Why do molecular clouds collapse

Blitz and Rosolowsky ; Krumholz et al. GMC formation may result from gravitational instability, or it may be seeded by turbulent motion or large-scale shocks see Sect. It is an important question which mechanisms trigger GMC formation as a function of the galactic environment e. Dobbs and Pringle ; Jeffreson and Kruijssen It is a major question why this efficiency is so low. It depends on the properties of the stars that form what happens to a GMC next.

If it forms massive stars, their radiation, stellar winds, and eventual detonation as SNe may disperse the cloud. In this case, the main question is which feedback mechanisms dominate GMC dispersal see Sect.

If it only forms low-mass stars, then it may eventually disperse under the influence of local dynamics if it is gravitationally unbound or galactic shear. In this case, the main questions are which fraction of GMCs disperses dynamically, and which dynamical mechanism is responsible.

The evolutionary cycling between these three phases is visualised in Fig. It remains a major open question how the physical mechanisms governing each of these phases may change with the galactic environment. Distribution of the gas in the galaxy simulation of Semenov et al. In the simulation, star formation is assumed to take place in gas at high density and low velocity dispersion reflecting the conditions expected in the real Universe , as indicated by the grey dashed line.

In the analytical model proposed by Semenov et al. Altogether, these rates characterise the matter flow between the three phases of the GMC lifecycle in galaxies. Figure taken from Semenov et al. While the above summary of the key phases in GMC evolution sketches a relatively comprehensive picture of the physical mechanisms that each must be understood in order to describe the molecular cloud lifecycle, the underlying timescales on which these phases proceed are not known a priori.

However, as discussed throughout this review, knowledge of these timescales holds the key to identifying several of the dominant physical processes and enables a comprehensive view of the GMC lifecycle. Initial studies of the GMC lifecycle often focused on a single and often differing evolutionary phase such as GMC assembly, low-mass star formation, or dispersal by feedback from massive stars, see e.

In addition, these studies generally adopted highly dissimilar methodological approaches to the problem, leading to greatly differing evolutionary timescales. While they all rely on some form of statistical inference, some previous works rely on object classification and number counts to infer timescales e.

Kawamura et al. Engargiola et al. Finally, the empirical constraints on the GMC lifecycle were not only limited by the lack of a single methodological framework, but also by the lack of large data sets enabling a systematic census of the GMC lifecycle as a function of the galactic environment. Thanks to the recent development of novel analysis frameworks e. In conjunction with the recent major progress in numerical simulations of cloud-scale star formation and feedback e.

Dale ; Walch et al. The first of these questions can help understand why the galaxy-wide gas depletion time is two orders of magnitude longer than the dynamical times of GMCs see Sects. In addition, it may help address what fraction of GMCs disperses without forming stars see below. The second of these questions can help understand which feedback mechanisms drive GMC dispersal, e.

These different cases are quite straightforward to distinguish observationally Schruba et al. If feedback operates slowly and GMCs are long-lived, we expect tracers of molecular gas and massive star formation to be co-spatial on the cloud scale. However, if molecular gas and massive stars represent distinct evolutionary phases of a rapid lifecycle, then they should not be correlated on small scales, but often be observed in isolation. The observations exhibit a universal decorrelation of molecular gas and massive stars on GMC scales, implying a rapid evolutionary lifecycle, with short-lived clouds and rapid GMC dispersal by pre-SN feedback.

Figure 7 shows the evolutionary timelines of GMC evolution, star formation, and feedback measured by Kruijssen et al. GMC lifetimes range from 10—30 Myr and exhibit a slight trend of lifetimes increasing with galaxy mass.

Once unembedded massive stars appear, GMCs are dispersed rapidly, within 1—5 Myr, often due to early, pre-SN feedback e. By measuring the GMC lifetime, it is possible to infer the integrated star formation efficiency per star formation event. This efficiency is otherwise inaccessible, because it is defined as the ratio between the GMC lifetime and the galaxy-wide molecular gas depletion time. The homogeneous census across eleven nearby star-forming galaxies shown in Fig.

Evolutionary timeline of the GMC lifecycle from molecular gas to star formation and feedback, for a sample of eleven nearby galaxies. During the second phase in maroon , gas and massive stars coexist. Galaxies are ordered from top to bottom by increasing stellar mass. This diagram is based on Fig. A key result of Fig. Chevance et al.

The obvious question is what drives this environmental variation. Previous studies had already argued that GMCs are dynamical entities, evolving either on an internal dynamical i. Elmegreen or on a dynamical time-scale set by galactic dynamical processes e.

Examples of galactic dynamical processes that have been proposed to set GMC lifetimes are free-fall collapse of the midplane gas e. Elmegreen ; Dobbs and Pringle , spiral arm passages e. Tan ; Takahira et al. Longmore et al. Jeffreson and Kruijssen derived an analytical model for GMC lifetimes under the influence of galactic dynamics that combines the timescales for the above processes through a harmonic sum and thus assumes that the corresponding rates can be linearly added or subtracted.

They find that evidence of two regimes of GMC lifetimes, separated by a critical kpc-scale mean gas surface density. The physical interpretation of this result is that GMCs in high surface density environments reside in a mostly molecular medium, such that the detectable, CO-bright part of the cloud can extend beyond its tidal radius and the visible part of the GMC is sensitive to galactic dynamics.

Interestingly, the fact that the observed GMC lifetime largely matches an internal or external dynamical time implies that GMCs in nearby galaxies on average do not undergo evolutionary cycles without massive star formation. The reason is that the methodology applied to measure the timescales in Fig.

If a GMC undergoes a lifecycle in which it does not form stars, disperses dynamically, forms again, and then does experience massive star formation, the starless cycle is added onto the measured total cloud lifetime. In such a scenario, the measured GMC lifetime would need to span at least three dynamical times one to form, one to disperse, and one to form again. While previous literature results did not provide as wide a variety of galactic environments or as homogeneous an analysis as in Fig.

Applying the same methodology, Corbelli et al. For the same galaxy, Hygate et al. Using evolutionary streamlines, Meidt et al.

While these measurements achieve broad consistency, the homogeneous application of a single analysis framework to a large sample of galaxies now rules out the possibility that differences between observed GMC lifetimes are caused by differences in methodology, and thus enables environmental trends to be cleanly identified. While it is now clear that part of the disagreement with other studies e. In the sample studied by Chevance et al. This example demonstrates the importance of both using homogenised methods and obtaining a sample large enough to reveal any environmental dependences.

The evolutionary timelines shown in Fig. In many cases, this requires early, pre-SN feedback. GMCs in galaxies with the longest feedback timescales 4—5 Myr may receive the final push towards dispersal from SNe. These results are consistent with those from previous observational studies, which used e. For instance, Hollyhead et al. For clusters in M51, Grasha et al. If star formation in GMCs typically accelerates with time as has been suggested by e. Murray , then the restriction of their lifetimes to 1—2 free-fall times also explains why the SFE per free-fall time is low.

The measured feedback timescales can be translated into characteristic velocities for GMC dispersal by dividing the GMC radius by the feedback timescale. The similarity between directly measured feedback velocities and those inferred from the feedback timescales is encouraging and shows that the measured feedback timescales are plausible. Numerical models for the GMC lifecycle reveal a similar picture of highly dynamical, feedback-regulated, short GMC lifecycles with low star formation efficiencies e.

In accordance with the interpretation of the observational measurements, these simulations highlight the importance of early, pre-SN feedback from photoionisation and stellar winds, as well as radiation pressure. These early feedback mechanisms are critical for reproducing the observed cloud lifecycle, but it is not clear how important they are for other GMC demographics such as masses, radii, and densities.

Fujimoto et al. The observed cloud-scale decorrelation between tracers of molecular gas and massive star formation Schruba et al. A major step made by numerical simulations during the last decade is to model the interplay between GMC-scale physics such as star formation and feedback and galaxy-scale processes such as galactic dynamics.

The ongoing growth of the spatial dynamic range spanned by simulations has recently made it possible to follow the galactic processes driving convergence e. Dobbs and Pringle ; Kim and Ostriker ; Tress et al. This is a major step towards understanding how the GMC lifecycle both drives and responds to galaxy evolution.

Taking together the results discussed above, the field has now reached the point at which the key phases of GMC formation, massive star formation, and feedback can be placed on an evolutionary timeline. Recent observations and simulations have made first steps towards understanding how this timeline may depend on the galactic environment. Across a wide variety of studies, the GMC lifecycle is now found to take place on a galactic or internal dynamical time mostly governed by gravitational free-fall and shear , after which it is truncated by early stellar feedback from massive stars mostly from photoionisation and stellar winds , resulting in low star formation efficiencies of up to a few percent both integrated and per unit free-fall time.

With large observational surveys and comprehensive numerical simulations that cover a wide parameter space of galactic environments at high spatial resolution, the community is very close to obtaining a systematic census of how the GMC lifecycle changes with the galactic environment, how it connects inflow and outflow processes in the ISM, and how it feeds the galactic baryon cycle. We have shown that observationally measuring the durations of the successive phases of the evolutionary cycle of molecular clouds and star formation, from cloud assembly to cloud collapse and dispersal allows us to identify the relevant physical mechanisms at play, on the cloud scale in nearby disc galaxies.

New theoretical developments combined with recent observations show that molecular clouds can be seen as the building blocks of galaxies. The cycle between molecular clouds and young stellar regions is rapid, driven by dynamics, self-gravity, and early stellar feedback e. In addition, this cycle is not universal, but the physical mechanisms controlling the different phases of this process likely depend on the environmental conditions.

We have shown in particular that cloud lifetime may be set by the galactic dynamical timescale at high kpc-scale gas surface densities, whereas at low kpc-scale gas surface densities, GMCs appear to decouple from the galactic dynamics and their lifetime is regulated by internal dynamical processes.

To comprehensively constrain the relative roles of these mechanisms, and determine quantitatively how they depend on galactic structure and properties, future high-resolution, high-sensitivity, multi-wavelength observations across a large range of environments will be necessary, from the most quiescent e.

These observations will enable building a multi-scale model for star formation and feedback in galaxies, applicable across cosmic time. Constructing a multi-scale model for star formation and feedback is becoming critical, because galaxy formation and evolution simulations are starting to reach these small cloud scale resolutions, even in large cosmological volumes see e.

However, it remains computationally too demanding to treat the actual mechanisms of star-formation and feedback, which happen on the scales of individual stars at sub-pc resolution , from first principles. Therefore, these simulations need to use sub-grid models for describing how gas is converted into stars and how energy and momentum is deposited by stellar feedback in the surrounding ISM. Additionally, in order to make reliable predictions for the demographics of the observed galaxy population at large, the cloud-scale predictions of simulations also need to be tested against similarly high resolution observations, as a function of the galactic environment.

These cloud-scale predictions need to replicate specific observables, and most prominently the observed molecular cloud lifecycle. The recent study by Fujimoto et al. Comparing the observed and simulated molecular cloud lifecycles will make major contributions to better constraining the sub-grid physics used in galaxy formation and evolution simulations.

This dynamical vision of star formation and feedback in galaxies can be extended to larger scales. The next challenge is to characterise the physical processes driving the mass flows coupling the small-scale molecular cloud lifecycle to the galactic-scale baryon cycle, as a function of the environment. Eventually, combining all of these different elements will allow us to construct a multi-scale description of star formation across cosmic history.

Various definitions of GMCs can be adopted, based on observational or physical criteria. We specify our working definition of GMCs in Sect. Adamo, P. Zeidler et al. Space Sci. Adamo, G. Bastian et al. ADS Google Scholar. Agertz, A. Kravtsov, S. Leitner et al. Audit, P. Hennebelle, Thermal condensation in a turbulent atomic hydrogen flow.

Hennebelle, On the structure of the turbulent interstellar clouds. Influence of the equation of state on the dynamics of 3D compressible flows. Balbus, Local interstellar gasdynamical stability and substructure in spiral arms. Balbus, L. Cowie, On the gravitational stability of the interstellar medium in spiral arms. Ballesteros-Paredes, L. Hartmann, Remarks on rapid vs. Ballesteros-Paredes, A. Gazol, J. Kim et al. Hartmann, E. Velocity dispersion-size relation. Ballesteros-Paredes, E.

Palau et al. Collapsing cores in out-of-virial disguise. Barnes, A. Hernandez, E. Muller et al. Molecular clump radiative transfer, mass distributions, kinematics, and dynamical evolution. Bash, E. Green, I. Peters, The galactic density wave, molecular clouds, and star formation. Bergin, L. Hartmann, J. Raymond et al. Bertoldi, C.

McKee, The photoevaporation of interstellar clouds. II — equilibrium cometary clouds. McKee, Pressure-confined clumps in magnetized molecular clouds. Bigiel, A. Leroy, F. Walter et al. Bisbas, K. Tanaka, J. Tan et al.

Observational signatures. Bisbas, J. Tan, T. Csengeri et al. Blanc, A. Heiderman, K. Gebhardt et al. Blitz, E. Blitz, Y. Fukui, A. Reipurth, D. Jewitt, K. Keil , pp. Google Scholar. Bolatto, M. Wolfire, A. Leroy, The CO-to-H 2 conversion factor. Brandl, B. Sams, F. Bertoldi et al. Butterfield, C. Lang, M. Morris et al. Camacho, E. Canto, A. Raga, L. Rodriguez, The hot, diffuse gas in a dense cluster of massive stars.

Carroll-Nellenback, A. Frank, F. Heitsch, The effects of flow-inhomogeneities on molecular cloud formation: local versus global collapse. Cava, D. Schaerer, J. Richard et al. Chen, Q. Zhang, M. Wright et al. Chevalier, A. Clegg, Wind from a starburst galaxy nucleus. Nature , 44—45 Chevance, J. Kruijssen, A. Hygate et al. Clark, I. Bonnell, The onset of collapse in turbulently supported molecular clouds. Clark, S. Glover, R. Cloud destruction by stellar feedback. Colombo, A. Hughes, E. Schinnerer et al.

Corbelli, J. Braine, R. Bandiera et al. Crutcher, Magnetic fields in molecular clouds. Dale, The modelling of feedback in star formation simulations. New Astron. Dale, B. Ercolano, I. Bonnell, Ionizing feedback from massive stars in massive clusters — II. Disruption of bound clusters by photoionization.

Dale, J. Ngoumou, B. Ercolano et al. Dessauges-Zavadsky, A. Adamo, First constraints on the stellar mass function of star-forming clumps at the peak of cosmic star formation.

Dessauges-Zavadsky, D. Schaerer, A. Dessauges-Zavadsky, J. Richard, F. Combes et al. Dobashi, T. Shimoikura, S. Katakura et al. Dobbs, J. Baba, Dawes review 4: spiral structures in disc galaxies. Pringle, The exciting lives of giant molecular clouds. Dobbs, A. Burkert, J. Pringle, Why are most molecular clouds not gravitationally bound? Dobbs, M. Krumholz, J. Beuther, R.

Klessen, C. Dullemond et al. Pringle, A. Draine, Physics of the Interstellar and Intergalactic Medium Duarte-Cabral, C. Dobbs, N. Peretto et al. Dunne, Y.

Chu, C. Chen et al. Efremov, B. The compressed clouds therefore convert a larger fraction of their mass into stars over the cloud lifetime, and produce clusters that are initially more compact.

Neither cloud rotation nor shear against the ISM affect this result significantly, unless the shear velocity is higher than the sound speed in the confining ISM. We conclude that external pressure is an important element in the star formation process, provided that it dominates over the internal pressure of the cloud.

On the other hand, the hierarchical structure of the interstellar medium ISM implies that stars forming in the densest parts of the molecular clouds are more bound than the cloud as a whole, and so can form bound star clusters even though the global star formation efficiency SFE stays low Kruijssen , However, these suggestions are unlikely to be correct, since they predict a cutoff of cluster populations at low masses, which is not observed Bastian et al.

Overall, a picture emerges wherein regions of high density are important for the formation of bound star clusters, but those regions do not necessarily encompass whole clouds. Several authors have suggested that the star formation rate is ultimately governed by self-regulation, for example a balance between pressure created by stellar feedback and self-gravity of the gas e.

If that is the case, then an increase in the pressure of the ISM surrounding the cloud should result in an increase of the star formation rate, as suggested recently by e. Zubovas et al. Drouart et al. Although the connection between higher ambient pressure and an increased star formation rate seems robust, it rests on several assumptions.

First of all, it assumes that external pressure creates more favourable conditions for star formation, i. The second assumption is that there is enough material that can readily react to an increase in pressure by forming stars, so that external pressure accelerates ongoing star formation. Finally, the connection requires a steady state to be established: star formation must not increase to such rates that molecular gas is exhausted before feedback can establish self-regulation.

These assumptions cannot be tested in large-scale models, because they require analysis of gas dynamics and fragmentation on scales of molecular clouds, below the typical resolution of galaxy-wide numerical simulations. In this paper, we present results of numerical SPH simulations of spherically symmetric turbulent clouds embedded in a hot ISM. We track the collapse and fragmentation of the clouds, showing that under pressure, fragmentation is caused by a combined action of a shockwave driven into the cloud and instabilities behind it, leading to much higher fragmentation rates than in uncompressed cloud.

Furthermore, external pressure may confine even gravitationally unbound clouds, suggesting that compressed clouds may survive for longer even in the presence of stellar feedback or following the passage of an AGN shockwave.

The resulting cluster of sink particles forming in our compressed cloud simulations is more massive and compact than the cluster born in uncompressed models. We conclude that external pressure enhances star formation in the cold ISM and produces clusters that are likely to survive for longer periods of time.

The paper is organized as follows. We begin by describing in more detail the physical basis of the connection between AGN activity and enhanced star formation Section 2. Next, in Section 3 , we present analytical estimates of the effect of external pressure on the cloud.

In Section 4 , we describe the setup of numerical simulations, while their results are shown in Section 5. Discussion of our findings and their implications is presented in Section 6. Finally, we summarize and conclude in Section 7. The general picture of AGN effect upon the host galaxy is that of negative feedback.

There is growing evidence, however, that the real picture is more complex, and that star formation can be enhanced by AGN activity as well.

A more promising approach for increasing the ISM pressure is shock heating. Shocks can be caused by a variety of processes, such as tidal interactions with companion galaxies Ricker and ram-pressure stripping in galaxy clusters Bekki et al. A high-pressure outflow created by either wind or jet can overtake dense clumps of gas and compress them. The interaction between a dense cloud and the outflow is very similar to the interaction of a cloud with a passing shockwave e.

These clouds form in pressure equilibrium with the surrounding flow and hence are not necessarily bound by their own gravity. Finally, the outflow expanding in the diffuse gas of the galactic bulge and halo compresses the galactic disc. This creates a secondary shockwave passing into the disc and compressing the clouds there Zubovas et al. The shockwave can develop a complex morphology due to the uneven density distribution of the disc ISM and therefore the clouds experience a wide range of shockwave velocities passing through them.

In the next section, we make analytic estimates of the effect of external pressure in these three situations, starting with the simplest, if somewhat unrealistic, scenario of negligible lateral velocity of the shockwave.

Here, we make rough estimates regarding the effect that external pressure has on the internal dynamics of a giant molecular cloud GMC in the various configurations discussed above. Using the Larson and the Solomon et al. The velocity dispersion of a cloud of this size should be 3. For simplicity, we consider the cloud to be spherical with uniform density.

If the external pressure increases above this value, the cloud is compressed. The exact situation depends on the dynamics of the surrounding ISM. In the simplest case, the external pressure around the molecular cloud increases isotropically and homogeneously. This is an unlikely scenario, since typically high pressure is caused by a shockwave enveloping the cloud. There are, however, a few situations where the velocity of the cloud with respect to its surroundings is low.

Thompson et al. It is therefore possible that in the reference frame moving with the cloud, the shockwave is much slower than the sound speed of the shocked gas behind it. The direct interaction between the molecular cloud and the passing shockwave is mitigated by the atomic hydrogen envelope around the cloud see also Section 6. In another case, a cloud that forms due to cooling of gas inside the fast hot outflow also experiences high external pressure without significant lateral motion.

If the cloud forms under conditions of high external pressure, its turbulent velocity should have a value as given by equation 6. Such a cloud would not be bound by its own gravity, but as long as the external pressure persists, it is able to fragment and form stars.

Only a weak shockwave is driven into the cloud, so star formation starts in the central parts of the cloud, where the local dynamical time is shortest. The low integrated that is, calculated over the lifetime of the cloud rather than its dynamical time efficiency of mass conversion into stars in a GMC suggests that the clouds are rapidly destroyed by stellar feedback.

It is not well understood which of the many feedback processes are most important. The calculations above reveal three major effects that confining external pressure has on a molecular cloud. The cloud is compressed, reducing the effective dynamical time-scale and thus increasing the rates of fragmentation and star formation.

This should be a general effect of higher ambient pressure, independent of its source, the time-scale over which the pressure increases or the shear velocity of the hot ISM w. As a result, stars form more rapidly in compressed clouds than in undisturbed ones, so that the cloud evolves on the effective dynamical time-scale.

The shockwave is approximately spherical if the lateral motion of the ISM past the cloud is slow. If this velocity is large, the cloud is destroyed by the shockwave in a few effective dynamical times. The presence of the shockwave is guaranteed only if the external pressure increases around the cloud on a time-scale shorter than the cloud dynamical time; otherwise, the cloud has time to establish virial equilibrium with the higher surrounding pressure.

As long as the total external plus gravitational pressure confining the cloud exceeds the pressure created by stellar feedback, the cloud is not disrupted and can continue to form stars. The fraction of gas converted into stars is larger in confined clouds than in uncompressed ones, leading to formation of more tightly bound clusters. Although these conclusions seem robust based on analytical calculations alone, we wish to investigate the evolution of compressed clouds in more detail.

Therefore, we turn to numerical simulations. We employ the fourth-order HOCT4 kernel with neighbours, and use adaptive smoothing and gravitational softening lengths. We assume the cloud to have uniform density initially; we comment on this assumption in the Discussion Section 6. This means that turbulence is incompressible; another extreme would be a purely compressive curl-free turbulence. Although supersonic turbulence is generally at least partially compressive, a large fraction of the turbulent energy is expected to be in solenoidal modes Federrath et al.

Furthermore, solenoidal turbulence has a shallower power spectrum than compressive one. Numerical simulations tend to steepen the spectrum as time goes by, since turbulence decays artificially starting from the smallest length-scales highest wavenumbers ; therefore, our choice of turbulent power spectrum should produce more realistic results than the opposite extreme.

Once the turbulent velocities are set up, we scale them to give the desired characteristic velocity and hence turbulent energy.

The high external pressure is also higher than the ISM pressure necessary to prevent cloud dispersal by photoionization see equation The whole system is set up in a periodic box of side length 80 pc models with shearing motion use a box of side length pc.

This resolution is good enough to resolve very massive stars and small stellar associations. With this prescription, the cloud gas is modelled with reasonable accuracy, while the surrounding ISM stays isothermal.

This mass is similar to that of pre-stellar cores, so our simulations should not overproduce the number and total mass of fragments.

The models analysed are listed in Table 1. We first consider models with zero lateral velocity — t4T5, t4T7, t10T5, t10T7, t2. These simulations are designed to show the basic behaviour of clouds compressed by the hot ISM. Next, we model the more realistic cases of non-zero shear, with relative velocities of the ISM w.

Parameters of the numerical models and most important results. The first column shows the model ID. The next four columns give the parameters: cloud turbulent velocity, confining ISM temperature, angular velocity of cloud rotation and linear velocity of shearing cloud motion, respectively. Numbers with asterisks are extrapolated from earlier snapshots. We divide the result presentation into two parts. First, we analyse the effects of external pressure without shear, including cases of static gravitationally bound clouds models t4T5 and t4T7 , static gravitationally unbound clouds models t10T5 and t10T7 and rotating gravitationally bound clouds t2.

Next, we consider the effects of progressively stronger shear upon self-gravitating clouds models t4vXT5 and t4vXT7. For each model, we derive four parameters which allow for easy quantitative comparison of their progress.

The first parameter is the time when the first sink particle forms, which we use as a proxy for the onset of star formation. The numerical values of these parameters are given in the last three columns of Table 1. Figs 1 and 2 show the column density plots which depict the evolution of the models t4T5 and t4T7, respectively.

The uncompressed model quickly develops an uneven density structure and expands slightly, before starting to collapse as the turbulence decays. Star formation begins in the central regions, where the density is highest due to convergent turbulent flows.

Sink particles form along filaments and are gradually absorbed into a central elliptical cluster right-hand panel. Furthermore, the decay of turbulence in the cloud is partly responsible for the high fragmentation rate see Section 6. Evolution of the uncompressed cloud model, t4T5. Stars form in the densest regions of the interstellar medium, or ISM, called molecular clouds. The ISM is the name given to the gas and dust that exists between the stars within a galaxy. Molecular clouds are perfect star-forming regions because the combination of these atoms into molecules is much more likely in very dense regions.

This photograph shows the Orion Nebula, an interstellar cloud in which star systems - and possibly planets - are forming. Our own solar system presumably formed as gravity caused the collapse of a similar large cloud of gas. The piece of cloud that formed our Solar System is known as the solar nebula. Click the photo to the left to see more images of the Orion Nebula.

A star forms when a molecular cloud collapses under its own gravity forming a dense core sustained by nuclear fusion. This happens only when the force of gravity pulling in exceeds the outward push of pressure. High-density molecular clouds have stronger forces of gravity pushing in, making it easier to overcome the total pressure within the cloud.

Once started, the collapse of the solar nebula continues because the force of gravity exerted on the cloud grows stronger as the cloud shrinks in size. But, when gas and dust start to collapse in a region within the molecular cloud, it slowly heats up.

This is a consequence of a law of physics, which tells us that, when matter is squeezed together, the density of the matter will increase and the matter will start to heat up. When the collapsing region has reached a size of nearly 10, AU, it is called a pre-stellar core Figure 1B and is officially a star in-the-making. Also, this pre-stellar core will later become the interior core of the star. Over the next 50, years or so, the pre-stellar core contracts. This might sound like a long time, but on an astronomical timescale it is considered a fairly swift process compared, for instance, to the age of the Universe, which is almost 14 billion years.

The core contracts until it is around 1, AU Figure 1C. After 50, years has passed, the system will have formed a disk around the central core, and excess material will be ejected outward from the poles of the star. A pole on a star is like those on the Earth, namely defined as the axis that the star spins around. In Figure 1C , you can see two fountain-like structures where this excess material is ejected.

These structures are called jets, and they obey the laws of physics. The random motion of the gas and dust that we described earlier, combined with the system's contraction as the pre-stellar core forms, will cause the whole system to rotate. This process causes a flat disk to form around the pre-stellar core.

This is similar to the way a dress forms a flat disk around a spinning ice-skater. If the skater was not rotating, the dress would not be a flat disk around her, but instead would hang along her sides. The jets at the poles arise to keep the system in balance. The system is now called a proto-star, which means it is at its very first stage of becoming a real star.

The disk is crucial for the proto-star to grow into a properly sized star. The disk is mainly composed of gas, which rotates with the disk and slowly approaches the surface of the proto-star. When the gas comes close enough to the star, it falls onto the surface of the star because of gravity, and the star grows.