态激光器是其应用技术领域中关键的、基础的核心器件，因此一直倍受关注。近年来，由于大功率半导体激光器迅速发展，促成全固态激光器的研发工作得以卓有成效地展开，并取得了诸多显赫成果。已经确认，传统灯泵浦固体激光器的赖以占据世界激光器市场主导地位的所有运转方式，均可以通过半导体激光器泵浦成功地加以实现。通常应用在激光打标机、激光划片机、激光切割机、激光焊接机、激光去重平衡、激光蚀刻等系统中。由于全固态激光器具有高光电转换效率、高功率、高稳定性、高可靠性、寿命长、体积小等优势，采用全固态激光器已成为激光加工设备的趋势和主流方向。全固态激光器的研发与应用概况近几年，美国、德国、特别是日本都在加大力量发展全固态紫外激光器，特别是中大功率全固态紫外激光器的开发应用。由于1064nm或532nm波长激光对材料的加工主要是产生气化或熔融等热作用，所以加工出的产品往往很难达到精细、光滑，甚至有些材料（如陶瓷、硅片等）在加工时会引起碎裂，因此，全固态紫外激光器在激光微加工、激光精密加工有着广泛推广应用的趋势。目前国外工业发达国家，全固态紫外激光器已开始成为工业用标准激光器。据文献报道：日本M.Nishioka公司已研发出40W的266nm全固态紫外激光器；三菱公司也在市场上推出了18W 355nm 25kHz全固态紫外激光器产品；另外相干公司的A VIV系列激光器已做到在266nm,30kHz时，平均功率大于3W，在355nm,40kHz时，平均功率大于10W；光谱物理公司的YHP-series系列激光器也达到在266nm,20kHz时，平均功率大于1.5W，在355nm,20kHz 时，平均功率大于3.5W；Lightwave electronics公司所推出的Q301-SM激光器也达到了在355nm,10kHz时，平均功率大于10W的技术指标。总体来说，国外全固态紫外激光器技术及应用设备已趋向成熟，但价格昂贵。高功率半导体激光列阵单光纤耦合模块可直接作为光源广泛应用于激光医疗、信息产业、激光加工、国防工业、激光武器和战术装备等领域。作为泵浦光源将是泵浦全固态激光器的核心器件，是一种高光－光转换效率（大于30％）的高功率泵浦全固态激光器的商用半导体激光光源模块，是替代灯泵浦激光器的理想产品。目前，国外半导体激光器单根光纤耦合模块的最高研究水平是耦合进入1个芯径400μm，输出功率200W。耦合进入1根800μm的光纤，输出功率700W；耦合进入1根1.5mm的光纤，输出功率超过2000W。国外出售的单光纤耦合模块产品水平如：Apollo公司产品为