&A~5647-MSI Informal Report

/

J

A Short Account ofLos Alamos Theoretical Work on

Thermonuclear Weapons, 1946-1950

Prepared by

J. Carson Mark*

●LASL Consultant

1

@

$

10s slam

-..-h-

-SpeciakDi8tribtitionIssued: July 1974

scientific iaboraof the University of Calif~

LOS ALAM OS, NEW MEXIC O 87

DO NOT CIRCULATE

PERMANENT RETENTION

UNITED STATES

ATOMIC ENERGY COMMISSION

CONTRACT W-740B-ENG. 36

.

#- .. . .

. t-. ..,

.

This report was prepared as a n account of work sponsored by the UnitedStates Government. Neither the United States nor the United States AtomicEnergy Commission, nor any of their employees, nor eny of their contrac-tors, subcontractors, or their employees, makes any warranty, express or im-plied, or assumes any iagal liability or responsibility for the accuracy, com-pleteness or usefulness of any information, apparatus, product or processdis-closed, or represents that its use would not infringe privately owned ri@s.

.

. ..

In the interest of prompt distribution, this LAMS re-port was not edited by the Technical Information staff.

-..

-- .

I‘1

FOREWORD

This report is an unclassified—and consequently, somewhatabridged-vereion of a document prepared during the summer of 1954.Except ae required to remove claeeified references, and to restore con-tinuity, it follows the original.

The earlier document (iesued on October 1, 1954) waE the first draft ofa chapter for a proposed hietory of the technical work at Los Alamcafrom the end of the war up to 1954. This particular chapter wae to coverthe Los Alarnos work on thermonuclear weapona from 1946 to January1950-the time of President Truman’s decision concerning U.S. work onthe hydrogen bomb, Several other aectiona for such a history were alsodrafted (by other authors); but the project as a whole began to appear tobe too onerous to carry further-at least in the hands of a group of per-ecns already fully occupied (and much more intensely interested) in themore immediate undertakings of the Laboratory.

Thie unclaeeified version has been prepared in order to provide a fac-tual account of the Los Alamos work on thermonuclear weapons duringthis particular period. Because of neceseary claeeification restrictions itwould not have been possible many years ago to release an accountwhich was both factual and coherent. Many suck reatrictiona are, ofcourse, dill in effect; but the steady progress of declamification has nowreached the point (or may even have reached it several years ago) atwhich it is possible to make the present report available. Inaemuch craadistinctly erroneous impression of these matters is rather widely held, itwill seem worthwhile if this report should ultimately help establish abetter understanding of what actually took place.

J. Carson Mark

...Ill

A SHORT ACCOUNT OF LOS ALIAMOS THEORETICAIJ WORKON THERMONUCLEAR WEAPONS, 1946-1950

by

J. Carson Mark

A factual account of work (mainly theoretical) on ther-monuclear weapon development at the Los Alamos ScientificLaboratory from 1946 to 1950, this is an unclassified version of achapter (written in 1954) for a proposed technical history. It out-lines the computational and theoretical work devoted to study ofthe Super and other thermonuclear problems.

I. STATUS AS OF EARLY 1946

On April 18-20, 1946, there was a conference atLos Alamos on the subject of the Super. A largeproportion of the persons who had been investigatingthe possibility of thermonuclear weapons at LosAlamos had continued work on this problem up tothe time of this conference. A general description ofthe device then considered is given in reportsprepared for the conference. The results of work upto that time are indicated or embodied in thesereports, the unresolved problems as then perceivedare referred to, and the requirements, as they wereunderstood at that time, of further work along manydifferent lines were lieted.

The estimates available of the behavior of thevarious steps and links in the sort of device con-sidered were rather qualitative, and open to questionin detail. The main quedion of whether there was aspecific design of that type which would work wellwas not answered. Had the physical facts been suchthat there was a large factor to spare in attempting todemonstrate that such a device would detonate, thenthe type of considerations which it had been possibleto devote to this problem up to that time would havebeen sufficient to establish that fact. As it was, thestudies of this question had merely sufficed to showthat the problem was very difficult indeed; that themechanisms by which energy would be created in

the system and uselessly lost from it were com-parable; and that because of the great complexityand variety of the processes which were important, itwould require one of the most difficult and extensivemathematical analyses which had ever been con-templated to resolve the question-with no certaintythat even such an attempt could succeed in beingconclusive. The general belief of those working onthe problem at that time, however, was that somesuch design could be made to detonate, although itwas fully understood that much study would yet berequired to establish this fact and determine the mostfavorable pattern.

The requirements for materials, engineeringdevelopments, and more detailed understanding ofbasic physical processes were impressive, and (assaid at the time ) “would necessarily involve a con-siderable fraction of the resources which are likely tobe devoted to work on atomic developments in thenext years.” An active program to realize such adevice was then thought to require amounts of tritiumbeyond the reach of the Hanford plant to produce inany relevant time, so that the building of a reactor fortritium production was probably involved, It wassuggested that facilities for the production ofuranium-233 and/or the separation of plutonium-239would be desirable. The need of facilities for handl-ing deuterium was pointed out. Laboratory ex-periments, measurements of cross sections, and

f

—

studies of properties of materials were necessary. Thedevelopment of a large-yield fission device was ob-viously required.

The requirements, however, which werequalitatively most difficult to meet were those in-volving theoretical study of the behavior of thevarious steps in the process. The most difficult ofthese was, of course, the central problem (whichcame to be known as the “Super Problem”) ofwhether, and how, and under what conditions a bur-ning might proceed in thermonuclear fuel in thepattern envisaged at the Super Conference. In addi-tion, before the properties of any actual design couldbe discussed, it was necessary to obtain a much moredetailed picture than had yet been developed of thephenomena occurring in the immediate region of anexploding fission core. A successful treatment of thislast problem—which was also important for the fullerunderstanding of fission explosions-itself requiredthe results of laborious calculations of the propertiesof materials at the relevant temperatures which werethen being conducted by a small group which hadrecently moved from New York to Chicago. And, in-deed, each step in the sequence posed a family ofdifficult problems.

The prospects for realizing a thermonuclearweapon along these lines were problematical. An ac-tive program to establish what might be feasiblewould compete at many points for the resources of ef-fort and materials required for the immediatelynecessary program to improve and expand ourstockpile of fission bombs, and at some pointsdepended on advances in our understanding of fis-sion bombs. Itwas against this background that it wasproposed in a letter from Bradbury to Groves;November 23, 1945, that at least for the interimperiod, during which the future pattern of the LosAlamos Laboratory was being considered, the workon the thermonuclear program at Los Alamos consistof: several lines of laboratory experimentation,theoretical studies as practicable conducted in activeconsultation with Teller, and requests for smallamounts of tritium as needed for experimental pur-poses.

II. FROM 1946 TO END OF JANUARY 1950

A. General

Starting from the stage represented by the SuperConference, definitive progress towards obtaining or

-g out a model of the type discussed required as apreliminary step a very great extension and refine-

ment in the understanding of the theoretical andquantitative considerations involved. Mo

f

ver, thepossibility of opening up any radically erent ap-proach to a thermonuclear weapon als epended aalmost exclusively on further theoretical :ght. (Aslate as August 1950, in an appendix a “Ther-monuclear Status Report,” prepared at ~os Alamos efor the GAC, Teller and Wheeler, in discussing the“Scale of Theoretical Effort,” made the observationthat “The required scientific effort i$ clearly muchlarger than that needed for the first fission weapon . .. . Theoretical analysis is a major boffleneck to fasterprogress . . . .“) Consequen~y, the account of theprogress during this period will be given withprimary reference to the theoretical work onproblems of importance to the thermonuclear field.

Of course, some experimental studies (for ex-ample: cross-section studies, observation of behaviorof fast jets) were continued across this period and oc-cupied, on an average, the mqor part of the attentionof something like two of the fifty or so experimentaland engineering groups in the Laboratory. Suchwork, however, was mainly in the nature of acquiringdata which were believed would be needed in con-nection with any attempt to estimate the behavior of athermonuclear system. It was unlikely of itself toreduce the difficulty of undertaking a theoretical es-timate, or to suggest an essentially new approach to athermonuclear weapon. In addition~although workof an experimental and engineering Ed was knownto be a necessary and heavy compo~~nt of any ther-monuclear program, it could not rise~o the high levelof a full attack on the significtit outstandingquestions until the theoretical understanding of theprocesses involved in a particular system had ad-vanced to the stage at which such questions could beisolated and clearly defined.

There was also a considerable body of theoreticalwork which stood in a similar relationship to ther-monuclear studies. The work referred to could not beclassified as distinctively “thermonuclear,” nor was itconcerned with the details of any specified ther-monuclear system but it was background work whichit was recognized would have to be got in hand eitherbefore or while undertaking the detailed design ofany likely type of thermonuclear system. Amongsuch lines of necessary background theoretical workmay be mentioned (i) work on opacities and equa- -%.

tion of state of materials, (ii) great numerical refine-ment of the picture available of the processes oc-

‘9curring in a fission explosion, and (iii) advances inthe general area of computational ability, both in thematter of computing equipment and also in the fieldof computing technique and experience. Very

2

I

definite progress (some of which will be referred tobelow) was made along these various lines between1846 and 1950, and helped provide, by the end of1949, a very much greater theoretical capability withrespect to a thermonuclear (or any other) programthan was available at Los Alamos in 1946.

B. Resources for Theoretical Work

In this section it is proposed to discuss the growth ofthe Theoretical Division at Los Alamos, carrying thisthrough to the end of 1953, since this indicates thecontext in which the particular studies referred tolater were undertaken.

At Los Alarnos, the Theoretical Division, in addi-tion to the persons who might generally be con-sidered to be trained or capable in some branch orbranches of theoretical physics, has always includeda considerable number of persons acting in somefairly well-defined supporting role—such as com-putem, secretaries, assistants, computi-g machineoperators, and others. Something like two-thirds ofthe total personnel of the Division have normallybeen in this latter group. Although the conduct of anyappreciable program of theoretical work is veryheavily dependent on the ability and skill of the per-sons in this group, the content of the variousstudies-their quality, soundness, and degree ofnovelty, -is almost totally dependent on the ability ofthose identified as theoreticians. The separationsuggested here cannot always be made with absoluteprecision but it can be drawn sufficiently closely forthe purposes of the following Table, in which the totalnumber of persons in the Theoretical Division, whoby training or experience were in a position to helpdetermine the objectives and quality of thetheoretical program, is given at the end of each yearfrom 1946 to 1953. By no means all the persons in-dicated were (or could properly be) ever at one timefully engaged on immediate weapons problems,since studies similar to the background type of workreferred to above, as well as support of the activitiesof other sections of the Laboratory, not to mention theaspiration of everyone trained in pure science tomake his own recognized contributions to advancesin knowledge in those areas where he feels he has

<-ideas to contribute, together usually occupiedsomething like half of the attention of the group.

~t

END OF: 1946 1947 194s 1949 1%0 .1s51 1s2 1ss3—— —— —— ——

Los Alamos Staff 8 12 14 22 35 45 4s 51

Full-time Consul- - - - 2 3 1 1-tants at Los Alamos(See below)

At Matterhorn -.. -- 6(See below)

10 -

In addition to those holding a “permanent” ap-pointment to the staff at Los Alamos, the followinggroups should be mentioned:

1. Consultants. Ever since the war, thetheoretical program at Los Alamos has benefitedgreatly from the assistance of a large number of ableand distinguished consultants. This, for the most part,has been in the form of the consultant working withand among the regular staff for an extended periodof from a few weeks to three months or so during thesummer, sometimes with an additional period of afew weeks in December or January, and usuallycoupled with brief visits at other times, either of theconsultant to Los Alamos or of Los Alamos persons tothe consultant. The exact pattern has, of course,varied between various individuals and from year toyear with each individual. To give an indication ofthe quite impressive assistance obtained in this WOY

since the time of the Super Conference, the followingnotable instances are cited (though it would be easyto extend this list):

H. A. Bethe: brief vieits 1946, 1947, 1948; about twomonths each year, 1949, 1950, 1951; about eightmonths 1952; and three months in 1953.

E. Fermi: visited each year from 1946 to 1953 ex-cept 1949; about six weeks per year on the average(between two and ten weeks each year) for theseyears.

G. Gamow: about twelve months between June1949 and September 1950.

F. Hoyt: eight months between ]uly 1946 andJanuary 1948; brief visits January 1948 toDecember 1949; joined Los Alamos staff in July1950.

E. Konopinski: three weeks in 1946; four months in1950; three months in 1951.

3

L. Nordheim: one month in 1947; brief visits in ear-ly 1949 and 1950; full time September 1950 toSeptember 1952.

E. Teller: nine months between July 1946 and June1949; full time from July 1949 to October 1951.

J. von Neumann: two months per year on theaverage (between one and three months eachyear) from July 1946 to December 1953.

J. A. Wheeler: full time from March 1950 to June1951; after which continued to be heavily engagedin the program through Project Matterhom untilMarch 1953; two months at Los Alamos, July-August 1953.

2. Project Mattexhorn. In July 1951, J. A.Wheeler established and directed a Princeton agroup known as Project Matterhom to engage in theprogram of theoretical studies of thermonuclearweapons in the form then being considered. Thisgroup worked in collaboration with the work at LosAlamos. After the formation of the LivermoreLaboratory, Project Matterhom made plans to dis-continue its operation, and the contract was formallyterminated on March 1,1953. Several members of theProject continued work on the terminal and summaryreports of Matterhom work into the summer of 1953.(The present Project Matterhorn, working under L.Spitzer on the problem of controlled thermonuclearreactions, began its work about the same time andwas originally called Division S of ProjectMatterhorn. It was operated under direct contactwith the AEC and continued administratively un-affected by the termination of the group engaged onweapons studies.)

3. Opacity Group at Argonne Laboratory. Thewartime group working at Columbia under the direc-tion of Maria Mayer on the subject of opacity ofmaterials transferred its operation to Chicago at thebegimiing in 1946, and continued as a Los Alarnossponsored program in the Argonne Laboratory fromthen until 1952. In addition to continuing occasionalassistance from Teller and M. Mayer, this programoccupied on the average the attention of about twopersons (on the scale of the table above). At least un-til the advent of modem computing equipment,numerical calculation of opacity values was an enor-mously tedious undertaking and it was extremely dif-ficult to arouse, and particularly to sustain, the in-terest of capable persons in this program. This

program of study, which is yet (1954) by no meanecomplete, was contracted to the Rand Corporation inthe middle of 1953, where a considerably largergroup is attacking the problem with the aid ofmodem high-speed computing equipment.

4. Group at Yale. In March 195(J,amangementswere made to have Gregory Breit direct the part-timework of four or five senior graduate students at Yalein the study of some of the basic interactions betweennuclei, electrons, and radiation. This work which initself would be unclassified assumed importance asnecessary data for the detailed consideration ofvarious problems. Studies of this general type (re-quired in connection with improved ccdculations onthermonuclear weapons but not in themselves in-volving weapon design data) have been continuedunder Breit’s direction to the present. Incidentally,among the problems to which Breit has given con-siderable attention under this arrangement, has beenthat of checking, refining, and extending the con-siderations first applied by Teller and Konopinski tothe question of whether or not the concentrations ofenergy provided by possible thermonuclear ex-plosions would threaten to ignite the earth’s at-mosphere or the sea. Such consideration has, ofcourse, continued to show that such ignition isprobably impossible and that, even if possible, thereis a considerable number of orders of magnitudelacking between anything yet contemplated and theconditions which might be required.

One final comment on the subject of the Tableseems appropriate. Of the eight theorists at LosAlamos at the end of 1946, three had joined the staffafter July 1, 1946, and only one (Landshoff) had beenpreoccupied with work on the thermonuclearprogram before the time of the Super Conference. Bylate 1946, however, three of these eight were chieflyengaged on specifically thermonuclear studies.

C. A Brief Chronological Account

A partial calendar, with notations, is given belowto provide a picture of the time sequence of thedevelopments discussed. This “calendar” is largelyabstracted from the monthly progress reports of theTheoretical Division during this period, andreference is given only to items which appear to havehad a continuing significance in relation to the ther-monuclear field. There was, in addition, of course, alarge body of theoretical work involved in connection

.J

t

-..

‘$

4

with the developments in the fission weapon fielddescribed elsewhere in this account.* (In the follow-ing section, the progress of work along a number ofspecified lines will be traced across this period, andsome of the items merely referred to in the presentlisting will be discussed further.)

May-September 1946: All the individuals engagedon the studies and calculations discussed at theConference wind up work under way at that time andprepare final reports, with the exception ofLandshoff, who remains at Los Alamos and continuesstudies of aspects of the Super Problem. (Landehoffremained at this problem until the summer of 1947,)From mid-July to end of September, Hoyt works onsame problem.

September 1946: Teller proposes a new ther-monuclear system which later came to be called theTX-14, and Richtmyer takes up problem of estimatingperformance.

October 1946: Evans takes up studies related to theSuper Problem.

November 1946: First TX-14 report issued by Richt-myer. Report contains arguments of feasibility inprinciple, and rough estimates of efficiency andbehavior.

December 1946-January 1947: Landshoff, Mark,and Richtmyer propose possible experiment to checkpredicted features of thermonuclear burning, andmake estimates of feasibility of carrying this out inconjunction with some fission bomb test of moderateyield.

January-February 1947: Richtmyer starts to developan improved theory of efficiency of the TX-14.Discussion with Teller and von Neumann of possibleapplication of advanced electronic computing equip-ment (then in early stage of design at Princeton) toLos Alamos problems.

February-March 1947: Studies of equation of stateand related problems, pertaining to thermonuclear

..- as well as fission devices, reactivated at Los Alamosas H. Mayer joins staff. “Monte Carlo” method ofcomputation proposed by Ulam, and LAMS-551

>8●The reference here is to a more comprehensive ac-count of Los Alamos work in this period which wasdiscussed, and given preliminary attention, but neverpulled together into a fully coherent “history.”

(outlining prescription for application of method)prepared by von Neumann, with additionalsuggestions by Richtmyer.

March-April 1947: With Teller, planned programfor the summer of 1947, primary objectives being to:continue studies of burning of fusion fuel, developMonte Carlo method, initiate work on obtaining

detailed calculation of explosion of fission bomb,study the proposed experiment to check ideas onthermonuclear burning, and prepare status report onthermonuclear systems.

May-June f 947: A report on Improved Theory ofthe TX-14 issued by Richtmyer. Hoyt rejoins study ofprocesses in Super Problem.

July 1947: Nordheim joins Richtmyer in study ofTX-14.

A ugu.n 1947: Efficiency calculations for a numberof possible TX-14 configurations completed withRichtmyer’s improved theory. Landshoff takes upwork on fission bomb explosion calculation.

Sep(ember 1947: Further TX-14 examplescalculated. A report prepared by Teller* describesthe status of the studies of the Super and the TX-14. Itdiscusses a variety of possible thermonuclear fuels,and it urges consideration of tests of the sortsuggested in January 1947—but in an improvedfozm. Subsequent work on such experiments wasdirected at the new pattern for which the designation“boosting experiment” or “Booster” came intogeneral use.

October 1947: Richtrnyer starts to plan a fullydetailed machine calculation of the course of a fis-sion explosion. (T& turned out to be a two-yearprogram, and the first example was actuallycalculated only early in 1950.)

December 1947: Work started separately byLandahoff et al. on simpler and, hopefully, faster fie-sion explosion calculation. (Since Richtmyer’sproblem came to be known as “Hippo,” the work byLandshoff was known as “Baby Hippo.”)Preliminary consideration also given to preparing a

●The following entry appears on the title page of thisreport: “Work done by: F. Evans, F. Hoyt, R.Landshoff, M. Mayer, L. Nordheim, R. Richtmyer, E.Teller, E. Zudina.”

5

detailed calculation of the Super Problem for handl-ing on the electronic computer expected to be com-pleted at Princeton within a couple of years.

.lanuary 1948: The program of analytical study andattempts at numerical solutions (using only deskcomputing machines) of the Super Problem broughtto a close, and results written up. (From Januarythrough April, a considerable amount of effort wasrequired in connection with preparations for theSandstone tests and consideration of results.)

February 1948: First automatic machine calculationof Monte Carlo type prepared for handling on theENIAC. (Monte Carlo calculation techniques wereexpected to be required in the detailed calculationof thermonuclear burning, as well as other types ofproblems.)

March 1948: Richtmyer and von Neumann in-troduce so-called “viscosity treatment” of shocks.(This technique, which was devised to meet needsarising in connection with Hippo, reduced theproblem of calculating the progress of shock fronts inexplosion and implosion calculations to manageableproportions on automatic computing machines, andwas of profound value in very many of thecalculations undertaken subsequently.)

July 1948: Detailed study begins of behavior of aBooster system (considered either as a test of ther-monuclear principles or a possible weapon). Workbegins on equation of state of certain materials(wanted in connection with possible experimentalgadgets to test ideas in the thermonuclear field).

August 1948: Study of the scattering of neutrons bylight elements to obtain data required in connectionwith various calculations (Booster, hydrides, andthermonuclear burning).

September-October 1948: ‘TWO reports issued byReitz and Rosenbluth giving the results of detailedcalculations concerning the behavior of possiblespecified Booster systems. These are the first detailedstudies relevant to the proposal to include such adevice in the tests then scheduled for 1951. Logicallay-out of calculation of Super Problem begun byEvans, Metropolis, Teller, von Neumann, and Ulam.(From this point on, the planning and preparation ofthis calculation was continually kept in view, with theobjective of having it ready by the time the computerat Princeton should be ready to accept it. Dr. andMrs. Evans were mainly responsible for the prepara-

tion and eventual execution of the Super Problem. Inthis tremendous undertaking, they had, of course, thebenefit of suggestions and assistance from many per-sons on many aspects and details of the work. In par-ticular, they relied on the continuous and pervadinginterest of von Neumann, who advised on almostevery detail in the problem. As it turned out, the com-pletion of the machine was much later than had beenexpected in September 1948, and it was not in shapeto take this problem until about the end of 1952. Thefirst two examples were calculated in Princetonbetween February and July of 1953.)

November-December 1948: Detailed discussionstarts of plan to prepare a machine calculation (tnsimplified form) of a particular phase of the SuperProblem. (While this proposal would not providedefinitive answers to the main questions, it would atleast help establish the amount of tritium likely to berequired to provide possible ignition conditions forthe “classical” Super. By avoiding many of the enor-mous complications involved in the Super Problemproper, progress could be expected much morerapidly.) A report outlining the steps to be con-sidered in the calculation proposed prepared byEvans, von Neumann, and Ulam.

January 1949: Metropolis authorized to proceed toform group to build an electronic computer at LosAlamos along same general lines as computer atPrinceton. (Actual work on the machine was underway by the spring of 1949, and the computer beganeffective operation in the spring of 1952.)

January-June 1949: Work continues on many of theproblems mentioned above; in particular: Hippo andBaby Hippo, the detailed preparation of the SuperProblem, and various features of the Booster.

July 1949: Work starts on improved calculations ofthe equation of state for hydrogen, required in con-nection with all thermonuclear studies.

August 1949; Several calculations begin concer-ning details of behavior of thermonuclear fuel inBooster.

September 1949: Serious worries ratsed about possi-ble deleterious effects of extraneous processes onbehavior of boosting experiment. Bethe and othersinitiate study of these processes with intention of us-ing results to guide Booster design.

6

---

b-

October 1949: Work begun in July on equation ofstate for hydrogen is completed* and data madeavailable for relevant thermonuclear calculations.

Calculation starts for a device having a new anddifferent pattern. (This was in preparation for studiesof the sort which later led to the design of the GeorgeShot at Greenhouse.)

Baby’ Hippo calculation reaches stage at which itbegins to give results. These are of interest both inconnection with preparation of Hippo, and for detailsrequired in consideration of Booster.

November 1949: Full calculation of proposedBooster model gives dieappointing results, indicatingneed of seeking improved design, Many discussionsby Teller et al. of detciile connected with the SuperProblem.

December 1949: Preparations of the calculation ofthe Super Problem advance far enough that remti.ing detailed work can be completed much faster thanthe computing machine required to handle theproblem. Work on problem preparation consequent-ly suspended until such time as machine more nearlyavailable. Detailed work starts on preparation of amachine calculation of the simplified problemproposed in November-December 1948. Furthersimplified hand calculation of same problem begunby Ulam and Everett to provide information sooner,even though this information would be lees precise.

Results of first basic calculation of new patternproposed in October 1949 become available. Diecus-eion started of choice of parameters for further detail-ed consideration.

First model of IBM Company’s CPC delivered toLos Alamos. (Thie machine represented an enor-mous advance in capacity, flexibility, and speed overany computing equipment available at Los Alamosup to this time. It required, of course, several monthsto obtain experience needed to make full use of itscapabilities.)

Consideration of controlling various parameters ofBooster indicate ways to relieve difficulties met inNovember.

January f 950: Study status of range of designs inwhich boosting experiment might be applicable.

my Hippo calculation reaches stage about lmlf-way through explosion. Hippo calculation almost

●This work was embodied in a report written by J.Reitz; with work done by: Bethe, Longmire, M.Mayer, Reitz, M. Rosenbluth, Stemheimer, Teller.

ready to start in New York.Machine calculation for which preparations

started in December 1949 further trimmed to fit onthe ENIAC, with plan to prepare for first calculationduring the spring of 1950. Firet example of handcalculation continues, with results expected beforethe end of February,

H. Mayer completes “The Super Pocketbook,”chiefly a summary of two lectures delivered by Tellerto the Techniccd Council of the Laboratory a coupleof months previously. The report outlines principlesof the Super and gives basic formulae and upto-datephysical data, as well as estimates of damage from anassumed 40-megaton Super.

January 31, 1950: President Truman’s announce-ment concerning work on thermonuclear weapons.

D. Summary of Progress on Some ParticularProblems

In this section it is intended to identify the moresignificant problems or programs considered, in-dicate the progress made in the period 1946-1949,and describe the stage reached by the end ofJanuary 1950.

1. Calculation of Details of Fission Explosion.The need of such calculations was clearly stated inthe Super Conference reports of 1946. In a “Programfor the Theoretical Division,” drawn up by Fermi,Richtmyer, and Teller in August 1946, this problem isput in the foremost position as being necessary toprovide the basis for improving fission bomb designs,and also important to understanding the interactionbetween a fission explosion and the possible ignitionof thermonuclear fuel. The advent of the TX-14 inSeptember 1946 placed an additional, and evenmore specific, emphasis on the need for detailed un-derstanding of the processes involved in a fission ex-plosion.

Starting in the late summer and fall of 1947, twomqor calculations were undertaken on this problem.These calculations came to be known as “Hippo”and “Baby Hippo.” Baby Hippo was conducted byLan(dshoff, and relied on the computing facilities atLos Alamos-which at the time consisted of a groupof computers using desk calculators and anothergroup using the IBM equipment then available. Hip-po was conducted by Richtmyer, with the intention ofmaking as effective use as possible of advanced com-puting equipment. It was expected that Hippo wouldrequire about a year to prepare (since much new

7

I

ground in mathematical computing technique wouldhave to be broken), but when ready to run would gomuch faster than the other calculation and providemuch more detailed results. Baby Hippo on the otherhand, would get started sooner, probably give someresults faster, but, most particularly, give a foretasteof the nature of the difficulties not foreseen at thestart which could help guide the planning of Hippo,Things happened pretty much as expected exceptthat each approach was about twice as hard asoriginally supposed, and required twice as long toaccomplish. In January 1950, Baby Hippo had givena picture of the events in the core and tamper of theTrinity bomb up to about half-way through the explo-sion. Early in February 1950, Hippo was checked outon the IBM Company’s SSEC in New York and BabyHippo was discontinued. (By June 1950, two Hippoproblems were completed, and details of behaviorprovided by these were used as guides for estimatesrequired in studying designs of experiments propos-ed for Operation Greenhouse.)

2. Calculations on the Super. The main ques-tion, that of the burning of thermonuclear fuel, wasstudied from the time of the Conference up to the endof 1947. In all of these studies some of the relevanteffects were neglected so as to allow analytical, orsimple numerical, treatment. It then became clearthat only a full-scale treatment in which all the im-protant processes were simultaneously taken into ac-count could give significant information. This couldonly be approached by an elaborate numericalcalculation of a magnitude which would obviouslytax the resources of the most advanced machinesthen being designed. No adequate machine was ex-pected to be operating in less than a couple of yearsfrom that time, and in fact it was over four yearsbefore the first of these appeared. In the meantime,work continued on preparing the Super Problem soas to be able to take advantage of the machines aesoon as they became available. By January 1950,these preparations were in a stand-by status, stillwaiting for the difficulties in machine building to beovercome.

The much simpler question (analogous to that forwhich detailed preparations were begun inDecember 1949) was the one which had received themost specific study at the time of the Conference.Though subsidiary in principle to the main question,it had considerable importance in that it could belearned from study of this problem how much tritiummight be required in the bomb. From this informationa judgment could be made of whether a Super mightbe tolerably or prohibitively expensive. At the time of

the Conference it was believed that, though theamount of tritium required was likely to be large, itwas not prohibitively large, and that modifications ofdesign might enable this amount to be reduced ap-preciably. In September 1947, after mentioning someadverse effects not previously taken into account,Teller wrote, “Thus, I believe that a total amount of(so much) tritium suffices to set off the Super.” (Theamount mentioned was roughly twice as large as theamount envisaged at the time of the SuperConference.) This estimate was with respect to astraightforward, but probably wasteful, dieposition oftritium; and it was pointed out that by using a morefavorable disposition “. . . it ie very likely that con-siderably less tritium will suffice for ignition.” Star-ting late in 1948, plans for a calculation to give amore detailed and realistic picture of a process of thintype began to take shape.

In December, 1949, detailed work was started(chiefly by Calkin, Dr. and Mrs. Evans, Dr. and Mrs.von Neumann) on the preparation of a machinecalculation of the problem referred to above whichwas expected to require about six months to getready. (This calculation was started on the ENIAC atthe beginning of June 1950, and continued into thesummer,) In December 1949, aleo, Ulam and Everettstarted a simplified version of this calculation byhand. This would give less detailed results, but givethem sooner, and the difficulties encountered wouldprovide guidance in the preparation of the machineversion. Lacking calculations of this type, estimatesof the amount of tritium required were necessarily ona somewhat subjective basis. In many of the d.is-cuesione of the amount of tritium required, whichproceeded through the fall of 1949 both at LosAlamos and at other places, very small amountn wereconsidered likely to suffice. In this context, the firstcalculation of Ulam and Everett was started with arather modest amount. By the end of January 1850,this first calculation was about half complete. Beforethe end of February 1950, the results showed that theamount of tritium chosen was not nearly enough; sothe first calculation wa8 discontinued, it8 resultswritten up on March 9, 1950, and a second calcula-tion, with a larger amount of tritium, was started im-mediately.

3. TX-14 Studies. The TX-14 system wasproposed by Teller in September 1946. Work startedimmediately to obtain estimates of ignition conditionsand available efficiencies. In a report, issuedNovember 15, 1846, the conclusion is given, “At thepresent time it seems the proposal is entirely feasible,provided (some possible adverse effects are) not too

...

-I

8

.-

serious . . . .“ Measurements of relevant neutronreaction cross sections were undertaken, and im-provements in the theoretical treatment weredeveloped by June 1947 which would enable specificcalculations of assumed models to be made. Thesewould indicate the size of system required to producea stated yield, and, of more immediate interest, allowone to determine the size of the explosion required toget the reaction well started.

By the end of September 1947, calculations hadbeen made on several models. The results are die-cuszed by Teller in a report: “Because of the greattechnical difficulties that would be encountered inconstructing such a bomb, we have not further pur-sued this possibility . . . .” The most favorablecalculation available at that time indicated thepossibility of obtaining 10 megatons from a certainconfiguration weighing from about 40 to 100 tons.Although a megaton had been used in the calcula-tion to provide the initiating explosion, there werearguments to the effect that a smaller explosionmight suffice, possibly as small as 200 kilotons. (Thelargest explosion then realized had been about 20kilotons.) In the same report it was suggested that6LiD might be used as a fuel. This would simplifysome problems but require the production ofseparated lithium and leave the problem of the re-quired initiating explosion to be solved.

In proposing a program of research and develop-ment, Teller suggested a number of cross-sectionmeasurements; some tests to check predictions ofthermonuclear burning on a small scale (of the sortwhich subsequently became known as “boosting ex-periments”); and an attempt to make use of high-speed computing equipment when it should beavailable (then expected to be about two years off) toimprove the calculations of TX-14 behavior; and con-tinued “I think that the decision whether con-siderable effort is to be put on the development of theTX-14 or the Super should be postponed for ap-proximately two years; namely, until such time asthese experiments, tests, and calculations have beenearned out. ”

After September 1947, in consideration of the enor-mous difficulties of igniting a TX-14 system of thetype considered, or of achieving a practically usefulobject by any means then envisaged, further study ofTX-14 was soon laid aside. One of the persons, for in-stance, who through most of the preceding year hadparticipated in studies related to the TX-14 turned hisattention to work required in connection with Opera-tion Sandstone, As mentioned earlier, Richtmyer,who had conducted the detailed study of the TX-14,took Up the problem of obtaining a realistic

calculation of the behavior of a fission explosion.Among other things, experience with suchcalculations was a prerequisite to the improvedcalculations of behavior referred to by Teller.

At the end of January 1950, therefore, the un-derstanding and prospects of the TX-14 were in es-sentially the state indicated above. Systems of thiskind were believed to be feasible in principle, andcapable of providing arbitrarily large yields.However, the system required to obtain a significantamplification of the initiating yield was so large andheavy as tp appear to be of little practical value.

4. The Booster. The Booster, or the boosting pri-nciple, refers to the notion of using a fission bomb toinitate a small thermonuclear reaction with thepossibility that-in addition to being instructive withrespect to our understanding of the processes in-volved—the neutrons from this reaction might in-crease the efficiency of the use of the fide material.

Possibilities of this general type were recognized atleast as early as November 1945, when they were in-cluded in a patent application filed at Los Alamos.The designation “Booster” only became generalafter its use by Teller in September 1947.

In the summer of 1948a detailed study was begunto determine the necessary characteristics of a devicein which this interaction between fission and ther-monuclear processes might be realized. A full-scaletest of the model which would ultimately result fromthis program of study was put on the list of shots to bemade in the next overseas test operation which wasthen planned to be held in 1951.

These studies, which were in general directed byTeller, were earned out in their first stages byRosenbluth and Reitz. By the fall of 1948, a number ofpoints had been checked and the more promisinglines of approach had been identified.

Study of the Booster was continued through 1949and, starting early in the summer, was greatly inten-sified. Several unanticipated problems were turnedup and overcome.

A large number of people necessarily became in-volved in obtaining the information required for themany different aspects of the study of the Booster:Landshoff, because of his experience with Baby Hip-po; Evans, on the ignition and progress of ther-monuclear burning; Reitz and others, on equation ofstate problems; the members of the hydrodynamicscalculation group, under Hammer; the various per-sons who had experience with the neutronicscalculations required for standard fission bombs;Bethe, Longmire, and others, to consider possible ex-traneous processes and to make estimates of their

9

effects; and many others outside the TheoreticalDivision, to measure cross sections and other quan-tities required for the calculations, and to solve themechcmical problems involved. Almost all of theseaspects of the problem had been taken up before thetime of the first Russian test in September 1949. Byabout the end of January 1850, this work was farenough advanced to allow the choice of a model forwhich each step was to be calculated. This chain ofcalculations was expected to be completed sometimeduring the summer of 1950; and at that time, provid-ed no major surprises were encountered, it washoped to freeze the fine details of the design. (In theevent, things proceeded very much in this fashion ex-cept that it took a little longer than expected. The lastdetails of the design for the experiment were frozenlate in October 1950.)

5. Calculational Requirements. Already dur-ing the war, the Theoretical Division at Los Alamoshad been forced to make very heavy use of extensivenumerical calculation. There was a large group ofcomputers using desk calculators and there was aninstallation of IBM accounting equipment which wasbeing run twenty-four hours a day calculating implo-sion problems. At that time, this last was probably thelargest and most complex calculation being handledon a routine basis anywhere. This computing effort,which was considered very large in those days, wasrequired for the problems arising in connection withthe design of the first fission bombs and with ratherschematic calculations of the explosion of thosedevices. As mentioned above, a detailed calculationof the progress of a fission explosion (Baby Hippo)using these computing resources required many,many months to complete, even with the use of anumber of severely simplifying assumptions. Tocalculate this problem in noticeably more realistic(though still far from complete) detail was probablyphysically, and certainly psychologically, impossiblewithout the aid of computing devices such as the(now obsolete) SSEC which only began to appearabout 1948,

It was recognized very early that theoretical workon thermonuclear systems would, for comparablerealism, require enormously more arithmetical laborthan had the design of fission weapons, and that itwould also be necessary to rely much more heavilyon the information obtained by dead reckoning. Thismade itself evident in many ways. The first step inany thermonuclear explosion system yet consideredis a fission explosion. Somewhere in the middle of itshistory, it provides the energy required to induce thethermonuclear burning; that is, the starting con-

ditions for an estimate of thermonuclear behaviorrequire a picture of the date of things in an advancedstage of a fission explosion, which picture can itzelfonly be obtained by a calculation such ae Hippo orBaby Hippo. Thue, all the calculation normally re-quired for design of a tilon gadget, plus more ad-vanced calculations not absolutely requtred, zimplybring one to the start of an estimate of the behavior ofa thermonuclear zystem. No analogue of the ex-perimental checks which were available with rezpectto fission weapons (such as critical massmeasurements) nor the techniques used to explorethe progress of an implosion (such as measurementsof detonation velocity in high-explosive, pin-nhotstudies, or RaLa measurements) can be brought tobear in this field short of almost impossiblemeasurements on a full-scale nuclear detonation.Carrying on from there, the processes involved in theprogress of any thermonuclear reaction are in allrespects at least as complicated as those in a fissiondevice-involving the interplay of hydrodynamicmotions, transport of energy by heat flow and otherprocesses, and neutronics-and the variety of thedetails of thermonuclear reactions is in manyrespects much more complicated than the detailswhich have to be taken into account in connectionwith the fission reaction in estimating the propertiesof an explosion.

The final mqor indication of the qualitative shaltofemphasis towards calculation in going from fission tothermonuclear studies is the following. In the case ofa fission explosion, a modest number of experimentalfacts which could be determined in the laboratory(and had mostly been roughly ascertained before theManhattan District was formed), along with ratherelementary theoretical considerations, sufficed toshow that a fission explosion was feasible. The mqorpart of the wartime theoretical work at Los Alamoswas required to ascertain the detaile of a favorabledesign, the mechanics of its assembly, and estimatesof its performance. With respect to the classicalSuper in particular, the very proof of feasibility re-quired the fully detailed calculation of its behaviorduring an explosion. Without this, no conclusive ex-periment was possible short of a successful stab in thedark, since a failure would not necessarily establishunfeasibility, but possibly only that the system chosenwas unsuitable, or that the required ignition con-ditions had not been met. The fantastic requirementon calculation imposed by an attempt to explore thequestion of the classical Super as envisaged in 1946did not, of course, apply to the same extent withrespect to the thermonuclear devices in the form con-sidered since early 1951; but even those re -

4

T

. .

-.

10

.

quirementa still far exceeded the ones which had tobe met for the successful design of fission weapons.

The moat complicated calculations available at thetime of the Super Conference in 1946 were con-ducted on the ENIAC, the moat advanced computingmachine in the country at that time and, indeed, untilabout 1948. To trim the calculation to the capabilitiesof the machine as it then was, a number of quite im-portant effects and various complex physicalphenomena were almost necessarily ignored. Theresults of the calculations were promising; but, chief-ly because of having ignored these effects, any oneof which would have overloaded the calculation withrespect to that machine.

Between the time of that first ENIAC calculationand the present there has been a mqor revolution inthe facilities and technique of computing. Noqualitative change from the wartime situation in theresources available for Los Alamos work occurreduntil early in 1948, at which time Metropolis, of theLos Alamos staff, supervised changes on the ENIACat Aberdeen Proving Ground which transformed itfrom a somewhat inflexible machine to one of themodem type, capable of handling a long series ofcoded instructions without the heed of physical ad-justments on the machine to take account of eachseparate step. By modern standards, the ENIAC wasof very limited capacity. The SSEC (IBM in NewYork City) appeared the same year; but it wassomewhat slow and cumbersome. The SEAC(Bureau of Standards, Washington, D.C.), appeareda couple of years later and the UNIVAC in 1951.Then followed the Los Alamos MANIAC and thePrinceton machine in 1952, and the IBM 701 in 1953.Los Alamos problems were put on all these machinessoon after they became effective. At the present time(1954), the major computing equipment at LosAlamos consists of the MANIAC and two 701machines each running from eighteen to twenty-fourhours a day.

The effect of this revolution can be indicated inseveral ways. For example, it has been estimated thatin the course of running the Super Problem atPrinceton in 1953, which involved about three or fourmonths of effective computing time for eight hours aday, the number of basic arithmetic operations(multiplications, additiona, and so forth) performedwas of the same order of magnitude as the totalnumber of such operatiom performed at Los Alamos(excluding the arithmetic done on the Los AlamosMANIAC) from its beginning in 1843 up to that time.A similar indication ia given in the following Table inwhich the times required to compute an example ofthe implosion problem are indicated at various

periods. This problem, though improved in manyrespects and adapted to conform to the requirementsof the machines used, is basically the same as it wasin 1945 in that it is a calculation of the same physicalprocess, although in rather more detail now thanthen. (It should be noted that this calculation is com-parable in size to merely one of several basiccalculations required in connection with the designof a modern thermonuclear device.) The Table in-dicates the “elapsed time” (time from deciding tocalculate a particular example until the results areavailable), the “personnel time” (total man-months,etc., required to prepare and handle a single exam-ple) and the equipment used. The change between1945 and 1947 reflects improvements in technique ofhandling the problem, and not improvements inequipment.

PersonnelDats Equipmsnt Elapsed Time Time

1945 IBM 601 3 months 9 months1947 IBM 601 2 months 5 months1950 IBM 602 1-1/2 months 2 months1952 MANIAC 2 days 2-3 days1954 IBM 701 1-2 days 2-3 days1974a 30 min 15 min

aAe a matter of interest this entry hae been added tothe Table while preparing the unclassified version ofthis account (1974) to indicate the further progress onthis point since 1954. The mqor part of the elapsedtime indicated here is required to prepare and checkthe input numbers, gain acceas to the machine, andwait while the printer lists the results.

All through the period from 1946 to 1950, thephraee “when high-speed computing machinerybecomes available” keepa reappearing in reporta,usually in connection with thermonuclear problems.By the time (1951) when the present thermonuclearprogram began to emerge, the log-jam in computingresmurces was rapidly breaking. There was a periodin 1952 when the Los Alamos MANIAC, a model ofthe UNIVAC in Philadelphia, and the SEAC inWashington were all engaged essentially full-time onLoa Alamos (and Matterhorn) calculations for thenew thermonuclear program.

The essential pointa in this matter of computingrequirements are that: thermonuclear studies re-quired undertaking many more calculation andmuch more complicated calculations than had beenattempted in 1945 or could be sensibly handled withthe equipment then available; and, due to the revolu-tion in computing facilities which began to become

I

I

effective around 1950, the proposition of undertakingany particular complicated calculation has radicallychanged in character, in some cases to the extent ofbeing poaeible rather than irnposeible, and in allcases to being manageable in a time of the order of10 or more times less than before the appearance ofthese machines. The calculations made in connec-tion with the design of the Mike shot were essentiallyall made in the year between mid-1951 and mid-1952. With the computing resources available a cou-ple of years before, it would have been impomible tocomprees the same amount of work into anything likeas short a period.

It cannot be said that the present thermonuclearprogram could not poseibly have been handledwithout the revolution in computing equipment, orbefore the revolution. It ie clear, however, that itwould have required many more years than it did toaccomplish the same progrees.

6. Effort. If ope omits the work on Hippo andBaby Hippo (whose results were used at least asmuch in connection with thermonuclear con-siderations as others), there was about as much timedevoted in the Theoretical Division during thisperiod to studies of thermonuclear problems as tostudies of fission weapons. This situation did not app-ly to the other mqor Divisions of the Laboratory, withthe poesible exception of the Experimental Physics(P) Divinion.* The other Divisions had establishedprograms under way in connection with fissionweapona and there were not, until the Bomter beganto take shape, specific objects proposed in the ther-monuclear field requiring engineering studies ordevelopment of prcxwses or techniques. Some workrelated to the thermonuclear field (the work on jets inGMX, for example) did proceed; but, for the reasonsindicated, it was a emall fraction of the work of thelarger Divisions of the Laboratory. Indeed, it was feltby sections of the Laboratow which had fiaeionweapon work to accomplish but did not yet havework in connection with thermonuclear devices thatthey were not receiving as much detailed aeeietance

●Many instances of relevant work carried out in PDivision during thie period could be listed. The croeesections of many light element reactions weremeaeured to be oure that possible reactions ofweapon value were not overlooked. See, for exam-ple, the measurement of the croea section of the T+Treaction, performed by Agnew, Leland, Argo,Crews, Hemmendinger, Scott, and Taechek, andreported in Phys. Rev. 64, 862 (1951).

from the Theoretical Division as they needed.Because of this, in the fall of 1948 a new group wasformed in W Division specifically to carry out detail-ed analysis of problems arising in fission weaponengtneeringo

Au to the individuals on the Los Alamos Staff whocontributed, many names have been indicated inprevious sections in connection with particularstudies. Thin lint is by no mean~ exhaudive. In par-ticular, all the members of the computing groupdirected by Carlson have at one time or anotherbeen involved in executing calculations referred to.Their time would naturally be divided betweenvanouo programs roughly in the same proportion mthe time of the theoretician proper.

With respect to consultant, some—Hoyt andNordheim, in particular-worked only on ther-monuclear problems. Others—Fe;mi and Bethe, forexample-took an interest in any and everything, fis-eion or thermonuclear, that came to their attention,Von Neumann, also, followed many differentproblems; but partly because of his great interest inadvanced computing techniques, he gave most of Maattention to probleme where the computing dif-ficulties were severe. This naturally meant that hewae called on in connection with nearly every ther-monuclear investigation undertaken, at least in thisperiod. I% contributions to thinwork were direct andof enormous value, and, indeed, at many points maybe eaid to have made it peed.ble to undertake thecalculations required at the time they were done.

Finally, very special mention mud be mad. ofTeller’s contributions to this work. Although ho took adirect intereot in every aspect of the program, ffnfonan well m thermonuclear, his mod dfdinctivo fn- ‘fluence was tn the thermonuclear field. Ho &cussednearly every physical detail of almost every problomundertaken. He proponed many, though not all, ofthe problems. He called attention to possibilities. Heresolved difficulties, elucidated complicatedphenomena. Hia specukrtionn induced speculation inothers. The main thermonuclear studies would havecontinued even had he not kept in touch with them,not only because the theoretical problems themselveswere so challenging and interesting that peoplesimply couldn’t leave them alone, but alm becausethere was never a time at which the moral respon-sibility of determining whether or not the Super wasfeasible was not strongly felt at Los Alamos.However, they would have proceeded with leso in-genuity, and at a dower pace, without the benefit ofbin keen physical ineight and contagious enthusiasm.

t

1

t

.b

12 KT: 125(70)

,

?

---1

.1 w2-)

●

i